commit 57d42d92cc16bc5d94b409132b667d50aaaba57c Author: ModelHub XC Date: Sat Jun 6 12:06:39 2026 +0800 初始化项目,由ModelHub XC社区提供模型 Model: jackf857/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802 Source: Original Platform diff --git a/.gitattributes b/.gitattributes new file mode 100644 index 0000000..52373fe --- /dev/null +++ b/.gitattributes @@ -0,0 +1,36 @@ +*.7z filter=lfs diff=lfs merge=lfs -text +*.arrow filter=lfs diff=lfs merge=lfs -text +*.bin filter=lfs diff=lfs merge=lfs -text +*.bz2 filter=lfs diff=lfs merge=lfs -text +*.ckpt filter=lfs diff=lfs merge=lfs -text +*.ftz filter=lfs diff=lfs merge=lfs -text +*.gz filter=lfs diff=lfs merge=lfs -text +*.h5 filter=lfs diff=lfs merge=lfs -text +*.joblib filter=lfs diff=lfs merge=lfs -text +*.lfs.* filter=lfs diff=lfs merge=lfs -text +*.mlmodel filter=lfs diff=lfs merge=lfs -text +*.model filter=lfs diff=lfs merge=lfs -text +*.msgpack filter=lfs diff=lfs merge=lfs -text +*.npy filter=lfs diff=lfs merge=lfs -text +*.npz filter=lfs diff=lfs merge=lfs -text +*.onnx filter=lfs diff=lfs merge=lfs -text +*.ot filter=lfs diff=lfs merge=lfs -text +*.parquet filter=lfs diff=lfs merge=lfs -text +*.pb filter=lfs diff=lfs merge=lfs -text +*.pickle filter=lfs diff=lfs merge=lfs -text +*.pkl filter=lfs diff=lfs merge=lfs -text +*.pt filter=lfs diff=lfs merge=lfs -text +*.pth filter=lfs diff=lfs merge=lfs -text +*.rar filter=lfs diff=lfs merge=lfs -text +*.safetensors filter=lfs diff=lfs merge=lfs -text +saved_model/**/* filter=lfs diff=lfs merge=lfs -text +*.tar.* filter=lfs diff=lfs merge=lfs -text +*.tar filter=lfs diff=lfs merge=lfs -text +*.tflite filter=lfs diff=lfs merge=lfs -text +*.tgz filter=lfs diff=lfs merge=lfs -text +*.wasm filter=lfs diff=lfs merge=lfs -text +*.xz filter=lfs diff=lfs merge=lfs -text +*.zip filter=lfs diff=lfs merge=lfs -text +*.zst filter=lfs diff=lfs merge=lfs -text +*tfevents* filter=lfs diff=lfs merge=lfs -text +tokenizer.json filter=lfs diff=lfs merge=lfs -text diff --git a/README.md b/README.md new file mode 100644 index 0000000..08cc700 --- /dev/null +++ b/README.md @@ -0,0 +1,81 @@ +--- +library_name: transformers +base_model: W-61/llama-3-8b-base-sft-hh-helpful-4xh200 +tags: +- alignment-handbook +- new-dpo +- generated_from_trainer +datasets: +- Anthropic/hh-rlhf +model-index: +- name: llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802 + results: [] +--- + + + +# llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802 + +This model is a fine-tuned version of [W-61/llama-3-8b-base-sft-hh-helpful-4xh200](https://huggingface.co/W-61/llama-3-8b-base-sft-hh-helpful-4xh200) on the Anthropic/hh-rlhf dataset. +It achieves the following results on the evaluation set: +- Loss: 0.5131 +- Fcm Dpo/beta: 0.0057 +- Fcm Dpo/q T: 0.3538 +- Fcm Dpo/delta: 0.0491 +- Fcm Dpo/margin: 139.8486 +- Margin Dpo/margin Mean: 139.8486 +- Margin Dpo/margin Std: 209.5895 +- Logps/chosen: -511.8030 +- Logps/rejected: -659.3984 +- Logps/ref Chosen: -79.0510 +- Logps/ref Rejected: -86.7979 +- Logits/chosen: 0.2560 +- Logits/rejected: 0.2786 + +## Model description + +More information needed + +## Intended uses & limitations + +More information needed + +## Training and evaluation data + +More information needed + +## Training procedure + +### Training hyperparameters + +The following hyperparameters were used during training: +- learning_rate: 5e-07 +- train_batch_size: 8 +- eval_batch_size: 8 +- seed: 42 +- distributed_type: multi-GPU +- num_devices: 4 +- gradient_accumulation_steps: 2 +- total_train_batch_size: 64 +- total_eval_batch_size: 32 +- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments +- lr_scheduler_type: cosine +- lr_scheduler_warmup_ratio: 0.1 +- num_epochs: 1 + +### Training results + +| Training Loss | Epoch | Step | Validation Loss | Fcm Dpo/beta | Fcm Dpo/q T | Fcm Dpo/delta | Fcm Dpo/margin | Margin Dpo/margin Mean | Margin Dpo/margin Std | Logps/chosen | Logps/rejected | Logps/ref Chosen | Logps/ref Rejected | Logits/chosen | Logits/rejected | +|:-------------:|:------:|:----:|:---------------:|:------------:|:-----------:|:-------------:|:--------------:|:----------------------:|:---------------------:|:------------:|:--------------:|:----------------:|:------------------:|:-------------:|:---------------:| +| 0.8697 | 0.2937 | 200 | 0.5858 | 0.0198 | 0.3795 | 0.0634 | 38.2719 | 38.2719 | 74.7381 | -148.6721 | -194.6909 | -79.0510 | -86.7979 | -0.4905 | -0.4710 | +| 0.8159 | 0.5874 | 400 | 0.5362 | 0.0070 | 0.3603 | 0.0726 | 111.7486 | 111.7486 | 182.2796 | -360.3431 | -479.8386 | -79.0510 | -86.7979 | 0.0750 | 0.0972 | +| 0.8894 | 0.8811 | 600 | 0.5131 | 0.0057 | 0.3538 | 0.0491 | 139.8486 | 139.8486 | 209.5895 | -511.8030 | -659.3984 | -79.0510 | -86.7979 | 0.2560 | 0.2786 | + + +### Framework versions + +- Transformers 4.51.0 +- Pytorch 2.3.1+cu121 +- Datasets 2.21.0 +- Tokenizers 0.21.4 diff --git a/all_results.json b/all_results.json new file mode 100644 index 0000000..47d2b1f --- /dev/null +++ b/all_results.json @@ -0,0 +1,26 @@ +{ + "epoch": 1.0, + "eval_fcm_dpo/beta": 0.005598581861704588, + "eval_fcm_dpo/delta": 0.061232246458530426, + "eval_fcm_dpo/margin": 140.12789916992188, + "eval_fcm_dpo/q_t": 0.35402458906173706, + "eval_logits/chosen": 0.21182169020175934, + "eval_logits/rejected": 0.23247361183166504, + "eval_logps/chosen": -505.7603759765625, + "eval_logps/ref_chosen": -79.05104064941406, + "eval_logps/ref_rejected": -86.79793548583984, + "eval_logps/rejected": -653.6351928710938, + "eval_loss": 0.5076749920845032, + "eval_margin_dpo/margin_mean": 140.12789916992188, + "eval_margin_dpo/margin_std": 204.84616088867188, + "eval_runtime": 39.5033, + "eval_samples": 2339, + "eval_samples_per_second": 59.21, + "eval_steps_per_second": 1.873, + "total_flos": 0.0, + "train_loss": 0.8947697063024523, + "train_runtime": 1846.2605, + "train_samples": 43598, + "train_samples_per_second": 23.614, + "train_steps_per_second": 0.369 +} \ No newline at end of file diff --git a/config.json b/config.json new file mode 100644 index 0000000..5092b09 --- /dev/null +++ b/config.json @@ -0,0 +1,29 @@ +{ + "architectures": [ + "LlamaForCausalLM" + ], + "attention_bias": false, + "attention_dropout": 0.0, + "bos_token_id": 128000, + "eos_token_id": 128001, + "head_dim": 128, + "hidden_act": "silu", + "hidden_size": 4096, + "initializer_range": 0.02, + "intermediate_size": 14336, + "max_position_embeddings": 8192, + "mlp_bias": false, + "model_type": "llama", + "num_attention_heads": 32, + "num_hidden_layers": 32, + "num_key_value_heads": 8, + "pretraining_tp": 1, + "rms_norm_eps": 1e-05, + "rope_scaling": null, + "rope_theta": 500000.0, + "tie_word_embeddings": false, + "torch_dtype": "float32", + "transformers_version": "4.51.0", + "use_cache": true, + "vocab_size": 128256 +} diff --git a/eval_results.json b/eval_results.json new file mode 100644 index 0000000..e593efd --- /dev/null +++ b/eval_results.json @@ -0,0 +1,20 @@ +{ + "epoch": 1.0, + "eval_fcm_dpo/beta": 0.005598581861704588, + "eval_fcm_dpo/delta": 0.061232246458530426, + "eval_fcm_dpo/margin": 140.12789916992188, + "eval_fcm_dpo/q_t": 0.35402458906173706, + "eval_logits/chosen": 0.21182169020175934, + "eval_logits/rejected": 0.23247361183166504, + "eval_logps/chosen": -505.7603759765625, + "eval_logps/ref_chosen": -79.05104064941406, + "eval_logps/ref_rejected": -86.79793548583984, + "eval_logps/rejected": -653.6351928710938, + "eval_loss": 0.5076749920845032, + "eval_margin_dpo/margin_mean": 140.12789916992188, + "eval_margin_dpo/margin_std": 204.84616088867188, + "eval_runtime": 39.5033, + "eval_samples": 2339, + "eval_samples_per_second": 59.21, + "eval_steps_per_second": 1.873 +} \ No newline at end of file diff --git a/generation_config.json b/generation_config.json new file mode 100644 index 0000000..76247c9 --- /dev/null +++ b/generation_config.json @@ -0,0 +1,9 @@ +{ + "bos_token_id": 128000, + "do_sample": true, + "eos_token_id": 128001, + "max_length": 4096, + "temperature": 0.6, + "top_p": 0.9, + "transformers_version": "4.51.0" +} diff --git a/margin_logs/margins.jsonl b/margin_logs/margins.jsonl new file mode 100644 index 0000000..e050a7d --- /dev/null +++ b/margin_logs/margins.jsonl @@ -0,0 +1,681 @@ +{"epoch": 0.0, "step": 1, "batch_size": 64, "mean": -0.02287048101425171, "std": 0.42023447155952454, "min": -1.4034271240234375, "p10": -0.46674575805664065, "median": 0.04234886169433594, "p90": 0.4323463439941407, "max": 0.89263916015625, "pos_frac": 0.53125, "sample": [-0.06523895263671875, 0.436798095703125, 0.27811431884765625, -0.9194221496582031, 0.018890380859375, 0.20587158203125, 0.18878173828125, -0.3968696594238281, 0.26206207275390625, 0.2470550537109375, -0.040912628173828125, 0.4394989013671875, -0.44133758544921875, -0.39148712158203125, 0.2764854431152344, 0.89263916015625, -0.42584991455078125, -0.46125030517578125, -0.8638992309570312, -0.3508758544921875, 0.371368408203125, 0.887847900390625, -0.382904052734375, 0.36145782470703125, -0.4890003204345703, 0.052455902099609375, -0.036136627197265625, 0.23079299926757812, 0.2469482421875, 0.1643218994140625, -0.07129669189453125, 0.2790794372558594, 0.3637123107910156, -0.8916168212890625, 0.03298759460449219, -0.2790107727050781, -0.17860984802246094, 0.23892593383789062, 0.05171012878417969, -0.2564239501953125, -0.14655303955078125, 0.27777862548828125, 0.0810394287109375, -1.4034271240234375, -0.28739166259765625, -0.1489429473876953, 0.44918060302734375, 0.1693286895751953, 0.10933303833007812, -0.14766693115234375, -0.40944671630859375, -0.18532562255859375, 0.6261310577392578, -0.20856857299804688, 0.602569580078125, 0.05538177490234375, 0.1505279541015625, 0.1313800811767578, -0.006317138671875, 0.42195892333984375, -0.29936981201171875, -0.4691009521484375, 0.16705322265625, -0.5789260864257812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000001.npy"} +{"epoch": 0.0014684287812041115, "step": 2, "batch_size": 64, "mean": -0.06572240591049194, "std": 0.3523969054222107, "min": -0.9291305541992188, "p10": -0.46334152221679686, "median": -0.05502510070800781, "p90": 0.3672500610351563, "max": 1.0444793701171875, "pos_frac": 0.4375, "sample": [-0.2829437255859375, 0.3027191162109375, -0.19867706298828125, -0.3062286376953125, 0.10318756103515625, 0.20131683349609375, -0.34906005859375, 0.2802886962890625, 0.1914520263671875, -0.31072998046875, 0.08922195434570312, 0.10284614562988281, -0.03655242919921875, -0.0604095458984375, -0.06208038330078125, 0.32562255859375, -0.37982177734375, 0.2746162414550781, -0.049640655517578125, 0.3752174377441406, -0.103973388671875, 0.0699462890625, 0.36417388916015625, -0.033428192138671875, 0.37265777587890625, -0.3787078857421875, -0.6610565185546875, 0.4720420837402344, 0.47701263427734375, -0.27928924560546875, -0.44719696044921875, -0.0965118408203125, -0.7628555297851562, 0.046764373779296875, 0.06670379638671875, -0.9291305541992188, -0.7122802734375, -0.16554832458496094, 0.1485595703125, -0.07539939880371094, 0.2588920593261719, 0.039890289306640625, 0.201690673828125, 0.0623016357421875, 1.0444793701171875, -0.37696075439453125, -0.02794647216796875, -0.223297119140625, -0.35730743408203125, -0.1309051513671875, -0.3106689453125, -0.11409187316894531, -0.1669769287109375, 0.131317138671875, -0.2361297607421875, 0.4093780517578125, -0.6485977172851562, 0.36856842041015625, -0.1951904296875, -0.4702606201171875, -0.7624168395996094, 0.008928298950195312, -0.31630706787109375, 0.022550582885742188], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000002.npy"} +{"epoch": 0.002936857562408223, "step": 3, "batch_size": 64, "mean": -0.054007261991500854, "std": 0.3287087082862854, "min": -0.7089080810546875, "p10": -0.4666679382324218, "median": -0.057211875915527344, "p90": 0.3182075500488283, "max": 0.8736953735351562, "pos_frac": 0.421875, "sample": [-0.35785484313964844, -0.39081573486328125, -0.00292205810546875, 0.7697525024414062, -0.2964191436767578, 0.19768524169921875, 0.2404327392578125, 0.013631820678710938, -0.1414337158203125, 0.009889602661132812, -0.6505584716796875, -0.07275009155273438, 0.060943603515625, -0.5256423950195312, -0.08674240112304688, 0.040008544921875, 0.1096343994140625, 0.050159454345703125, -0.03302001953125, -0.029802322387695312, -0.6874618530273438, 0.239990234375, 0.8736953735351562, -0.1116180419921875, 0.35223388671875, -0.2932605743408203, 0.1155242919921875, 0.4285144805908203, 0.18306350708007812, -0.34081268310546875, 0.33766937255859375, 0.49463653564453125, -0.10449409484863281, -0.3104248046875, -0.04167366027832031, -0.1002960205078125, -0.499176025390625, -0.10048675537109375, -0.7089080810546875, -0.09004974365234375, -0.190826416015625, 0.10590362548828125, -0.3281707763671875, -0.02978515625, 0.272796630859375, 0.17899703979492188, -0.6053085327148438, 0.12556838989257812, -0.15384674072265625, -0.24903106689453125, 0.18095970153808594, -0.3004341125488281, 0.02069854736328125, -0.17620086669921875, -0.105560302734375, 0.27085113525390625, -0.35277557373046875, -0.301971435546875, -0.344085693359375, 0.6761932373046875, 0.012342453002929688, 0.078887939453125, -0.16149139404296875, -0.6210174560546875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000003.npy"} +{"epoch": 0.004405286343612335, "step": 4, "batch_size": 64, "mean": -0.03080245852470398, "std": 0.30792176723480225, "min": -0.8707122802734375, "p10": -0.35912189483642576, "median": -0.0220947265625, "p90": 0.2861873626708985, "max": 0.8557510375976562, "pos_frac": 0.46875, "sample": [-0.2910041809082031, 0.18427276611328125, -0.3132953643798828, -0.01235198974609375, -0.0220947265625, -0.127044677734375, 0.128509521484375, -0.051910400390625, -0.5478725433349609, -0.1121978759765625, -0.10576820373535156, 0.5728034973144531, -0.11403656005859375, -0.313446044921875, 0.113983154296875, -0.16777801513671875, 0.29250335693359375, 0.22510147094726562, -0.2919158935546875, -0.1295909881591797, -0.45014190673828125, -0.27887725830078125, -0.0220947265625, -0.2212371826171875, 0.160125732421875, 0.027069091796875, 0.131866455078125, 0.017070770263671875, -0.06982994079589844, -0.247222900390625, -0.24563217163085938, 0.019948959350585938, -0.2506866455078125, 0.14296722412109375, 0.03205680847167969, 0.17486953735351562, 0.08379364013671875, 0.176300048828125, -0.60943603515625, 0.11285400390625, -0.16248321533203125, -0.12499237060546875, 0.25785064697265625, 0.509796142578125, 0.00046539306640625, 0.22162246704101562, 0.0235595703125, 0.716094970703125, -0.18217086791992188, -0.107513427734375, 0.4100627899169922, 0.2714500427246094, -0.4070854187011719, 0.3345489501953125, 0.2141857147216797, 0.016988754272460938, 0.8557510375976562, -0.21773910522460938, -0.5431480407714844, 0.24904632568359375, -0.3292694091796875, -0.36658287048339844, -0.8707122802734375, -0.34171295166015625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000004.npy"} +{"epoch": 0.005873715124816446, "step": 5, "batch_size": 64, "mean": 0.020543158054351807, "std": 0.38211289048194885, "min": -0.769775390625, "p10": -0.48832702636718733, "median": -0.024078369140625, "p90": 0.5124046325683596, "max": 0.855224609375, "pos_frac": 0.46875, "sample": [-0.3475494384765625, 0.7018508911132812, 0.2114105224609375, 0.23133087158203125, -0.007568359375, -0.2471466064453125, -0.08893585205078125, -0.23287582397460938, -0.06856346130371094, -0.181732177734375, -0.24127960205078125, 0.2014312744140625, 0.19637298583984375, 0.3732414245605469, -0.2961082458496094, 0.0127105712890625, -0.03802490234375, -0.683685302734375, -0.29334259033203125, -0.6243972778320312, -0.2361583709716797, 0.36737060546875, -0.3142433166503906, 0.31531524658203125, 0.10400390625, 0.7077484130859375, 0.7427139282226562, -0.3224220275878906, -0.7108306884765625, 0.5307083129882812, 0.1433258056640625, -0.08245849609375, -0.1747283935546875, 0.18937301635742188, 0.1991252899169922, 0.2920379638671875, 0.6411590576171875, 0.23187637329101562, 0.41141510009765625, 0.29669189453125, 0.469696044921875, 0.06726837158203125, 0.22076416015625, -0.1538543701171875, 0.855224609375, -0.769775390625, -0.07804107666015625, -0.0545654296875, -0.6239013671875, -0.5486602783203125, -0.04031562805175781, -0.3456687927246094, 0.335174560546875, 0.2956581115722656, -0.09503936767578125, -0.18170166015625, 0.836669921875, -0.5613250732421875, -0.0345306396484375, 0.2732696533203125, -0.2932586669921875, 0.05161285400390625, -0.205474853515625, -0.0136260986328125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000005.npy"} +{"epoch": 0.007342143906020558, "step": 6, "batch_size": 64, "mean": 0.0022419095039367676, "std": 0.3720536231994629, "min": -0.951812744140625, "p10": -0.4138710021972656, "median": -0.01412200927734375, "p90": 0.5236434936523438, "max": 0.8499603271484375, "pos_frac": 0.484375, "sample": [-0.25848388671875, -0.10894775390625, -0.06250190734863281, 0.2006378173828125, -0.061901092529296875, 0.8499603271484375, -0.5646076202392578, -0.15706634521484375, 0.572601318359375, -0.951812744140625, 0.3001213073730469, 0.5337791442871094, -0.7843170166015625, 0.4170684814453125, -0.3717460632324219, -0.53485107421875, 0.6647491455078125, -0.1633586883544922, -0.13179397583007812, 0.46282196044921875, -0.45453643798828125, -0.3352241516113281, 0.1910400390625, -0.10908889770507812, -0.1317138671875, -0.2400054931640625, 0.5779953002929688, 0.28282928466796875, -0.21124267578125, 0.12249374389648438, -0.32888031005859375, 0.008264541625976562, 0.6839828491210938, 0.24327468872070312, 0.07695770263671875, 0.2558441162109375, 0.06804656982421875, -0.8971443176269531, -0.1562938690185547, -0.19244384765625, -0.21881103515625, 0.517608642578125, -0.1314239501953125, 0.10896492004394531, 0.4166221618652344, -0.02747344970703125, -0.08748245239257812, 0.10109329223632812, -0.383819580078125, -0.00077056884765625, 0.10398101806640625, 0.16522979736328125, 0.15970230102539062, 0.054203033447265625, 0.09356689453125, 0.013336181640625, 0.10859298706054688, -0.035980224609375, 0.489776611328125, -0.31616973876953125, -0.230987548828125, -0.42675018310546875, 0.5262298583984375, -0.1602630615234375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000006.npy"} +{"epoch": 0.00881057268722467, "step": 7, "batch_size": 64, "mean": -0.01150551438331604, "std": 0.39827245473861694, "min": -1.0072479248046875, "p10": -0.4448150634765625, "median": -0.06341266632080078, "p90": 0.5528564453125001, "max": 1.140899658203125, "pos_frac": 0.40625, "sample": [0.5157203674316406, -0.7199783325195312, -0.09095382690429688, 0.23048782348632812, -1.0072479248046875, -0.259613037109375, -0.2575492858886719, 0.207611083984375, -0.32892608642578125, 1.140899658203125, 0.523162841796875, -0.26335906982421875, 0.6297531127929688, 0.08685302734375, -0.124908447265625, -0.2187957763671875, 0.2299213409423828, -0.25237083435058594, 0.08934211730957031, -0.3401641845703125, 0.1680755615234375, 0.93145751953125, 0.0525665283203125, 0.05084991455078125, -0.15478515625, -0.09563064575195312, -0.007648468017578125, -0.06781959533691406, -0.12134933471679688, -0.15572738647460938, -0.12913131713867188, 0.7437591552734375, -0.09434890747070312, 0.4400177001953125, -0.1445159912109375, 0.1478424072265625, -0.38297271728515625, 0.565582275390625, 0.04754638671875, 0.025146484375, -0.5563201904296875, -0.544097900390625, 0.151275634765625, -0.07314300537109375, -0.02457427978515625, -0.0590057373046875, 0.1828460693359375, -0.3794403076171875, -0.3172149658203125, -0.5157852172851562, 0.688751220703125, -0.44647216796875, -0.14351654052734375, 0.6572189331054688, -0.0566253662109375, 0.10771942138671875, -0.2887840270996094, 0.21148681640625, 0.4867992401123047, -0.24550628662109375, -0.008453369140625, -0.67352294921875, -0.05783843994140625, -0.440948486328125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000007.npy"} +{"epoch": 0.010279001468428781, "step": 8, "batch_size": 64, "mean": -0.07862147688865662, "std": 0.36147069931030273, "min": -1.0023956298828125, "p10": -0.53790283203125, "median": -0.06868648529052734, "p90": 0.3224876403808594, "max": 0.8812408447265625, "pos_frac": 0.421875, "sample": [0.168365478515625, 0.0605316162109375, 0.02606964111328125, 0.06323814392089844, -0.114166259765625, 0.08950424194335938, 0.4373588562011719, -0.3906097412109375, -0.0355682373046875, -0.1189117431640625, -0.20315933227539062, -0.0645599365234375, -0.1782379150390625, 0.15294647216796875, 0.18624114990234375, -0.8432540893554688, -0.008249282836914062, -0.14990615844726562, 0.06516647338867188, -0.5357666015625, -0.3797149658203125, -0.94342041015625, -0.1507415771484375, 0.8812408447265625, -0.13065528869628906, -0.538818359375, -0.563873291015625, -0.07779693603515625, 0.2318878173828125, -0.35762786865234375, -0.12157440185546875, -0.672210693359375, -0.13236618041992188, 0.5421142578125, 0.04837799072265625, -0.49286651611328125, 0.0829620361328125, -0.06962394714355469, -0.004909515380859375, 0.38523101806640625, 0.5903587341308594, -0.38835906982421875, -0.2310638427734375, -0.0677490234375, -0.09082984924316406, -0.89080810546875, -1.0023956298828125, -0.11324310302734375, 0.06492805480957031, 0.33011627197265625, 0.00356292724609375, -0.085479736328125, 0.016357421875, 0.09577178955078125, 0.6521835327148438, -0.129638671875, -0.40885162353515625, 0.05451202392578125, 0.202728271484375, -0.2028961181640625, -0.10572433471679688, 0.2125396728515625, 0.014871597290039062, 0.3046875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000008.npy"} +{"epoch": 0.011747430249632892, "step": 9, "batch_size": 64, "mean": 0.04415696859359741, "std": 0.4679097831249237, "min": -1.2313919067382812, "p10": -0.6740959167480468, "median": 0.02852630615234375, "p90": 0.6785182952880862, "max": 1.14825439453125, "pos_frac": 0.546875, "sample": [-0.3190765380859375, -0.098358154296875, 0.6979789733886719, 0.19998931884765625, -0.00357818603515625, -0.7065353393554688, 0.103973388671875, 0.26795196533203125, 0.493682861328125, 0.2286224365234375, -0.5814590454101562, -0.09124755859375, -0.16048431396484375, -0.7819366455078125, -0.1530914306640625, -0.93878173828125, 0.6331100463867188, -0.7371978759765625, -0.0837860107421875, 0.23696136474609375, -0.08689498901367188, 0.03595161437988281, -0.18007659912109375, 0.02388763427734375, 0.0565032958984375, 0.36719512939453125, 0.5746726989746094, -0.200164794921875, -0.012403488159179688, 1.0236968994140625, -0.19893646240234375, 0.09629249572753906, 0.03316497802734375, 0.20938873291015625, 0.058441162109375, -0.6115951538085938, -0.038265228271484375, -0.10511970520019531, 0.5555267333984375, 0.290496826171875, -0.7008819580078125, 0.0232391357421875, -0.2205810546875, 0.2827320098876953, 0.16042709350585938, -0.3095245361328125, -0.1654815673828125, 0.1144256591796875, 0.7031173706054688, 0.020168304443359375, -0.2291717529296875, -0.01763916015625, 0.34844970703125, 0.788360595703125, 0.3181304931640625, 1.14825439453125, 0.2233428955078125, 0.8079833984375, 0.2971477508544922, 0.90325927734375, -1.2313919067382812, 0.324005126953125, -0.7117156982421875, -0.14910888671875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000009.npy"} +{"epoch": 0.013215859030837005, "step": 10, "batch_size": 64, "mean": -0.06056436896324158, "std": 0.338525652885437, "min": -1.1921157836914062, "p10": -0.37896194458007815, "median": -0.0813760757446289, "p90": 0.3085655212402345, "max": 0.8705291748046875, "pos_frac": 0.40625, "sample": [-0.04161834716796875, -0.2032470703125, 0.09479141235351562, -0.20221900939941406, -0.4154987335205078, 0.2415924072265625, -0.13744735717773438, 0.32427978515625, 0.32257080078125, -0.26653289794921875, 0.19310760498046875, 0.25408935546875, -1.1921157836914062, 0.214813232421875, -0.386260986328125, 0.232635498046875, -0.3019447326660156, -0.210601806640625, -0.04308891296386719, 0.15420150756835938, -0.03720855712890625, -0.20664215087890625, 0.08911514282226562, 0.12838363647460938, -0.3003044128417969, -0.051998138427734375, -0.13884735107421875, 0.27588653564453125, -0.07023048400878906, -0.15073394775390625, -0.9964599609375, -0.5188179016113281, -0.24144554138183594, -0.22415924072265625, -0.16220474243164062, -0.17933273315429688, -0.740142822265625, 0.4772834777832031, 0.3553447723388672, -0.12124252319335938, -0.249755859375, -0.3775787353515625, -0.2929401397705078, 0.22296142578125, 0.137481689453125, 0.0638580322265625, 0.41600799560546875, -0.24474334716796875, 0.03387451171875, 0.23766326904296875, 0.020416259765625, -0.09252166748046875, -0.37955474853515625, -0.22841262817382812, -0.1056365966796875, -0.2065143585205078, 0.146026611328125, -0.01392364501953125, 0.18885040283203125, 0.8705291748046875, 0.6917037963867188, -0.36573028564453125, -0.20185279846191406, 0.035923004150390625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000010.npy"} +{"epoch": 0.014684287812041116, "step": 11, "batch_size": 64, "mean": -0.005599021911621094, "std": 0.3497479259967804, "min": -0.8091049194335938, "p10": -0.4312568664550781, "median": -0.043953895568847656, "p90": 0.4825210571289063, "max": 0.87017822265625, "pos_frac": 0.4375, "sample": [-0.489410400390625, 0.6182861328125, 0.5292930603027344, -0.020803451538085938, -0.46250152587890625, 0.05240631103515625, -0.25724029541015625, -0.04931449890136719, -0.03694915771484375, 0.8560943603515625, 0.11139106750488281, -0.28675079345703125, 0.28490447998046875, 0.26250457763671875, -0.2862548828125, 0.4904632568359375, -0.0948028564453125, 0.21044921875, -0.686065673828125, 0.2037487030029297, 0.22017860412597656, 0.11993408203125, -0.19764328002929688, 0.25116729736328125, -0.09259033203125, -0.16728973388671875, 0.10826873779296875, -0.3539276123046875, -0.0621185302734375, 0.0863037109375, 0.4639892578125, -0.0921478271484375, -0.1262969970703125, -0.059783935546875, -0.17296218872070312, 0.134002685546875, 0.11891937255859375, 0.87017822265625, 0.0567779541015625, -0.2909832000732422, -0.038593292236328125, -0.17458724975585938, -0.49402618408203125, -0.0876312255859375, -0.14989852905273438, 0.07034683227539062, 0.7693939208984375, -0.021196365356445312, -0.438995361328125, -0.31098175048828125, -0.22362136840820312, 0.040252685546875, 0.29522705078125, -0.8091049194335938, -0.20067214965820312, 0.10305595397949219, -0.5944442749023438, -0.2534217834472656, -0.41320037841796875, 0.6884002685546875, 0.2653656005859375, 0.20299720764160156, -0.12476921081542969, -0.22165679931640625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000011.npy"} +{"epoch": 0.016152716593245228, "step": 12, "batch_size": 64, "mean": 0.0394621342420578, "std": 0.38682809472084045, "min": -1.349395751953125, "p10": -0.3883712768554687, "median": 0.07254362106323242, "p90": 0.43392524719238285, "max": 0.9300537109375, "pos_frac": 0.65625, "sample": [0.9300537109375, -0.319610595703125, 0.4080963134765625, -0.92486572265625, -0.28864288330078125, 0.0830078125, 0.2210845947265625, 0.20001220703125, 0.28607940673828125, 0.5530624389648438, 0.0323028564453125, -0.53179931640625, -0.13674163818359375, 0.1526031494140625, -0.21439361572265625, 0.489959716796875, 0.418731689453125, 0.36907958984375, 0.05377960205078125, -0.2490997314453125, -1.349395751953125, 0.05908966064453125, -0.080474853515625, -0.6827850341796875, -0.6369285583496094, 0.2434234619140625, 0.2045135498046875, 0.3230438232421875, -0.12347412109375, 0.07323455810546875, -0.411895751953125, 0.4368743896484375, 0.1625518798828125, 0.13218307495117188, 0.02420806884765625, 0.04364204406738281, -0.024133682250976562, 0.17020416259765625, -0.3334808349609375, -0.2030620574951172, -0.5508804321289062, 0.0547637939453125, -0.226318359375, 0.4270439147949219, -0.02599334716796875, 0.083221435546875, 0.634552001953125, -0.17155838012695312, 0.0718526840209961, 0.008453369140625, -0.24044227600097656, 0.1811656951904297, 0.34389495849609375, 0.32602691650390625, 0.169586181640625, 0.1695556640625, 0.7574615478515625, 0.08090591430664062, 0.017732620239257812, 0.03308868408203125, 0.6972427368164062, 0.244964599609375, -0.3070068359375, 0.18622589111328125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000012.npy"} +{"epoch": 0.01762114537444934, "step": 13, "batch_size": 64, "mean": -0.01988467574119568, "std": 0.41210681200027466, "min": -1.0084228515625, "p10": -0.3974853515625, "median": -0.042098045349121094, "p90": 0.45746879577636734, "max": 1.4728240966796875, "pos_frac": 0.453125, "sample": [-0.5523529052734375, -0.10229873657226562, 0.16802215576171875, -0.1060028076171875, -0.3049468994140625, 0.4944114685058594, 0.177703857421875, 0.03768157958984375, 0.420135498046875, -0.2886848449707031, 0.2235107421875, 0.83642578125, -0.28356170654296875, -0.39786529541015625, 0.15235519409179688, 1.4728240966796875, -0.007656097412109375, -0.11737060546875, -0.03460884094238281, -0.1336517333984375, -0.10317420959472656, 0.09178924560546875, 0.7403335571289062, -0.049587249755859375, -0.35439491271972656, -0.9836578369140625, 0.09437942504882812, -0.12921142578125, -0.030803680419921875, -0.31182861328125, -0.19889450073242188, 0.026275634765625, -0.103179931640625, 0.23130416870117188, -1.0084228515625, -0.28766632080078125, -0.39659881591796875, -0.25836753845214844, -0.09250831604003906, 0.170166015625, 0.2747764587402344, -0.06315040588378906, 0.08758544921875, 0.17322158813476562, 0.03995513916015625, 0.003875732421875, -0.15801429748535156, 0.13535308837890625, 0.3087577819824219, -0.08076858520507812, -0.635009765625, 0.54962158203125, -0.0708770751953125, 0.14691162109375, -0.28720855712890625, -0.9910964965820312, 0.4734687805175781, 0.17769622802734375, 0.6446533203125, -0.33339691162109375, 0.201385498046875, -0.653289794921875, -0.21307373046875, 0.29598236083984375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000013.npy"} +{"epoch": 0.01908957415565345, "step": 14, "batch_size": 64, "mean": -0.06844127178192139, "std": 0.4771553874015808, "min": -1.648223876953125, "p10": -0.6269660949707031, "median": -0.050022125244140625, "p90": 0.5469905853271485, "max": 0.90771484375, "pos_frac": 0.46875, "sample": [-0.103668212890625, 0.2606849670410156, 0.045207977294921875, -0.23883056640625, 0.90771484375, 0.030094146728515625, -1.648223876953125, -0.1492462158203125, 0.060455322265625, 0.2626819610595703, -0.3126068115234375, 0.19868850708007812, 0.6009063720703125, 0.39708709716796875, -0.3864707946777344, -0.219390869140625, -0.427337646484375, -0.10230255126953125, 0.06984710693359375, -0.1806964874267578, 0.5489692687988281, -0.176422119140625, 0.0248565673828125, 0.07408523559570312, -0.17144775390625, -0.2723121643066406, -0.4044342041015625, 0.20851898193359375, -0.036346435546875, 0.34124755859375, 0.28841400146484375, -0.1104736328125, -0.242462158203125, 0.17470169067382812, 0.0031261444091796875, 0.6184883117675781, 0.08418655395507812, -0.89251708984375, -0.1674518585205078, -0.12652969360351562, -1.0576171875, -0.06369781494140625, -0.8992385864257812, -0.257415771484375, 0.5423736572265625, 0.0270843505859375, 0.09642982482910156, -0.5792160034179688, -1.113189697265625, -0.17399978637695312, -0.647430419921875, 0.6532821655273438, 0.02787017822265625, -0.4218330383300781, -0.2953147888183594, 0.7397994995117188, 0.502685546875, 0.795166015625, -1.0017852783203125, -0.01766204833984375, -0.3697700500488281, 0.20029258728027344, 0.36264801025390625, -0.2604942321777344], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000014.npy"} +{"epoch": 0.020558002936857563, "step": 15, "batch_size": 64, "mean": 0.010961085557937622, "std": 0.34748896956443787, "min": -1.1977920532226562, "p10": -0.3552051544189453, "median": 0.06501483917236328, "p90": 0.3883094787597656, "max": 0.7895278930664062, "pos_frac": 0.546875, "sample": [-0.01406097412109375, -0.09747314453125, 0.17753028869628906, -0.0022735595703125, 0.33202362060546875, -0.0003814697265625, 0.12354850769042969, -0.05991363525390625, 0.3895111083984375, 0.3594970703125, -0.1399383544921875, 0.09351539611816406, 0.49945831298828125, -1.1615753173828125, 0.1421070098876953, 0.094757080078125, 0.021091461181640625, -0.0706939697265625, 0.16885757446289062, 0.7895278930664062, 0.28998565673828125, -1.1977920532226562, -0.27867889404296875, -0.1659698486328125, -0.6829051971435547, -0.315887451171875, 0.15036773681640625, 0.09710693359375, -0.394683837890625, 0.25475311279296875, -0.11993408203125, -0.2224273681640625, -0.08763885498046875, -0.11790847778320312, 0.1582183837890625, 0.6202392578125, 0.07442474365234375, 0.142181396484375, 0.1305999755859375, -0.3361473083496094, -0.014989852905273438, 0.23781585693359375, 0.05560493469238281, 0.4190082550048828, -0.01847076416015625, 0.046337127685546875, 0.1443653106689453, -0.2859477996826172, 0.10140228271484375, -0.03491973876953125, 0.17668914794921875, -0.040119171142578125, 0.38550567626953125, -0.01718902587890625, -0.363372802734375, 0.45652008056640625, -0.618133544921875, 0.14020538330078125, 0.12415695190429688, 0.0778656005859375, -0.06968498229980469, -0.5526657104492188, 0.2588081359863281, 0.44969940185546875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000015.npy"} +{"epoch": 0.022026431718061675, "step": 16, "batch_size": 64, "mean": 0.05033496022224426, "std": 0.40199267864227295, "min": -1.2354812622070312, "p10": -0.405999755859375, "median": 0.0968942642211914, "p90": 0.46608657836914064, "max": 0.82342529296875, "pos_frac": 0.671875, "sample": [0.32388877868652344, -0.08953857421875, -0.31542205810546875, -0.24246978759765625, 0.050960540771484375, 0.5462112426757812, 0.584075927734375, 0.140625, -0.16163253784179688, 0.35283660888671875, -0.4131317138671875, 0.46715545654296875, 0.6364974975585938, 0.3205413818359375, -0.07417106628417969, -0.14099884033203125, 0.1220550537109375, -0.700225830078125, -1.0898818969726562, 0.17730712890625, 0.4094085693359375, 0.2363739013671875, 0.07704544067382812, 0.026092529296875, 0.001708984375, -0.4185009002685547, -0.913330078125, -0.1468505859375, 0.009794235229492188, 0.024524688720703125, 0.4275474548339844, 0.1364593505859375, 0.23894500732421875, -0.3437347412109375, 0.15386199951171875, 0.11578369140625, 0.3625335693359375, 0.090789794921875, -0.3893585205078125, 0.36989402770996094, 0.463592529296875, -0.09633636474609375, 0.10299873352050781, 0.30804443359375, 0.7288818359375, 0.255859375, 0.08945465087890625, -0.2008056640625, -0.106048583984375, 0.151336669921875, -0.07103347778320312, 0.27425384521484375, -0.372406005859375, 0.1743621826171875, 0.17247772216796875, -0.6949081420898438, 0.04644966125488281, 0.82342529296875, 0.6511993408203125, 0.411468505859375, 0.30718994140625, 0.00086212158203125, 0.07292938232421875, -1.2354812622070312], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000016.npy"} +{"epoch": 0.023494860499265784, "step": 17, "batch_size": 64, "mean": 0.08410793542861938, "std": 0.3371948301792145, "min": -1.03509521484375, "p10": -0.287961196899414, "median": 0.09944534301757812, "p90": 0.5082920074462892, "max": 1.1090850830078125, "pos_frac": 0.640625, "sample": [-0.21877288818359375, 0.19531822204589844, 0.0641937255859375, 0.01068115234375, -0.452789306640625, -0.21195030212402344, 0.6009063720703125, -1.03509521484375, 0.096710205078125, 0.03388214111328125, 0.635894775390625, 0.1300506591796875, -0.0441741943359375, -0.1226348876953125, 0.022220611572265625, 0.669952392578125, -0.1868438720703125, -0.06398773193359375, 0.15408897399902344, 0.38300323486328125, 0.08017730712890625, 0.05437278747558594, -0.3417091369628906, 0.3083038330078125, 0.1474151611328125, -0.5133285522460938, 0.3016796112060547, -0.11600494384765625, 0.3194427490234375, 0.1912841796875, 0.02941131591796875, 0.1161041259765625, 0.199432373046875, 0.230316162109375, 0.2834796905517578, 0.480865478515625, -0.005893707275390625, 0.104583740234375, 0.2831687927246094, 0.10858154296875, 0.5433120727539062, 0.10218048095703125, 0.20026016235351562, 0.08436203002929688, -0.147552490234375, -0.3706207275390625, -0.23100662231445312, -0.11181068420410156, -0.10092926025390625, -0.0229949951171875, -0.31237030029296875, 0.18343734741210938, -0.01702880859375, -0.19033432006835938, 0.20503997802734375, -0.219329833984375, 0.47918701171875, 0.38771820068359375, 1.1090850830078125, -0.6377716064453125, 0.2455005645751953, 0.16466522216796875, 0.5975265502929688, 0.5200462341308594], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000017.npy"} +{"epoch": 0.024963289280469897, "step": 18, "batch_size": 64, "mean": 0.055854856967926025, "std": 0.28332436084747314, "min": -0.5408172607421875, "p10": -0.25344295501708985, "median": -0.012065887451171875, "p90": 0.4075723648071289, "max": 0.907318115234375, "pos_frac": 0.484375, "sample": [0.2548789978027344, -0.5408172607421875, 0.514801025390625, 0.3292083740234375, -0.16318893432617188, -0.4192047119140625, -0.23700904846191406, -0.23345947265625, 0.1124725341796875, 0.4069995880126953, -0.03826904296875, -0.03143882751464844, -0.06760406494140625, 0.2649192810058594, -0.235198974609375, 0.27605438232421875, -0.24010086059570312, 0.11837005615234375, -0.0054779052734375, -0.04238128662109375, 0.12439346313476562, 0.0953521728515625, 0.282501220703125, 0.13428878784179688, 0.6298446655273438, -0.06212043762207031, 0.238189697265625, 0.28220367431640625, -0.25916099548339844, -0.16381454467773438, 0.37477874755859375, 0.2682609558105469, -0.28353118896484375, 0.06513214111328125, 0.907318115234375, -0.12839508056640625, -0.01865386962890625, 0.4078178405761719, -0.48200225830078125, -0.3900909423828125, 0.397064208984375, -0.12560272216796875, -0.3011474609375, 0.49736785888671875, 0.535247802734375, -0.17653656005859375, 0.09488296508789062, -0.1727142333984375, 0.153076171875, 0.16756439208984375, 0.2078399658203125, -0.10448074340820312, -0.02291107177734375, -0.07967376708984375, -0.16566848754882812, 0.43309783935546875, -0.1533050537109375, -0.058330535888671875, 0.2190093994140625, -0.043804168701171875, -0.12445449829101562, 0.06451797485351562, 0.31432342529296875, -0.026517868041992188], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000018.npy"} +{"epoch": 0.02643171806167401, "step": 19, "batch_size": 64, "mean": 0.07485997676849365, "std": 0.3466305136680603, "min": -0.7751617431640625, "p10": -0.34803009033203125, "median": 0.039886474609375, "p90": 0.6309961318969728, "max": 0.8965682983398438, "pos_frac": 0.546875, "sample": [0.1310596466064453, -0.07623481750488281, -0.3740501403808594, 0.09336471557617188, -0.01049041748046875, 0.33941650390625, 0.607269287109375, -0.16613006591796875, 0.03753089904785156, 0.8965682983398438, -0.242095947265625, -0.20026397705078125, 0.3518562316894531, -0.00933837890625, 0.5137481689453125, -0.7751617431640625, -0.4098968505859375, 0.2850608825683594, -0.712432861328125, 0.181365966796875, 0.3296966552734375, 0.10649871826171875, -0.04204368591308594, -0.29058837890625, -0.06493186950683594, 0.2940540313720703, -0.07308197021484375, 0.073089599609375, -0.06581497192382812, -0.0942230224609375, 0.1280364990234375, 0.6411647796630859, 0.08115386962890625, -0.3701763153076172, 0.028961181640625, -0.15227890014648438, -0.395751953125, 0.04224205017089844, -0.0536956787109375, -0.332916259765625, 0.1980133056640625, -0.3043365478515625, 0.6775436401367188, 0.011077880859375, 0.19000625610351562, 0.64617919921875, 0.3704833984375, 0.35271453857421875, 0.324981689453125, -0.17610931396484375, 0.080078125, 0.13721084594726562, -0.0146636962890625, -0.0067691802978515625, 0.7278060913085938, 0.1609344482421875, -0.18792724609375, 0.22687911987304688, -0.3545074462890625, 0.804107666015625, 0.696075439453125, 0.2716789245605469, -0.07173538208007812, -0.2192230224609375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000019.npy"} +{"epoch": 0.027900146842878122, "step": 20, "batch_size": 64, "mean": 0.03958427906036377, "std": 0.3709593415260315, "min": -1.4266738891601562, "p10": -0.3748468399047852, "median": 0.10810470581054688, "p90": 0.426272201538086, "max": 0.7168197631835938, "pos_frac": 0.609375, "sample": [0.10558700561523438, 0.07603836059570312, 0.08405113220214844, 0.303955078125, -0.04474639892578125, 0.0476531982421875, -0.232147216796875, 0.2103595733642578, -0.4635162353515625, 0.41551971435546875, -0.7858734130859375, -0.188995361328125, 0.16222381591796875, -0.21170997619628906, -0.1183319091796875, 0.18761062622070312, 0.23051834106445312, 0.4797935485839844, 0.1437530517578125, 0.3086395263671875, 0.6286087036132812, -0.5650177001953125, -0.2457275390625, -0.3728160858154297, -0.2521209716796875, -0.1948089599609375, -0.4180908203125, 0.3497161865234375, -0.17130661010742188, -0.23442459106445312, 0.30246925354003906, 0.11062240600585938, -0.5044403076171875, 0.103179931640625, 0.13252639770507812, -0.2576866149902344, -0.002288818359375, -1.4266738891601562, -0.12873458862304688, -0.3757171630859375, 0.7168197631835938, 0.5823745727539062, 0.142242431640625, 0.7105979919433594, -0.22301864624023438, 0.3610248565673828, 0.177490234375, 0.359588623046875, 0.5551605224609375, 0.4196739196777344, 0.1595611572265625, 0.03784942626953125, 0.3381214141845703, -0.15254974365234375, -0.33280181884765625, -0.3002147674560547, 0.15936279296875, 0.3529033660888672, 0.19229888916015625, 0.027860641479492188, 0.12456893920898438, 0.42910003662109375, 0.2374267578125, 0.27030181884765625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000020.npy"} +{"epoch": 0.02936857562408223, "step": 21, "batch_size": 64, "mean": 0.14244650304317474, "std": 0.405142217874527, "min": -1.0787200927734375, "p10": -0.34736175537109376, "median": 0.1759805679321289, "p90": 0.6664443969726562, "max": 1.1633377075195312, "pos_frac": 0.6875, "sample": [0.24810028076171875, -0.34078216552734375, 0.27307891845703125, 0.6650238037109375, 0.7908172607421875, 0.18990421295166016, -0.14279747009277344, 0.28237152099609375, 0.66705322265625, 0.5200958251953125, 0.219512939453125, -0.470062255859375, 0.07285881042480469, 0.182281494140625, 0.2359466552734375, 0.42150115966796875, -0.1686992645263672, 0.10596656799316406, -0.1734771728515625, 0.44503021240234375, -0.049896240234375, 0.7184600830078125, 0.5527191162109375, -0.07081985473632812, -0.12520790100097656, 1.1633377075195312, 0.2486572265625, 0.13419151306152344, 0.18595504760742188, 0.27397918701171875, -0.6294784545898438, 0.22994232177734375, 0.4596977233886719, 0.3016338348388672, -0.35018157958984375, 0.05494117736816406, -0.2185516357421875, 0.08622550964355469, 0.866241455078125, 0.283721923828125, -0.04925537109375, 0.1696796417236328, 0.4514617919921875, 0.06337928771972656, 0.08224868774414062, -0.50848388671875, 0.3583335876464844, -0.12384033203125, -0.261993408203125, -0.7001495361328125, 0.14087486267089844, 0.64093017578125, 0.7589111328125, 0.2002716064453125, 0.03440666198730469, -0.08097457885742188, 0.0082550048828125, -0.51336669921875, 0.7750930786132812, 0.43342018127441406, -1.0787200927734375, 0.0539093017578125, 0.3652801513671875, -0.2423877716064453], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000021.npy"} +{"epoch": 0.030837004405286344, "step": 22, "batch_size": 64, "mean": 0.17688971757888794, "std": 0.45676496624946594, "min": -0.8770294189453125, "p10": -0.4336486816406249, "median": 0.17938995361328125, "p90": 0.7593101501464845, "max": 1.172149658203125, "pos_frac": 0.671875, "sample": [0.15705108642578125, 0.447357177734375, -0.0457916259765625, 0.24930191040039062, 0.511077880859375, -0.792510986328125, -0.8770294189453125, 0.39710235595703125, -0.36376190185546875, 0.623260498046875, 0.9050636291503906, 0.17437744140625, 0.12433624267578125, 0.8671607971191406, 0.1881256103515625, 0.07421875, 0.26788330078125, 0.504486083984375, 0.13794708251953125, -0.551055908203125, 0.0963287353515625, 0.9771575927734375, -0.6728515625, 0.7282867431640625, -0.1184539794921875, 0.5280799865722656, 0.39605712890625, 0.1844024658203125, -0.1059417724609375, 0.1296234130859375, 0.21857643127441406, 0.109527587890625, -0.2169342041015625, -0.5891151428222656, 0.5073356628417969, 0.42728424072265625, 0.6415290832519531, -0.21905517578125, -0.07723426818847656, 0.4710693359375, 0.30047607421875, 0.5068397521972656, 0.1635417938232422, 0.4354972839355469, -0.11382865905761719, 0.8921432495117188, -0.60064697265625, -0.34676361083984375, 0.9577484130859375, 0.34394073486328125, -0.19228553771972656, -0.46360015869140625, -0.2755584716796875, -0.07221031188964844, -0.3397216796875, 0.720001220703125, -0.2021942138671875, 0.21779823303222656, 0.7726058959960938, 1.172149658203125, 0.420135498046875, 0.5360565185546875, 0.049835205078125, 0.024709701538085938], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000022.npy"} +{"epoch": 0.032305433186490456, "step": 23, "batch_size": 64, "mean": 0.2380504608154297, "std": 0.4865148067474365, "min": -1.021575927734375, "p10": -0.25638351440429685, "median": 0.19916534423828125, "p90": 0.7492404937744142, "max": 2.062591552734375, "pos_frac": 0.75, "sample": [-0.10336685180664062, -0.11876296997070312, 0.31551361083984375, 0.09902572631835938, 0.705535888671875, 0.7550621032714844, 0.16794204711914062, -0.1604156494140625, -0.28978729248046875, -1.021575927734375, 0.8819427490234375, 1.5090408325195312, 0.73565673828125, 0.281005859375, 0.4674873352050781, 0.0437469482421875, 0.06876564025878906, 0.5354232788085938, 0.269012451171875, 0.021001815795898438, 0.8785247802734375, -0.3993492126464844, 2.062591552734375, 0.21345901489257812, -0.10522842407226562, 0.0993499755859375, -0.04828643798828125, -0.9067115783691406, 0.09737396240234375, 0.4319610595703125, -0.27584075927734375, 0.5211334228515625, -0.3596038818359375, 0.45354461669921875, 0.28197479248046875, 0.10057830810546875, 0.162445068359375, 0.04400634765625, 0.32084083557128906, 0.29070091247558594, 0.4531974792480469, 0.33565521240234375, 0.18718719482421875, 0.4746208190917969, -0.05704498291015625, 1.484161376953125, 0.6154518127441406, -0.1256103515625, 0.4581146240234375, -0.3100433349609375, 0.2332916259765625, 0.1251087188720703, 0.21114349365234375, -0.2109832763671875, 0.13786697387695312, 0.36846160888671875, 0.237884521484375, 0.8331298828125, 0.012081146240234375, 0.30780029296875, 0.03014373779296875, -0.1258087158203125, 0.08295059204101562, 0.45075035095214844], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000023.npy"} +{"epoch": 0.033773861967694566, "step": 24, "batch_size": 64, "mean": 0.1957438886165619, "std": 0.40492355823516846, "min": -0.79827880859375, "p10": -0.24687805175781247, "median": 0.1578845977783203, "p90": 0.6911228179931642, "max": 1.432220458984375, "pos_frac": 0.671875, "sample": [0.449981689453125, 0.6304740905761719, 0.33293914794921875, -0.0034027099609375, 0.23419761657714844, -0.029937744140625, 0.2387847900390625, 0.1327972412109375, 0.5838394165039062, 0.1959686279296875, 0.3569202423095703, -0.1192474365234375, -0.1500530242919922, 0.5550537109375, 0.41042327880859375, -0.2929859161376953, -0.30159759521484375, 0.017526626586914062, 0.826812744140625, -0.076080322265625, 0.6594505310058594, -0.2283782958984375, -0.2548065185546875, 0.04632568359375, 0.7046966552734375, 0.7280197143554688, 0.3013496398925781, 0.028594970703125, -0.1318511962890625, 0.2197723388671875, 1.432220458984375, 0.710723876953125, 0.22117996215820312, -0.33228302001953125, 1.003448486328125, -0.79827880859375, -0.1903228759765625, -0.073089599609375, -0.22782135009765625, 0.06975555419921875, 1.227386474609375, 0.11011123657226562, -0.12574386596679688, 0.5330963134765625, 0.30129241943359375, 0.5793533325195312, -0.47515869140625, -0.013502120971679688, 0.6329345703125, 0.15381240844726562, 0.27874755859375, 0.1699237823486328, 0.11223983764648438, 0.13022613525390625, 0.1407928466796875, 0.252838134765625, 0.22896194458007812, -0.1107177734375, 0.3004798889160156, -0.01752471923828125, 0.103759765625, -0.6119232177734375, 0.161956787109375, 0.5831451416015625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000024.npy"} +{"epoch": 0.03524229074889868, "step": 25, "batch_size": 64, "mean": 0.20315146446228027, "std": 0.46423274278640747, "min": -0.9643955230712891, "p10": -0.3405040740966796, "median": 0.19600296020507812, "p90": 0.6777240753173829, "max": 1.42364501953125, "pos_frac": 0.703125, "sample": [-0.25238800048828125, -0.07486343383789062, 0.565948486328125, 0.5224876403808594, 0.08370208740234375, -0.10531425476074219, -0.26778411865234375, 0.8656768798828125, -0.439239501953125, -0.039203643798828125, 0.2617950439453125, 0.680572509765625, -0.058177947998046875, 0.5842819213867188, 1.2248306274414062, -0.20592689514160156, -0.9643955230712891, 0.631500244140625, 0.037723541259765625, 0.44159698486328125, -0.3716697692871094, 0.04596710205078125, -0.03483772277832031, 0.20174407958984375, 0.6335296630859375, 0.074371337890625, 0.1902618408203125, 1.42364501953125, -0.07275390625, -0.0329742431640625, 0.28305816650390625, 0.02197265625, 0.0623016357421875, 0.4864654541015625, 0.11637115478515625, 0.4549369812011719, -0.05587577819824219, 1.3585662841796875, -0.6731414794921875, 0.3240814208984375, -0.19698715209960938, 0.37042236328125, 1.198028564453125, 0.3176994323730469, 0.4294319152832031, 0.24550628662109375, -0.5656967163085938, 0.0883331298828125, 0.46303558349609375, 0.4784049987792969, 0.2115478515625, 0.5486679077148438, 0.31920623779296875, -0.706695556640625, 0.22136688232421875, 0.14633750915527344, -0.5891571044921875, 0.6710777282714844, 0.296905517578125, 0.00554656982421875, 0.2567787170410156, 0.6884307861328125, 0.0789337158203125, 0.09572601318359375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000025.npy"} +{"epoch": 0.03671071953010279, "step": 26, "batch_size": 64, "mean": 0.3847799003124237, "std": 0.5604045391082764, "min": -0.557464599609375, "p10": -0.15171451568603517, "median": 0.2830314636230469, "p90": 1.2627754211425783, "max": 2.2513427734375, "pos_frac": 0.734375, "sample": [0.7786102294921875, 0.0880126953125, 0.3644256591796875, 0.35985374450683594, 0.209716796875, 0.22214508056640625, -0.01947021484375, 0.5086936950683594, -0.28363037109375, -0.14916038513183594, 0.4593353271484375, 0.36022186279296875, 1.391326904296875, -0.07437896728515625, 0.1056976318359375, 0.691802978515625, 0.576629638671875, 0.28998565673828125, 0.433685302734375, -0.07051467895507812, 0.15188217163085938, -0.011249542236328125, 0.4205474853515625, 0.636993408203125, 0.12380218505859375, 0.6611404418945312, 0.5869293212890625, -0.3547210693359375, 0.042469024658203125, 1.3692779541015625, -0.557464599609375, -0.13304901123046875, 0.39157676696777344, -0.13900375366210938, -0.15280914306640625, -0.0948333740234375, 1.44573974609375, -0.13329315185546875, 0.5505485534667969, 2.2513427734375, 0.16089630126953125, 0.2189178466796875, 1.77105712890625, 0.7855224609375, 0.4930839538574219, 0.746185302734375, -0.4532051086425781, 0.9789581298828125, 0.18302154541015625, 0.14050674438476562, -0.2301921844482422, 0.18283843994140625, 0.10518646240234375, 0.39693450927734375, -0.5205917358398438, 1.366851806640625, 1.161041259765625, 1.2928619384765625, 1.1925735473632812, 0.028308868408203125, 0.2760772705078125, -0.07764816284179688, 0.712371826171875, 0.4155406951904297], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000026.npy"} +{"epoch": 0.0381791483113069, "step": 27, "batch_size": 64, "mean": 0.4595893323421478, "std": 0.5447052717208862, "min": -0.53155517578125, "p10": -0.12427864074707023, "median": 0.3883171081542969, "p90": 1.0885353088378906, "max": 2.2380752563476562, "pos_frac": 0.828125, "sample": [-0.054645538330078125, 0.9925193786621094, 0.6375961303710938, 0.0941314697265625, 0.4665260314941406, 0.8581695556640625, 0.70831298828125, -0.499053955078125, 0.8649444580078125, -0.3678703308105469, 0.5112590789794922, 0.24895095825195312, 0.6404533386230469, 0.0645751953125, 1.286376953125, 0.904632568359375, 2.2380752563476562, 0.19953155517578125, 0.3631629943847656, 0.7631607055664062, 0.07838058471679688, 0.521820068359375, 0.6425704956054688, -0.29553794860839844, 0.11248016357421875, 0.26992225646972656, 0.1262187957763672, 0.03878021240234375, 0.34874725341796875, 2.0396575927734375, 0.11328125, 0.22291946411132812, -0.026002883911132812, 1.39312744140625, 0.7127609252929688, 0.7261810302734375, 0.5414390563964844, 1.09710693359375, -0.027820587158203125, 0.132568359375, -0.43076324462890625, 0.2506866455078125, 0.2992095947265625, 0.9171218872070312, 0.36699676513671875, 1.202850341796875, -0.53155517578125, 0.4365406036376953, 0.9747085571289062, 0.8537521362304688, 0.6205291748046875, 0.23202133178710938, 0.6703643798828125, 0.467681884765625, 0.409637451171875, -0.15412139892578125, 1.0685348510742188, -0.2534065246582031, 0.12083625793457031, -0.03823089599609375, 0.03144073486328125, 0.785919189453125, 0.095977783203125, 1.32757568359375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000027.npy"} +{"epoch": 0.039647577092511016, "step": 28, "batch_size": 64, "mean": 0.3094020485877991, "std": 0.6136017441749573, "min": -2.183502197265625, "p10": -0.2957344055175781, "median": 0.30254173278808594, "p90": 1.090586280822754, "max": 1.4806480407714844, "pos_frac": 0.734375, "sample": [1.1034507751464844, -0.18316650390625, 7.62939453125e-05, 0.8487281799316406, 1.0867290496826172, -0.2315673828125, 0.008146286010742188, 0.2619762420654297, -0.015336990356445312, 0.20539283752441406, -0.2782440185546875, 0.9527587890625, 0.36163330078125, 0.7285003662109375, 0.38336944580078125, -0.1855182647705078, 0.3078804016113281, 0.805450439453125, 0.10138320922851562, 0.262115478515625, -0.18914794921875, -0.2643890380859375, 0.2269744873046875, 1.2371826171875, 0.1599578857421875, 0.47064971923828125, 0.5204620361328125, 0.483978271484375, 1.4806480407714844, 0.49855804443359375, 0.27819061279296875, 0.132049560546875, 0.10448455810546875, -0.6641616821289062, 0.4388999938964844, 0.5110416412353516, 0.6163234710693359, 0.8791313171386719, 0.29720306396484375, -0.30323028564453125, 0.31343841552734375, 0.38077735900878906, -0.015604019165039062, 1.2728424072265625, 1.062591552734375, 1.224090576171875, 1.0922393798828125, -0.06595420837402344, 0.8320770263671875, 0.09261322021484375, 0.7892665863037109, -0.8852081298828125, -0.42214202880859375, 0.5969009399414062, -0.3049888610839844, 0.39788818359375, -0.1310272216796875, 1.370849609375, 1.02740478515625, 0.1891021728515625, 0.36443328857421875, -2.183502197265625, -0.747161865234375, 0.11223983764648438], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000028.npy"} +{"epoch": 0.041116005873715125, "step": 29, "batch_size": 64, "mean": 0.5431155562400818, "std": 0.571643590927124, "min": -0.5013580322265625, "p10": -0.12165832519531247, "median": 0.49868297576904297, "p90": 1.2525711059570312, "max": 2.5500030517578125, "pos_frac": 0.8125, "sample": [1.252777099609375, -0.05209922790527344, 0.43109130859375, 0.2758903503417969, 0.2293548583984375, 0.48482322692871094, -0.09233856201171875, 1.1807098388671875, 0.898773193359375, 0.217041015625, 0.22708892822265625, -0.09037590026855469, 1.543182373046875, 1.0910186767578125, -0.19113922119140625, 1.5285797119140625, 1.1169891357421875, 0.13894271850585938, -0.06008338928222656, 1.435791015625, -0.5013580322265625, 0.6231956481933594, 0.4576301574707031, 0.29845619201660156, 0.7440032958984375, 0.8698959350585938, -0.25848388671875, 0.7596893310546875, 0.52862548828125, 0.26922607421875, 0.18678665161132812, 0.8608474731445312, 2.5500030517578125, -0.296600341796875, 0.19927978515625, 0.014818191528320312, 0.15866851806640625, -0.2950325012207031, 1.5724334716796875, 0.9136962890625, 0.6259765625, 0.7565765380859375, 0.6087722778320312, 0.512542724609375, 0.4591655731201172, 0.27053070068359375, 0.8869857788085938, 0.3045501708984375, 0.7409820556640625, -0.40074920654296875, 0.21396827697753906, 0.2147369384765625, 1.1451034545898438, 0.23438262939453125, 0.8002891540527344, 1.2520904541015625, -0.13422393798828125, 0.799285888671875, 0.6061592102050781, 0.5752143859863281, 1.522705078125, 0.72894287109375, -0.0010223388671875, 0.8146324157714844], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000029.npy"} +{"epoch": 0.042584434654919234, "step": 30, "batch_size": 64, "mean": 0.6291906237602234, "std": 0.6676943898200989, "min": -0.77874755859375, "p10": -0.21032142639160156, "median": 0.4987039566040039, "p90": 1.4855880737304688, "max": 2.736297607421875, "pos_frac": 0.828125, "sample": [0.9634628295898438, 0.428375244140625, -0.22627639770507812, 0.39532470703125, 2.015472412109375, -0.21318817138671875, -0.4283332824707031, 0.304718017578125, -0.2679729461669922, 0.4059906005859375, 0.5007476806640625, 0.6555404663085938, 0.7018165588378906, 1.285400390625, -0.77874755859375, 1.0001258850097656, 0.7221755981445312, 0.34816741943359375, 0.003520965576171875, 1.450897216796875, 0.25262451171875, 0.963409423828125, 2.736297607421875, 0.8809661865234375, 1.3196182250976562, 0.3961906433105469, 0.84368896484375, 0.9291152954101562, -0.09893798828125, 0.16450119018554688, 0.5540695190429688, 0.2280101776123047, 0.86907958984375, -0.17641639709472656, 0.10850143432617188, 0.9309616088867188, 1.825531005859375, 1.280029296875, 1.688568115234375, 0.9492340087890625, -0.007808685302734375, 1.240203857421875, 0.20308494567871094, 0.4240264892578125, -0.46372222900390625, 1.488861083984375, 1.4779510498046875, 0.4966602325439453, 0.982757568359375, 0.8555908203125, 1.5467529296875, 1.048370361328125, 0.4736499786376953, -0.2826557159423828, 0.11974143981933594, 1.181884765625, 0.4401588439941406, -0.20363235473632812, 0.42549896240234375, 0.27577972412109375, 0.108612060546875, 1.6604461669921875, 0.6256256103515625, 0.23810195922851562], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000030.npy"} +{"epoch": 0.04405286343612335, "step": 31, "batch_size": 64, "mean": 0.49011051654815674, "std": 0.6812103986740112, "min": -0.6348876953125, "p10": -0.25500831604003904, "median": 0.34700965881347656, "p90": 1.4084609985351566, "max": 2.8839035034179688, "pos_frac": 0.765625, "sample": [-0.14581298828125, 1.160919189453125, -0.33148193359375, 0.4228477478027344, 2.0401840209960938, 0.6115264892578125, -0.2700996398925781, -0.2780723571777344, 0.31072998046875, -0.2150726318359375, 2.8839035034179688, 0.014862060546875, 0.221527099609375, 0.31768035888671875, 0.009334564208984375, 0.16177940368652344, 0.973876953125, 0.45648956298828125, 1.438568115234375, 0.8102130889892578, 0.08074951171875, -0.026714324951171875, -0.1740570068359375, 0.160308837890625, -0.09785079956054688, 0.6618595123291016, 0.5987663269042969, 0.19492721557617188, 0.8116436004638672, -0.4502449035644531, 1.975372314453125, 0.489288330078125, 1.805450439453125, -0.49383544921875, 0.2585487365722656, 0.268310546875, -0.13298797607421875, 0.5499343872070312, -0.21979522705078125, 0.602020263671875, 1.1156539916992188, 1.481475830078125, -0.18723297119140625, 0.3213081359863281, 0.330078125, -0.6348876953125, 0.8862686157226562, 0.8207550048828125, 0.27608489990234375, 0.0784912109375, 0.3639411926269531, 1.12347412109375, 1.3382110595703125, -0.5015716552734375, 0.17089080810546875, 0.7722244262695312, 0.420867919921875, 0.6768951416015625, 0.5996246337890625, 0.22406005859375, 0.9321517944335938, 1.7686080932617188, 0.847900390625, 0.6862030029296875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000031.npy"} +{"epoch": 0.04552129221732746, "step": 32, "batch_size": 64, "mean": 0.7347670793533325, "std": 0.8586938381195068, "min": -0.6252632141113281, "p10": -0.0667327880859375, "median": 0.5780048370361328, "p90": 1.8761985778808594, "max": 4.128265380859375, "pos_frac": 0.84375, "sample": [0.0480804443359375, -0.039581298828125, 0.48883819580078125, 1.0599899291992188, 0.6305770874023438, 1.420257568359375, -0.0543212890625, 0.7167816162109375, 0.5452156066894531, 1.0095291137695312, 1.3017387390136719, 0.025104522705078125, 0.010990142822265625, 0.16621017456054688, -0.4012298583984375, 1.152069091796875, -0.6252632141113281, 1.0188484191894531, -0.4369392395019531, 2.4418182373046875, 0.5901298522949219, 1.6966476440429688, 2.1020965576171875, -0.23166656494140625, 1.0082588195800781, 0.8891334533691406, 0.1465301513671875, 0.21533966064453125, -0.3135490417480469, 0.6851043701171875, 0.5502090454101562, 0.15439224243164062, 0.321319580078125, 2.88031005859375, 1.2653770446777344, 0.8070697784423828, 0.795196533203125, 0.7910079956054688, 0.02550506591796875, 0.18463897705078125, 1.844940185546875, 0.7407608032226562, -0.072052001953125, 0.4673614501953125, 4.128265380859375, 0.9865341186523438, 0.3628997802734375, 1.8895950317382812, 2.00537109375, -0.04231452941894531, 0.2731647491455078, 0.3957366943359375, 2.5788192749023438, 1.0144195556640625, 0.027904510498046875, -0.11999130249023438, 1.118408203125, 1.382781982421875, 0.15756988525390625, 0.692657470703125, 0.4877471923828125, 0.5658798217773438, 0.9030609130859375, 0.19380569458007812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000032.npy"} +{"epoch": 0.04698972099853157, "step": 33, "batch_size": 64, "mean": 0.626814067363739, "std": 0.6711933016777039, "min": -1.1497650146484375, "p10": -0.1309272766113281, "median": 0.5337791442871094, "p90": 1.476387023925782, "max": 2.476226806640625, "pos_frac": 0.859375, "sample": [-0.06000518798828125, 0.514404296875, 0.19453811645507812, 0.3795051574707031, 2.0328216552734375, 0.11623001098632812, 0.4195709228515625, 0.16265869140625, 0.00018310546875, 0.7499752044677734, 0.26090240478515625, 0.942474365234375, 1.74371337890625, 0.4691619873046875, 1.54974365234375, 0.67236328125, 0.9819831848144531, -0.5419425964355469, 0.17120361328125, -1.1497650146484375, 0.1473236083984375, 1.3052215576171875, 0.9968948364257812, 0.7546157836914062, 0.6001968383789062, -0.1130523681640625, 1.7777938842773438, -0.173065185546875, 0.2105865478515625, 1.666351318359375, 0.43422698974609375, -0.16359710693359375, 2.277484893798828, 0.13628768920898438, 0.0815887451171875, 1.0082588195800781, 0.354827880859375, 0.847259521484375, 1.21917724609375, 1.0640335083007812, 0.2812080383300781, 1.2717361450195312, 0.010101318359375, 0.4501152038574219, 0.6529159545898438, 1.1521759033203125, 0.485626220703125, 2.476226806640625, 0.8836269378662109, -0.21000289916992188, 0.746826171875, -0.143218994140625, 0.655731201171875, 0.027036666870117188, 0.271636962890625, 1.2786827087402344, 1.2564010620117188, -0.13858795166015625, 1.1418685913085938, 0.5531539916992188, 0.5836677551269531, 0.666534423828125, 0.45951271057128906, 1.2609901428222656], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000033.npy"} +{"epoch": 0.048458149779735685, "step": 34, "batch_size": 64, "mean": 0.868897557258606, "std": 0.8507899045944214, "min": -0.611297607421875, "p10": -0.05736808776855467, "median": 0.6554279327392578, "p90": 2.0385807037353514, "max": 3.075439453125, "pos_frac": 0.859375, "sample": [2.0754165649414062, -0.4013023376464844, 1.0074386596679688, 0.142547607421875, 2.8334197998046875, 0.5604419708251953, 1.31768798828125, 0.2322235107421875, -0.11065673828125, 0.28421783447265625, 0.6563339233398438, 0.020717620849609375, 0.5497207641601562, 0.18425750732421875, 1.5977630615234375, 1.9054412841796875, 0.4641265869140625, 0.16187286376953125, 1.632354736328125, 0.48490142822265625, 1.69744873046875, -0.1020050048828125, 1.1741085052490234, 2.0300865173339844, 0.84307861328125, 1.8893585205078125, 1.9091720581054688, -0.611297607421875, 1.0250244140625, 1.0902252197265625, 0.50189208984375, 1.0201644897460938, 1.2892951965332031, 1.0159683227539062, 2.902587890625, 0.7106361389160156, 0.9857025146484375, 2.660308837890625, 0.30597496032714844, -0.028614044189453125, 0.3199272155761719, 0.5484962463378906, 1.0412216186523438, 1.6646041870117188, 1.1231193542480469, 0.6464881896972656, 0.21750450134277344, 0.3398475646972656, 1.4839706420898438, -0.1358795166015625, -0.06392669677734375, 0.033935546875, 0.8018951416015625, 0.5661773681640625, 0.8415374755859375, 0.20776939392089844, 2.0422210693359375, -0.042064666748046875, -0.23366546630859375, 0.6545219421386719, 2.1411361694335938, 3.075439453125, 0.1785449981689453, 0.2485809326171875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000034.npy"} +{"epoch": 0.049926578560939794, "step": 35, "batch_size": 64, "mean": 1.1832659244537354, "std": 1.0637937784194946, "min": -0.6867599487304688, "p10": 0.08377552032470703, "median": 0.9696636199951172, "p90": 2.3740386962890625, "max": 6.04541015625, "pos_frac": 0.9375, "sample": [1.1980133056640625, 1.1295204162597656, 2.98175048828125, 1.5851936340332031, -0.029926300048828125, 0.4383506774902344, 1.136810302734375, -0.01616668701171875, -0.043792724609375, 1.678863525390625, 1.582427978515625, 0.04064369201660156, 0.7729263305664062, 0.7206363677978516, 0.4930152893066406, 0.0895233154296875, 1.2392158508300781, 3.1859054565429688, 0.35906219482421875, 0.30750083923339844, 0.8756599426269531, 2.0703125, 2.64837646484375, 0.345855712890625, 0.14452362060546875, 0.7805404663085938, 0.6028175354003906, 1.1417007446289062, 0.541748046875, 0.9028778076171875, 6.04541015625, 1.5438766479492188, 0.877197265625, 1.5248794555664062, 0.5437736511230469, 3.34368896484375, 0.12240982055664062, 2.208221435546875, 2.68707275390625, 2.3044357299804688, 0.45545196533203125, 2.3244781494140625, 1.0761947631835938, 1.2695999145507812, 0.9838829040527344, 0.0076694488525390625, 2.0473480224609375, 2.096649169921875, 1.0848541259765625, 0.6728172302246094, 1.0397529602050781, 0.48012542724609375, 0.8130588531494141, 2.0998687744140625, 0.9208526611328125, 1.0632286071777344, 0.9554443359375, 0.08131217956542969, 1.588287353515625, -0.6867599487304688, 0.6517810821533203, 2.3952789306640625, 1.766937255859375, 0.4360504150390625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000035.npy"} +{"epoch": 0.0513950073421439, "step": 36, "batch_size": 64, "mean": 1.1588000059127808, "std": 1.283197045326233, "min": -2.1429367065429688, "p10": -0.20764389038085931, "median": 0.9866952896118164, "p90": 3.095497131347657, "max": 4.21484375, "pos_frac": 0.78125, "sample": [1.0730476379394531, -0.23109054565429688, 0.9762363433837891, 0.7499618530273438, 1.1120758056640625, 0.9971542358398438, 1.2802276611328125, -0.6147079467773438, -0.1521453857421875, 2.891326904296875, 2.5435638427734375, 0.13928604125976562, -0.5312728881835938, 0.3210487365722656, -0.019990921020507812, 1.6189956665039062, -0.2588958740234375, 1.6222763061523438, 1.398681640625, -0.0196075439453125, -0.1043853759765625, -0.6592273712158203, 0.8359451293945312, 0.2876110076904297, 3.2343902587890625, 3.816783905029297, 1.9355888366699219, -0.47322845458984375, 3.1829986572265625, 0.56201171875, 0.9173583984375, 1.6933441162109375, -0.15293502807617188, 3.7094039916992188, 0.198883056640625, -0.08962059020996094, 3.42987060546875, 0.577545166015625, 2.5438919067382812, 0.2851409912109375, 2.4285430908203125, 1.0582351684570312, 2.2385330200195312, 0.8137550354003906, 2.610382080078125, 1.063873291015625, -0.09324264526367188, 4.21484375, 0.9526443481445312, -2.1429367065429688, 2.6054115295410156, 0.67266845703125, 1.9035682678222656, 0.9493026733398438, 2.4198074340820312, 1.3696403503417969, 0.187286376953125, 0.8343658447265625, 2.21551513671875, 0.5905094146728516, 1.1184158325195312, 3.502044677734375, 1.0214309692382812, 1.0010566711425781], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000036.npy"} +{"epoch": 0.05286343612334802, "step": 37, "batch_size": 64, "mean": 1.056151032447815, "std": 1.418853759765625, "min": -2.406890869140625, "p10": -0.11274795532226561, "median": 0.6869678497314453, "p90": 2.787006378173829, "max": 6.7253570556640625, "pos_frac": 0.796875, "sample": [2.1200408935546875, -0.010578155517578125, 0.9430007934570312, 0.66986083984375, 0.2832794189453125, -0.09761810302734375, -0.230712890625, 0.482666015625, 2.13800048828125, -0.09376716613769531, 0.29559326171875, 3.43634033203125, 0.6522140502929688, -0.831634521484375, -0.0518798828125, 0.29244232177734375, 0.7857208251953125, 0.08198356628417969, 1.9052543640136719, 6.7253570556640625, 0.8033065795898438, 4.31207275390625, 0.9062232971191406, 0.7696647644042969, 0.8811492919921875, 1.4188804626464844, 0.6477432250976562, 0.4470481872558594, 1.1083145141601562, 1.7057418823242188, 0.6751213073730469, 0.5223846435546875, 0.3129005432128906, -0.7981414794921875, 0.5329208374023438, 0.12774658203125, 0.17719650268554688, 2.1676025390625, 2.369110107421875, 3.62542724609375, -0.34993934631347656, 2.5657730102539062, 0.6988143920898438, 1.371185302734375, -0.25722503662109375, 0.7027473449707031, 0.06680107116699219, -0.119232177734375, 1.4378280639648438, 1.6978950500488281, 2.442291259765625, -0.0942535400390625, 2.8818206787109375, 2.32061767578125, 0.5690937042236328, 3.5769424438476562, 1.7626419067382812, -2.406890869140625, 0.12119674682617188, -0.0275726318359375, 2.3885650634765625, 3.2993011474609375, 0.01993560791015625, 0.7173538208007812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000037.npy"} +{"epoch": 0.05433186490455213, "step": 38, "batch_size": 64, "mean": 1.427796721458435, "std": 1.7502321004867554, "min": -0.89593505859375, "p10": -0.1701406478881835, "median": 0.925628662109375, "p90": 3.3566390991210944, "max": 8.0670166015625, "pos_frac": 0.859375, "sample": [0.478179931640625, 8.0670166015625, 3.940460205078125, 1.82391357421875, 0.5245018005371094, 0.014308929443359375, 1.0908050537109375, -0.6056442260742188, 0.8060760498046875, 2.93646240234375, 2.0436630249023438, 1.2148208618164062, 0.466278076171875, 2.69873046875, 1.504638671875, -0.21280479431152344, 0.4198150634765625, 0.15095138549804688, -0.07059097290039062, 2.24609375, 0.8190116882324219, 1.5447845458984375, 3.2130050659179688, 0.2632179260253906, 3.421661376953125, 0.027750015258789062, 2.315277099609375, 2.9469833374023438, 1.4818687438964844, 0.14388656616210938, 0.7375984191894531, 3.2430343627929688, -0.8013763427734375, 0.8099517822265625, 0.9697971343994141, 0.09696578979492188, 0.5967864990234375, 6.017791748046875, 0.6528472900390625, 0.41131591796875, 3.4053268432617188, 6.328216552734375, 1.9586868286132812, 0.9239349365234375, 0.9273223876953125, -0.89593505859375, 0.047992706298828125, 0.4417915344238281, 0.4553070068359375, 1.4201736450195312, -0.66253662109375, -0.033660888671875, -0.22275924682617188, 1.4590339660644531, 1.5000419616699219, 2.158233642578125, 1.7656097412109375, 5.943206787109375, 1.0260391235351562, 0.6548843383789062, 1.963958740234375, -0.45525169372558594, 0.7977752685546875, 2.0217628479003906], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000038.npy"} +{"epoch": 0.055800293685756244, "step": 39, "batch_size": 64, "mean": 1.6471561193466187, "std": 1.4889471530914307, "min": -3.1726837158203125, "p10": 0.11662006378173839, "median": 1.4190359115600586, "p90": 3.9510879516601567, "max": 5.295196533203125, "pos_frac": 0.90625, "sample": [2.1599044799804688, 1.9598236083984375, 4.6347503662109375, 2.1966171264648438, 1.5152587890625, -0.6734848022460938, 3.176910400390625, 2.84100341796875, 1.3797054290771484, 1.9786968231201172, 0.6707077026367188, 3.9887008666992188, 1.343963623046875, 1.0114994049072266, 3.8908462524414062, 1.0625343322753906, -0.022518157958984375, 0.5466842651367188, -0.177459716796875, 1.4691085815429688, 0.32293701171875, 1.8361663818359375, 2.4561233520507812, 0.7205314636230469, 0.9311943054199219, 2.1279678344726562, -0.097320556640625, 1.5358963012695312, 4.897064208984375, 0.774810791015625, 0.6200027465820312, 0.06645774841308594, 3.0504226684570312, 2.2541732788085938, -3.1726837158203125, 1.9895515441894531, 0.4601573944091797, 2.0932788848876953, 4.068044662475586, 5.295196533203125, 1.0741920471191406, 1.8125762939453125, 4.2624664306640625, 0.8619232177734375, 1.1670417785644531, 0.5323028564453125, 0.8248291015625, 0.8832511901855469, 1.7965774536132812, 1.1245479583740234, 0.9259185791015625, 3.5836143493652344, 2.40533447265625, 3.062042236328125, 0.6685123443603516, 3.9769058227539062, 3.1651458740234375, 0.7749137878417969, 1.4583663940429688, 0.23366546630859375, 0.7287483215332031, -0.09630775451660156, 0.5705146789550781, 2.4376869201660156], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000039.npy"} +{"epoch": 0.05726872246696035, "step": 40, "batch_size": 64, "mean": 1.6004118919372559, "std": 1.7148033380508423, "min": -0.8911819458007812, "p10": -0.23456707000732419, "median": 1.3475418090820312, "p90": 3.7123779296875, "max": 7.477264404296875, "pos_frac": 0.828125, "sample": [-0.20017242431640625, 0.1740875244140625, 1.1544723510742188, 1.2819843292236328, 1.5887184143066406, 0.13636016845703125, 0.821441650390625, 0.24790191650390625, -0.04022789001464844, 2.3791027069091797, 6.8939666748046875, 3.7242431640625, 1.1697444915771484, 3.1844100952148438, 1.7036285400390625, -0.8911819458007812, 2.4655895233154297, 1.842041015625, 0.2922401428222656, -0.3199462890625, 2.8681259155273438, -0.24930763244628906, 1.596221923828125, -0.0743408203125, 2.5037002563476562, 2.1160888671875, 0.13872909545898438, 0.9812374114990234, 0.489776611328125, -0.53741455078125, 4.525825500488281, 0.0005283355712890625, 3.7668838500976562, -0.3030242919921875, 0.4611091613769531, 4.337806701660156, 0.12000656127929688, 3.6846923828125, 0.55938720703125, 1.3068313598632812, 4.6953125, 1.0296268463134766, 1.3882522583007812, 7.477264404296875, -0.49658966064453125, 1.9349594116210938, -0.376708984375, 2.309284210205078, 2.6001739501953125, 2.113727569580078, 2.147186279296875, 2.7616233825683594, 3.5539169311523438, 1.4235210418701172, 1.1302337646484375, 1.8028755187988281, 3.5781707763671875, -0.09264373779296875, 2.10992431640625, 0.18257904052734375, 0.11670684814453125, 1.98291015625, 2.5998001098632812, 0.5529823303222656], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000040.npy"} +{"epoch": 0.05873715124816446, "step": 41, "batch_size": 64, "mean": 1.8981152772903442, "std": 2.0470831394195557, "min": -2.8131866455078125, "p10": -0.010087966918945222, "median": 1.3352813720703125, "p90": 4.331360626220704, "max": 8.972244262695312, "pos_frac": 0.890625, "sample": [0.9963035583496094, 1.0763816833496094, 3.524505615234375, 1.3338470458984375, 0.36614990234375, 1.5833892822265625, 3.3264007568359375, 2.337177276611328, -0.6380767822265625, 3.2923355102539062, 5.2939453125, 0.8154678344726562, 2.3428955078125, 8.972244262695312, 0.2475128173828125, 0.46435546875, 0.5523681640625, 1.2992630004882812, 0.07646942138671875, -1.177703857421875, 1.4933929443359375, 3.5752601623535156, 4.402565002441406, 1.1847820281982422, 7.489166259765625, 1.9374427795410156, 4.125885009765625, 4.198150634765625, 1.0036869049072266, 0.16113662719726562, 2.424510955810547, 0.6383056640625, 7.48175048828125, 0.46323394775390625, 1.0758285522460938, 3.5998382568359375, 2.0543136596679688, 2.9980926513671875, -0.0659942626953125, 2.2855606079101562, 4.388450622558594, 1.9408035278320312, -2.8131866455078125, 3.2575225830078125, 0.6634674072265625, 0.6317138671875, 2.0398178100585938, 1.91204833984375, 2.209360122680664, 0.6079025268554688, 1.3367156982421875, 0.87261962890625, 0.5778045654296875, 0.7656784057617188, 0.8032512664794922, 1.4814071655273438, -0.204681396484375, 3.4702415466308594, 1.1709518432617188, 0.4750022888183594, -0.1463642120361328, -0.047183990478515625, 2.338001251220703, 5.135894775390625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000041.npy"} +{"epoch": 0.06020558002936858, "step": 42, "batch_size": 64, "mean": 2.5605297088623047, "std": 2.5344948768615723, "min": -1.6491012573242188, "p10": 0.1341655731201172, "median": 2.012348175048828, "p90": 5.9212570190429705, "max": 12.229705810546875, "pos_frac": 0.90625, "sample": [1.1319427490234375, 0.281341552734375, 2.4615650177001953, 9.85516357421875, 12.229705810546875, 2.55755615234375, -0.24715423583984375, 6.2303466796875, 1.9098052978515625, 3.346435546875, 2.2311477661132812, 6.795982360839844, 2.9099159240722656, 8.0567626953125, 0.7855567932128906, 1.6838607788085938, 2.6220626831054688, 0.12397003173828125, 3.107818603515625, 2.6121826171875, 1.4549751281738281, 1.571249008178711, 3.2002487182617188, 0.4659271240234375, 1.1869544982910156, 2.7231483459472656, -0.3237152099609375, 2.3444061279296875, 1.8283882141113281, -0.6138839721679688, 2.234294891357422, 6.0891876220703125, 0.4692363739013672, 5.5294189453125, 2.1148910522460938, 1.4958419799804688, 0.9595222473144531, 1.5854320526123047, 2.3759765625, -1.076904296875, 4.23406982421875, -0.04296875, 1.7852401733398438, 6.1732025146484375, 1.2626380920410156, 3.9023284912109375, 1.025674819946289, 0.15795516967773438, 1.2382011413574219, 1.1490707397460938, 0.769134521484375, 5.499298095703125, -1.6491012573242188, 0.7676601409912109, 2.4562530517578125, 4.158393859863281, 0.18602371215820312, 2.1579132080078125, 5.142852783203125, 1.7999725341796875, 4.76885986328125, 5.0537261962890625, 1.011861801147461, 4.5650634765625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000042.npy"} +{"epoch": 0.06167400881057269, "step": 43, "batch_size": 64, "mean": 2.4469194412231445, "std": 2.06974720954895, "min": -1.3701400756835938, "p10": 0.1600811004638672, "median": 2.315117835998535, "p90": 5.114734649658204, "max": 10.623153686523438, "pos_frac": 0.9375, "sample": [-0.2529335021972656, 0.15781402587890625, 0.0062808990478515625, 3.3516464233398438, 4.58599853515625, 0.7471160888671875, 4.040809631347656, 3.246307373046875, 2.542999267578125, 6.073966979980469, 0.700042724609375, 1.9620952606201172, 0.7172431945800781, 2.3089981079101562, 3.1865501403808594, 1.3043441772460938, 3.427215576171875, 5.5067138671875, 1.2544174194335938, 0.30915069580078125, 6.65032958984375, 3.027729034423828, 5.317604064941406, 0.8745307922363281, 3.0599403381347656, 2.4149093627929688, 0.16537094116210938, 2.321237564086914, -0.3336982727050781, 4.89569091796875, 1.3632583618164062, 2.9081573486328125, 4.151161193847656, 3.543487548828125, 2.4247894287109375, 3.297901153564453, 0.6636581420898438, 2.1571121215820312, 1.6875629425048828, 0.23958587646484375, 1.310150146484375, 1.4999198913574219, 6.8206787109375, 3.228240966796875, 0.7496185302734375, -1.3701400756835938, 1.1738643646240234, 2.579803466796875, 2.1942138671875, 3.53753662109375, 2.5710983276367188, 3.959716796875, 1.880859375, 0.5386333465576172, 2.186767578125, 10.623153686523438, 3.745849609375, 1.6049842834472656, -0.149749755859375, 0.3452892303466797, 2.6488571166992188, 5.208610534667969, 3.6184921264648438, 0.08931159973144531], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000043.npy"} +{"epoch": 0.0631424375917768, "step": 44, "batch_size": 64, "mean": 2.763123035430908, "std": 2.2371573448181152, "min": -0.96392822265625, "p10": 0.47458915710449223, "median": 2.2803306579589844, "p90": 5.646308898925781, "max": 9.45220947265625, "pos_frac": 0.953125, "sample": [3.66455078125, 2.2901229858398438, 1.8496437072753906, 0.5626506805419922, 7.121318817138672, 1.9715805053710938, 3.0626487731933594, 1.3195571899414062, 2.1796340942382812, 5.536827087402344, 6.414375305175781, 0.4207649230957031, 4.364006042480469, 4.975006103515625, 4.285614013671875, 3.8930130004882812, 0.6687698364257812, 1.44317626953125, 2.3161239624023438, 5.368194580078125, 0.4583396911621094, 2.4612274169921875, 2.522571563720703, 7.9404754638671875, 3.562103271484375, 2.5270004272460938, 0.5575790405273438, 1.2895660400390625, 5.693229675292969, 1.4231414794921875, 2.746715545654297, -0.033084869384765625, 1.8833999633789062, 1.1351165771484375, 1.2896919250488281, 3.4080429077148438, 1.1880722045898438, 1.8528594970703125, 4.962799072265625, 2.1512451171875, 1.0608673095703125, 2.8077392578125, 4.3144683837890625, 3.8663787841796875, 1.2384529113769531, 3.655303955078125, 2.86309814453125, 2.270538330078125, 0.5259838104248047, 0.4503898620605469, 7.0361328125, 3.8182754516601562, 1.0564727783203125, 4.1956329345703125, 0.970916748046875, 9.45220947265625, 0.5125045776367188, -0.96392822265625, 3.5774078369140625, 2.0720481872558594, -0.9606132507324219, 0.24054527282714844, 1.4936370849609375, 8.557746887207031], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000044.npy"} +{"epoch": 0.06461086637298091, "step": 45, "batch_size": 64, "mean": 2.311833381652832, "std": 2.6114349365234375, "min": -1.5460968017578125, "p10": 0.026502990722656293, "median": 1.6491336822509766, "p90": 6.039569473266601, "max": 12.33929443359375, "pos_frac": 0.90625, "sample": [3.790069580078125, 1.2447052001953125, 0.6747817993164062, 2.960735321044922, 0.12833786010742188, 1.2823410034179688, 3.5754318237304688, 7.1083984375, 5.690582275390625, 3.3235397338867188, 0.6265106201171875, 8.045028686523438, 3.6223316192626953, 0.17029190063476562, 0.24135971069335938, 1.8168220520019531, 0.06781768798828125, 1.8745040893554688, 0.12137413024902344, 6.043926239013672, 0.2239227294921875, 0.18771743774414062, 2.629302978515625, 0.7079505920410156, 0.6486072540283203, 4.675020217895508, 6.596199035644531, 4.216819763183594, 0.30200767517089844, 0.36520957946777344, 5.1138916015625, 2.7173385620117188, 5.85589599609375, 0.13782882690429688, -0.0030670166015625, 1.7941055297851562, 6.3264617919921875, 3.6157913208007812, 0.00879669189453125, 1.6910400390625, 1.2239665985107422, 0.2869720458984375, -1.5460968017578125, -1.2475814819335938, 1.37066650390625, 12.33929443359375, 1.2112579345703125, 7.271339416503906, 3.1016387939453125, -1.0208930969238281, 2.405975341796875, 1.6303024291992188, 1.6679649353027344, 0.4401416778564453, 1.2663497924804688, 6.0294036865234375, 1.1850128173828125, 2.123626708984375, 2.370746612548828, 2.2275314331054688, -0.25478172302246094, 3.6517066955566406, 0.7173995971679688, -0.7143325805664062], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000045.npy"} +{"epoch": 0.06607929515418502, "step": 46, "batch_size": 64, "mean": 2.9520645141601562, "std": 3.2337348461151123, "min": -3.058441162109375, "p10": 0.024174118041992213, "median": 2.241929054260254, "p90": 6.717332839965821, "max": 14.9425048828125, "pos_frac": 0.90625, "sample": [0.05277824401855469, 0.507110595703125, 5.44927978515625, 7.572540283203125, 4.321281433105469, -2.3551483154296875, 1.5207672119140625, 1.5692062377929688, 6.5448455810546875, 3.3555068969726562, 6.791255950927734, 1.4419422149658203, 0.44574737548828125, 2.4295902252197266, 4.711517333984375, 0.049106597900390625, 0.0966033935546875, 0.9877090454101562, 4.770149230957031, 12.63885498046875, -0.4865550994873047, 0.01348876953125, 0.11182022094726562, 3.9119720458984375, 0.4163703918457031, 4.5396728515625, 4.1094970703125, 1.1997966766357422, 3.8977813720703125, 2.508514404296875, 4.427375793457031, 0.5441341400146484, -3.058441162109375, 1.01007080078125, 3.3309783935546875, 6.124931335449219, 5.799339294433594, -0.02553558349609375, 8.167312622070312, 0.15177154541015625, 0.27509307861328125, 3.6448516845703125, 4.9827880859375, -0.3752899169921875, 14.9425048828125, 0.3522186279296875, 6.2105560302734375, 6.310295104980469, 0.669525146484375, 2.854949951171875, 2.0542678833007812, 8.346267700195312, 2.8882503509521484, 0.2869415283203125, 1.9715919494628906, 7.395408630371094, 1.4385147094726562, 2.648059844970703, 3.1512527465820312, 5.042243957519531, 1.9547500610351562, 0.9671459197998047, 1.5436592102050781, -0.21866607666015625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000046.npy"} +{"epoch": 0.06754772393538913, "step": 47, "batch_size": 64, "mean": 2.978566884994507, "std": 2.8161396980285645, "min": -2.537933349609375, "p10": -0.024801635742187425, "median": 2.358759880065918, "p90": 6.988553619384767, "max": 10.81494140625, "pos_frac": 0.890625, "sample": [2.2332305908203125, 3.0919036865234375, 0.818115234375, 1.3022994995117188, 5.204715728759766, 1.891885757446289, 0.8523788452148438, 2.1287841796875, 7.1858978271484375, 2.871776580810547, 0.36932373046875, 6.6516571044921875, 6.6899566650390625, 1.1314010620117188, 3.3834228515625, 2.630889892578125, 0.30290985107421875, 1.4103546142578125, 2.38482666015625, 5.211769104003906, 7.116523742675781, 1.4363479614257812, 0.5534229278564453, 3.6938724517822266, 3.4267196655273438, 1.257345199584961, -1.3680648803710938, 6.5376129150390625, 7.751495361328125, 2.878925323486328, -0.31107139587402344, 3.2661876678466797, 6.536212921142578, 4.0899200439453125, -2.537933349609375, -0.08019638061523438, 0.05132293701171875, 1.32183837890625, 1.131591796875, 10.81494140625, 7.25048828125, 2.332693099975586, 1.083200454711914, 2.94464111328125, 9.07916259765625, 6.682464599609375, -0.05742645263671875, 3.2574844360351562, 10.1988525390625, -1.8091506958007812, 1.06219482421875, 1.9987335205078125, -0.1060638427734375, 1.3598785400390625, 2.5389785766601562, 1.1842803955078125, 4.9412384033203125, 1.2805938720703125, 3.6232147216796875, 2.5683116912841797, 5.56268310546875, 3.8506011962890625, 2.2039642333984375, 2.2827491760253906], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000047.npy"} +{"epoch": 0.06901615271659324, "step": 48, "batch_size": 64, "mean": 3.124472141265869, "std": 3.1642394065856934, "min": -7.7147064208984375, "p10": 0.24423561096191412, "median": 2.519071578979492, "p90": 8.008029174804689, "max": 11.776229858398438, "pos_frac": 0.90625, "sample": [3.70245361328125, 0.504852294921875, 8.141448974609375, 9.193496704101562, 7.3412322998046875, 2.8506546020507812, -0.736541748046875, 2.7550201416015625, 5.564765930175781, 4.620384216308594, 1.526885986328125, 6.671775817871094, 2.08544921875, 0.6380157470703125, 0.657318115234375, 1.2102775573730469, 7.69671630859375, -0.12103843688964844, 1.3249053955078125, -0.3470268249511719, 0.8069267272949219, 9.038261413574219, -0.3246269226074219, 2.4878692626953125, 5.862579345703125, 1.9868240356445312, 3.7966842651367188, 3.9187450408935547, 0.21718978881835938, 3.661724090576172, 2.164661407470703, 2.249225616455078, 3.144256591796875, 2.390453338623047, 8.412689208984375, 4.344142913818359, 4.9032745361328125, 9.585212707519531, 3.006866455078125, 1.0286445617675781, 0.7298469543457031, 1.8357868194580078, 1.141815185546875, 3.1540489196777344, 0.8123779296875, 4.232015609741211, 3.1310081481933594, 2.1488189697265625, -1.8886547088623047, 8.366771697998047, 3.0056533813476562, 3.158346176147461, 2.18499755859375, 11.776229858398438, 4.96551513671875, -7.7147064208984375, 5.837596893310547, 1.674713134765625, 0.8060836791992188, 0.307342529296875, 5.5874481201171875, 2.021404266357422, 2.550273895263672, 2.1788253784179688], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000048.npy"} +{"epoch": 0.07048458149779736, "step": 49, "batch_size": 64, "mean": 3.746046543121338, "std": 3.9682528972625732, "min": -3.1914596557617188, "p10": -0.014555168151855348, "median": 3.230091094970703, "p90": 8.591242980957032, "max": 16.029754638671875, "pos_frac": 0.890625, "sample": [3.4810409545898438, 0.6170578002929688, 0.14373779296875, 3.175811767578125, 2.5507354736328125, 6.068897247314453, 0.6811752319335938, 2.3927841186523438, -2.2673187255859375, 0.5876617431640625, 0.2416362762451172, 1.9700164794921875, 2.9412002563476562, 8.783958435058594, 8.069015502929688, 2.316375732421875, 4.365489959716797, 2.1505699157714844, 3.6798248291015625, 1.5158843994140625, 1.9437026977539062, -3.1914596557617188, 3.82208251953125, -0.08896827697753906, 15.164520263671875, 10.465347290039062, 0.729522705078125, 5.377838134765625, 4.0894927978515625, 0.6634674072265625, 8.343376159667969, 16.029754638671875, 5.128837585449219, 0.929656982421875, -1.39654541015625, -3.0952224731445312, 4.010734558105469, 3.6083736419677734, 3.9159622192382812, 4.964653015136719, 3.0979061126708984, 0.10529136657714844, -0.06591796875, 4.735443115234375, 5.4504852294921875, 14.768264770507812, 2.3092575073242188, 2.5628204345703125, 8.697471618652344, 6.749847412109375, 7.1940765380859375, 3.3428611755371094, 5.2300872802734375, 5.167121887207031, 12.52239990234375, 0.6049404144287109, -0.6474151611328125, 3.2843704223632812, 2.1023597717285156, 4.290473937988281, 5.54962158203125, 1.9120674133300781, 4.753101348876953, 1.14935302734375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000049.npy"} +{"epoch": 0.07195301027900147, "step": 50, "batch_size": 64, "mean": 4.05955696105957, "std": 3.979112386703491, "min": -3.064352035522461, "p10": -0.3599624633789059, "median": 3.149643898010254, "p90": 9.502223205566407, "max": 15.4556884765625, "pos_frac": 0.875, "sample": [6.944854736328125, -0.0076446533203125, -2.5531005859375, 10.023025512695312, 1.0272331237792969, 2.6744766235351562, 7.8139801025390625, 1.8851356506347656, 3.950397491455078, 1.7714414596557617, -1.0883636474609375, 7.035408020019531, 3.6255645751953125, 15.4556884765625, 0.5800399780273438, 3.00189208984375, 7.39697265625, 2.7331809997558594, 9.643218994140625, 4.8212432861328125, 0.7900962829589844, 5.3536376953125, 10.193473815917969, 2.7758026123046875, 2.127239227294922, 4.377309799194336, 0.01760101318359375, -0.510955810546875, 3.143087387084961, 9.173233032226562, 6.507820129394531, 5.68292236328125, 9.747623443603516, 14.238327026367188, -3.064352035522461, 0.28105926513671875, 1.0077972412109375, 2.4828643798828125, 0.11114120483398438, 4.175102233886719, 1.8932037353515625, 1.6992778778076172, 6.581947326660156, 4.076530456542969, 7.791088104248047, 7.50225830078125, -1.0800323486328125, 1.8687858581542969, 5.878501892089844, 0.0036468505859375, 6.175506591796875, 3.156200408935547, 4.890445709228516, 6.339256286621094, 2.8562469482421875, 6.583976745605469, 7.9885101318359375, 1.894622802734375, 12.405380249023438, -0.72100830078125, 6.906764984130859, 2.0449905395507812, -2.3276748657226562, 0.05773353576660156], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000050.npy"} +{"epoch": 0.07342143906020558, "step": 51, "batch_size": 64, "mean": 3.9906833171844482, "std": 5.138714790344238, "min": -6.1121978759765625, "p10": -0.839967918395996, "median": 2.5547027587890625, "p90": 10.247237396240235, "max": 20.077362060546875, "pos_frac": 0.828125, "sample": [10.361007690429688, 4.223838806152344, 0.9162979125976562, -1.1332931518554688, 0.856231689453125, -0.281158447265625, 8.898704528808594, -0.7626190185546875, 2.0711288452148438, -0.081329345703125, 11.8922119140625, 7.567195892333984, 1.9219684600830078, 4.790069580078125, 2.328868865966797, 5.406982421875, 4.012996673583984, -0.8731174468994141, 9.851959228515625, 6.798423767089844, 4.300594329833984, -1.17510986328125, 0.20631027221679688, 3.8508071899414062, 1.7488937377929688, 8.02914810180664, 0.6522407531738281, 5.8334503173828125, 0.9030380249023438, 5.058555603027344, 20.077362060546875, 3.033367156982422, 0.36989593505859375, 5.71136474609375, 3.3101577758789062, 15.067947387695312, 0.957977294921875, 16.203414916992188, 5.084220886230469, 2.350811004638672, 2.568267822265625, 0.7808151245117188, 3.8847808837890625, 9.981773376464844, 1.1217079162597656, 0.283111572265625, 5.1366729736328125, -3.7221450805664062, -0.4839954376220703, 18.38861083984375, 14.746871948242188, 7.234382629394531, 7.618480682373047, -0.9925575256347656, 1.3869285583496094, 6.57513427734375, 0.12921905517578125, 0.343841552734375, 2.5394744873046875, -6.1121978759765625, 2.6317367553710938, -1.634490966796875, 2.5411376953125, 0.1153564453125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000051.npy"} +{"epoch": 0.07488986784140969, "step": 52, "batch_size": 64, "mean": 6.104278564453125, "std": 5.0724334716796875, "min": -2.5556182861328125, "p10": 0.9509189605712891, "median": 4.9931182861328125, "p90": 12.863924407958987, "max": 21.927154541015625, "pos_frac": 0.90625, "sample": [1.9188423156738281, 12.543067932128906, 4.406944274902344, 12.412940979003906, 7.679862976074219, 10.457321166992188, 3.7232513427734375, 7.665702819824219, 8.32562255859375, 14.525222778320312, -0.5724563598632812, 13.106590270996094, 5.326240539550781, 5.588508605957031, 9.436262130737305, 8.262140274047852, 2.5784988403320312, 5.160501480102539, 18.448394775390625, 9.230262756347656, 3.525646209716797, 1.1657028198242188, 12.288177490234375, 2.9440441131591797, -0.014080047607421875, 7.895687103271484, 3.1459426879882812, 2.074373245239258, 8.767232894897461, 21.927154541015625, 9.82159423828125, 1.8724822998046875, 10.255455017089844, 4.982051849365234, 4.474952697753906, 13.97918701171875, 4.259372711181641, -1.3806304931640625, 4.8493194580078125, -2.5556182861328125, -2.3176498413085938, 2.4966354370117188, 6.358064651489258, 6.225547790527344, 1.3481597900390625, 10.074638366699219, 4.403217315673828, 13.001434326171875, 6.923946380615234, 2.222949981689453, 0.9098701477050781, 3.64154052734375, 3.3878936767578125, 5.004184722900391, 3.9346389770507812, 9.98870849609375, 6.276237487792969, 1.0466995239257812, 17.583404541015625, -1.3561477661132812, 6.676605224609375, 4.299587249755859, 2.608539581298828, 1.4333457946777344], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000052.npy"} +{"epoch": 0.0763582966226138, "step": 53, "batch_size": 64, "mean": 6.129757881164551, "std": 6.641849994659424, "min": -3.3241424560546875, "p10": 0.2226364135742192, "median": 4.2077178955078125, "p90": 15.421270751953127, "max": 29.880218505859375, "pos_frac": 0.90625, "sample": [-1.5995311737060547, 3.5353164672851562, 3.8451766967773438, 3.12310791015625, 0.810272216796875, 4.104711532592773, 13.711540222167969, -0.2339324951171875, 7.6504669189453125, -1.7758026123046875, 1.5332183837890625, 27.5810546875, -2.7600631713867188, 0.9270401000976562, 6.5805511474609375, 7.926586151123047, 1.4229583740234375, 11.372188568115234, 2.5865097045898438, 12.764816284179688, 4.532173156738281, 6.108650207519531, 13.024085998535156, 7.421394348144531, 15.810344696044922, 1.5555343627929688, 5.076145172119141, 0.6626358032226562, 6.0860595703125, 1.7167434692382812, 18.323089599609375, 9.577461242675781, 7.562671661376953, 8.431961059570312, 3.80328369140625, 0.03406524658203125, 0.8003463745117188, 4.565135955810547, 1.203390121459961, 4.309991836547852, 3.9082565307617188, 8.131711959838867, 1.4983177185058594, 15.067710876464844, -3.3241424560546875, 17.792236328125, 18.596282958984375, 3.1798439025878906, 5.5034637451171875, 3.123941421508789, 2.0317745208740234, 4.289939880371094, 2.3709716796875, 2.4965362548828125, 4.125495910644531, 4.390003204345703, 3.074981689453125, 29.880218505859375, 2.8069686889648438, 15.571746826171875, 4.772357940673828, 4.68853759765625, 15.070159912109375, -0.4541473388671875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000053.npy"} +{"epoch": 0.07782672540381791, "step": 54, "batch_size": 64, "mean": 5.31174373626709, "std": 4.997278690338135, "min": -1.0948562622070312, "p10": 0.27296009063720705, "median": 4.084638595581055, "p90": 12.65085144042969, "max": 21.477981567382812, "pos_frac": 0.96875, "sample": [8.049644470214844, 5.4061126708984375, 8.17563247680664, 1.0960254669189453, 0.26869773864746094, 3.2255477905273438, 0.2667388916015625, 2.863077163696289, 5.621437072753906, 8.35696792602539, 0.8673954010009766, 5.498382568359375, 12.93487548828125, 2.001279830932617, 4.0380401611328125, 12.111656188964844, 0.09605216979980469, 4.709415435791016, 5.269536972045898, 5.705513000488281, 0.12250328063964844, 1.561126708984375, 17.19513702392578, 0.328857421875, 7.6915740966796875, 0.5852737426757812, 5.327472686767578, 5.21044921875, 5.9564208984375, 6.147369384765625, 1.5527114868164062, -0.15037155151367188, 8.402137756347656, 0.2672157287597656, 0.8798007965087891, 3.493885040283203, 12.881935119628906, 2.679393768310547, 6.2371368408203125, 13.503814697265625, 1.9333076477050781, 2.6500473022460938, 21.477981567382812, 11.066413879394531, -1.0948562622070312, 2.927915573120117, 0.28290557861328125, 13.13287353515625, 0.8294086456298828, 4.4554290771484375, 1.5142059326171875, 4.131237030029297, 11.184959411621094, 2.1389999389648438, 2.681255340576172, 18.887481689453125, 1.1826133728027344, 5.333581924438477, 2.418701171875, 10.614151000976562, 2.357870101928711, 11.093109130859375, 1.4799766540527344, 10.836135864257812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000054.npy"} +{"epoch": 0.07929515418502203, "step": 55, "batch_size": 64, "mean": 6.309366226196289, "std": 7.277461051940918, "min": -11.411056518554688, "p10": -1.2211782455444335, "median": 5.667240142822266, "p90": 16.113076782226564, "max": 28.875564575195312, "pos_frac": 0.8125, "sample": [7.643150329589844, 9.386070251464844, 5.458106994628906, -4.130126953125, 24.506195068359375, -2.016387939453125, 7.220104217529297, 4.457775115966797, 16.69739532470703, -1.1750411987304688, 6.5214691162109375, 5.939460754394531, 2.4359970092773438, 11.18853759765625, 20.15704345703125, 10.259384155273438, 4.5404510498046875, 9.275390625, 28.875564575195312, 3.5669174194335938, 10.806915283203125, -1.7750244140625, 4.286933898925781, 8.359779357910156, 23.276336669921875, 0.5869197845458984, 16.3460693359375, 0.6544036865234375, 0.478118896484375, 8.164203643798828, 4.614585876464844, -1.9689788818359375, 5.998943328857422, 0.19164466857910156, 12.097976684570312, 4.983970642089844, -1.2391300201416016, -0.283172607421875, 0.3141002655029297, -4.89031982421875, 9.827926635742188, 3.4142627716064453, 2.302215576171875, 15.569427490234375, 7.0887298583984375, -1.179290771484375, 1.0883216857910156, 10.989730834960938, 5.52313232421875, -11.411056518554688, 2.897226333618164, 15.499908447265625, 6.744590759277344, 3.8246803283691406, 7.241619110107422, 5.811347961425781, -0.989715576171875, 8.064445495605469, 7.4840545654296875, 13.4656982421875, 5.966270446777344, -0.20434951782226562, 5.0596160888671875, 17.908905029296875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000055.npy"} +{"epoch": 0.08076358296622614, "step": 56, "batch_size": 64, "mean": 5.844947814941406, "std": 6.9918212890625, "min": -9.123844146728516, "p10": -1.7078151702880857, "median": 5.24588680267334, "p90": 14.809692382812507, "max": 26.302780151367188, "pos_frac": 0.796875, "sample": [-1.92620849609375, 7.115608215332031, -2.9527740478515625, 8.314144134521484, -0.5884475708007812, -9.123844146728516, 10.242500305175781, -1.4154586791992188, 3.5845565795898438, -4.281459808349609, 3.7652816772460938, 2.087657928466797, 5.433418273925781, 15.98577880859375, 2.4891414642333984, 0.0820465087890625, -0.29923439025878906, 11.972526550292969, 9.423038482666016, 7.571647644042969, -1.8331108093261719, 11.403839111328125, 3.3527984619140625, 21.1072998046875, 8.50433349609375, 15.528564453125, 0.9636440277099609, -0.200897216796875, 9.077169418334961, 2.475248336791992, 7.474639892578125, 8.05499267578125, 8.936729431152344, 1.6960811614990234, 10.455890655517578, 11.841327667236328, -0.055957794189453125, 19.831893920898438, 6.645896911621094, 6.549278259277344, 5.404382705688477, 4.9934234619140625, 2.974414825439453, 20.137451171875, 6.339988708496094, 13.089569091796875, 6.2952728271484375, 10.86785888671875, 0.501434326171875, 26.302780151367188, 1.244659423828125, 0.1311626434326172, 3.9364776611328125, 13.13232421875, -7.3892822265625, 3.0558910369873047, -0.269622802734375, 21.423797607421875, 5.087390899658203, -3.8167800903320312, 3.37811279296875, 1.7750701904296875, 6.236568450927734, 9.954742431640625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000056.npy"} +{"epoch": 0.08223201174743025, "step": 57, "batch_size": 64, "mean": 6.6771626472473145, "std": 6.340317249298096, "min": -9.918731689453125, "p10": 0.20937538146972665, "median": 6.738365173339844, "p90": 14.740353775024415, "max": 23.677352905273438, "pos_frac": 0.90625, "sample": [8.685245513916016, 3.4127864837646484, 0.307342529296875, -2.700653076171875, 10.218856811523438, -9.918731689453125, 11.733612060546875, 0.2832221984863281, 13.4722900390625, 6.804821014404297, 17.252700805664062, 3.640583038330078, 23.677352905273438, 3.37237548828125, 7.316276550292969, 6.7573699951171875, -1.4554882049560547, 3.325002670288086, 7.039131164550781, 20.603927612304688, 10.852607727050781, 16.796798706054688, 12.019683837890625, 4.2759552001953125, 7.135669708251953, 1.5008220672607422, 14.672183990478516, 5.453519821166992, 4.612161636352539, 2.5224075317382812, 0.9403533935546875, 3.8353214263916016, 3.926055908203125, 0.17772674560546875, 6.7193603515625, 7.483154296875, -0.17234039306640625, 1.4618949890136719, 2.6931381225585938, 8.881294250488281, 12.383880615234375, 2.073772430419922, 10.363304138183594, 5.0467681884765625, 9.413925170898438, -2.7040977478027344, 8.605178833007812, 6.981529235839844, 9.538665771484375, 3.489107131958008, 2.581981658935547, 17.880233764648438, 5.545188903808594, 2.9262771606445312, 8.421829223632812, 3.5916366577148438, 0.9369640350341797, 12.620193481445312, 12.388954162597656, -5.9500885009765625, 14.769569396972656, 20.598342895507812, 7.875555038452148, 8.343963623046875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000057.npy"} +{"epoch": 0.08370044052863436, "step": 58, "batch_size": 64, "mean": 7.373539924621582, "std": 7.8252387046813965, "min": -9.213172912597656, "p10": 0.016627120971679857, "median": 5.23388671875, "p90": 18.18810272216798, "max": 27.771026611328125, "pos_frac": 0.890625, "sample": [15.109878540039062, -2.2897777557373047, 0.5063819885253906, 4.946269989013672, 3.6486358642578125, 4.523891448974609, 20.87036895751953, 2.4817047119140625, 15.672187805175781, 2.8177947998046875, 5.521503448486328, 7.245513916015625, 7.9729766845703125, 3.9000282287597656, 2.5744094848632812, 12.415229797363281, 10.936721801757812, 9.295455932617188, 4.788234710693359, 3.3356151580810547, 8.710556030273438, 14.952522277832031, 14.288490295410156, -1.8879776000976562, 12.248336791992188, 2.1437149047851562, 9.789970397949219, 6.805475234985352, 23.100570678710938, 4.636985778808594, 27.771026611328125, 8.486408233642578, 6.496120452880859, 0.1833038330078125, 15.555435180664062, 0.9544200897216797, 14.055435180664062, 3.671478271484375, 2.496673583984375, -1.7128753662109375, 2.1130924224853516, 2.5339622497558594, 24.089126586914062, -2.4076309204101562, 10.06671142578125, -0.054805755615234375, 1.2765750885009766, 6.6685028076171875, -6.03253173828125, 25.13336181640625, 2.0805721282958984, 6.655124664306641, 1.0032234191894531, 15.274890899658203, 11.789634704589844, -9.213172912597656, 22.997451782226562, 0.931884765625, 2.7795944213867188, 7.693656921386719, 15.977745056152344, 1.7983589172363281, 0.5967636108398438, 19.135398864746094], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000058.npy"} +{"epoch": 0.08516886930983847, "step": 59, "batch_size": 64, "mean": 7.913702011108398, "std": 8.097885131835938, "min": -7.3852386474609375, "p10": -0.9979976654052733, "median": 6.664695739746094, "p90": 20.364333343505862, "max": 32.84602355957031, "pos_frac": 0.875, "sample": [8.684616088867188, 9.170143127441406, 16.01177978515625, 21.20416259765625, 16.283889770507812, 4.785552978515625, 32.84602355957031, 21.029708862304688, 7.57056999206543, 16.864654541015625, 0.6223678588867188, 7.62762451171875, -3.9962844848632812, 2.691568374633789, 22.99224853515625, -0.9065055847167969, 2.467597007751465, 13.980606079101562, 6.25653076171875, 20.691696166992188, 1.5936965942382812, 2.440399169921875, 1.1524581909179688, 1.7701835632324219, 4.32177734375, 26.27032470703125, 2.645242691040039, 1.9169845581054688, 8.572742462158203, -7.3852386474609375, 9.614601135253906, -1.6221427917480469, 8.720100402832031, 7.869968414306641, 10.012046813964844, 19.600486755371094, 3.5487899780273438, 12.287307739257812, 7.5623321533203125, 4.695320129394531, -5.661506652832031, -3.62091064453125, 5.8612823486328125, -1.0372085571289062, 4.70256233215332, 10.138420104980469, 7.379840850830078, 2.7569580078125, 7.7519378662109375, 2.0018138885498047, 2.018594741821289, 6.328559875488281, 3.2355213165283203, 10.236408233642578, 5.660829544067383, 15.6009521484375, 9.763969421386719, 4.261787414550781, 3.4655685424804688, -2.2630462646484375, 18.115032196044922, 19.4923095703125, 7.000831604003906, 20.82049560546875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000059.npy"} +{"epoch": 0.08663729809104258, "step": 60, "batch_size": 64, "mean": 6.335684776306152, "std": 7.137468338012695, "min": -15.9505615234375, "p10": -1.5225479125976562, "median": 6.631529808044434, "p90": 13.186691284179688, "max": 26.3946533203125, "pos_frac": 0.828125, "sample": [8.167675018310547, -4.7094573974609375, 1.9207782745361328, 12.434822082519531, 0.32213592529296875, 3.1896591186523438, 11.737251281738281, 8.615226745605469, 1.898712158203125, -15.9505615234375, 7.133884429931641, -1.4530029296875, 2.209928512573242, 3.8953380584716797, 11.343948364257812, 5.7664031982421875, 5.694549560546875, 11.009521484375, 4.317909240722656, 10.865013122558594, 13.81964111328125, 13.22637939453125, 19.3992919921875, 8.836883544921875, 9.201072692871094, 7.9520111083984375, -0.7952880859375, 6.468467712402344, 11.445816040039062, 0.2159137725830078, 9.403972625732422, 19.5821533203125, 3.914417266845703, 6.486968994140625, 7.963695526123047, -1.9938850402832031, 11.206527709960938, 9.450958251953125, 2.7403564453125, -5.043548583984375, 12.082805633544922, 3.388355255126953, 5.6794586181640625, 8.437309265136719, 4.627540588378906, -0.93243408203125, -10.389034271240234, 16.889297485351562, 20.571945190429688, 11.788314819335938, 13.094085693359375, 2.747222900390625, -2.0617904663085938, -1.4232254028320312, 11.274497985839844, 6.776090621948242, 26.3946533203125, 9.828094482421875, 2.346282958984375, 9.955179214477539, -1.5523529052734375, 8.427963256835938, 4.577123641967773, 1.064889907836914], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000060.npy"} +{"epoch": 0.0881057268722467, "step": 61, "batch_size": 64, "mean": 7.6252336502075195, "std": 10.420119285583496, "min": -10.752296447753906, "p10": -0.9404190063476563, "median": 5.569454193115234, "p90": 18.476559448242188, "max": 59.3463134765625, "pos_frac": 0.8125, "sample": [2.9283905029296875, 2.1788158416748047, -0.9452667236328125, 1.0420379638671875, 15.090225219726562, -0.1197509765625, 0.04311180114746094, 8.106552124023438, -0.653045654296875, 26.59600830078125, 21.61968231201172, 20.372817993164062, 27.82799530029297, 13.843070983886719, 8.095932006835938, 9.237586975097656, -2.277484893798828, 2.104085922241211, 12.158550262451172, 11.810691833496094, 3.110260009765625, 5.4883575439453125, 16.905319213867188, 8.34273910522461, 6.2987823486328125, 7.00184440612793, 5.69854736328125, 5.650550842285156, 9.243507385253906, 9.054283142089844, -0.1170654296875, 0.740447998046875, 12.567398071289062, 16.763702392578125, 15.908416748046875, 22.837493896484375, 18.338516235351562, 2.0007781982421875, -1.646942138671875, 13.369400024414062, 3.090606689453125, 0.38756752014160156, 12.994674682617188, 2.812549591064453, 1.1779346466064453, 18.535720825195312, 0.7722091674804688, 3.4376449584960938, 59.3463134765625, 13.230598449707031, 13.120506286621094, -0.7886886596679688, -10.752296447753906, 12.435813903808594, 2.290557861328125, -5.5872344970703125, 5.428558349609375, 3.40740966796875, -7.568939208984375, 2.155162811279297, 6.8053436279296875, -5.777259826660156, -0.929107666015625, 1.3729820251464844], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000061.npy"} +{"epoch": 0.08957415565345081, "step": 62, "batch_size": 64, "mean": 5.733864784240723, "std": 7.378479957580566, "min": -9.817169189453125, "p10": -2.9650098800659177, "median": 5.636591911315918, "p90": 14.959680175781251, "max": 30.464073181152344, "pos_frac": 0.84375, "sample": [8.596637725830078, 1.6554336547851562, 0.7916126251220703, 10.966796875, 5.868200302124023, 11.567344665527344, 7.279668807983398, 1.0330657958984375, 9.250579833984375, -5.593658447265625, 14.610748291015625, 6.19854736328125, 2.022695541381836, -0.12752342224121094, 9.632774353027344, 3.3666229248046875, 9.5931396484375, 5.81634521484375, 0.09864425659179688, 5.729217529296875, 7.379344940185547, 7.680873870849609, 2.368560791015625, 5.239795684814453, -4.120212554931641, 8.773818969726562, 4.1197967529296875, 8.23703384399414, -0.15338516235351562, 30.464073181152344, 17.584747314453125, 9.008598327636719, 6.502449035644531, 2.509063720703125, 1.06103515625, 5.739509582519531, 1.4861679077148438, 18.498687744140625, -9.817169189453125, -3.9676132202148438, -6.924407958984375, 1.1170330047607422, 2.027057647705078, 0.6493415832519531, 1.6058197021484375, 9.163171768188477, 17.025039672851562, 9.59658432006836, -3.058778762817383, 15.109222412109375, 8.705795288085938, 5.543966293334961, -6.783355712890625, 3.6213302612304688, 13.316303253173828, 3.9264755249023438, 3.2532310485839844, 3.8028106689453125, 27.07861328125, 19.358139038085938, 7.038209915161133, 1.2645339965820312, -2.7462158203125, 6.3253631591796875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000062.npy"} +{"epoch": 0.09104258443465492, "step": 63, "batch_size": 64, "mean": 7.697388648986816, "std": 8.136672973632812, "min": -13.64959716796875, "p10": -0.36238975524902284, "median": 6.920219421386719, "p90": 16.602493286132816, "max": 34.11505126953125, "pos_frac": 0.890625, "sample": [24.971725463867188, 7.523246765136719, 5.030977249145508, 9.5784912109375, 3.968109130859375, 0.8371124267578125, 4.8286285400390625, -6.7433929443359375, 6.703788757324219, 6.803977966308594, -13.64959716796875, 16.149078369140625, 8.9296875, 7.046287536621094, 2.919168472290039, -1.0167407989501953, 16.79681396484375, 4.410123825073242, 12.562515258789062, 21.630889892578125, 12.172134399414062, 14.388664245605469, 6.3239593505859375, 0.2871417999267578, -0.60516357421875, 11.613554000854492, -3.395477294921875, 0.20408248901367188, 5.427467346191406, 3.542470932006836, 12.165725708007812, 7.8935699462890625, 7.036460876464844, 9.596446990966797, 11.54949951171875, 11.73944091796875, 14.110450744628906, 13.358238220214844, 5.3864593505859375, 3.2205963134765625, 0.457061767578125, 13.956497192382812, 2.4812870025634766, 6.070888519287109, -10.752685546875, 13.734888076782227, 7.386859893798828, 1.4008026123046875, 2.612030029296875, 6.0240631103515625, 2.59686279296875, 14.509475708007812, 18.534408569335938, 8.393672943115234, 17.656455993652344, 0.6463241577148438, 9.266691207885742, 34.11505126953125, -3.8214645385742188, 14.976024627685547, 4.367485046386719, 25.747116088867188, 5.494598388671875, 11.481857299804688], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000063.npy"} +{"epoch": 0.09251101321585903, "step": 64, "batch_size": 64, "mean": 8.871689796447754, "std": 10.070149421691895, "min": -16.116500854492188, "p10": -0.880234146118164, "median": 7.301671981811523, "p90": 21.162796020507816, "max": 48.28704833984375, "pos_frac": 0.8125, "sample": [0.6451568603515625, 17.476524353027344, 27.297897338867188, 7.811435699462891, 5.165855407714844, 12.469245910644531, 3.067525863647461, 15.734874725341797, -0.9062786102294922, 6.824901580810547, 1.9159393310546875, 10.265899658203125, 0.9648818969726562, 5.091056823730469, 11.758163452148438, 29.7569580078125, 4.110618591308594, -1.791107177734375, 18.64739990234375, 7.487737655639648, 11.207000732421875, -2.328521728515625, 29.198623657226562, 7.497440338134766, -0.3842887878417969, 6.3560943603515625, -0.8194637298583984, 6.812261581420898, -16.116500854492188, -0.25095558166503906, 7.4733734130859375, -0.43517303466796875, 48.28704833984375, 5.381019592285156, 2.6618404388427734, 10.823780059814453, 16.746658325195312, 20.077056884765625, 11.04815673828125, 6.008655548095703, 5.919258117675781, 22.72247314453125, 11.746665954589844, 8.891471862792969, -1.02777099609375, -9.695938110351562, 23.60052490234375, 2.834136962890625, 21.62811279296875, 0.321746826171875, 15.929328918457031, 13.349472045898438, 12.547264099121094, -1.7155113220214844, 12.997413635253906, -0.34192848205566406, 4.733283996582031, 7.129970550537109, 13.3826904296875, 12.444808959960938, 2.305126190185547, 11.514144897460938, 18.683021545410156, 4.849601745605469], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000064.npy"} +{"epoch": 0.09397944199706314, "step": 65, "batch_size": 64, "mean": 8.778661727905273, "std": 9.676823616027832, "min": -21.94733428955078, "p10": -0.8601259231567377, "median": 7.490543365478516, "p90": 21.965017700195318, "max": 29.574386596679688, "pos_frac": 0.84375, "sample": [29.574386596679688, 24.450210571289062, 15.747909545898438, 11.879302978515625, 5.661016464233398, 8.974273681640625, 0.6630935668945312, 12.824295043945312, 3.5936241149902344, 20.63629913330078, 15.78472900390625, 1.5812320709228516, -0.1344585418701172, 22.396331787109375, 27.592971801757812, 0.4578666687011719, 10.143978118896484, 2.579263687133789, -4.761878967285156, 4.760490417480469, 1.2899971008300781, 18.197784423828125, 5.753143310546875, -21.94733428955078, 16.595413208007812, 8.28546142578125, -4.268463134765625, 6.939308166503906, 13.580429077148438, 5.6577606201171875, 5.7980804443359375, 5.679492950439453, -1.9341506958007812, 5.078742980957031, 28.29522705078125, 7.468818664550781, 20.9586181640625, 15.702327728271484, 25.260276794433594, 8.023731231689453, -0.325653076171875, -10.577705383300781, 8.84185791015625, 2.3283748626708984, 4.790252685546875, 18.076438903808594, 23.355934143066406, -1.2692413330078125, 5.565399169921875, 20.354156494140625, 11.734214782714844, 4.1002197265625, 19.998611450195312, -1.0891857147216797, 12.377273559570312, 7.51226806640625, 5.9833984375, 1.504913330078125, 16.50927734375, 10.633567810058594, -0.3026275634765625, 8.704166412353516, 0.4409809112548828, 7.767885208129883], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000065.npy"} +{"epoch": 0.09544787077826726, "step": 66, "batch_size": 64, "mean": 8.181053161621094, "std": 9.840169906616211, "min": -9.0028076171875, "p10": -0.542322731018066, "median": 6.773188591003418, "p90": 19.24979934692383, "max": 46.63514709472656, "pos_frac": 0.875, "sample": [7.8272552490234375, 46.63514709472656, 1.28021240234375, 14.557075500488281, 13.580474853515625, 11.987548828125, 2.3037586212158203, 0.0308380126953125, 10.739706039428711, 6.98626708984375, -9.0028076171875, 2.4901866912841797, 6.04481315612793, -0.6868324279785156, 12.318584442138672, 2.975444793701172, 4.297613143920898, 0.14610862731933594, 2.427165985107422, 19.356277465820312, 6.862007141113281, 10.272483825683594, 0.6155319213867188, 10.662338256835938, 17.166412353515625, 2.4599151611328125, 4.0697479248046875, -3.0934410095214844, -3.895538330078125, -7.331390380859375, 9.297588348388672, 17.426246643066406, 19.00135040283203, 7.0559539794921875, 3.105466842651367, 11.22113037109375, 6.684370040893555, -0.20513343811035156, 2.0605316162109375, 14.034156799316406, 10.170692443847656, 3.8331832885742188, 0.445281982421875, 7.36309814453125, 5.832511901855469, -8.155570983886719, 9.162406921386719, 1.3161449432373047, -3.1167068481445312, 12.638580322265625, 4.229404449462891, 20.036697387695312, 26.856491088867188, 21.876502990722656, 3.872650146484375, 16.597488403320312, 2.5576629638671875, 7.926021575927734, 15.257377624511719, 2.2163124084472656, 33.2481689453125, 1.578704833984375, 31.529022216796875, 12.550651550292969], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000066.npy"} +{"epoch": 0.09691629955947137, "step": 67, "batch_size": 64, "mean": 8.340764999389648, "std": 9.678102493286133, "min": -10.356231689453125, "p10": -1.047050476074218, "median": 5.9323272705078125, "p90": 20.788732910156252, "max": 44.978363037109375, "pos_frac": 0.84375, "sample": [21.104568481445312, 2.7864723205566406, 20.957794189453125, 0.579620361328125, 1.9456043243408203, -0.43373870849609375, 16.244918823242188, 4.907949447631836, -6.3461151123046875, 1.9830608367919922, 16.57292938232422, 9.579788208007812, 3.9935684204101562, 15.978469848632812, -0.18128204345703125, -4.581089019775391, 9.05633544921875, 9.499534606933594, 1.3660736083984375, -10.356231689453125, 20.394256591796875, 8.127883911132812, 0.3798027038574219, -0.25516510009765625, 4.4338531494140625, 2.2634506225585938, 21.790069580078125, 6.373416900634766, -1.3098983764648438, 7.455207824707031, 4.3515167236328125, 14.286441802978516, 5.905143737792969, 3.145132064819336, 0.9060745239257812, 9.715591430664062, 15.048871994018555, 5.238630294799805, 1.6161918640136719, -4.95263671875, 4.1810760498046875, 17.426361083984375, 4.221088409423828, 3.270639419555664, 26.725387573242188, 14.279335021972656, 6.535675048828125, 16.783828735351562, 13.786163330078125, 13.829032897949219, 11.91717529296875, 5.458833694458008, 12.325952529907227, 32.24688720703125, 5.959510803222656, 1.5276012420654297, 44.978363037109375, 3.3993453979492188, -2.063385009765625, 24.534385681152344, 20.346054077148438, 7.015033721923828, -2.69366455078125, 8.24622917175293], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000067.npy"} +{"epoch": 0.09838472834067548, "step": 68, "batch_size": 64, "mean": 8.047855377197266, "std": 7.964416980743408, "min": -5.63958740234375, "p10": 0.6021533966064457, "median": 5.907436370849609, "p90": 19.90957946777344, "max": 35.86796569824219, "pos_frac": 0.90625, "sample": [-4.915306091308594, 3.6143112182617188, 22.5662841796875, 7.474090576171875, 5.177059173583984, -5.63958740234375, 8.531303405761719, 22.48206329345703, 1.0107269287109375, 0.4323310852050781, 12.637954711914062, 15.224868774414062, 15.839668273925781, 11.286575317382812, 11.619239807128906, 1.047271728515625, 6.096714019775391, 11.90401840209961, 2.8964614868164062, 8.037590026855469, -0.1060638427734375, 4.2926788330078125, 4.937229156494141, 2.687835693359375, 6.7784271240234375, 13.134307861328125, 18.9671630859375, 9.232131958007812, 8.987129211425781, 19.5179443359375, 20.077423095703125, 22.34829330444336, 2.563568115234375, 5.956787109375, 9.476119995117188, 1.7099685668945312, -0.179962158203125, 6.014177322387695, 2.8065414428710938, 5.0573577880859375, 15.802127838134766, 2.5509185791015625, 10.427383422851562, -5.325897216796875, 5.858085632324219, 5.723297119140625, 3.7623748779296875, 20.574119567871094, 2.8002243041992188, 1.4123039245605469, 2.0612850189208984, 35.86796569824219, 24.196365356445312, 16.947067260742188, 12.829071044921875, 0.9984054565429688, 2.1111488342285156, 5.274238586425781, 11.565788269042969, -1.852874755859375, 4.508934020996094, 1.9859886169433594, 5.2431793212890625, 8.158523559570312], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000068.npy"} +{"epoch": 0.09985315712187959, "step": 69, "batch_size": 64, "mean": 9.200922012329102, "std": 8.942436218261719, "min": -4.143226623535156, "p10": -0.1697189331054687, "median": 6.77076530456543, "p90": 19.37670364379883, "max": 42.397247314453125, "pos_frac": 0.875, "sample": [4.743408203125, 6.371601104736328, 27.995925903320312, 8.14312744140625, 11.618621826171875, 15.568191528320312, -2.0603790283203125, -0.6816940307617188, -0.3740253448486328, 3.2762908935546875, 5.5908050537109375, 8.113323211669922, -1.1353759765625, 8.376419067382812, 7.400402069091797, 13.856090545654297, 23.381622314453125, 7.06939697265625, 16.173126220703125, 14.536605834960938, 0.12738800048828125, 22.355361938476562, 13.0635986328125, 10.538810729980469, 17.37409210205078, 12.85528564453125, 4.517112731933594, 14.699485778808594, 12.798980712890625, 4.852680206298828, 0.6520156860351562, 1.065185546875, 17.988143920898438, 42.397247314453125, 4.021144866943359, 18.843345642089844, 5.3950347900390625, 16.500869750976562, 15.128036499023438, 0.9095382690429688, 1.7760543823242188, 14.611572265625, 10.67081069946289, 0.5269966125488281, -0.1920623779296875, 13.179096221923828, 6.472133636474609, 35.120574951171875, 4.2044219970703125, 7.176477432250977, 6.058950424194336, -4.143226623535156, 5.107696533203125, 3.74786376953125, 2.2443008422851562, 6.1602020263671875, 19.60528564453125, 20.169097900390625, -0.117584228515625, 15.658164978027344, 1.5765857696533203, 1.3087730407714844, 4.531148910522461, -0.6411666870117188], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000069.npy"} +{"epoch": 0.1013215859030837, "step": 70, "batch_size": 64, "mean": 9.938135147094727, "std": 10.198431968688965, "min": -6.6180877685546875, "p10": -0.49107093811035096, "median": 9.665985107421875, "p90": 24.196446990966805, "max": 38.3231201171875, "pos_frac": 0.890625, "sample": [5.893899917602539, 2.970367431640625, 10.729839324951172, 17.801040649414062, 21.788536071777344, 11.405939102172852, -6.6180877685546875, 0.3973541259765625, 13.225631713867188, 3.9919052124023438, -1.0473175048828125, 10.950035095214844, -1.813161849975586, 7.51995849609375, 35.54429626464844, 2.261383056640625, 11.418182373046875, 25.228408813476562, 10.51469612121582, 38.3231201171875, 10.09442138671875, 9.676788330078125, 1.9061126708984375, 27.777740478515625, 0.4811058044433594, 16.296630859375, 17.333847045898438, 12.073446273803711, -0.8350353240966797, 6.979984283447266, 4.555763244628906, 28.277481079101562, 1.6139450073242188, 0.8197822570800781, 1.1709556579589844, 0.10338592529296875, 12.637432098388672, 2.242095947265625, 9.655181884765625, -1.1818218231201172, 2.9809112548828125, 3.3085289001464844, 0.5800933837890625, 2.136810302734375, 10.677057266235352, 1.555328369140625, 18.245765686035156, 21.598915100097656, -6.51617431640625, 4.621345520019531, 16.145790100097656, 1.5701141357421875, 1.6638545989990234, 30.321800231933594, 11.766590118408203, -0.7458381652832031, 31.321975708007812, 10.143112182617188, 19.65581512451172, 15.35405158996582, 4.00883674621582, 21.6319580078125, 12.434051513671875, 19.41468048095703], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000070.npy"} +{"epoch": 0.1027900146842878, "step": 71, "batch_size": 64, "mean": 10.965913772583008, "std": 10.64250659942627, "min": -8.724960327148438, "p10": 1.021372222900391, "median": 8.891914367675781, "p90": 20.67743148803712, "max": 44.175994873046875, "pos_frac": 0.921875, "sample": [-5.217613220214844, 36.88128662109375, 6.5072021484375, 0.8880233764648438, 6.326778411865234, 12.045330047607422, 6.239627838134766, 11.072021484375, 16.00611114501953, 4.467662811279297, 18.824310302734375, 5.456441879272461, 15.47406005859375, 4.106807708740234, 2.4808120727539062, 8.318222045898438, 11.014240264892578, 13.0892333984375, -0.7369461059570312, 1.9612674713134766, 4.283348083496094, 39.06111145019531, 6.632545471191406, 5.674762725830078, 15.024398803710938, 15.680870056152344, 5.885475158691406, 10.647048950195312, 12.163721084594727, -7.13226318359375, 4.853446960449219, 14.8114013671875, 8.2684326171875, 0.10861396789550781, 5.177988052368164, 15.910400390625, 17.179428100585938, 1.33251953125, 18.50042724609375, 12.427177429199219, 5.908355712890625, 11.388885498046875, 8.271530151367188, 18.242523193359375, 13.01495361328125, 15.545768737792969, 4.5790252685546875, 32.92637634277344, 44.175994873046875, 42.254425048828125, 2.0548095703125, 21.47162628173828, -8.724960327148438, -0.930511474609375, 11.220890045166016, 4.4313507080078125, 7.382610321044922, 9.818000793457031, 18.221481323242188, 13.597869873046875, 27.42987060546875, 2.8765296936035156, 5.49970817565918, 9.465606689453125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000071.npy"} +{"epoch": 0.10425844346549193, "step": 72, "batch_size": 64, "mean": 11.478952407836914, "std": 10.859827995300293, "min": -9.482696533203125, "p10": -0.7972778320312489, "median": 8.833353996276855, "p90": 27.16513671875001, "max": 45.729461669921875, "pos_frac": 0.890625, "sample": [16.548011779785156, 13.216451644897461, 5.642448425292969, 11.813873291015625, 8.11594009399414, 2.0067081451416016, 7.091156005859375, 24.093536376953125, -9.161727905273438, 22.7744140625, 12.87343978881836, 12.313526153564453, 8.232236862182617, 0.280364990234375, 15.802677154541016, 8.646991729736328, 10.164047241210938, 22.2991943359375, 11.834869384765625, 29.149070739746094, 35.22526550292969, -1.261892318725586, 6.297996520996094, 8.880258560180664, 14.486602783203125, 4.006034851074219, 16.639930725097656, 7.9504241943359375, 8.039031982421875, 14.544105529785156, 2.3309478759765625, 6.518684387207031, 8.244918823242188, -9.482696533203125, 29.574600219726562, 24.986373901367188, 16.852622985839844, 6.481645584106445, 19.87969970703125, -2.2668075561523438, 6.85759162902832, 34.647216796875, 45.729461669921875, -6.250152587890625, 8.887128829956055, 24.265426635742188, 8.000930786132812, 7.163066864013672, 0.9102077484130859, 28.098892211914062, -1.259124755859375, 3.035564422607422, 2.7665481567382812, 16.851913452148438, 11.772209167480469, 16.499053955078125, 8.786449432373047, 14.453712463378906, 31.73876953125, 10.85692024230957, 5.750709533691406, 2.974029541015625, -1.4612350463867188, 1.9126777648925781], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000072.npy"} +{"epoch": 0.10572687224669604, "step": 73, "batch_size": 64, "mean": 11.041110038757324, "std": 15.013915061950684, "min": -20.2982177734375, "p10": -5.8714065551757795, "median": 8.033514022827148, "p90": 28.75206565856934, "max": 55.14604187011719, "pos_frac": 0.796875, "sample": [27.5091552734375, -8.235801696777344, 5.31658935546875, 12.629058837890625, 5.5034942626953125, 29.408523559570312, -3.5560226440429688, 23.105121612548828, 6.1159515380859375, -1.4164581298828125, 22.842132568359375, 28.968502044677734, 19.749755859375, -0.624908447265625, 2.344207763671875, -12.582611083984375, 47.340789794921875, -0.9599456787109375, 9.72024917602539, 15.679351806640625, 15.06243896484375, 6.701808929443359, 8.36968994140625, 0.8197784423828125, 1.6287841796875, 7.697338104248047, -12.58810806274414, 9.664413452148438, 25.758941650390625, 12.860187530517578, 11.247200012207031, 1.9928550720214844, 45.646820068359375, -8.611663818359375, 5.338191986083984, 24.59821319580078, 23.767127990722656, -8.566556930541992, 3.049135208129883, 3.1628894805908203, 0.25307464599609375, -4.0094146728515625, -6.669403076171875, 6.527372360229492, 15.476509094238281, 13.133102416992188, 2.336568832397461, 4.983175277709961, 18.928314208984375, 36.51405334472656, 11.210037231445312, 5.673748016357422, 17.637542724609375, 5.370433807373047, 11.099876403808594, -2.0973358154296875, 13.905975341796875, 46.60289001464844, 19.444015502929688, 55.14604187011719, 3.2751407623291016, 28.247047424316406, 17.483863830566406, -20.2982177734375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000073.npy"} +{"epoch": 0.10719530102790015, "step": 74, "batch_size": 64, "mean": 11.332639694213867, "std": 16.051288604736328, "min": -31.561988830566406, "p10": -1.9190547943115226, "median": 8.826419830322266, "p90": 26.262744140625006, "max": 91.2896728515625, "pos_frac": 0.8125, "sample": [24.901206970214844, 8.572097778320312, 9.250118255615234, 0.6233730316162109, 20.817787170410156, 4.777549743652344, 7.850982666015625, 44.262550354003906, 2.0397911071777344, 7.502861022949219, 11.025245666503906, 9.396331787109375, 16.64229965209961, 2.9122161865234375, 15.029899597167969, 91.2896728515625, 7.614410400390625, 23.63622283935547, 13.003009796142578, 1.82073974609375, 1.748830795288086, 41.01066589355469, 1.7667617797851562, 26.84626007080078, 23.01752471923828, 8.717041015625, -3.0444469451904297, 2.477294921875, -0.7300872802734375, -15.362060546875, 12.528236389160156, -0.2928009033203125, -0.77984619140625, 8.088577270507812, -31.561988830566406, 18.22600555419922, -0.3670806884765625, 20.03519058227539, 4.828191757202148, -2.2864341735839844, -7.544136047363281, 9.834964752197266, 15.009750366210938, 23.314285278320312, -1.0618362426757812, 32.337677001953125, 12.1689453125, 4.8986358642578125, 5.8359375, 6.645618438720703, 8.935798645019531, 31.305206298828125, 14.444259643554688, 19.03704833984375, 30.924400329589844, 16.65985870361328, 4.731624603271484, 12.495445251464844, 18.741912841796875, 23.13042449951172, 1.7811203002929688, -3.5747528076171875, -3.6148109436035156, 11.017398834228516], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000074.npy"} +{"epoch": 0.10866372980910426, "step": 75, "batch_size": 64, "mean": 15.949441909790039, "std": 15.212785720825195, "min": -8.0726318359375, "p10": -2.2375640869140607, "median": 13.285959243774414, "p90": 37.887567901611334, "max": 59.310028076171875, "pos_frac": 0.875, "sample": [-0.38582611083984375, 17.174209594726562, 13.567020416259766, 14.60760498046875, 15.584442138671875, 11.545782089233398, 9.309829711914062, 5.78564453125, 32.580078125, 48.26904296875, 26.72623062133789, 12.798049926757812, -3.6228179931640625, 21.43238067626953, 59.310028076171875, 12.001533508300781, 30.286895751953125, 50.42915344238281, 8.281320571899414, 0.4329490661621094, 35.27226257324219, 19.320846557617188, 12.733024597167969, 9.823867797851562, 19.942581176757812, 37.30689239501953, 38.13642883300781, 3.403463363647461, 6.577430725097656, 26.05276870727539, 14.840774536132812, 13.25802993774414, 5.27947998046875, 52.215545654296875, -3.3818817138671875, -3.0311660766601562, 20.641265869140625, -8.0726318359375, 17.00426483154297, 16.037025451660156, 3.938709259033203, 11.142784118652344, -5.367027282714844, 5.2424468994140625, -6.526937484741211, 21.77728271484375, 2.2480621337890625, 7.827838897705078, 29.015365600585938, 20.88909912109375, 3.545534133911133, -4.3985137939453125, 7.3231964111328125, 4.500099182128906, 12.633163452148438, 3.958498001098633, 10.201736450195312, 15.657669067382812, 14.780101776123047, 14.191703796386719, 28.34514617919922, 39.579620361328125, 47.470977783203125, 13.313888549804688], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000075.npy"} +{"epoch": 0.11013215859030837, "step": 76, "batch_size": 64, "mean": 10.794473648071289, "std": 11.741714477539062, "min": -28.325302124023438, "p10": -0.4881864547729492, "median": 9.970804214477539, "p90": 23.70474548339844, "max": 53.193878173828125, "pos_frac": 0.859375, "sample": [11.233148574829102, 7.780082702636719, 2.111928939819336, 19.043594360351562, 2.2808303833007812, 16.940109252929688, 13.9605712890625, 21.495086669921875, 23.985088348388672, 9.148002624511719, 9.815204620361328, 15.617427825927734, 8.004446029663086, 27.136703491210938, 4.1522674560546875, 4.151388168334961, 7.892723083496094, 53.193878173828125, 24.44324493408203, 4.046089172363281, -1.1536483764648438, 10.42396354675293, 0.49649620056152344, 9.114593505859375, -11.6561279296875, 6.699274063110352, 4.331747055053711, 10.511116027832031, 5.895118713378906, 8.865589141845703, 23.911842346191406, 17.707054138183594, 16.899169921875, 7.795919418334961, 23.221519470214844, 3.47998046875, -0.6138458251953125, 19.762969970703125, 3.5747146606445312, -4.5376129150390625, 6.511234283447266, -0.10187721252441406, -3.985525131225586, -0.46221923828125, 13.545333862304688, 21.879440307617188, 40.75456237792969, 22.657447814941406, 7.816951751708984, 3.60040283203125, 10.736885070800781, 0.3541908264160156, 14.222557067871094, 11.043052673339844, 12.5682373046875, 10.789474487304688, -0.4993152618408203, 10.12640380859375, 28.409835815429688, 16.294212341308594, 17.092605590820312, 22.830902099609375, 11.825141906738281, -28.325302124023438], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000076.npy"} +{"epoch": 0.11160058737151249, "step": 77, "batch_size": 64, "mean": 16.856914520263672, "std": 16.083045959472656, "min": -23.4449462890625, "p10": 1.6757415771484376, "median": 17.168073654174805, "p90": 40.33541259765625, "max": 55.193206787109375, "pos_frac": 0.90625, "sample": [12.537025451660156, 38.410972595214844, 25.286602020263672, 30.686771392822266, 33.280296325683594, 1.9412841796875, 47.99674987792969, 17.988460540771484, 8.146656036376953, 8.752601623535156, 3.9782962799072266, 28.442031860351562, 28.67603302001953, 5.556049346923828, 7.74554443359375, 18.00537872314453, -23.4449462890625, 15.349746704101562, 25.0460205078125, 26.40899658203125, 12.652824401855469, -1.7427558898925781, 6.992656707763672, -0.0578155517578125, 2.0016326904296875, 39.461822509765625, 1.58795166015625, 18.73680877685547, 8.479804992675781, -4.474189758300781, 27.002674102783203, 22.95684814453125, 11.541650772094727, 50.020294189453125, 26.442138671875, 8.29850959777832, 40.709808349609375, 55.193206787109375, 16.82819366455078, 8.156341552734375, -22.096511840820312, 5.902191162109375, -5.599273681640625, 3.4259796142578125, 9.675790786743164, 23.22125244140625, 26.125274658203125, 21.777517318725586, 21.246536254882812, 21.291236877441406, 5.652936935424805, 1.880584716796875, 4.382665634155273, 19.346839904785156, 22.739341735839844, 19.60882568359375, 2.5594635009765625, 42.402496337890625, 52.27622985839844, 8.76788330078125, 17.843524932861328, 40.85881042480469, 17.507953643798828, 6.465972900390625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000077.npy"} +{"epoch": 0.1130690161527166, "step": 78, "batch_size": 64, "mean": 13.276593208312988, "std": 13.305554389953613, "min": -12.21185302734375, "p10": -0.5054374694824219, "median": 13.539135932922363, "p90": 32.75639991760255, "max": 56.34375, "pos_frac": 0.859375, "sample": [7.947479248046875, 1.3447074890136719, 29.874664306640625, 10.92178726196289, 0.12663650512695312, 35.02162170410156, 14.73208236694336, 56.34375, 21.381134033203125, 18.27066421508789, 17.086139678955078, 6.854579925537109, -4.183712005615234, 14.148735046386719, 19.335620880126953, 17.89645004272461, -0.4796905517578125, 6.996898651123047, -5.183845520019531, 0.4544830322265625, 6.6735687255859375, 14.03049087524414, 20.61629867553711, 14.793373107910156, 16.07131576538086, 13.889202117919922, 14.19510269165039, 12.2900390625, 14.71429443359375, 4.604297637939453, 33.65043640136719, 13.78007698059082, 5.090492248535156, -5.572460174560547, 21.03350067138672, -4.673431396484375, 36.997467041015625, 16.808605194091797, 34.931915283203125, 5.055442810058594, 30.67031478881836, 4.342170715332031, 38.73560333251953, 28.430896759033203, 20.70379638671875, 2.55413818359375, 9.780902862548828, 3.1417770385742188, 11.203460693359375, 6.906368255615234, 4.398445129394531, -0.21705055236816406, 13.298194885253906, -0.5164718627929688, 22.857650756835938, -12.21185302734375, 41.542510986328125, 4.194042205810547, -10.302154541015625, 23.289047241210938, 5.8611297607421875, 21.486282348632812, 1.95184326171875, 19.730712890625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000078.npy"} +{"epoch": 0.1145374449339207, "step": 79, "batch_size": 64, "mean": 14.89985179901123, "std": 17.347658157348633, "min": -10.869247436523438, "p10": -3.2406295776367187, "median": 10.711711883544922, "p90": 41.50805740356448, "max": 60.223388671875, "pos_frac": 0.75, "sample": [-0.1070556640625, 43.91608428955078, 5.457187652587891, 53.45965576171875, 7.300575256347656, -0.7525787353515625, -1.5120868682861328, -6.441673278808594, 35.88932800292969, 33.22308349609375, 16.552207946777344, -10.869247436523438, 21.183277130126953, 53.92889404296875, 10.179428100585938, 20.33448028564453, 60.223388671875, 45.150238037109375, 15.093969345092773, -2.0009002685546875, 57.77459716796875, 14.158546447753906, -1.156402587890625, -7.401824951171875, 21.251739501953125, 0.8770828247070312, 4.1433563232421875, 12.62213134765625, 11.243995666503906, 46.186180114746094, 3.4916305541992188, 8.236183166503906, 9.6549072265625, 24.3280029296875, -5.130126953125, 8.753814697265625, -2.242748260498047, 5.80889892578125, 34.45411682128906, 29.627456665039062, 14.323707580566406, 25.408836364746094, 26.0751953125, 17.22814178466797, 22.227508544921875, -3.2675209045410156, 2.9736328125, 22.121795654296875, 27.2874755859375, 9.893960952758789, 12.361810684204102, -4.293386459350586, -1.4576911926269531, 31.723953247070312, -2.0744857788085938, 15.606719970703125, 5.480289459228516, 5.008796691894531, 0.7288360595703125, -3.543243408203125, 8.009296417236328, 32.597808837890625, -3.1778831481933594, 15.4571533203125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000079.npy"} +{"epoch": 0.11600587371512482, "step": 80, "batch_size": 64, "mean": 13.337163925170898, "std": 16.962793350219727, "min": -14.432174682617188, "p10": -3.7804450988769522, "median": 11.359651565551758, "p90": 32.078581237792974, "max": 65.6929931640625, "pos_frac": 0.78125, "sample": [26.842857360839844, 7.147178649902344, -2.1399612426757812, 21.586997985839844, 4.137872695922852, 2.4351158142089844, -1.1971778869628906, -4.1114654541015625, 16.56903076171875, 10.601295471191406, 3.251863479614258, 13.329448699951172, -12.009960174560547, -14.432174682617188, -2.03863525390625, 18.362106323242188, 63.559234619140625, 28.200119018554688, 65.6929931640625, 40.039031982421875, 15.189079284667969, 6.1126251220703125, 33.48973083496094, 30.75366973876953, 28.304336547851562, 12.10787582397461, -0.8152103424072266, 22.159149169921875, 12.022716522216797, 12.650344848632812, 0.7386665344238281, -11.241790771484375, 5.9251708984375, -3.0080642700195312, 21.833160400390625, -7.499977111816406, 47.43324279785156, 24.10028076171875, -5.161090850830078, 60.04193115234375, 8.565179824829102, 3.22259521484375, 3.7342491149902344, 0.00150299072265625, 32.646400451660156, 8.293407440185547, 19.415077209472656, 19.252090454101562, 14.032865524291992, 10.696586608886719, -0.8171520233154297, -2.55975341796875, 0.08483314514160156, 2.83221435546875, 20.530885696411133, 17.608558654785156, 4.3721160888671875, 4.885303497314453, 25.967315673828125, 18.786529541015625, 15.62652587890625, 26.431182861328125, -4.318098068237305, 13.326515197753906], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000080.npy"} +{"epoch": 0.11747430249632893, "step": 81, "batch_size": 64, "mean": 18.248262405395508, "std": 19.949203491210938, "min": -13.4688720703125, "p10": -5.237692260742184, "median": 14.913415908813477, "p90": 41.80066528320313, "max": 86.28912353515625, "pos_frac": 0.875, "sample": [30.904197692871094, 21.74700927734375, -13.4688720703125, 31.782630920410156, 18.61984634399414, 40.813385009765625, 14.302188873291016, -8.455677032470703, 5.698909759521484, 26.360931396484375, 22.722518920898438, -8.58355712890625, 1.327789306640625, 17.960296630859375, 63.330413818359375, 5.043205261230469, 15.524642944335938, 22.876060485839844, 42.223785400390625, -9.112556457519531, 0.8443107604980469, 12.641311645507812, 18.409622192382812, 53.51744842529297, 19.40515899658203, 10.617462158203125, 12.081451416015625, 4.921863555908203, 3.8786067962646484, -1.7852020263671875, 19.446762084960938, 10.137741088867188, 3.652231216430664, 11.170982360839844, 27.20124053955078, 82.0313720703125, 32.88203811645508, 13.334976196289062, 3.0439987182617188, 43.885894775390625, 30.077880859375, 35.15936279296875, 10.14410400390625, 2.901165008544922, 4.3909759521484375, 1.0436973571777344, -11.00741195678711, 11.428173065185547, 21.984153747558594, 20.599502563476562, 13.919376373291016, 12.891170501708984, 17.24938201904297, -7.2887420654296875, 33.0506591796875, 35.714942932128906, 86.28912353515625, 48.07696533203125, 6.5315704345703125, 29.59185028076172, 4.478096008300781, 20.90576171875, -6.7173309326171875, 23.537948608398438], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000081.npy"} +{"epoch": 0.11894273127753303, "step": 82, "batch_size": 64, "mean": 14.88681697845459, "std": 21.69973373413086, "min": -40.995513916015625, "p10": -6.990193557739257, "median": 13.761377334594727, "p90": 38.09659957885742, "max": 87.09140014648438, "pos_frac": 0.8125, "sample": [-1.7459793090820312, 23.623497009277344, -9.69439697265625, 16.06525421142578, 18.32254409790039, 16.333864212036133, -16.295501708984375, 12.595592498779297, 18.84925079345703, 47.53453826904297, 51.57691955566406, 12.04489517211914, -7.185771942138672, 4.705711364746094, 12.000284194946289, 79.56578063964844, 9.390731811523438, -40.995513916015625, -6.533843994140625, 27.155853271484375, 0.973876953125, 28.138092041015625, 36.66075897216797, 3.0692081451416016, -12.865592956542969, -2.8252487182617188, 3.8417816162109375, 15.943191528320312, 16.06011962890625, 10.050796508789062, 13.842445373535156, 28.626251220703125, 26.280662536621094, -0.6428031921386719, 53.70964050292969, 14.863601684570312, 37.861480712890625, 13.680309295654297, 8.752201080322266, 1.3537139892578125, -0.5910720825195312, 25.096267700195312, 43.618377685546875, -10.181922912597656, 24.274398803710938, 16.779197692871094, 6.945486068725586, 37.435916900634766, 22.041954040527344, 1.355865478515625, 6.368824005126953, 5.8872222900390625, -40.1553955078125, 5.907432556152344, 24.88378143310547, 18.38922119140625, 7.390953063964844, 22.477294921875, 15.773162841796875, 87.09140014648438, 11.355171203613281, 17.36979103088379, 0.3573760986328125, 38.197364807128906], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000082.npy"} +{"epoch": 0.12041116005873716, "step": 83, "batch_size": 64, "mean": 16.931299209594727, "std": 18.86237907409668, "min": -19.659801483154297, "p10": -4.538430023193358, "median": 14.621905326843262, "p90": 43.15030136108399, "max": 65.51423645019531, "pos_frac": 0.8125, "sample": [14.868659973144531, 18.063671112060547, 12.914691925048828, 54.10890197753906, 34.86852264404297, 10.7470703125, -3.6432266235351562, 43.42303466796875, 40.490203857421875, 6.7411651611328125, 4.992523193359375, 14.375150680541992, 6.60101318359375, -2.0623931884765625, 15.811172485351562, 54.01380920410156, -10.024795532226562, 65.51423645019531, 7.3436279296875, 53.32501983642578, 30.751754760742188, 9.081539154052734, 42.51392364501953, -11.649070739746094, -19.659801483154297, -4.922088623046875, 39.721527099609375, -12.911300659179688, 16.029932022094727, 48.472564697265625, 16.19261932373047, 10.971878051757812, 4.415063858032227, 3.8901519775390625, 16.532386779785156, 10.916671752929688, 22.464275360107422, -0.7022628784179688, 28.2060546875, 7.708534240722656, 15.780738830566406, 11.849929809570312, 33.64665222167969, 54.509246826171875, 17.2779541015625, 36.447723388671875, 5.537868499755859, -13.07012939453125, -1.1981849670410156, 0.2825126647949219, -5.802398681640625, 32.436920166015625, 7.962093353271484, 32.72703552246094, 19.51844024658203, 19.088218688964844, -1.2371597290039062, 6.209007263183594, 15.180850982666016, 24.861099243164062, 32.90644073486328, 13.375045776367188, 21.730194091796875, 3.0866470336914062], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000083.npy"} +{"epoch": 0.12187958883994127, "step": 84, "batch_size": 64, "mean": 15.714971542358398, "std": 14.936293601989746, "min": -27.043960571289062, "p10": -0.5931720733642577, "median": 13.85467529296875, "p90": 33.480149078369145, "max": 63.37611389160156, "pos_frac": 0.859375, "sample": [8.507225036621094, 11.81402587890625, 3.8859176635742188, 3.4129199981689453, 31.953155517578125, 3.4806060791015625, 3.7481155395507812, 17.848892211914062, 11.064943313598633, 10.425010681152344, 24.293880462646484, 25.54352569580078, -11.069931030273438, 14.653846740722656, 21.251758575439453, 26.659881591796875, 8.06634521484375, 63.37611389160156, 24.814800262451172, 0.978912353515625, 6.793621063232422, 25.98955535888672, 11.031623840332031, 5.636043548583984, -1.1460990905761719, 34.13457489013672, 20.35363006591797, 30.569747924804688, -27.043960571289062, -0.90032958984375, 42.376625061035156, 4.6319122314453125, 29.939163208007812, 13.621322631835938, 13.308311462402344, 2.159271240234375, 36.962249755859375, 30.7291259765625, 21.38683319091797, 20.81719970703125, 4.714439392089844, 25.837234497070312, -1.0459709167480469, 13.582658767700195, 23.88128662109375, 23.23659896850586, -0.5327301025390625, -0.6190757751464844, 25.27234649658203, 25.455642700195312, 3.71368408203125, 35.98457336425781, 40.81488037109375, 18.480064392089844, 43.03179931640625, 14.088027954101562, 24.528579711914062, 4.762939453125, -6.7207183837890625, 25.477195739746094, 10.121746063232422, 15.1241455078125, 10.653327941894531, -0.1448802947998047], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000084.npy"} +{"epoch": 0.12334801762114538, "step": 85, "batch_size": 64, "mean": 15.736580848693848, "std": 17.872610092163086, "min": -19.35204315185547, "p10": -1.6058658599853515, "median": 13.566448211669922, "p90": 41.76849441528321, "max": 58.607940673828125, "pos_frac": 0.75, "sample": [32.73036193847656, 9.044898986816406, 25.683692932128906, 2.5433616638183594, 5.055458068847656, 19.6829833984375, 16.794815063476562, 3.248912811279297, 48.013938903808594, 6.477073669433594, -2.509347915649414, 6.953010559082031, 45.52433776855469, 18.161941528320312, 17.721494674682617, 16.649192810058594, -1.4957733154296875, -9.648574829101562, 16.128517150878906, -0.7666854858398438, -19.35204315185547, 52.343658447265625, 12.985801696777344, -1.5485401153564453, 1.649749755859375, 35.16454315185547, 10.309722900390625, -0.14447021484375, 14.1470947265625, 8.393028259277344, 36.11962890625, -0.4396820068359375, -7.502616882324219, -1.6304340362548828, 12.63665771484375, 5.004987716674805, 53.55418395996094, 14.849102020263672, 28.126708984375, 0.9837932586669922, -1.259307861328125, 58.607940673828125, -1.3066139221191406, 32.716949462890625, 38.59385681152344, 11.797859191894531, 40.165611267089844, 42.4554443359375, 5.971527099609375, 4.046773910522461, 21.97968292236328, 39.702613830566406, -0.8966522216796875, -2.5912704467773438, 15.080001831054688, 14.578060150146484, 44.533905029296875, 15.23984146118164, 27.110794067382812, 29.960124969482422, -12.883773803710938, 16.462764739990234, -0.461212158203125, 35.89177703857422], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000085.npy"} +{"epoch": 0.12481644640234948, "step": 86, "batch_size": 64, "mean": 16.63092613220215, "std": 25.570661544799805, "min": -31.574264526367188, "p10": -7.825500488281249, "median": 11.202838897705078, "p90": 46.96224670410159, "max": 95.68026733398438, "pos_frac": 0.78125, "sample": [-6.951194763183594, 13.481201171875, 49.474151611328125, -10.350151062011719, 30.708417892456055, 21.5572509765625, 22.43320655822754, -4.463462829589844, 72.01010131835938, 8.213207244873047, 4.591974258422852, 1.4800682067871094, 3.3783950805664062, 15.173812866210938, -3.361419677734375, 15.565322875976562, 37.68548583984375, 17.549713134765625, -2.2912330627441406, -28.46600341796875, 37.909210205078125, 13.62752914428711, 23.835186004638672, 8.924476623535156, 8.675811767578125, -11.185646057128906, 2.6655960083007812, 52.83137512207031, -4.419338226318359, 5.185997009277344, 28.990497589111328, 17.360271453857422, 5.5831451416015625, -31.574264526367188, 33.69306945800781, 41.10113525390625, 19.992340087890625, -20.035499572753906, 49.79817199707031, 3.652437210083008, 94.58001708984375, 16.61095428466797, 21.974456787109375, 8.33663558959961, -1.3881759643554688, 95.68026733398438, 6.109697341918945, 86.65005493164062, 33.90574645996094, 3.2969818115234375, -21.48957061767578, 33.78050231933594, 8.465499877929688, 28.122726440429688, 21.083633422851562, 30.714427947998047, 32.844932556152344, 2.6387767791748047, 5.251316070556641, 5.698659896850586, -1.728750228881836, -8.200202941894531, 3.4194183349609375, 13.990894317626953], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000086.npy"} +{"epoch": 0.1262848751835536, "step": 87, "batch_size": 64, "mean": 18.636001586914062, "std": 22.742950439453125, "min": -40.022125244140625, "p10": -2.6016275405883773, "median": 16.24578857421875, "p90": 47.269987487792974, "max": 96.83235168457031, "pos_frac": 0.828125, "sample": [19.323076248168945, 24.230239868164062, 0.7821216583251953, 7.86015510559082, 27.86785888671875, 1.113861083984375, 19.0289306640625, -0.5642757415771484, 1.3411712646484375, 0.887451171875, -8.569229125976562, 28.4071044921875, -3.174285888671875, 45.04374694824219, 20.62334442138672, 36.76506042480469, 42.22039794921875, 7.748083114624023, 71.5501708984375, 36.47283935546875, 16.085853576660156, 16.268775939941406, 5.717384338378906, 30.68225860595703, 26.44818115234375, 1.8114604949951172, 26.480323791503906, 96.83235168457031, 13.889091491699219, -0.14498138427734375, 12.879404067993164, 51.702301025390625, 48.576446533203125, 6.300571441650391, 43.85383605957031, 19.98419952392578, 11.715494155883789, 29.113346099853516, 15.512855529785156, 44.05192565917969, 0.41002464294433594, 56.96717834472656, 18.35460662841797, 19.627525329589844, -40.022125244140625, 6.855995178222656, 28.81067657470703, -25.683183670043945, 29.13849639892578, 51.80572509765625, -7.747802734375, -7.58441162109375, 6.608318328857422, 16.222801208496094, -0.6548843383789062, 48.224090576171875, 25.893905639648438, 26.040428161621094, -23.09711456298828, 15.846939086914062, 14.452125549316406, 1.5238876342773438, 35.257476806640625, -1.2654247283935547], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000087.npy"} +{"epoch": 0.1277533039647577, "step": 88, "batch_size": 64, "mean": 14.273067474365234, "std": 21.207107543945312, "min": -32.361572265625, "p10": -3.417302131652831, "median": 11.08305835723877, "p90": 39.01131134033204, "max": 83.79324340820312, "pos_frac": 0.828125, "sample": [9.892936706542969, 10.623970031738281, 51.60643005371094, 40.51716613769531, 27.564437866210938, 31.982772827148438, 2.1710643768310547, 45.42231750488281, 1.933349609375, 11.112451553344727, 23.274017333984375, 5.3656463623046875, 6.779182434082031, 31.327804565429688, -16.767410278320312, 21.549942016601562, -15.782882690429688, 15.944351196289062, -29.812728881835938, -31.90289306640625, -2.623525619506836, 51.566864013671875, 25.64057159423828, -3.7574920654296875, 33.07939910888672, 4.107856750488281, 19.39959716796875, -2.4423675537109375, 17.326772689819336, 22.031890869140625, 31.87848663330078, 5.665855407714844, -0.6569099426269531, 24.64862823486328, 0.867462158203125, 63.80950927734375, 16.27904510498047, 35.497650146484375, 3.2439804077148438, 83.79324340820312, 3.0735321044921875, 28.79875946044922, 5.641712188720703, 12.174718856811523, 11.053665161132812, 15.91323471069336, 3.691844940185547, 0.083099365234375, 24.10834503173828, 54.27618408203125, 25.316200256347656, 11.730697631835938, 2.655384063720703, 6.0008697509765625, -32.361572265625, 0.0316162109375, 7.541015625, -1.3362979888916016, 6.972272872924805, 14.327083587646484, 16.563451766967773, -15.741767883300781, 10.364944458007812, 30.43889617919922], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000088.npy"} +{"epoch": 0.12922173274596183, "step": 89, "batch_size": 64, "mean": 15.564217567443848, "std": 17.921138763427734, "min": -33.4310302734375, "p10": -0.6797523498535153, "median": 14.175106048583984, "p90": 35.32193679809571, "max": 81.32298278808594, "pos_frac": 0.875, "sample": [2.20062255859375, 2.122112274169922, 3.6808624267578125, 17.002836227416992, -5.000728607177734, 32.91778564453125, 24.265419006347656, 14.119789123535156, -0.8038177490234375, 45.970314025878906, 41.49676513671875, 19.329662322998047, 16.161666870117188, 31.82762908935547, 21.21894073486328, 26.416091918945312, 12.411788940429688, 8.283485412597656, -1.8209514617919922, 14.281547546386719, 10.324661254882812, 2.086812973022461, 13.81243896484375, 8.729621887207031, -4.254119873046875, 3.8215789794921875, -0.39026641845703125, 7.5108489990234375, 46.966400146484375, 14.560890197753906, 15.617576599121094, 62.907623291015625, 13.645160675048828, 36.35228729248047, 16.589073181152344, 4.5582275390625, 14.230422973632812, 81.32298278808594, 1.3441429138183594, 0.019304275512695312, 22.24441909790039, 4.488838195800781, 15.444501876831055, 21.166545867919922, 14.939092636108398, 6.171960830688477, 9.011238098144531, 5.150505065917969, 5.71881103515625, 24.6031494140625, 15.03500747680664, 25.743820190429688, -1.2550086975097656, 10.23775863647461, 28.929176330566406, -33.4310302734375, -11.694618225097656, 19.129226684570312, 15.101617813110352, 12.936798095703125, 21.7066650390625, 20.048507690429688, 64.26237487792969, 4.583095550537109], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000089.npy"} +{"epoch": 0.13069016152716592, "step": 90, "batch_size": 64, "mean": 19.70709991455078, "std": 26.67934799194336, "min": -29.327377319335938, "p10": -2.906845664978027, "median": 12.183979034423828, "p90": 61.78670349121095, "max": 95.47296142578125, "pos_frac": 0.828125, "sample": [43.30290222167969, -27.888702392578125, 27.655838012695312, 59.0860595703125, 18.997928619384766, 27.500999450683594, 0.39579200744628906, -1.488128662109375, 43.17144775390625, -0.7319927215576172, 33.13074493408203, 12.693916320800781, 1.8224449157714844, 11.064414978027344, 24.609405517578125, 65.82892608642578, -1.4734306335449219, 95.47296142578125, 34.90232849121094, 2.2379226684570312, 3.055910110473633, 5.738094329833984, 11.674041748046875, -3.0880508422851562, 76.7479248046875, 15.801015853881836, 22.217613220214844, 78.76010131835938, 4.581239700317383, 20.170459747314453, 92.57110595703125, 13.695564270019531, -7.3030548095703125, 6.934759140014648, 6.683860778808594, -29.327377319335938, 20.07650375366211, 5.740104675292969, 42.45295715332031, 4.082221984863281, 8.239692687988281, 9.798019409179688, -14.251319885253906, -2.4840335845947266, 11.560806274414062, 52.95860290527344, 62.944122314453125, -24.539047241210938, -14.053451538085938, 8.834068298339844, 21.539222717285156, 1.6611194610595703, 14.918537139892578, 21.08917999267578, 42.457061767578125, 22.375770568847656, 9.20114517211914, 6.612276077270508, 28.74078369140625, 8.927257537841797, 14.752098083496094, 10.525896072387695, 66.83847045898438, 31.05133819580078], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000090.npy"} +{"epoch": 0.13215859030837004, "step": 91, "batch_size": 64, "mean": 18.46245765686035, "std": 22.674222946166992, "min": -20.74683380126953, "p10": -7.50035400390625, "median": 15.630769729614258, "p90": 46.871373748779305, "max": 80.0819091796875, "pos_frac": 0.75, "sample": [14.816802978515625, 41.40858459472656, -16.079322814941406, 6.129581451416016, 48.00386047363281, -0.1520061492919922, -3.3658523559570312, 35.2280387878418, 5.058160781860352, 14.995872497558594, 31.298301696777344, 28.060501098632812, 30.436058044433594, 11.740619659423828, -20.74683380126953, 80.0819091796875, -8.101341247558594, 2.30023193359375, 5.4343414306640625, 58.73014831542969, 10.750476837158203, 23.832233428955078, 70.6903076171875, -6.7382354736328125, 24.025375366210938, 44.228904724121094, -7.597404479980469, 0.6631755828857422, 15.921554565429688, -3.3268203735351562, -3.1601104736328125, 17.000492095947266, 42.006561279296875, -10.609283447265625, 17.345428466796875, 15.339984893798828, 40.845916748046875, -6.86993408203125, 16.93758773803711, -2.2903499603271484, 6.242889404296875, 31.55908203125, 44.081634521484375, 3.013591766357422, -8.915233612060547, 25.945968627929688, 19.9510498046875, 57.595611572265625, 29.63654327392578, 27.000320434570312, 16.183937072753906, -15.927261352539062, 0.5052776336669922, 14.098762512207031, -1.340585708618164, 37.226470947265625, 0.2501811981201172, 61.2669677734375, 32.899444580078125, 15.266563415527344, 36.961151123046875, -7.273902893066406, 32.269378662109375, 58.825965881347656], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000091.npy"} +{"epoch": 0.13362701908957417, "step": 92, "batch_size": 64, "mean": 17.94965362548828, "std": 22.629735946655273, "min": -30.397415161132812, "p10": -6.422637939453125, "median": 15.801023483276367, "p90": 43.16379356384277, "max": 86.227783203125, "pos_frac": 0.765625, "sample": [-0.8304462432861328, 4.80596923828125, 9.6708984375, 32.20960998535156, -7.447746276855469, 5.561042785644531, -30.397415161132812, -10.258996963500977, 30.298179626464844, 18.504432678222656, 45.532649993896484, 1.1716270446777344, 30.522567749023438, 2.2103633880615234, 1.4792709350585938, 77.27278137207031, 43.242523193359375, 86.227783203125, -16.157379150390625, -6.4084320068359375, 18.281055450439453, -6.766559600830078, 23.187164306640625, -3.6523780822753906, -6.4287261962890625, 35.997283935546875, -5.158058166503906, 54.66748046875, 25.269935607910156, 6.897775650024414, 57.772605895996094, -1.0798492431640625, 27.407196044921875, 1.9762516021728516, 40.47676086425781, 4.40660285949707, 18.93952178955078, 67.49531555175781, 8.722877502441406, 39.13731384277344, 32.03307342529297, 38.94535827636719, -1.342926025390625, 0.6204147338867188, 5.42034912109375, 15.935466766357422, -0.0164337158203125, 27.604705810546875, 28.468711853027344, 7.268823623657227, 0.1890106201171875, -7.965019226074219, 42.9800910949707, 11.762226104736328, 15.666580200195312, 23.17565155029297, 28.984764099121094, 31.119338989257812, 32.467567443847656, 17.79492950439453, 9.437459945678711, -4.8047027587890625, 41.93821716308594, 26.33531951904297], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000092.npy"} +{"epoch": 0.13509544787077826, "step": 93, "batch_size": 64, "mean": 18.3045597076416, "std": 20.217966079711914, "min": -23.93082046508789, "p10": -1.9872312545776358, "median": 13.155380249023438, "p90": 43.99145965576172, "max": 105.44619750976562, "pos_frac": 0.875, "sample": [20.095916748046875, 7.690673828125, 18.309593200683594, 48.292930603027344, 46.84034729003906, 35.01692199707031, 5.2556610107421875, 14.737865447998047, 29.90088653564453, -3.9537582397460938, 51.14373779296875, 8.487373352050781, 37.226497650146484, 39.17180633544922, 105.44619750976562, 8.788230895996094, 16.370399475097656, 7.9356842041015625, 7.3572998046875, 62.82410430908203, -5.298530578613281, 11.246028900146484, 16.501995086669922, -1.1099014282226562, 22.435821533203125, 12.242019653320312, 15.978736877441406, 6.877037048339844, -2.363229751586914, 12.646354675292969, 0.4973468780517578, 12.889274597167969, 7.0620880126953125, 19.626445770263672, 27.286697387695312, 34.10682678222656, 45.76017761230469, -8.939910888671875, -3.3742408752441406, 33.6156005859375, 30.16211700439453, 30.382587432861328, 16.173744201660156, 4.763134002685547, 8.56485366821289, 6.11090087890625, 20.343528747558594, 2.3773059844970703, 44.089988708496094, 5.896154403686523, 0.8550567626953125, 43.761558532714844, 13.421485900878906, 9.606826782226562, 4.815990447998047, 4.772510528564453, -10.807929992675781, 30.874923706054688, 32.68220520019531, 28.98662567138672, 28.478622436523438, 12.815116882324219, -23.93082046508789, 1.670297622680664], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000093.npy"} +{"epoch": 0.13656387665198239, "step": 94, "batch_size": 64, "mean": 20.046058654785156, "std": 20.48992919921875, "min": -10.213058471679688, "p10": -4.034367561340331, "median": 16.81078338623047, "p90": 49.3896354675293, "max": 75.498291015625, "pos_frac": 0.8125, "sample": [19.902923583984375, -8.140304565429688, 33.46674346923828, 13.588798522949219, 7.708763122558594, -6.93768310546875, 6.430564880371094, 45.52263259887695, 5.158149719238281, -8.918930053710938, 21.134336471557617, 10.64129638671875, 18.44455337524414, 3.141357421875, 8.362064361572266, 0.39171600341796875, 40.15525817871094, 9.57919692993164, -4.450580596923828, 75.498291015625, 5.251420974731445, 15.177013397216797, 21.454635620117188, 41.56414794921875, 53.871002197265625, 19.784820556640625, 12.651763916015625, 21.69898223876953, 15.098773956298828, -1.1641998291015625, 0.9474353790283203, 14.393022537231445, 27.219118118286133, 36.943389892578125, 33.333980560302734, 13.827713012695312, 26.5203857421875, 47.72925567626953, 28.962448120117188, 20.612192153930664, 3.6844558715820312, 34.53941345214844, 4.889900207519531, -6.459968566894531, 50.101226806640625, 54.431915283203125, -3.063203811645508, 51.23858642578125, 65.75726318359375, -1.2610855102539062, -10.213058471679688, 68.67781066894531, 18.79930877685547, 37.095947265625, -1.3713912963867188, 11.335784912109375, -7.5366668701171875, -0.27286529541015625, 27.7440185546875, 7.557319641113281, 27.551959991455078, 36.56996154785156, 38.006683349609375, 28.587982177734375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000094.npy"} +{"epoch": 0.13803230543318648, "step": 95, "batch_size": 64, "mean": 16.312501907348633, "std": 17.811023712158203, "min": -36.86506652832031, "p10": -1.5782560348510728, "median": 13.49206256866455, "p90": 41.437361907958994, "max": 56.86180114746094, "pos_frac": 0.875, "sample": [7.5630340576171875, 27.46788787841797, 13.086219787597656, 16.1083984375, 15.077642440795898, 45.7679443359375, 8.29738998413086, 4.475011825561523, -2.9514846801757812, 17.10509490966797, 17.871185302734375, 2.7776107788085938, 21.511043548583984, 53.11689758300781, 39.30632781982422, 4.4262847900390625, 26.93963623046875, 4.390546798706055, 8.573003768920898, 26.181602478027344, 9.309066772460938, 24.212982177734375, 15.636251449584961, 11.641748428344727, 8.53862190246582, -2.79302978515625, 6.753173828125, 31.474395751953125, 16.287172317504883, 22.301467895507812, -2.9571914672851562, 1.0257720947265625, 42.35066223144531, 56.42289733886719, 56.86180114746094, 37.61407470703125, 19.3583984375, 2.597057342529297, 1.8495903015136719, 38.475799560546875, 32.4057502746582, 49.59907531738281, 51.46406555175781, 4.834403991699219, 13.248125076293945, 8.898508071899414, -0.06783294677734375, 23.68889045715332, 12.599433898925781, -36.86506652832031, 0.6918487548828125, 2.1409435272216797, -10.388397216796875, 26.138336181640625, 32.03790283203125, 12.28841781616211, 29.470062255859375, 10.474578857421875, 13.736000061035156, 14.24847412109375, -2.2255802154541016, 0.9267024993896484, 14.45452880859375, -13.850982666015625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000095.npy"} +{"epoch": 0.1395007342143906, "step": 96, "batch_size": 64, "mean": 21.81195640563965, "std": 23.026262283325195, "min": -32.77302551269531, "p10": 0.0763788223266606, "median": 17.84748649597168, "p90": 55.18889236450198, "max": 103.98396301269531, "pos_frac": 0.890625, "sample": [37.92120361328125, 19.189239501953125, 25.384979248046875, 20.023601531982422, 25.0513916015625, 41.950653076171875, 8.671646118164062, 1.9157638549804688, -0.10909843444824219, 27.124954223632812, 73.95240783691406, 10.338096618652344, 15.292633056640625, 1.3469886779785156, -0.8580474853515625, 18.023326873779297, 23.182601928710938, 35.32140350341797, 66.13489532470703, 60.0274658203125, 8.953216552734375, 29.08838653564453, 24.23675537109375, 103.98396301269531, 47.32262420654297, 13.694267272949219, 21.232986450195312, 17.671646118164062, 14.06793212890625, 14.418655395507812, 7.074382781982422, 81.95486450195312, 25.902624130249023, 21.686187744140625, 3.2562408447265625, 58.560150146484375, 33.827735900878906, 4.7830963134765625, 26.760398864746094, 2.6048412322998047, 31.57354736328125, 24.28025245666504, 12.39630126953125, -1.9592514038085938, 10.647485733032227, -0.46428489685058594, 6.877288818359375, 26.235084533691406, 0.9373207092285156, 13.866111755371094, 11.20684814453125, 34.91221618652344, -8.633569717407227, -0.651153564453125, 36.651123046875, 6.305103302001953, 13.803770065307617, 33.24104309082031, 68.56893920898438, 12.345626831054688, 0.5091590881347656, 20.951759338378906, -32.77302551269531, 4.170463562011719], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000096.npy"} +{"epoch": 0.14096916299559473, "step": 97, "batch_size": 64, "mean": 17.63809585571289, "std": 17.887022018432617, "min": -33.620094299316406, "p10": -2.8540473937988273, "median": 15.715002059936523, "p90": 38.478690719604494, "max": 73.16928100585938, "pos_frac": 0.84375, "sample": [12.6806640625, 36.97990417480469, 16.359832763671875, 7.4633941650390625, -2.0479354858398438, 5.230434417724609, 14.70077133178711, 20.60662841796875, 0.8383216857910156, 28.92254638671875, 14.72091293334961, 28.925376892089844, 0.0386962890625, 15.24409294128418, 59.13389587402344, 36.49031066894531, -4.128623962402344, 11.55914306640625, 1.5111083984375, 10.585996627807617, 20.507415771484375, 6.170051574707031, 12.224983215332031, -4.101409912109375, 48.417572021484375, 17.53690528869629, 18.748626708984375, 20.838363647460938, 32.480712890625, 11.718254089355469, 19.25365447998047, -0.9453048706054688, 10.132476806640625, -4.305673599243164, 29.874183654785156, 20.729705810546875, 37.88083267211914, 21.818424224853516, -33.620094299316406, -3.5238895416259766, 36.11756896972656, 28.930362701416016, -3.19952392578125, 4.298252105712891, 38.59446716308594, 22.780879974365234, 2.4192867279052734, 16.185911178588867, 16.422456741333008, 12.408767700195312, -1.5634441375732422, 36.02912902832031, -5.0494232177734375, 38.20854568481445, 8.237071990966797, 40.80633544921875, 2.9583663940429688, 41.69483947753906, 35.04591369628906, 43.99937438964844, 25.93079376220703, 3.0810585021972656, 73.16928100585938, 13.680557250976562], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000097.npy"} +{"epoch": 0.14243759177679882, "step": 98, "batch_size": 64, "mean": 19.38921356201172, "std": 22.25815200805664, "min": -53.513397216796875, "p10": -6.882363891601561, "median": 18.903423309326172, "p90": 47.28157196044923, "max": 69.05014038085938, "pos_frac": 0.828125, "sample": [20.59449005126953, 7.029870986938477, 52.91961669921875, 25.41053009033203, 4.311288833618164, 15.285806655883789, 15.972541809082031, 19.37329864501953, 18.433547973632812, 4.211540222167969, 21.6462459564209, 8.63861083984375, 15.634742736816406, 12.629585266113281, 48.195465087890625, 36.957130432128906, -7.563793182373047, 9.590866088867188, 4.072273254394531, -14.30206298828125, 31.575695037841797, 27.408836364746094, -11.380813598632812, -9.368743896484375, 45.14915466308594, 38.18449783325195, 26.118682861328125, 49.819801330566406, 44.38627624511719, 21.26289176940918, 23.734359741210938, 1.4221267700195312, 44.002784729003906, 69.05014038085938, -53.513397216796875, 53.22285461425781, 43.296112060546875, 19.9603271484375, 28.904129028320312, 10.534431457519531, -5.292362213134766, 30.230628967285156, -18.08233642578125, 17.230072021484375, 24.214035034179688, 2.5544509887695312, 31.601158142089844, 59.74684143066406, 41.5723876953125, 28.971752166748047, 9.126327514648438, 0.21318817138671875, 41.00236511230469, 65.2955322265625, 25.610671997070312, 16.836894989013672, 13.563148498535156, 8.06287956237793, -2.8442115783691406, -3.0585556030273438, 41.27705383300781, -1.6382369995117188, 3.0243892669677734, -11.120124816894531], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000098.npy"} +{"epoch": 0.14390602055800295, "step": 99, "batch_size": 64, "mean": 22.337358474731445, "std": 24.817840576171875, "min": -14.571784973144531, "p10": 0.31524200439453176, "median": 14.999102592468262, "p90": 50.64691772460938, "max": 106.66763305664062, "pos_frac": 0.90625, "sample": [0.8693256378173828, -1.1491622924804688, 4.229532241821289, 16.568523406982422, 4.615449905395508, 32.760379791259766, 13.462825775146484, 15.318252563476562, 70.47616577148438, 14.679952621459961, 13.706138610839844, 72.92318725585938, 13.649402618408203, -14.571784973144531, 48.72247314453125, 5.453399658203125, 11.989006042480469, 27.68292236328125, 0.8081855773925781, 9.162208557128906, 36.283782958984375, 9.383468627929688, 6.605386734008789, 106.66763305664062, 4.511510848999023, 33.649932861328125, 5.8266754150390625, -8.737174987792969, 11.165964126586914, -13.926963806152344, 0.1044464111328125, 41.29219055175781, 27.282470703125, 17.693920135498047, 38.72233581542969, 104.50724792480469, 51.4716796875, 20.133747100830078, 17.03821563720703, 12.856391906738281, 2.505756378173828, 19.183338165283203, 23.238555908203125, 45.31430435180664, 72.03926849365234, 2.6116867065429688, 11.299549102783203, 46.16340637207031, -1.6525421142578125, 3.799325942993164, 23.200546264648438, 52.18408203125, 20.42647933959961, 8.692893981933594, 35.14039611816406, 0.807098388671875, 29.661357879638672, 12.98038101196289, 43.317230224609375, 46.086631774902344, 18.171855926513672, 33.492462158203125, 2.1281471252441406, -5.090492248535156], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000099.npy"} +{"epoch": 0.14537444933920704, "step": 100, "batch_size": 64, "mean": 14.920074462890625, "std": 22.134889602661133, "min": -35.230106353759766, "p10": -6.737463188171387, "median": 12.058629989624023, "p90": 45.57449684143067, "max": 93.57876586914062, "pos_frac": 0.8125, "sample": [39.86106872558594, 26.33616828918457, -4.538860321044922, 10.925376892089844, 14.241497039794922, 4.370155334472656, 26.042829513549805, 11.442195892333984, 8.424182891845703, 13.1580810546875, 9.105308532714844, 0.6436824798583984, 24.854949951171875, 15.158008575439453, 1.579010009765625, 54.26502990722656, 15.078475952148438, -26.005172729492188, -6.842041015625, -21.874114990234375, -6.493448257446289, -3.653064727783203, 2.7339553833007812, -35.230106353759766, 42.71238327026367, 19.95520782470703, 36.82182312011719, 18.661041259765625, 0.8584251403808594, 5.535703659057617, 29.32769012451172, 1.3611679077148438, 33.842559814453125, 0.6607627868652344, 48.54447937011719, 23.7454833984375, 6.491460800170898, 10.369659423828125, 37.86663055419922, 13.455657958984375, 27.564693450927734, 11.051666259765625, 9.732025146484375, 2.1073665618896484, 20.98033905029297, 25.813034057617188, 6.403312683105469, -17.489463806152344, -3.423309326171875, 48.48896026611328, 23.78400421142578, 49.50489044189453, 19.677040100097656, -33.68196105957031, 93.57876586914062, -1.4207706451416016, 12.675064086914062, 52.22320556640625, -9.009628295898438, 46.801116943359375, 26.536331176757812, 5.377811431884766, 23.45362091064453, 10.363384246826172], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000100.npy"} +{"epoch": 0.14684287812041116, "step": 101, "batch_size": 64, "mean": 17.44969367980957, "std": 24.121967315673828, "min": -24.59949493408203, "p10": -9.0709659576416, "median": 14.65107536315918, "p90": 50.38092117309572, "max": 100.2740478515625, "pos_frac": 0.71875, "sample": [52.63282775878906, 100.2740478515625, 23.76690673828125, 21.153884887695312, -3.7876815795898438, 64.22029876708984, 69.90325927734375, 4.1862640380859375, 60.22673034667969, 68.48130798339844, -11.889556884765625, 15.912948608398438, -13.820869445800781, 34.30810546875, 9.896787643432617, -2.4077529907226562, -4.079032897949219, 10.376367568969727, 33.36597442626953, 23.68610382080078, -3.7823638916015625, 14.376617431640625, 34.39616394042969, 35.619171142578125, 18.52056884765625, 14.161943435668945, 5.301662445068359, 45.12647247314453, 67.757080078125, 15.077110290527344, 6.22705078125, -4.611392974853516, -9.443492889404297, 34.511905670166016, 15.738004684448242, 24.364959716796875, 40.16880798339844, 26.70374298095703, -2.993896484375, 7.020599365234375, -0.6707515716552734, -11.68191146850586, 14.161060333251953, 3.2335586547851562, 7.943702697753906, -5.382902145385742, 18.07146453857422, 21.252578735351562, 6.209133148193359, 17.431564331054688, -24.59949493408203, 29.045440673828125, 12.195968627929688, -13.209247589111328, 20.955036163330078, 14.925533294677734, -2.6865882873535156, 27.030567169189453, 4.03363037109375, -1.2413005828857422, 38.986602783203125, -16.17212677001953, 24.502979278564453, -8.201736450195312], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000101.npy"} +{"epoch": 0.14831130690161526, "step": 102, "batch_size": 64, "mean": 13.28602409362793, "std": 22.03500747680664, "min": -31.111175537109375, "p10": -11.06923027038574, "median": 13.829014778137207, "p90": 34.153566741943365, "max": 91.43623352050781, "pos_frac": 0.6875, "sample": [2.2234878540039062, 35.03233337402344, 29.98664093017578, 3.6754302978515625, 15.647430419921875, 3.4527664184570312, -16.147109985351562, -24.5230712890625, 18.69879722595215, -29.676788330078125, 24.174224853515625, 51.96638488769531, -0.0921173095703125, 25.47412872314453, 16.64004898071289, -3.8228626251220703, 11.026336669921875, 24.390487670898438, 31.42150115966797, -1.1721458435058594, 2.9122161865234375, -2.468221664428711, 23.829627990722656, -2.0491256713867188, 31.018463134765625, 15.289804458618164, 25.891584396362305, 91.43623352050781, -1.3249893188476562, 0.24760818481445312, 65.02396392822266, -0.1959514617919922, 27.849445343017578, 30.91270637512207, 2.8902530670166016, -7.370689392089844, 17.74162483215332, -31.111175537109375, 2.02154541015625, -14.956146240234375, 0.7388248443603516, -18.3631591796875, -11.927616119384766, 6.236228942871094, 32.103111267089844, 12.36822509765625, 23.865745544433594, 51.47682189941406, 28.057777404785156, 17.885223388671875, 45.041343688964844, -8.077949523925781, -0.04531097412109375, 9.435302734375, 40.07520294189453, -4.4351806640625, 25.98297119140625, 20.27410125732422, 16.28952980041504, 28.825218200683594, -6.2067718505859375, -9.066329956054688, 24.56780242919922, 29.239757537841797], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000102.npy"} +{"epoch": 0.14977973568281938, "step": 103, "batch_size": 64, "mean": 24.238311767578125, "std": 23.312124252319336, "min": -10.852645874023438, "p10": -1.001807212829589, "median": 21.000038146972656, "p90": 56.47448501586916, "max": 94.65411376953125, "pos_frac": 0.875, "sample": [0.7557411193847656, 5.749267578125, 10.941768646240234, 17.003013610839844, 3.759857177734375, -10.852645874023438, 28.615535736083984, 27.989761352539062, 9.320043563842773, 16.333492279052734, 20.972640991210938, 28.36850929260254, 9.224319458007812, 21.027435302734375, 0.5994796752929688, 47.24348449707031, 71.35865783691406, 24.85552978515625, 32.78990173339844, 33.207847595214844, -2.248748779296875, 5.649383544921875, 26.493392944335938, 29.802627563476562, -9.732421875, 47.979713439941406, 24.97856903076172, -9.619415283203125, 5.311515808105469, 17.781455993652344, 17.805992126464844, 85.17840576171875, 15.789695739746094, 23.231525421142578, 44.728240966796875, 31.244415283203125, -2.3991851806640625, -10.461990356445312, 15.744401931762695, 15.95530891418457, -0.0893096923828125, 23.834457397460938, 32.30213928222656, 5.13031005859375, 15.000991821289062, 73.02075958251953, 51.002899169921875, 22.50897979736328, 94.65411376953125, 63.54597473144531, 70.46215057373047, 47.31866455078125, 12.244247436523438, 21.74676513671875, 49.0032958984375, -1.3928775787353516, 26.48131561279297, 2.3504505157470703, 4.3799896240234375, 39.724769592285156, 58.81945037841797, 15.236486434936523, 19.99676513671875, 31.492612838745117], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000103.npy"} +{"epoch": 0.1512481644640235, "step": 104, "batch_size": 64, "mean": 22.85425567626953, "std": 27.10721778869629, "min": -31.403854370117188, "p10": -2.8268835067749007, "median": 15.992990493774414, "p90": 60.14978523254395, "max": 98.45657348632812, "pos_frac": 0.875, "sample": [51.23943328857422, 1.311614990234375, 8.419700622558594, 6.695213317871094, 39.718017578125, 4.2624664306640625, -12.21380615234375, 1.7550029754638672, 10.58245849609375, 25.90514373779297, -7.004190444946289, 18.701183319091797, 90.94808959960938, 38.30445098876953, 51.418479919433594, 0.48050689697265625, 61.112152099609375, 26.44087791442871, -6.783683776855469, 73.18621826171875, 18.078384399414062, 12.687042236328125, 19.080081939697266, 6.776302337646484, 11.007184982299805, 2.8513622283935547, 98.45657348632812, -17.36083221435547, 36.9505615234375, 81.90953063964844, 84.57405090332031, 18.664932250976562, 8.284690856933594, 37.44306945800781, 7.757287979125977, 6.838813781738281, -3.495655059814453, -9.050857543945312, 58.968505859375, 12.412925720214844, 38.97952651977539, 7.21893310546875, 45.91474914550781, 7.317108154296875, 60.65604782104492, 32.803504943847656, 23.943138122558594, 28.718730926513672, 15.916458129882812, 29.616378784179688, 1.2065982818603516, 53.52268981933594, 8.232810974121094, 16.59423065185547, 49.08186340332031, 46.368995666503906, -1.2664165496826172, 23.256391525268555, 0.029163360595703125, 16.069522857666016, 0.8360748291015625, 2.624645233154297, -31.403854370117188, 9.121780395507812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000104.npy"} +{"epoch": 0.1527165932452276, "step": 105, "batch_size": 64, "mean": 31.581022262573242, "std": 24.970497131347656, "min": -19.6710205078125, "p10": 2.552611923217776, "median": 30.048053741455078, "p90": 66.30653381347656, "max": 118.79176330566406, "pos_frac": 0.953125, "sample": [26.998153686523438, 23.907791137695312, 36.221946716308594, 26.088760375976562, 31.789390563964844, 26.35039520263672, 47.103424072265625, 74.18004608154297, 11.772832870483398, -12.228836059570312, 1.2708206176757812, 55.647857666015625, 44.59480285644531, 11.244813919067383, 30.68067169189453, 31.850360870361328, -19.6710205078125, 7.687278747558594, 1.2429656982421875, 66.96722412109375, 22.30394744873047, 70.96513366699219, 67.02070617675781, 12.809837341308594, 10.007568359375, 54.465904235839844, 41.452239990234375, 25.78807830810547, 6.3563385009765625, 31.21649932861328, 36.08428955078125, 90.01889038085938, 11.224227905273438, 35.16065979003906, 11.412303924560547, 16.182899475097656, 42.49720001220703, 60.43736267089844, 29.415435791015625, 13.091983795166016, 37.09888458251953, 71.10069274902344, 24.61488151550293, 25.73518180847168, 16.041900634765625, 36.97328186035156, 1.3557624816894531, 46.31256103515625, 48.431827545166016, 53.383087158203125, 20.68557357788086, 19.380630493164062, -4.6515655517578125, 5.3452606201171875, 38.86724853515625, 38.43158721923828, 34.97401428222656, 64.76492309570312, 37.31352233886719, 118.79176330566406, 12.617630004882812, 0.7799224853515625, 9.935722351074219, 51.28999328613281], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000105.npy"} +{"epoch": 0.15418502202643172, "step": 106, "batch_size": 64, "mean": 20.225627899169922, "std": 21.619686126708984, "min": -26.936073303222656, "p10": -4.063160896301266, "median": 19.78085708618164, "p90": 49.98634872436524, "max": 96.2394027709961, "pos_frac": 0.84375, "sample": [19.17896270751953, 2.981189727783203, 3.4836349487304688, 0.2690887451171875, 16.151538848876953, -0.7242259979248047, 35.861595153808594, -5.494132995605469, 24.82738494873047, 45.70217514038086, 28.69733428955078, 26.66086196899414, 3.8336315155029297, 54.27191162109375, 16.664276123046875, 52.14508056640625, 1.2780017852783203, 12.392509460449219, 39.22612762451172, -0.3661479949951172, 47.1888427734375, 23.253376007080078, 4.566535949707031, 51.452880859375, 15.448667526245117, 27.151512145996094, 10.821388244628906, -10.398284912109375, 27.649559020996094, -12.411735534667969, 29.28293800354004, 15.207221984863281, 25.083560943603516, 29.858680725097656, 1.3858299255371094, 8.330860137939453, 17.488832473754883, 3.7698822021484375, 53.52964782714844, 51.185279846191406, 30.315078735351562, 20.38275146484375, -26.936073303222656, 0.37155914306640625, 96.2394027709961, 18.275611877441406, 21.765361785888672, 31.065826416015625, 13.100051879882812, 5.454139709472656, 2.6219635009765625, 31.992813110351562, 65.6867904663086, 27.140016555786133, 34.819671630859375, 34.62701416015625, -7.3169708251953125, 22.097808837890625, -0.653778076171875, 24.709579467773438, 44.50794982910156, -15.514434814453125, -8.253242492675781, 31.055030822753906], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000106.npy"} +{"epoch": 0.15565345080763582, "step": 107, "batch_size": 64, "mean": 22.89925765991211, "std": 23.956438064575195, "min": -36.90065002441406, "p10": -4.423463821411129, "median": 16.02847671508789, "p90": 55.92364463806153, "max": 82.80653381347656, "pos_frac": 0.875, "sample": [45.923797607421875, 63.03315734863281, 30.7869873046875, 47.14222717285156, 16.01665496826172, 19.258193969726562, 33.45172119140625, -0.4079246520996094, -36.90065002441406, 3.853628158569336, 11.752546310424805, 48.66877746582031, -12.661231994628906, 25.380992889404297, 8.72207260131836, 8.725456237792969, 9.941656112670898, 59.79912567138672, 1.8038349151611328, -11.816455841064453, 56.619728088378906, 45.68074035644531, 50.9918212890625, 3.7200374603271484, -6.1444091796875, 5.0450439453125, 5.610542297363281, 4.6776580810546875, 82.80653381347656, 43.08885955810547, 19.709705352783203, 15.998268127441406, 52.01280975341797, 24.32080841064453, 41.07350158691406, 63.15752410888672, 12.352128982543945, 19.329360961914062, 9.147848129272461, -10.926071166992188, 38.89237976074219, -21.587936401367188, 40.393470764160156, 54.2994499206543, 12.400970458984375, 56.93689727783203, 5.555747985839844, 60.06489562988281, 39.47331237792969, 34.89231872558594, 7.6217193603515625, 11.892904281616211, 15.3304443359375, -8.75799560546875, 8.676948547363281, 5.58367919921875, 9.769371032714844, 45.815887451171875, 16.040298461914062, 22.07422637939453, 37.47845458984375, 45.473388671875, 5.336181640625, 11.148574829101562], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000107.npy"} +{"epoch": 0.15712187958883994, "step": 108, "batch_size": 64, "mean": 24.657085418701172, "std": 31.04875373840332, "min": -25.988018035888672, "p10": -8.773519897460938, "median": 19.635475158691406, "p90": 71.36209487915039, "max": 106.261474609375, "pos_frac": 0.765625, "sample": [25.004486083984375, 73.24278259277344, 27.96422576904297, -17.856048583984375, -23.964996337890625, 54.399810791015625, 22.4344482421875, 4.8252716064453125, 106.261474609375, -2.5728836059570312, 4.149688720703125, 75.36811065673828, 4.397369384765625, 0.5215988159179688, -0.2677803039550781, 38.743919372558594, -2.147092819213867, 51.71051025390625, 84.20576477050781, 70.51793670654297, 71.723876953125, 45.160064697265625, -1.1944866180419922, -14.146158218383789, 26.431055068969727, 8.590185165405273, 18.918983459472656, 11.885608673095703, 57.211700439453125, 6.201930999755859, 11.228374481201172, -0.6204414367675781, 52.637786865234375, -8.724578857421875, 103.0811767578125, -8.79449462890625, 50.39466094970703, -12.75191879272461, -6.3073883056640625, 3.9494152069091797, 21.67139434814453, 8.18292236328125, 28.19208526611328, 78.513427734375, 21.33612823486328, 5.331939697265625, 52.047584533691406, 40.63099670410156, 16.869619369506836, -23.416366577148438, 7.006019592285156, 37.777565002441406, 69.80776977539062, 34.167686462402344, 3.379486083984375, -2.053384780883789, 33.86095428466797, 12.295244216918945, 49.28862762451172, -25.988018035888672, 20.351966857910156, 36.7963752746582, 27.617965698242188, 12.571544647216797], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000108.npy"} +{"epoch": 0.15859030837004406, "step": 109, "batch_size": 64, "mean": 27.51488494873047, "std": 25.898794174194336, "min": -15.993053436279297, "p10": -3.7237657546997065, "median": 24.35595703125, "p90": 62.97060623168946, "max": 98.60818481445312, "pos_frac": 0.84375, "sample": [35.32129669189453, 4.665241241455078, 63.86363220214844, 49.518890380859375, 11.47174072265625, 37.17960739135742, -13.286956787109375, 12.057254791259766, 11.704330444335938, 36.147247314453125, 70.49462890625, 23.731319427490234, 42.495758056640625, 23.881484985351562, 40.20838928222656, 6.467260360717773, 18.310640335083008, 28.75419044494629, -4.515220642089844, -3.919431686401367, -2.2851638793945312, 98.60818481445312, -0.7245121002197266, -5.3489990234375, 22.96175193786621, 3.305805206298828, 76.39942932128906, 24.26712417602539, 12.795278549194336, -15.993053436279297, 33.35777282714844, 41.779563903808594, 4.409460067749023, 30.80859375, 39.7469482421875, 26.785551071166992, 54.224700927734375, 49.719825744628906, 60.886878967285156, 65.5066146850586, 7.494544982910156, 7.542076110839844, -11.104736328125, 19.705726623535156, 2.497772216796875, 48.11396026611328, 29.550079345703125, 27.573261260986328, 16.763212203979492, 56.67010498046875, -13.677459716796875, 25.3406982421875, 24.44478988647461, 65.48493957519531, 75.84226989746094, 15.395721435546875, 56.427650451660156, 19.090621948242188, 54.25495910644531, 59.92504119873047, 0.182525634765625, -3.2672119140625, 6.672313690185547, 54.266761779785156], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000109.npy"} +{"epoch": 0.16005873715124816, "step": 110, "batch_size": 64, "mean": 25.2530517578125, "std": 29.23346710205078, "min": -48.30418395996094, "p10": -1.123815059661865, "median": 21.407949447631836, "p90": 64.96130142211916, "max": 107.66473388671875, "pos_frac": 0.875, "sample": [51.729248046875, 25.62872314453125, 22.04092788696289, 40.33258056640625, 34.326210021972656, 25.48944854736328, 30.356342315673828, 16.766027450561523, -48.30418395996094, 26.403961181640625, 36.99156188964844, 12.174423217773438, -13.803482055664062, 12.403100967407227, 13.150920867919922, 0.12286186218261719, 20.77497100830078, 107.66473388671875, 69.08309936523438, 17.973007202148438, 3.2179222106933594, -1.2363090515136719, 104.45791625976562, 73.62130737304688, 57.787506103515625, 67.87899780273438, 7.843940734863281, 7.685081481933594, 0.5198745727539062, 44.58892059326172, -15.847412109375, -10.113670349121094, 5.839691162109375, 14.33369255065918, 4.183929443359375, 90.30795288085938, 32.30015563964844, 58.153343200683594, 52.9346923828125, 2.102458953857422, 43.304298400878906, -0.8613290786743164, 14.819103240966797, 29.703773498535156, 12.163871765136719, 23.024314880371094, 48.24492645263672, -33.23565673828125, 39.84881591796875, -9.058441162109375, 19.22740936279297, 1.7633228302001953, 30.125831604003906, 19.584522247314453, 36.319461822509766, 25.98401641845703, 16.92432975769043, 22.302093505859375, 12.786554336547852, 36.868370056152344, 35.651458740234375, 1.2194271087646484, 7.631862640380859, 79.988525390625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000110.npy"} +{"epoch": 0.16152716593245228, "step": 111, "batch_size": 64, "mean": 26.988754272460938, "std": 30.760452270507812, "min": -43.498504638671875, "p10": -8.643358612060544, "median": 23.711994171142578, "p90": 67.39702301025393, "max": 121.2945556640625, "pos_frac": 0.8125, "sample": [121.2945556640625, -17.12628173828125, -43.498504638671875, 51.198944091796875, -4.873931884765625, 30.69744110107422, 4.226722717285156, 22.296279907226562, 12.912195205688477, 78.39828491210938, 20.730819702148438, 50.64915466308594, 26.428455352783203, 31.196823120117188, 19.152664184570312, 47.93061828613281, 69.80980682373047, 2.96881103515625, 57.366943359375, 7.84423828125, 40.84077453613281, -10.258827209472656, -4.353496551513672, 1.03076171875, 39.04939270019531, 62.10420227050781, 7.885444641113281, 61.191680908203125, -12.406280517578125, 9.418075561523438, -1.9254150390625, 21.885881423950195, 69.66537475585938, 14.193679809570312, 82.67457580566406, 26.519683837890625, 18.92376708984375, 32.45552444458008, 39.37071990966797, 85.02967071533203, 16.7823486328125, -12.37420654296875, 13.665445327758789, 16.772480010986328, 83.79412841796875, 28.754470825195312, 23.00439453125, 24.712478637695312, 58.541259765625, 1.9990406036376953, 62.03215026855469, 26.925064086914062, 53.00750732421875, 50.00303649902344, 30.179977416992188, 6.692378997802734, -29.110809326171875, 24.419593811035156, 16.046918869018555, 30.013397216796875, -3.2492237091064453, 52.112884521484375, -17.51030731201172, -2.8333053588867188], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000111.npy"} +{"epoch": 0.16299559471365638, "step": 112, "batch_size": 64, "mean": 18.578956604003906, "std": 23.828439712524414, "min": -37.089134216308594, "p10": -8.418171691894528, "median": 15.188496589660645, "p90": 56.21775360107423, "max": 74.04031372070312, "pos_frac": 0.8125, "sample": [22.896007537841797, 74.04031372070312, 44.19325256347656, -3.5941905975341797, -19.493629455566406, 17.6734676361084, 22.621204376220703, 15.806137084960938, 9.373931884765625, -6.312980651855469, 65.371826171875, 53.38755798339844, 1.8668575286865234, 1.5219573974609375, 57.430694580078125, 43.679134368896484, 27.355052947998047, 38.813018798828125, 28.81359100341797, 22.903564453125, 34.017181396484375, 33.24658203125, 43.88340759277344, 1.3419914245605469, 18.59136962890625, 21.137296676635742, 21.62310791015625, 10.200023651123047, -12.457496643066406, 14.672256469726562, 3.320558547973633, -15.56180191040039, 15.612640380859375, 64.4172134399414, 71.82476043701172, -13.462051391601562, 26.17229461669922, 11.8675537109375, 41.53877258300781, 4.687221527099609, -37.089134216308594, 61.010406494140625, 10.646549224853516, 6.4982147216796875, 0.874267578125, 10.414173126220703, 34.5753173828125, 3.1596298217773438, 15.130805969238281, 57.56117248535156, 8.428611755371094, 4.831043243408203, 31.14378547668457, -12.029388427734375, 5.148187637329102, -9.320396423339844, -5.8450469970703125, -2.003559112548828, 43.70891571044922, 0.9270401000976562, 15.246187210083008, 5.9287872314453125, -2.9853363037109375, 28.07327651977539], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000112.npy"} +{"epoch": 0.1644640234948605, "step": 113, "batch_size": 64, "mean": 20.957496643066406, "std": 21.205978393554688, "min": -15.378616333007812, "p10": -5.094252014160156, "median": 19.849526405334473, "p90": 48.63909416198731, "max": 65.76575469970703, "pos_frac": 0.8125, "sample": [-15.378616333007812, 43.532562255859375, -7.295278549194336, 47.849029541015625, 12.140670776367188, 17.859359741210938, 18.373430252075195, 30.39745330810547, 10.50700569152832, 3.5550994873046875, 5.161338806152344, 40.87474060058594, 35.65398406982422, 61.851898193359375, 21.32562255859375, -10.380929946899414, 6.215507507324219, -3.4777450561523438, 12.656356811523438, 55.06562805175781, -9.822158813476562, 28.073020935058594, 44.84196472167969, -2.667266845703125, 33.06037139892578, 7.727569580078125, 8.831829071044922, 65.76575469970703, 25.16027069091797, -10.579185485839844, 9.621208190917969, 56.480194091796875, 23.041339874267578, -4.507408142089844, 23.90814208984375, 38.85419464111328, 9.583702087402344, 47.563446044921875, 33.41368865966797, 4.44573974609375, 33.501792907714844, 44.98845672607422, -3.553804397583008, 3.041015625, -12.569183349609375, 33.694644927978516, 22.248573303222656, 23.486282348632812, -5.345756530761719, 6.140527725219727, 64.12873840332031, 3.3661460876464844, 24.375049591064453, 34.48230743408203, 55.446922302246094, 48.88127517700195, 26.048721313476562, 48.074005126953125, -1.2404956817626953, 9.259078979492188, 8.299652099609375, 8.546894073486328, 16.203346252441406, 30.49212646484375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000113.npy"} +{"epoch": 0.16593245227606462, "step": 114, "batch_size": 64, "mean": 30.63623809814453, "std": 35.19435119628906, "min": -16.83875846862793, "p10": -6.049532127380371, "median": 28.81706714630127, "p90": 61.54848861694336, "max": 182.393798828125, "pos_frac": 0.78125, "sample": [96.41690063476562, 61.86023712158203, 39.51312255859375, 51.18498992919922, 15.433357238769531, 79.36940002441406, 9.876880645751953, 29.59429931640625, -2.8576812744140625, 35.071922302246094, -16.83875846862793, 34.088409423828125, 39.44226837158203, 4.307838439941406, -12.758049011230469, 47.48176574707031, 7.657070159912109, -5.955921173095703, 24.838985443115234, 60.821075439453125, 13.939960479736328, 27.045042037963867, -7.7911529541015625, 65.5182113647461, 38.39286804199219, 45.3211669921875, 19.04345703125, 17.838924407958984, 14.4697265625, 41.45896530151367, 43.45793151855469, -0.372406005859375, -2.8231353759765625, 41.06785583496094, -6.089651107788086, 8.103082656860352, 22.205080032348633, 34.20832824707031, 46.42271423339844, 48.876373291015625, 40.65134811401367, 50.34119415283203, 3.7537612915039062, 28.129899978637695, 21.89548110961914, -0.0045623779296875, 141.91136169433594, 3.847341537475586, 29.504234313964844, 95.27413177490234, 1.5442428588867188, 46.93440246582031, -8.536872863769531, -7.950542449951172, 182.393798828125, 46.09836959838867, -0.5736846923828125, 39.797271728515625, 32.30476379394531, -16.185428619384766, 14.64385986328125, 57.13978576660156, -0.5919151306152344, 49.55546569824219], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000114.npy"} +{"epoch": 0.16740088105726872, "step": 115, "batch_size": 64, "mean": 22.12807273864746, "std": 33.63416290283203, "min": -62.191131591796875, "p10": -6.801375770568847, "median": 14.208854675292969, "p90": 75.30417404174807, "max": 135.60556030273438, "pos_frac": 0.765625, "sample": [3.6142730712890625, -2.9851036071777344, 76.96849060058594, 29.3145751953125, -8.899612426757812, 2.0688095092773438, -9.632484436035156, 9.054569244384766, 4.3526763916015625, 17.783340454101562, 12.746932983398438, -5.053253173828125, 32.444007873535156, 34.17042541503906, 25.284326553344727, 93.84352111816406, 18.458131790161133, 1.6615753173828125, 135.60556030273438, -6.985219955444336, 38.21265411376953, 2.2063064575195312, 77.24369812011719, 40.968605041503906, 86.50454711914062, -4.079887390136719, 23.427902221679688, 28.383464813232422, 3.656370162963867, 48.78691101074219, 92.82487487792969, 52.553863525390625, 10.529670715332031, -58.73316192626953, 6.887615203857422, 26.073043823242188, 33.316490173339844, 12.193361282348633, 12.700712203979492, 29.029701232910156, 20.094985961914062, 21.516647338867188, 24.674354553222656, 13.857654571533203, 83.16546630859375, -6.372406005859375, 12.510734558105469, 7.868865966796875, -1.0462799072265625, 7.409828186035156, 14.560054779052734, 46.476898193359375, -0.44971466064453125, 22.33303451538086, -10.893770217895508, 71.42076873779297, 44.674705505371094, -3.569398880004883, 43.14628601074219, -62.191131591796875, 50.39312744140625, -9.314613342285156, 5.337717056274414, -5.9095306396484375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000115.npy"} +{"epoch": 0.16886930983847284, "step": 116, "batch_size": 64, "mean": 23.02229881286621, "std": 24.178129196166992, "min": -14.026798248291016, "p10": -3.473207283020019, "median": 18.192030906677246, "p90": 59.38792114257813, "max": 128.3798065185547, "pos_frac": 0.828125, "sample": [7.130769729614258, 3.295787811279297, -14.026798248291016, -5.366264343261719, 18.768285751342773, 25.443313598632812, 25.914077758789062, 16.325843811035156, 32.80604553222656, 35.635292053222656, 14.707839965820312, 2.376222610473633, 24.64514923095703, 20.33001708984375, 2.7755165100097656, -4.3856353759765625, 29.27582550048828, -1.7930145263671875, 45.60993957519531, 57.78175354003906, 21.88848114013672, 10.692626953125, 13.169097900390625, 5.751441955566406, 29.595748901367188, 15.13818359375, 32.903228759765625, 62.529693603515625, 11.857170104980469, -2.8606338500976562, 68.38764953613281, 28.39067840576172, 15.838310241699219, 23.350509643554688, -3.735738754272461, -10.096435546875, 15.582088470458984, 14.792579650878906, 8.613943099975586, 61.9989013671875, 3.397357940673828, 62.45521545410156, 17.057418823242188, 44.618896484375, 37.05833435058594, 12.462757110595703, 41.363502502441406, -1.545562744140625, -4.226356506347656, 42.964088439941406, 23.85245132446289, -4.875614166259766, 128.3798065185547, 35.64009094238281, 17.61577606201172, 27.734718322753906, 30.23772430419922, 30.190311431884766, -1.2594451904296875, 9.431800842285156, 0.975677490234375, 37.884552001953125, 60.07627868652344, 60.89979553222656], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000116.npy"} +{"epoch": 0.17033773861967694, "step": 117, "batch_size": 64, "mean": 21.231441497802734, "std": 26.494613647460938, "min": -38.673309326171875, "p10": -3.04555206298828, "median": 19.07328701019287, "p90": 55.94088211059574, "max": 112.11105346679688, "pos_frac": 0.84375, "sample": [12.542232513427734, 29.478471755981445, 28.670562744140625, 4.942569732666016, 15.106685638427734, 5.525856018066406, 18.72151756286621, 13.074928283691406, 12.524948120117188, 23.736669540405273, 31.662384033203125, 43.966766357421875, 61.95111083984375, 91.35464477539062, 24.05498504638672, 20.277587890625, -17.054824829101562, 43.75640869140625, 30.814422607421875, -38.673309326171875, -12.269989013671875, 7.087066650390625, 15.359966278076172, 36.454647064208984, 33.509159088134766, 2.89581298828125, 5.915863037109375, 1.885049819946289, 23.861839294433594, 0.5788803100585938, 24.404273986816406, 70.92355346679688, -3.4685897827148438, 27.8145751953125, 71.50403594970703, 13.563228607177734, 112.11105346679688, 16.805831909179688, 11.731876373291016, 35.24755096435547, 6.249210357666016, 2.2432823181152344, 5.676435470581055, -2.0584640502929688, -37.70762634277344, 15.132247924804688, 47.788177490234375, 23.851898193359375, 59.434898376464844, 23.186548233032227, -15.855697631835938, -0.9536705017089844, 5.417903900146484, 36.50192642211914, 62.615882873535156, 43.704620361328125, 7.437959671020508, 31.19165802001953, 31.17864227294922, 22.832122802734375, 19.42505645751953, -0.4074554443359375, -12.660324096679688, 32.2366943359375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000117.npy"} +{"epoch": 0.17180616740088106, "step": 118, "batch_size": 64, "mean": 22.107864379882812, "std": 23.960716247558594, "min": -18.210128784179688, "p10": -2.4709346771240215, "median": 15.931841850280762, "p90": 51.76665573120118, "max": 88.39950561523438, "pos_frac": 0.859375, "sample": [-3.3052902221679688, 12.747306823730469, 6.192628860473633, -11.90216064453125, 15.748559951782227, 15.327033996582031, 41.81327438354492, 31.650712966918945, 35.029632568359375, 52.5992431640625, 47.175628662109375, 15.9178466796875, 69.81549072265625, 19.59714698791504, -0.2240142822265625, 28.67487144470215, 1.0388984680175781, 1.5323486328125, 21.44442367553711, 88.39950561523438, 56.69915771484375, -3.790761947631836, -15.092788696289062, 4.4081268310546875, -18.210128784179688, 14.677255630493164, 3.6537418365478516, 18.017995834350586, 2.187122344970703, 4.6600799560546875, 10.54262924194336, 31.530166625976562, -7.988960266113281, 47.9752197265625, 15.455360412597656, 25.97832489013672, 15.945837020874023, 21.613548278808594, 77.38131713867188, 2.0727500915527344, 12.052608489990234, -6.578849792480469, 6.633995056152344, 22.651290893554688, 74.44781494140625, 45.010963439941406, 79.4875717163086, 33.34510040283203, 49.823951721191406, 11.902772903442383, 2.544506072998047, 35.42787170410156, 46.1602783203125, 37.19817352294922, 30.61559295654297, 20.012588500976562, 24.15186882019043, 9.923080444335938, 2.771800994873047, 23.34636688232422, 9.041191101074219, 48.18272399902344, 0.28515625, -0.5241050720214844], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000118.npy"} +{"epoch": 0.17327459618208516, "step": 119, "batch_size": 64, "mean": 22.46298599243164, "std": 28.279319763183594, "min": -25.472152709960938, "p10": -9.353214263916014, "median": 19.30974292755127, "p90": 59.73944892883301, "max": 112.32621765136719, "pos_frac": 0.8125, "sample": [18.82926368713379, 23.00311279296875, 13.456298828125, 76.45980834960938, 35.5228271484375, -4.0385284423828125, 1.3531646728515625, 31.832870483398438, -5.370513916015625, 38.83258819580078, 5.090087890625, 1.6234512329101562, 103.57037353515625, 35.81769561767578, 4.07427978515625, 69.5621337890625, 17.731327056884766, 9.131301879882812, 112.32621765136719, 34.00225830078125, 37.23472595214844, 43.967445373535156, 42.72200012207031, 4.516330718994141, 18.75177764892578, 24.619369506835938, -19.008621215820312, 8.042854309082031, 35.072265625, 34.64299011230469, 5.8781890869140625, -7.0077667236328125, 59.028076171875, 18.709125518798828, 78.83029174804688, -10.358406066894531, 33.769798278808594, 28.870033264160156, 4.553792953491211, 12.171760559082031, -3.707763671875, 19.814620971679688, -18.4608154296875, 68.07791900634766, 23.2674560546875, 34.08637237548828, 3.6354827880859375, -16.098739624023438, 47.315635681152344, 29.560901641845703, -16.74303436279297, -25.472152709960938, 10.843971252441406, 38.83905029296875, 22.71166229248047, 30.931964874267578, 30.43445587158203, 7.398885726928711, 60.0443229675293, -1.8797492980957031, -24.106426239013672, 4.4228363037109375, 19.79022216796875, 15.107963562011719], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000119.npy"} +{"epoch": 0.17474302496328928, "step": 120, "batch_size": 64, "mean": 29.621591567993164, "std": 30.679790496826172, "min": -27.02978515625, "p10": -7.253680419921868, "median": 25.559982299804688, "p90": 77.37682495117188, "max": 106.48202514648438, "pos_frac": 0.875, "sample": [91.35191345214844, 45.143096923828125, 25.3321533203125, 68.64884948730469, 24.010417938232422, 12.917417526245117, -17.98822021484375, 64.93023681640625, 15.914300918579102, 39.09349822998047, 41.909278869628906, 90.99392700195312, 44.238258361816406, 21.500823974609375, 8.190399169921875, 14.159591674804688, 64.75863647460938, 23.840232849121094, 36.70458221435547, -17.02985382080078, -24.06414794921875, 80.293701171875, 25.787811279296875, 29.60137176513672, 28.02680206298828, 15.783683776855469, 22.21875, 10.234363555908203, 79.07623291015625, -27.02978515625, -10.19873046875, 36.847991943359375, 29.518043518066406, 3.0727767944335938, 61.34843444824219, 2.7847862243652344, 101.0142822265625, 37.67500686645508, 37.90992736816406, 1.490234375, 1.9181404113769531, 3.8260860443115234, 40.57378387451172, 76.73223876953125, 22.227996826171875, 29.3079833984375, 77.653076171875, 106.48202514648438, 18.8106689453125, 28.992515563964844, 0.42826080322265625, 44.143089294433594, -12.925155639648438, 17.535842895507812, 31.65550994873047, 11.597061157226562, 4.916553497314453, 55.242584228515625, 34.61625671386719, -0.38189697265625, 46.53887176513672, -10.521800994873047, 22.776016235351562, 3.6251754760742188], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000120.npy"} +{"epoch": 0.1762114537444934, "step": 121, "batch_size": 64, "mean": 32.329833984375, "std": 35.23461151123047, "min": -38.066192626953125, "p10": -3.9701419830322267, "median": 29.961124420166016, "p90": 85.15222473144533, "max": 119.83024597167969, "pos_frac": 0.828125, "sample": [4.645046234130859, 4.476112365722656, 0.8394927978515625, 110.44770812988281, -3.885234832763672, 5.402442932128906, 49.5322265625, -4.333759307861328, 60.89170837402344, 43.39160919189453, 9.543121337890625, 38.413551330566406, 80.57496643066406, 70.94275665283203, 36.62348937988281, -7.312156677246094, 41.78755187988281, 44.30475997924805, 40.65336608886719, -11.510137557983398, 17.88441276550293, 76.23553466796875, -18.87238311767578, 56.99491882324219, 89.2041015625, 32.96797180175781, 18.327987670898438, 5.224880218505859, 67.10578918457031, 5.243371963500977, 36.4930419921875, -6.019584655761719, 36.813255310058594, 35.73443603515625, 6.420995712280273, 89.15890502929688, 4.269218444824219, 93.48696899414062, 24.855812072753906, 118.94773864746094, 2.7291641235351562, 75.02266693115234, -3.697021484375, 8.715171813964844, -0.4147186279296875, 18.89319610595703, 55.47097396850586, 9.502151489257812, 10.068954467773438, 29.503013610839844, 6.675548553466797, 7.385078430175781, 37.61701965332031, 66.15234375, 87.11390686035156, 52.04486846923828, -38.066192626953125, 34.847869873046875, -4.00653076171875, 47.630218505859375, 119.83024597167969, 13.250104904174805, -3.4839954376220703, 30.419235229492188], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000121.npy"} +{"epoch": 0.1776798825256975, "step": 122, "batch_size": 64, "mean": 40.240089416503906, "std": 33.22461700439453, "min": -26.074356079101562, "p10": 6.9684442520141605, "median": 33.99272155761719, "p90": 83.07159118652345, "max": 120.84577941894531, "pos_frac": 0.921875, "sample": [70.46763610839844, 9.611946105957031, 28.63397979736328, 52.523338317871094, 13.396554946899414, 22.157981872558594, 48.049346923828125, 58.25568389892578, 21.005172729492188, 62.82417297363281, 10.198326110839844, 65.24707794189453, 65.06298065185547, 118.2984619140625, 67.09896087646484, 28.56561279296875, 10.8125, 47.05426025390625, 80.06185913085938, 38.385498046875, 7.020193099975586, 81.22146606445312, 16.602466583251953, 7.94866943359375, 31.839401245117188, 21.12541961669922, 31.53478240966797, 77.8157958984375, 46.17217254638672, 43.36176300048828, 49.365318298339844, -10.441871643066406, 102.9496078491211, -2.069784164428711, 32.244873046875, 83.864501953125, 41.097076416015625, 34.449134826660156, 67.92626190185547, 90.947265625, 8.635042190551758, 35.38385009765625, -14.260953903198242, 120.84577941894531, 24.397010803222656, 28.31185531616211, 23.421470642089844, 15.12027359008789, 34.673133850097656, -26.074356079101562, -3.4126129150390625, 18.50994873046875, 10.492456436157227, 59.01545715332031, 117.48867797851562, 7.889003753662109, 66.62376403808594, 33.53630828857422, 2.4035682678222656, 6.946266174316406, 90.31192016601562, 46.34858703613281, 18.606796264648438, 77.46641540527344], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000122.npy"} +{"epoch": 0.17914831130690162, "step": 123, "batch_size": 64, "mean": 27.071605682373047, "std": 37.76756286621094, "min": -75.4800796508789, "p10": -10.66026611328125, "median": 21.266407012939453, "p90": 76.76828536987306, "max": 138.21096801757812, "pos_frac": 0.78125, "sample": [40.53504180908203, 46.089874267578125, 2.4540481567382812, 5.8955841064453125, 34.456443786621094, 34.88671875, 56.49501037597656, 51.910858154296875, 45.582557678222656, 32.5577392578125, 1.1378707885742188, 9.814834594726562, 12.147441864013672, -10.014518737792969, 96.07844543457031, 45.39447021484375, 14.618366241455078, -1.0857467651367188, -1.7473602294921875, 32.015380859375, 1.2786712646484375, 82.19680786132812, 34.15052795410156, -39.382659912109375, 19.87934112548828, 48.53996276855469, 138.21096801757812, 11.938003540039062, 75.20370483398438, 79.49090576171875, -12.212064743041992, 54.22930908203125, 18.21198272705078, 6.2393341064453125, 55.454200744628906, 68.75656127929688, 22.653472900390625, 37.94336700439453, 103.61949157714844, -10.549919128417969, 43.49609375, 38.259300231933594, -22.885608673095703, -14.446250915527344, 46.15592956542969, 2.476715087890625, -24.306732177734375, -0.8318729400634766, -10.707557678222656, 126.58291625976562, -8.387306213378906, 28.852615356445312, 24.469886779785156, 77.4388198852539, 7.41314697265625, 17.619876861572266, 70.02680969238281, 25.17487144470215, -75.4800796508789, 17.901504516601562, 9.537117004394531, -1.7398414611816406, 0.9975681304931641, 9.889921188354492], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000123.npy"} +{"epoch": 0.18061674008810572, "step": 124, "batch_size": 64, "mean": 37.067848205566406, "std": 47.40776824951172, "min": -45.247711181640625, "p10": -10.89631118774414, "median": 27.020426750183105, "p90": 91.90823516845704, "max": 210.29086303710938, "pos_frac": 0.84375, "sample": [74.89996337890625, -9.581787109375, 31.2967529296875, 24.478424072265625, 3.907367706298828, 87.12791442871094, -45.247711181640625, -4.722799301147461, 71.54837036132812, 105.46638488769531, 2.4294891357421875, 22.693504333496094, 3.9271507263183594, 35.162025451660156, 33.65221405029297, 147.93954467773438, 77.36062622070312, 25.663394927978516, 7.399971008300781, 46.38921356201172, -36.35113525390625, -19.257156372070312, 74.98190307617188, 5.636241912841797, 114.91671752929688, 76.63188171386719, 41.45329284667969, 7.048004150390625, 38.413604736328125, 63.364906311035156, -11.459678649902344, 27.06458282470703, 30.103302001953125, 26.97627067565918, 4.707275390625, 167.43701171875, 59.46223831176758, 93.10081481933594, 13.496528625488281, 83.75735473632812, 55.521629333496094, -13.764381408691406, 24.022945404052734, -8.985305786132812, 28.41624641418457, 64.3245849609375, -34.52435302734375, 26.337738037109375, 14.575582504272461, 16.77606201171875, 27.280155181884766, 32.713172912597656, -22.183212280273438, 108.94003295898438, 19.689605712890625, 210.29086303710938, 13.710968017578125, 89.12554931640625, 2.391571044921875, 21.896888732910156, 32.78997802734375, 35.654701232910156, 10.26845932006836, 13.798763275146484], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000124.npy"} +{"epoch": 0.18208516886930984, "step": 125, "batch_size": 64, "mean": 34.27327346801758, "std": 35.13113021850586, "min": -25.636573791503906, "p10": -0.5164741516113275, "median": 20.904332160949707, "p90": 83.0122039794922, "max": 125.68569946289062, "pos_frac": 0.890625, "sample": [0.4978065490722656, 54.81559753417969, 33.793304443359375, 8.271865844726562, 7.5350799560546875, 8.960369110107422, 61.32014465332031, -5.382194519042969, 83.48193359375, 18.41563606262207, 2.0750656127929688, -15.195144653320312, 15.174901962280273, 14.662277221679688, 56.26659393310547, 48.9176025390625, 39.98887634277344, 34.91681671142578, 8.651554107666016, 9.86606216430664, 80.32203674316406, 13.369400024414062, -3.7103729248046875, 14.889373779296875, 58.83306884765625, 125.68569946289062, -20.416748046875, 20.758432388305664, 42.31336212158203, 12.098312377929688, 11.166374206542969, -25.636573791503906, 21.10538101196289, 8.587589263916016, 20.727458953857422, 111.48162841796875, 3.376239776611328, 104.60759735107422, 69.17800903320312, 0.0967254638671875, 7.749320983886719, 69.27291870117188, 81.91616821289062, 58.63947296142578, 99.9283447265625, 48.73698425292969, 45.86884307861328, 26.14028549194336, 87.337890625, 39.422119140625, 14.348846435546875, 66.30899047851562, 21.05023193359375, 1.8216896057128906, 58.853904724121094, 23.825679779052734, 13.41851806640625, 62.52442932128906, 19.845439910888672, -2.9364013671875, 113.73255920410156, 68.4557876586914, -0.7792739868164062, 12.139625549316406], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000125.npy"} +{"epoch": 0.18355359765051396, "step": 126, "batch_size": 64, "mean": 31.7573299407959, "std": 34.08960723876953, "min": -44.141510009765625, "p10": -1.4073322296142576, "median": 25.674497604370117, "p90": 81.91134490966797, "max": 124.27336120605469, "pos_frac": 0.875, "sample": [62.965972900390625, 40.45740509033203, 8.703964233398438, 54.530853271484375, 10.485458374023438, 22.538543701171875, -14.0770263671875, 2.0723228454589844, 27.887184143066406, 98.7039794921875, 6.69036865234375, 13.890247344970703, 82.00448608398438, 27.00495147705078, 47.979156494140625, -1.1199264526367188, 71.34806060791016, 25.276317596435547, 61.35337829589844, 33.17636489868164, 32.98249053955078, 31.742992401123047, -1.5305061340332031, 19.15877914428711, 35.12109375, 87.9910888671875, -3.8025588989257812, 81.69401550292969, 24.934772491455078, 30.562713623046875, 15.644599914550781, 13.92690658569336, 26.072677612304688, 8.721969604492188, 87.27711486816406, 33.23585510253906, 44.2420654296875, -28.180641174316406, 59.00028991699219, 13.113697052001953, -12.449050903320312, 3.2479705810546875, 113.786376953125, 9.548643112182617, 15.935991287231445, 5.846370697021484, 5.387012481689453, 11.624979019165039, 66.7042236328125, 62.58546447753906, -44.141510009765625, 124.27336120605469, 43.250762939453125, 16.70977783203125, 18.310462951660156, 47.73292541503906, 26.90802001953125, 46.14459991455078, 14.450069427490234, -17.57598876953125, 4.1258087158203125, 55.93842315673828, 13.28594970703125, 107.05693054199219], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000126.npy"} +{"epoch": 0.18502202643171806, "step": 127, "batch_size": 64, "mean": 26.10796546936035, "std": 28.82771873474121, "min": -30.827194213867188, "p10": -10.350395202636717, "median": 24.951108932495117, "p90": 64.61497573852539, "max": 89.47059631347656, "pos_frac": 0.796875, "sample": [31.2545166015625, 48.85327911376953, -1.0550498962402344, 29.52057647705078, 2.3649673461914062, -25.064697265625, 82.25129699707031, 10.350616455078125, 63.74065399169922, -22.96856689453125, 7.706211090087891, 31.142166137695312, 48.59803771972656, -14.399747848510742, 17.050111770629883, 18.8852481842041, 41.622581481933594, 8.003387451171875, -30.827194213867188, 52.97184753417969, 20.324119567871094, 52.008087158203125, 64.98968505859375, 18.285629272460938, 46.9454460144043, 84.73041534423828, 25.023815155029297, 41.631500244140625, -0.7047653198242188, 20.287738800048828, 81.6605453491211, 27.998191833496094, 55.91996765136719, 15.794303894042969, 20.177078247070312, 73.84300231933594, -1.8853836059570312, -3.3970603942871094, 21.69364356994629, -1.8802318572998047, 9.734203338623047, 62.72178649902344, 6.977277755737305, 34.8795166015625, 28.0318603515625, 41.460411071777344, 32.60504913330078, 55.95379638671875, 52.02173614501953, 35.80254364013672, -20.933731079101562, 26.39470672607422, 13.896930694580078, 89.47059631347656, -25.08789825439453, 24.878402709960938, 10.690773010253906, 70.06304168701172, 30.76507568359375, 7.703786849975586, 27.321861267089844, 11.713462829589844, -11.17498779296875, -8.426345825195312], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000127.npy"} +{"epoch": 0.18649045521292218, "step": 128, "batch_size": 64, "mean": 27.133769989013672, "std": 35.993370056152344, "min": -64.5511474609375, "p10": -11.141885757446289, "median": 19.69703483581543, "p90": 83.01674194335938, "max": 137.37452697753906, "pos_frac": 0.78125, "sample": [25.790203094482422, 24.915252685546875, 77.94265747070312, -16.203170776367188, 67.58793640136719, -14.610061645507812, 71.25263977050781, 85.95575714111328, 14.746055603027344, 20.000778198242188, 83.66793823242188, 24.598892211914062, 32.28375244140625, 18.14093780517578, -19.275497436523438, 67.00235748291016, 62.486541748046875, 5.502786636352539, 5.344778060913086, 48.438472747802734, -29.61834716796875, -1.039011001586914, 137.37452697753906, -64.5511474609375, 81.49728393554688, 19.908470153808594, 61.12542724609375, 85.5197525024414, 21.7996826171875, 30.065196990966797, -4.380619049072266, -10.567306518554688, 67.8626937866211, 7.941127777099609, -0.17266082763671875, 16.854148864746094, -16.948532104492188, 9.876689910888672, 19.485599517822266, 35.7890510559082, 36.67027282714844, -9.275146484375, 43.4136962890625, -11.388134002685547, -0.23987960815429688, 89.06993103027344, 9.779699325561523, 37.685302734375, 15.900999069213867, 26.452342987060547, 14.246681213378906, 5.248403549194336, 37.76611328125, -1.2694377899169922, 11.068117141723633, 10.930648803710938, 9.088157653808594, 37.53235626220703, 41.9368896484375, 1.770425796508789, 3.645751953125, 87.06813049316406, 85.3791275024414, 0.6897869110107422], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000128.npy"} +{"epoch": 0.18795888399412627, "step": 129, "batch_size": 64, "mean": 27.928403854370117, "std": 39.61906051635742, "min": -65.8392333984375, "p10": -12.445752525329588, "median": 23.039745330810547, "p90": 86.5402168273926, "max": 158.46636962890625, "pos_frac": 0.703125, "sample": [-7.290014266967773, 22.689842224121094, -17.647296905517578, 69.96490478515625, -13.397157669067383, 43.241119384765625, 9.351188659667969, -7.333038330078125, 88.41372680664062, 23.837329864501953, 19.106063842773438, 56.332763671875, 15.895803451538086, 28.368988037109375, 35.5767936706543, 42.40922546386719, 90.30279541015625, -2.45562744140625, 6.6938629150390625, 48.35669708251953, 97.2574462890625, 51.96257019042969, 88.634521484375, -38.95543670654297, 29.462509155273438, -30.606781005859375, -1.9253463745117188, 25.877338409423828, 158.46636962890625, 82.16869354248047, 21.85017967224121, 20.544387817382812, -9.299224853515625, 47.78797149658203, -4.8291473388671875, 2.4352359771728516, 10.941436767578125, 67.15029907226562, 10.376663208007812, -8.21063232421875, -22.578819274902344, 9.074197769165039, 23.3896484375, 52.40730667114258, 104.098388671875, -65.8392333984375, 12.947885513305664, 12.868751525878906, -2.5563716888427734, -3.727367401123047, 67.84417724609375, 100.3304443359375, 34.107364654541016, 29.754844665527344, 48.08381652832031, -3.620025634765625, 35.89305877685547, 30.3419189453125, -10.225807189941406, 45.68523406982422, 66.3221435546875, -15.260543823242188, 66.0270004272461, -1.4572162628173828], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000129.npy"} +{"epoch": 0.1894273127753304, "step": 130, "batch_size": 64, "mean": 39.8369140625, "std": 41.80584716796875, "min": -61.585540771484375, "p10": -5.451068115234374, "median": 33.224220275878906, "p90": 101.06760025024414, "max": 142.1688232421875, "pos_frac": 0.828125, "sample": [46.43153762817383, 35.36250686645508, 13.502601623535156, 102.83671569824219, -5.615226745605469, 17.570465087890625, 86.81817626953125, 63.08789825439453, -5.068031311035156, 1.7863655090332031, 17.570287704467773, 24.00051498413086, 25.66469955444336, 79.56137084960938, 115.8727798461914, 13.597846984863281, 64.65155029296875, -3.755178451538086, 26.90125274658203, 53.477577209472656, 3.900188446044922, 17.579147338867188, -0.34336090087890625, 51.87238311767578, 63.304771423339844, 13.132942199707031, 32.89521789550781, 71.0093765258789, 39.83281707763672, 103.05551147460938, 86.20976257324219, 101.02511596679688, -9.62744140625, 13.490066528320312, 70.37759399414062, -61.585540771484375, 23.838623046875, 18.986448287963867, 20.528507232666016, -8.547500610351562, 44.388572692871094, 34.371700286865234, 53.968055725097656, 127.50846862792969, 101.08580780029297, -7.080970764160156, 142.1688232421875, -14.079917907714844, -2.227407455444336, 44.659812927246094, 33.55322265625, 26.68698501586914, 102.90740203857422, 1.0870132446289062, 81.45086669921875, -52.16028594970703, 77.22125244140625, 3.506532669067383, 49.32557678222656, 80.58880615234375, 95.66790771484375, 46.73582458496094, 32.38486862182617, 20.653152465820312], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000130.npy"} +{"epoch": 0.19089574155653452, "step": 131, "batch_size": 64, "mean": 37.46085739135742, "std": 35.37492370605469, "min": -45.970184326171875, "p10": 4.357853889465332, "median": 31.528228759765625, "p90": 83.1317222595215, "max": 123.1685791015625, "pos_frac": 0.921875, "sample": [34.71719741821289, 28.7457218170166, 34.285804748535156, 25.35006332397461, 117.44869995117188, -44.83056640625, 17.613353729248047, 21.81024932861328, 32.88771057128906, 53.008094787597656, 73.08834838867188, 16.72716522216797, 21.11346435546875, 8.207168579101562, 42.16961669921875, 17.484222412109375, 21.649555206298828, -2.309995651245117, 71.76832580566406, 37.82634353637695, 33.216392517089844, 85.00819396972656, 9.679634094238281, -3.838409423828125, 29.455169677734375, 14.926422119140625, 64.01872253417969, 7.743797302246094, 91.19216918945312, 30.168746948242188, 25.51938247680664, -32.856353759765625, 78.75328826904297, 4.305042266845703, 123.1685791015625, 1.8697376251220703, 40.399070739746094, 12.418807983398438, 86.30525207519531, 33.55165481567383, 60.76434326171875, 24.598512649536133, 117.27484130859375, 16.668758392333984, -45.970184326171875, 62.184600830078125, 8.243804931640625, 64.1592788696289, 4.481081008911133, 22.146595001220703, 5.1305389404296875, 60.87517547607422, 12.503532409667969, 102.60301208496094, 67.31344604492188, 77.92601013183594, 50.770660400390625, 39.63988494873047, 17.01071548461914, 50.23847961425781, 51.067527770996094, 67.84395599365234, 71.23309326171875, 25.021270751953125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000131.npy"} +{"epoch": 0.19236417033773862, "step": 132, "batch_size": 64, "mean": 33.759395599365234, "std": 40.23978042602539, "min": -50.10394287109375, "p10": -1.5763715744018547, "median": 28.88606357574463, "p90": 77.56826934814454, "max": 172.04135131835938, "pos_frac": 0.875, "sample": [13.796010971069336, 9.443401336669922, 88.52163696289062, -22.541893005371094, 33.01124572753906, 36.02830505371094, 41.88164138793945, 151.53036499023438, 146.29440307617188, 7.3480987548828125, 34.26739501953125, 52.771080017089844, 5.382184982299805, 22.725372314453125, -50.10394287109375, 14.696708679199219, 87.64556884765625, -3.6064376831054688, 31.200599670410156, -1.8896427154541016, 4.891054153442383, -0.8454055786132812, 59.784156799316406, 59.25062561035156, 6.13531494140625, 17.091156005859375, 6.156049728393555, 59.684104919433594, 3.74981689453125, 11.106399536132812, 8.027320861816406, 9.66766357421875, 11.636581420898438, 52.32469177246094, 58.669498443603516, 72.31867218017578, 25.283477783203125, 32.37721252441406, 51.80814743041992, 39.441165924072266, 28.449583053588867, 29.629436492919922, 37.84447479248047, 76.68020629882812, -35.069976806640625, 29.32254409790039, 77.94886779785156, -17.35468292236328, 8.226619720458984, 55.66571044921875, -10.00056266784668, 44.60205078125, 73.16290283203125, 8.191680908203125, 30.373321533203125, 5.78729248046875, 21.107284545898438, 57.0260009765625, 21.286209106445312, 11.952791213989258, 102.53717041015625, 12.181447982788086, 32.04978942871094, 172.04135131835938], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000132.npy"} +{"epoch": 0.19383259911894274, "step": 133, "batch_size": 64, "mean": 30.422595977783203, "std": 39.1685791015625, "min": -36.584747314453125, "p10": -6.878535461425776, "median": 22.276723861694336, "p90": 77.74162063598634, "max": 150.19952392578125, "pos_frac": 0.84375, "sample": [17.96192169189453, 64.68223571777344, 12.823089599609375, 53.49559020996094, 9.394039154052734, -12.680870056152344, 4.6267547607421875, 14.011177062988281, -30.457740783691406, 119.46656799316406, 31.001510620117188, 50.19207763671875, 0.5622482299804688, 76.4198226928711, -33.171478271484375, 3.521942138671875, 39.094268798828125, -1.7266845703125, 115.41859436035156, 12.960344314575195, 54.33910369873047, 27.988067626953125, -1.4214248657226562, 67.91587829589844, 2.7337112426757812, 63.629913330078125, 28.318885803222656, 48.47322463989258, 49.83678436279297, 23.793991088867188, 16.643417358398438, 7.8317718505859375, 16.319860458374023, 131.63385009765625, 3.5534286499023438, 39.120201110839844, 41.226585388183594, -15.82244873046875, 110.14767456054688, 28.438655853271484, 39.55964279174805, 22.834632873535156, 24.107072830200195, 150.19952392578125, -36.584747314453125, -1.0287322998046875, 42.15824890136719, 95.83406066894531, 1.396026611328125, 21.718814849853516, 6.888694763183594, 59.068328857421875, -24.320770263671875, 78.30810546875, 48.58606719970703, -9.086471557617188, 61.46681213378906, 27.855804443359375, 15.965805053710938, 13.922962188720703, 5.3367156982421875, 0.24546432495117188, 6.424535751342773, 3.8930511474609375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000133.npy"} +{"epoch": 0.19530102790014683, "step": 134, "batch_size": 64, "mean": 33.73609161376953, "std": 33.706565856933594, "min": -13.716178894042969, "p10": 1.2485540390014653, "median": 26.16773796081543, "p90": 81.15214843750002, "max": 143.86465454101562, "pos_frac": 0.90625, "sample": [6.821533203125, -13.716178894042969, 1.0635757446289062, 71.65357971191406, 123.13226318359375, 17.84088897705078, 27.41425323486328, 17.200153350830078, 2.8011703491210938, 27.877307891845703, 60.64814758300781, 14.072872161865234, 8.814836502075195, 8.517158508300781, 36.53602600097656, 25.551712036132812, -6.641714096069336, 32.37787628173828, 46.75352478027344, 9.740676879882812, 6.71479606628418, 6.230377197265625, 38.199493408203125, 28.69458770751953, 107.40029907226562, 35.68501281738281, 68.77716064453125, 69.55961608886719, 33.8311882019043, 13.786972045898438, 15.754318237304688, 84.92335510253906, 4.95311164855957, 24.757064819335938, 78.22149658203125, 77.00846099853516, 88.33370208740234, -5.285531997680664, 48.822078704833984, 5.316009521484375, 22.067108154296875, 6.285057067871094, 26.783763885498047, 6.850732803344727, 143.86465454101562, 22.93114471435547, 53.81840515136719, 22.54058074951172, 44.53819274902344, 1.6801700592041016, 29.826370239257812, -5.408027648925781, -9.58740234375, 105.16448974609375, 51.394187927246094, 27.15837860107422, 41.47951126098633, 18.93252944946289, 22.602401733398438, 45.43560028076172, -11.997657775878906, 41.505435943603516, 18.692821502685547, 82.40814208984375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000134.npy"} +{"epoch": 0.19676945668135096, "step": 135, "batch_size": 64, "mean": 40.12480163574219, "std": 44.5528678894043, "min": -43.28460693359375, "p10": -5.356583976745604, "median": 29.052888870239258, "p90": 106.08762664794924, "max": 166.87615966796875, "pos_frac": 0.859375, "sample": [28.520065307617188, 57.473907470703125, 15.45541000366211, 107.818115234375, 46.78692626953125, 47.282142639160156, 142.60960388183594, 28.968448638916016, 122.078369140625, 115.57209777832031, 88.88497924804688, 23.83740234375, 101.610595703125, 79.44636535644531, 132.17608642578125, 55.7319221496582, 23.039764404296875, 90.80775451660156, 11.123577117919922, -5.931753158569336, 4.489917755126953, 78.21539306640625, 75.08676147460938, 51.02351379394531, 19.733694076538086, 61.38377380371094, 12.04327392578125, 166.87615966796875, -16.983638763427734, 68.83387756347656, -14.34417724609375, 110.81192016601562, 102.04981994628906, -43.28460693359375, 23.57132339477539, -4.014522552490234, 19.810155868530273, 35.314170837402344, 51.36878967285156, 4.977691650390625, -32.541419982910156, 14.031478881835938, 46.031654357910156, 18.670982360839844, 11.669620513916016, 31.726760864257812, 15.450794219970703, -2.9029159545898438, 30.16663360595703, 12.35023307800293, 14.508922576904297, 48.86024475097656, 49.92717742919922, 40.162269592285156, 2.7544403076171875, 50.160186767578125, 29.1373291015625, 82.59747314453125, 12.767974853515625, -31.18140411376953, 0.00359344482421875, 0.01446533203125, -8.302963256835938, 11.668594360351562], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000135.npy"} +{"epoch": 0.19823788546255505, "step": 136, "batch_size": 64, "mean": 29.32502555847168, "std": 43.02831268310547, "min": -22.45627212524414, "p10": -9.923112487792968, "median": 17.988224029541016, "p90": 70.93488922119141, "max": 209.40875244140625, "pos_frac": 0.75, "sample": [47.1588249206543, 32.056617736816406, 50.22039794921875, 49.95310974121094, 3.6918792724609375, -21.838577270507812, 146.2808380126953, 74.61537170410156, -12.563644409179688, 32.01578140258789, 51.13069152832031, 37.3270149230957, 29.800750732421875, 7.5721435546875, 49.590858459472656, 24.810165405273438, 43.088470458984375, 15.901941299438477, 71.37945556640625, 151.47518920898438, -3.853668212890625, 53.04487609863281, 209.40875244140625, 11.583183288574219, 50.73193359375, 6.236541748046875, 104.20089721679688, 8.818134307861328, 118.5999755859375, 9.237350463867188, 16.32024383544922, 6.925086975097656, 40.69239044189453, -18.711517333984375, 19.656204223632812, -10.220535278320312, 15.124275207519531, -7.5631256103515625, 39.74208068847656, 43.657203674316406, 15.60235595703125, -9.194286346435547, -7.5604705810546875, 43.81935119628906, -2.8097267150878906, 1.1270198822021484, -2.657978057861328, 30.215251922607422, 50.3674201965332, 58.32775115966797, -15.805946350097656, -5.681465148925781, 23.64044189453125, 26.08086395263672, -9.2291259765625, 8.106483459472656, 21.742767333984375, -4.562740325927734, -22.45627212524414, 8.209299087524414, 11.911540985107422, 69.89756774902344, 8.680107116699219, -18.26612091064453], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000136.npy"} +{"epoch": 0.19970631424375918, "step": 137, "batch_size": 64, "mean": 44.24881362915039, "std": 54.50262451171875, "min": -74.56221008300781, "p10": -5.84192771911621, "median": 26.571552276611328, "p90": 111.45529556274414, "max": 210.0108642578125, "pos_frac": 0.84375, "sample": [15.554458618164062, 20.65353012084961, 21.080337524414062, 60.120033264160156, 84.38921356201172, 14.049720764160156, 19.89337158203125, 44.391845703125, -5.186748504638672, 76.02376556396484, 8.045339584350586, 58.80242156982422, 210.0108642578125, 10.967348098754883, 25.335613250732422, -36.43040084838867, 60.338134765625, 16.27947998046875, 22.173654556274414, 94.15528869628906, 108.44708251953125, 50.047401428222656, 3.9365921020507812, 166.25677490234375, 55.584259033203125, 146.0054168701172, 98.10263061523438, 49.08619689941406, 9.721780776977539, 110.45305633544922, -35.79517364501953, 24.895065307617188, 43.36909484863281, 11.013397216796875, 85.625732421875, 3.5650062561035156, 88.45635986328125, 42.512107849121094, -6.122718811035156, 29.341041564941406, 40.501190185546875, 22.751220703125, 23.62957000732422, -74.56221008300781, -15.288124084472656, -25.71331024169922, 111.88482666015625, -2.995189666748047, 27.270545959472656, 183.79312133789062, 29.866928100585938, 65.8875732421875, 5.090398788452148, 69.3424072265625, -6.217376708984375, 5.783172607421875, 25.87255859375, 5.946895599365234, 4.934106826782227, 60.327674865722656, 150.2345733642578, -2.0645599365234375, 77.26744842529297, 143.2322998046875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000137.npy"} +{"epoch": 0.2011747430249633, "step": 138, "batch_size": 64, "mean": 33.17308044433594, "std": 37.315956115722656, "min": -64.93374633789062, "p10": -3.996244812011716, "median": 25.310094833374023, "p90": 79.5873390197754, "max": 145.92294311523438, "pos_frac": 0.859375, "sample": [82.82037353515625, -5.117889404296875, 55.88722610473633, 80.13482666015625, 23.154449462890625, 43.47667694091797, 1.4868621826171875, -1.1580791473388672, 1.4618453979492188, 72.777587890625, 23.460201263427734, 17.36968994140625, 44.580474853515625, 6.314949035644531, 21.99559211730957, -64.93374633789062, 12.034383773803711, -10.394538879394531, 1.6090431213378906, 10.563308715820312, 90.88201141357422, 95.22643280029297, 8.794378280639648, 4.365684509277344, 87.67868041992188, 74.36691284179688, 32.628318786621094, 7.9475860595703125, 56.694854736328125, 65.67381286621094, 37.581878662109375, 13.376947402954102, 61.39937210083008, 10.819351196289062, 78.30986785888672, 70.61953735351562, 22.05859375, 38.17483139038086, -35.20989990234375, 20.65404510498047, 43.47064208984375, 20.855545043945312, 27.159988403320312, 5.258777618408203, 1.701019287109375, -19.41387176513672, 66.95527648925781, 76.81465911865234, 15.542247772216797, 32.008819580078125, 63.53581237792969, 22.56791877746582, 65.84815979003906, -29.97118377685547, -19.04738426208496, 41.88737869262695, -1.3790740966796875, 40.015625, 9.733322143554688, 145.92294311523438, 60.172630310058594, 36.915557861328125, 60.76667022705078, 96.189208984375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000138.npy"} +{"epoch": 0.2026431718061674, "step": 139, "batch_size": 64, "mean": 41.65619659423828, "std": 41.745506286621094, "min": -33.271881103515625, "p10": -3.7912788391113272, "median": 35.80196762084961, "p90": 88.26908111572266, "max": 174.48629760742188, "pos_frac": 0.84375, "sample": [-14.025291442871094, 73.11438751220703, 59.75231170654297, 87.65129089355469, -0.8709869384765625, 49.541656494140625, 81.8268814086914, -0.679412841796875, 39.959774017333984, 82.37157440185547, -33.271881103515625, 37.225440979003906, -11.067638397216797, 139.72897338867188, 88.03089141845703, 40.735984802246094, 63.923736572265625, 33.00090026855469, 25.409408569335938, 30.400165557861328, 88.37116241455078, 43.72743225097656, -20.821990966796875, 81.3203353881836, 27.13360595703125, 2.594297409057617, 89.774169921875, 52.12284851074219, 6.320035934448242, 32.91210174560547, 82.25221252441406, 8.121912002563477, 51.74635314941406, 52.72792053222656, 37.151458740234375, 44.219154357910156, 13.922828674316406, 1.5994396209716797, 2.136474609375, 7.537590026855469, 28.933258056640625, 2.2230148315429688, 8.526260375976562, 82.93356323242188, 22.431194305419922, 119.50613403320312, 68.05844116210938, -12.392486572265625, 34.452476501464844, -2.697021484375, 53.217689514160156, -4.260246276855469, 23.58601188659668, 45.48039627075195, 174.48629760742188, 24.824649810791016, 148.9359588623047, 21.303421020507812, 47.37163543701172, -9.664749145507812, 95.02685546875, 23.33587646484375, 27.286109924316406, 65.46455383300781], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000139.npy"} +{"epoch": 0.20411160058737152, "step": 140, "batch_size": 64, "mean": 38.573219299316406, "std": 46.32626724243164, "min": -50.06488800048828, "p10": -10.568501281738278, "median": 32.7358283996582, "p90": 100.84234161376955, "max": 169.35836791992188, "pos_frac": 0.765625, "sample": [-2.789447784423828, 81.69480895996094, 169.35836791992188, -23.50958251953125, -4.87994384765625, 71.48126220703125, 62.930458068847656, 103.05270385742188, 45.27125549316406, 93.3238525390625, 67.57986450195312, 37.21238708496094, 76.777099609375, 19.5589599609375, 22.2802734375, 40.430641174316406, 144.93087768554688, 6.418109893798828, -0.3230628967285156, 40.09931945800781, 11.949043273925781, 46.464271545410156, 51.76557540893555, 63.055091857910156, 3.3486709594726562, -11.392196655273438, 129.4014129638672, -5.708984375, 20.16376495361328, 62.7055778503418, 19.029653549194336, 13.748023986816406, 8.908418655395508, 68.79869079589844, -8.64654541015625, 39.00782775878906, 63.30014419555664, 22.28076934814453, 3.23297119140625, 80.12074279785156, 75.62860870361328, -15.530563354492188, 95.68482971191406, -42.4764404296875, -7.244720458984375, 43.552642822265625, 24.239700317382812, 55.16602325439453, 115.82418060302734, 9.226417541503906, -8.246639251708984, 82.63369750976562, -5.358192443847656, 29.802520751953125, -50.06488800048828, 0.8226737976074219, -23.484512329101562, 14.6143798828125, 128.34796142578125, 103.80493927001953, -13.081764221191406, 35.66913604736328, 15.733810424804688, 70.9911117553711], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000140.npy"} +{"epoch": 0.2055800293685756, "step": 141, "batch_size": 64, "mean": 37.88555145263672, "std": 42.04969024658203, "min": -89.07626342773438, "p10": -5.375123596191404, "median": 33.36971855163574, "p90": 94.24749145507813, "max": 129.42990112304688, "pos_frac": 0.84375, "sample": [73.689453125, 28.229881286621094, 11.426517486572266, 65.4267807006836, 92.28816223144531, -6.468666076660156, 46.63031005859375, 43.87782287597656, 12.926254272460938, -0.5853996276855469, 36.82662582397461, 107.29962158203125, -89.07626342773438, 30.06761932373047, -2.4221954345703125, 36.489959716796875, 1.5755271911621094, -16.36090087890625, 122.42884826660156, -33.606201171875, 33.383506774902344, 95.08720397949219, 33.35593032836914, 55.5430908203125, 88.06413269042969, 48.040870666503906, 48.40489196777344, 13.697395324707031, 22.073394775390625, 38.18136215209961, 73.2688980102539, 20.67293930053711, 15.436418533325195, 88.438232421875, 129.42990112304688, 23.778867721557617, 55.38732147216797, 14.52618408203125, -15.743757247924805, 14.431989669799805, -2.8235244750976562, 38.25166320800781, 3.2661056518554688, 114.54293823242188, 10.89015007019043, 76.23477172851562, 82.2535629272461, 31.328628540039062, 10.633199691772461, -47.50604248046875, 18.069984436035156, 60.79212951660156, 32.09272766113281, 27.213523864746094, 50.3394775390625, 77.98868560791016, 113.8282470703125, 67.10884857177734, 5.388261795043945, 103.29362487792969, 41.36956024169922, 48.978851318359375, -19.679779052734375, 24.6971435546875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000141.npy"} +{"epoch": 0.20704845814977973, "step": 142, "batch_size": 64, "mean": 42.89756774902344, "std": 54.31058883666992, "min": -107.74713134765625, "p10": -11.764151573181152, "median": 28.56437110900879, "p90": 125.53383865356447, "max": 162.36593627929688, "pos_frac": 0.8125, "sample": [-6.288368225097656, 114.587890625, 9.46949577331543, 22.309799194335938, -4.312614440917969, 106.18003845214844, -17.002178192138672, -4.14637565612793, -32.0169677734375, 24.384777069091797, 63.53472137451172, 75.68209838867188, 110.14875030517578, 21.65813636779785, 33.11189270019531, 108.44184875488281, 5.377147674560547, -107.74713134765625, -27.594650268554688, 39.84156799316406, 39.27916717529297, 30.209877014160156, 14.141979217529297, 133.2115478515625, 109.43154907226562, 27.91236114501953, 73.55661010742188, 28.289567947387695, 69.79011535644531, -11.11503791809082, 45.735015869140625, 86.39283752441406, 89.08393859863281, 54.92503356933594, 0.22490692138671875, 113.01271057128906, 149.25582885742188, 20.785303115844727, 8.982671737670898, 16.237197875976562, -12.042343139648438, -16.224788665771484, 5.822998046875, 128.93983459472656, 38.034812927246094, 127.30549621582031, 10.411266326904297, 0.6798934936523438, 32.436004638671875, 2.26739501953125, 43.74269485473633, -3.107696533203125, 17.4960880279541, -35.020538330078125, 140.6571502685547, 162.36593627929688, 26.026466369628906, 121.39997100830078, 8.694084167480469, 2.6353206634521484, 74.28223419189453, 128.43328857421875, 28.839174270629883, 76.40653991699219], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000142.npy"} +{"epoch": 0.20851688693098386, "step": 143, "batch_size": 64, "mean": 27.756986618041992, "std": 44.83174514770508, "min": -70.82395935058594, "p10": -22.38443660736084, "median": 24.566336631774902, "p90": 78.84068984985352, "max": 179.57388305664062, "pos_frac": 0.703125, "sample": [41.438446044921875, 1.0815887451171875, 35.50469207763672, 30.356727600097656, 126.44793701171875, 33.99628448486328, -24.263784408569336, 58.80451202392578, 79.5274658203125, 15.442022323608398, 66.67138671875, -56.14286804199219, -25.795249938964844, 58.65888977050781, 52.27104949951172, -41.447349548339844, -20.41014289855957, -0.24286651611328125, 67.86564636230469, -10.592529296875, 14.863899230957031, 107.95944213867188, -27.32907485961914, 56.87733459472656, 28.31542205810547, -15.825603485107422, 37.65184020996094, 75.22713470458984, 23.345474243164062, 31.029930114746094, -12.04306411743164, -22.26799201965332, -7.9900970458984375, 7.748699188232422, 68.990234375, 24.751739501953125, 22.203880310058594, 32.90441131591797, 24.933364868164062, 179.57388305664062, -0.36199188232421875, -5.235963821411133, 28.898941040039062, 68.18325805664062, 0.14635086059570312, 47.225608825683594, 77.23821258544922, 87.6028823852539, -3.6373519897460938, 5.685577392578125, -22.434341430664062, 130.2686767578125, -5.623348236083984, 44.896705627441406, 30.788360595703125, 22.224191665649414, 54.00920104980469, 20.92572784423828, 86.03958892822266, -70.82395935058594, 5.631877899169922, 24.38093376159668, -1.126220703125, 11.451568603515625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000143.npy"} +{"epoch": 0.20998531571218795, "step": 144, "batch_size": 64, "mean": 36.99311065673828, "std": 37.970924377441406, "min": -43.939064025878906, "p10": -7.6031888961791925, "median": 35.11686706542969, "p90": 82.65870590209963, "max": 139.1918487548828, "pos_frac": 0.875, "sample": [-10.480010986328125, 59.795677185058594, 11.584352493286133, 92.91070556640625, -43.939064025878906, 8.048294067382812, 52.820518493652344, 71.06177520751953, 35.742332458496094, 27.937610626220703, 54.566314697265625, 3.2862892150878906, 37.80169677734375, 46.21839904785156, 48.575035095214844, 75.73639678955078, 39.07203674316406, 70.32666778564453, 74.20716094970703, 83.7190170288086, 72.22412109375, 97.69535064697266, -0.8906040191650391, -31.8045654296875, 56.629844665527344, 0.3280029296875, 37.788330078125, 8.383522033691406, 51.92698669433594, 10.45494270324707, -16.32621192932129, 59.78672790527344, 44.279701232910156, 27.8675537109375, -15.121726989746094, 88.36996459960938, 29.952648162841797, 17.92599868774414, 17.679534912109375, 50.153228759765625, 21.270492553710938, 35.826255798339844, 136.78396606445312, 34.49140167236328, 22.48413848876953, 50.63824462890625, -22.17115020751953, 19.569076538085938, 34.07023620605469, 68.72071838378906, 60.728965759277344, 125.66612243652344, 2.870574951171875, 1.5235595703125, 8.286170959472656, 3.5033798217773438, 139.1918487548828, 12.49605941772461, 17.458723068237305, 19.426727294921875, 36.31388854980469, 26.731544494628906, 80.18464660644531, -14.801246643066406], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000144.npy"} +{"epoch": 0.21145374449339208, "step": 145, "batch_size": 64, "mean": 38.81694412231445, "std": 43.2185173034668, "min": -42.60990905761719, "p10": -11.279090499877928, "median": 29.226322174072266, "p90": 97.0341178894043, "max": 173.40573120117188, "pos_frac": 0.828125, "sample": [17.138885498046875, -1.3093414306640625, -20.082138061523438, 173.40573120117188, 85.38064575195312, 79.35778045654297, 103.01504516601562, 5.770030975341797, -42.60990905761719, 37.46628189086914, 7.046875, 30.533103942871094, 85.51748657226562, -13.3134765625, 15.48452377319336, 63.959983825683594, 53.823646545410156, 23.449045181274414, 49.009761810302734, -17.280675888061523, 1.2827129364013672, 22.790924072265625, 27.919540405273438, 98.89070892333984, 52.056640625, -33.357276916503906, 13.800567626953125, 8.400196075439453, 1.628082275390625, 97.25711059570312, 24.966848373413086, 17.988525390625, 73.31034851074219, -13.300830841064453, -12.042854309082031, 68.72114562988281, 6.158843994140625, 10.49809455871582, 44.466285705566406, -5.1110992431640625, 48.147762298583984, 11.886402130126953, 96.51380157470703, -1.1446800231933594, 79.04122161865234, 108.72804260253906, 56.49578857421875, -9.49697494506836, 33.41780471801758, 106.44043731689453, 15.37533187866211, 27.44186782836914, 85.92631530761719, 93.7192153930664, 50.31681442260742, 12.10714340209961, 3.262603759765625, 38.27880859375, 48.193450927734375, 70.71979522705078, 132.78170776367188, 41.23835372924805, 8.886497497558594, 83.91912841796875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000145.npy"} +{"epoch": 0.21292217327459617, "step": 146, "batch_size": 64, "mean": 31.292926788330078, "std": 46.48145294189453, "min": -104.98843383789062, "p10": -28.5396499633789, "median": 29.80483627319336, "p90": 88.11047515869143, "max": 136.487060546875, "pos_frac": 0.78125, "sample": [33.61779022216797, -3.1437721252441406, 83.53089904785156, -38.00801467895508, 14.557550430297852, 7.1869354248046875, 67.70812225341797, 34.5726318359375, 44.68278503417969, 30.213287353515625, 129.51202392578125, 29.396385192871094, 19.8005313873291, 10.748687744140625, 25.09467315673828, 13.259513854980469, 54.357093811035156, 9.382553100585938, 9.049423217773438, -5.051176071166992, 0.10393524169921875, 22.887039184570312, 51.581939697265625, 90.07315063476562, -41.953216552734375, 44.494873046875, 77.48326110839844, -104.98843383789062, -31.509254455566406, 94.3251953125, -36.370582580566406, 37.483551025390625, 6.368217468261719, 51.499237060546875, 136.487060546875, -35.709373474121094, 46.291648864746094, 20.113149642944336, 72.75235748291016, 131.74070739746094, 20.64276123046875, 48.861183166503906, 60.80104446411133, 38.51258087158203, 33.13795471191406, 55.555877685546875, 78.8031997680664, -40.41148376464844, 21.421829223632812, -21.610572814941406, 60.53788757324219, 78.51807403564453, 79.23416137695312, 30.4145450592041, 0.1992645263671875, 20.27965545654297, 16.13502311706543, -11.054443359375, 50.980224609375, -19.745628356933594, 119.16307067871094, -13.16261100769043, 107.17637634277344, -15.265167236328125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000146.npy"} +{"epoch": 0.2143906020558003, "step": 147, "batch_size": 64, "mean": 42.82188415527344, "std": 51.198089599609375, "min": -66.15969848632812, "p10": -8.853072357177732, "median": 34.195159912109375, "p90": 121.46343078613283, "max": 175.42877197265625, "pos_frac": 0.828125, "sample": [103.80929565429688, -35.29155349731445, 38.455413818359375, 132.44837951660156, -4.065773010253906, 2.1166000366210938, 57.731292724609375, 73.8316650390625, 22.423904418945312, -16.17394256591797, 19.017126083374023, 49.399269104003906, 40.169532775878906, 81.63224792480469, 27.297893524169922, 124.1314926147461, 75.16656494140625, 21.443317413330078, -4.4852447509765625, 115.23795318603516, 124.47300720214844, 14.161720275878906, 99.28971862792969, 20.958826065063477, 159.79090881347656, 30.563804626464844, 159.1600341796875, 68.37632751464844, 7.0795745849609375, 37.939292907714844, 19.924427032470703, -66.15969848632812, 7.223995208740234, 83.69802856445312, 59.20024108886719, -16.5364990234375, 47.23683166503906, 175.42877197265625, 40.90760040283203, 16.880184173583984, 95.08597564697266, 11.663223266601562, -1.3204593658447266, 11.85955810546875, 23.109420776367188, 43.92298889160156, 8.033462524414062, 44.67905044555664, 47.84117126464844, -7.1446380615234375, 20.794952392578125, 37.445404052734375, 21.353633880615234, 87.36030578613281, -9.585258483886719, 35.16604232788086, 137.35781860351562, 3.4139175415039062, 23.50609588623047, -49.52838134765625, 103.39749145507812, -30.663970947265625, 35.73612594604492, 33.22427749633789], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000147.npy"} +{"epoch": 0.21585903083700442, "step": 148, "batch_size": 64, "mean": 33.76416015625, "std": 43.62861251831055, "min": -53.28977966308594, "p10": -4.502249526977536, "median": 20.93646240234375, "p90": 85.18871383666993, "max": 176.43792724609375, "pos_frac": 0.84375, "sample": [-2.0862770080566406, 31.389617919921875, 6.224700927734375, 75.48199462890625, -1.8679428100585938, 47.40947723388672, 71.305908203125, 27.59964370727539, -15.644416809082031, 114.65870666503906, 13.968124389648438, 86.23184967041016, 19.075326919555664, 13.983757019042969, 82.75473022460938, 124.51882934570312, 0.3336029052734375, 1.388519287109375, -13.073410034179688, -24.054351806640625, 87.06056213378906, 77.40008544921875, 1.3485469818115234, 54.725494384765625, 75.72986602783203, 33.27381134033203, 45.16352844238281, 25.989891052246094, 75.39410400390625, -1.339944839477539, 81.19916534423828, 15.1337890625, 25.194040298461914, 34.85223388671875, -7.608642578125, 22.339427947998047, 13.221702575683594, 20.543899536132812, 70.43299102783203, 15.59859848022461, 86.2965316772461, 26.933807373046875, -53.28977966308594, 19.47509002685547, 24.941320419311523, 17.24695587158203, 76.49020385742188, 20.56261444091797, 14.212249755859375, -5.537666320800781, -47.35588073730469, 32.203216552734375, 15.394607543945312, 9.135091781616211, 23.883079528808594, 5.784055709838867, 21.31031036376953, 170.5423583984375, 176.43792724609375, 6.030622482299805, 10.451467514038086, 11.511665344238281, 66.49327087402344, 6.501626968383789], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000148.npy"} +{"epoch": 0.2173274596182085, "step": 149, "batch_size": 64, "mean": 43.64375305175781, "std": 46.22150802612305, "min": -77.37828063964844, "p10": 1.0940605163574255, "median": 34.36161422729492, "p90": 95.90419082641603, "max": 194.58494567871094, "pos_frac": 0.890625, "sample": [74.07939147949219, 45.182373046875, 133.52847290039062, 33.702274322509766, 74.77706146240234, 27.878738403320312, 34.78367614746094, -11.80441665649414, -17.466087341308594, 9.82529067993164, 13.295570373535156, 11.026344299316406, 50.300655364990234, 33.939552307128906, 28.930557250976562, -27.18450927734375, 87.05525207519531, 58.27252197265625, 19.073562622070312, 71.82176971435547, 53.036865234375, 103.34304809570312, 70.5351791381836, 19.587080001831055, 13.231767654418945, 54.77281188964844, 26.820329666137695, 81.89432525634766, 57.3224983215332, 76.180419921875, 50.72663116455078, -0.4552726745605469, 194.58494567871094, 25.18560218811035, 96.84247589111328, 23.11395263671875, 133.63119506835938, 40.7137451171875, 14.12762451171875, 72.58744049072266, 22.986713409423828, 65.33406066894531, 13.016069412231445, 35.001930236816406, -5.847679138183594, 40.93592834472656, 20.900753021240234, 56.222007751464844, 7.92608642578125, -77.37828063964844, 43.51531982421875, 26.651824951171875, 11.432647705078125, 91.12393188476562, 18.138769149780273, 116.70382690429688, 14.4732666015625, 29.1865177154541, 93.71485900878906, 172.90667724609375, 12.373771667480469, 4.709171295166016, 59.17946243286133, -38.80829620361328], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000149.npy"} +{"epoch": 0.21879588839941264, "step": 150, "batch_size": 64, "mean": 41.41212463378906, "std": 54.889278411865234, "min": -61.64812469482422, "p10": -6.394661712646482, "median": 24.790668487548828, "p90": 110.95131301879886, "max": 229.8280029296875, "pos_frac": 0.859375, "sample": [160.22418212890625, 211.3678741455078, 19.275489807128906, 18.03778839111328, 45.272430419921875, -14.725822448730469, 95.78278350830078, 30.677942276000977, -61.64812469482422, 23.44506072998047, 229.8280029296875, -15.18275260925293, 8.712791442871094, 15.299797058105469, 8.788482666015625, 70.61248779296875, 24.534400939941406, 38.65251922607422, 42.21672058105469, 43.35961151123047, 27.49721908569336, 63.242408752441406, 118.479736328125, 23.462711334228516, 80.53933715820312, 60.14054870605469, 115.85093688964844, 76.89732360839844, 14.447128295898438, 58.08498001098633, 7.52374267578125, -29.443466186523438, 57.963462829589844, 105.42640686035156, 21.870208740234375, 5.490839004516602, 113.31912994384766, 17.791494369506836, 79.81047821044922, 2.8946151733398438, 60.13018798828125, -7.161949157714844, -10.511985778808594, -2.848175048828125, 84.16960144042969, 4.901821136474609, 13.387199401855469, 40.3016357421875, 14.137687683105469, 23.785499572753906, 25.04693603515625, 35.686885833740234, 44.42023468017578, 10.655841827392578, 32.980438232421875, 197.87997436523438, 13.209932327270508, 11.447418212890625, 0.37386131286621094, 39.817413330078125, 42.585777282714844, -4.6043243408203125, -37.2462158203125, 2.009288787841797], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000150.npy"} +{"epoch": 0.22026431718061673, "step": 151, "batch_size": 64, "mean": 43.702308654785156, "std": 44.7911262512207, "min": -62.06097412109375, "p10": -11.192253112792969, "median": 37.699127197265625, "p90": 116.85092163085942, "max": 163.7373809814453, "pos_frac": 0.875, "sample": [-62.06097412109375, 163.7373809814453, 44.76316833496094, 1.6971206665039062, 18.495529174804688, 28.225494384765625, 48.622283935546875, 67.3623046875, 56.78147888183594, 23.178955078125, 107.864013671875, -20.926679611206055, -11.434745788574219, 73.4398193359375, 34.022705078125, -11.424671173095703, 8.54293441772461, 27.242942810058594, 33.09141540527344, 37.561485290527344, 133.07095336914062, 50.99429702758789, 64.43070983886719, 37.836769104003906, -24.31005096435547, 122.20795440673828, 16.751609802246094, 20.71697998046875, -10.649944305419922, 32.8101806640625, 38.309112548828125, 21.088979721069336, 20.746692657470703, 18.12810707092285, 48.31559753417969, 120.70245361328125, -39.44530487060547, 61.233970642089844, 51.70014190673828, 59.867340087890625, 9.86320686340332, 45.398841857910156, 44.78428649902344, 84.96858215332031, 21.075111389160156, 126.83468627929688, 76.66557312011719, 34.119163513183594, 84.93218994140625, 25.107955932617188, 77.33735656738281, 13.477092742919922, 1.8316287994384766, 88.60028839111328, 17.773590087890625, 8.913673400878906, 125.8136978149414, 77.52741241455078, 129.79684448242188, 58.18408203125, -21.989822387695312, 52.89061737060547, 18.55340576171875, 81.19979858398438], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000151.npy"} +{"epoch": 0.22173274596182085, "step": 152, "batch_size": 64, "mean": 48.08686447143555, "std": 51.47636795043945, "min": -44.105194091796875, "p10": -5.483287811279294, "median": 37.395416259765625, "p90": 123.63268737792974, "max": 179.17306518554688, "pos_frac": 0.859375, "sample": [26.045196533203125, 22.548667907714844, 108.24163818359375, 110.63005065917969, 44.27061080932617, 102.34722137451172, 45.899593353271484, 67.4681396484375, 160.8790740966797, 24.21112060546875, 102.98981475830078, 81.33680725097656, 25.897296905517578, -23.877830505371094, 17.088172912597656, 139.26104736328125, 78.2361831665039, 20.072898864746094, 46.482601165771484, 25.491531372070312, 47.21118927001953, 29.484092712402344, -33.85283660888672, 40.95933532714844, 9.655563354492188, 177.8021697998047, 70.68157958984375, 134.6014404296875, 26.090347290039062, 6.18035888671875, 51.31565856933594, -44.105194091796875, 32.81119155883789, 1.0011444091796875, 6.478504180908203, 45.636878967285156, 26.643081665039062, 97.48707580566406, 26.854705810546875, 28.540863037109375, 179.17306518554688, 0.6973190307617188, 40.61981201171875, 72.45162963867188, 38.666900634765625, -3.107666015625, 153.8428497314453, 20.196992874145508, -28.13172149658203, 75.41653442382812, 20.18681526184082, 55.46148681640625, 36.123931884765625, 22.969947814941406, 129.2052459716797, -19.363197326660156, 81.10455322265625, -32.67901611328125, 10.900184631347656, 78.6192626953125, -1.870269775390625, -6.501411437988281, 83.74942779541016, 62.82958984375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000152.npy"} +{"epoch": 0.22320117474302498, "step": 153, "batch_size": 64, "mean": 40.949493408203125, "std": 52.211910247802734, "min": -75.9012451171875, "p10": -12.208909225463865, "median": 31.822195053100586, "p90": 113.85601272583011, "max": 221.45962524414062, "pos_frac": 0.796875, "sample": [63.79884338378906, 23.90655517578125, 12.891326904296875, 4.132316589355469, 9.184986114501953, 22.18195343017578, 20.659339904785156, 73.7191162109375, -9.817516326904297, 128.8942413330078, 55.86207580566406, 27.381011962890625, -5.023586273193359, 53.44198989868164, 149.2860565185547, -42.109458923339844, 66.82951354980469, 8.903057098388672, 40.73115539550781, 74.30695343017578, 36.345794677734375, 221.45962524414062, 27.45465850830078, 46.65806579589844, 50.541900634765625, 118.18096160888672, -13.095985412597656, -10.13906478881836, 131.7086944580078, 70.59661865234375, 100.1763916015625, 11.063957214355469, 103.76446533203125, 6.977479934692383, 0.8092498779296875, 5.973052978515625, 65.8950424194336, 49.06348419189453, -18.407825469970703, -14.562793731689453, 51.06007385253906, 31.963912963867188, 10.077468872070312, 120.133056640625, 16.22265625, -7.4253692626953125, -37.107276916503906, 38.729034423828125, 24.02509307861328, -75.9012451171875, 59.71868896484375, 168.81068420410156, 54.46794128417969, -14.504796981811523, 42.895530700683594, -0.45810699462890625, 54.8765983581543, 26.150909423828125, 66.4600601196289, 97.02876281738281, 31.680477142333984, -9.645748138427734, 19.675270080566406, 82.21014404296875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000153.npy"} +{"epoch": 0.22466960352422907, "step": 154, "batch_size": 64, "mean": 40.03327941894531, "std": 53.553897857666016, "min": -72.23016357421875, "p10": -23.05820522308349, "median": 29.889728546142578, "p90": 105.25945053100587, "max": 161.88809204101562, "pos_frac": 0.75, "sample": [-2.7738571166992188, 18.22936248779297, -72.23016357421875, 5.412408828735352, 161.88809204101562, 43.197669982910156, 0.9957332611083984, 61.21107482910156, 93.26087951660156, -18.771686553955078, 41.142982482910156, 135.39027404785156, -31.880462646484375, 59.143028259277344, 75.40611267089844, 39.18002700805664, -0.5525684356689453, -7.814117431640625, 25.14029312133789, -0.8485183715820312, 57.55931854248047, 59.42930221557617, -0.9619789123535156, 34.639163970947266, -1.0980072021484375, 91.58236694335938, 1.5078086853027344, 99.3358154296875, 82.88787078857422, -25.19500732421875, 3.9669456481933594, 110.2546157836914, -43.04875946044922, 104.70365142822266, -24.89528465270996, 118.7301025390625, 159.648681640625, 105.49765014648438, 6.51934814453125, 97.45259094238281, 104.18989562988281, 77.47920227050781, 12.963512420654297, 49.82905578613281, 98.62507629394531, 6.386253356933594, 15.603263854980469, 18.09193229675293, -39.935340881347656, 67.45235443115234, 77.97486877441406, 17.965667724609375, -6.179677963256836, 22.409912109375, 13.536773681640625, -48.71458053588867, 70.79718017578125, 65.96473693847656, 64.80245208740234, -18.06494140625, 3.6394500732421875, 15.595541000366211, 48.67961120605469, 159.79486083984375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000154.npy"} +{"epoch": 0.2261380323054332, "step": 155, "batch_size": 64, "mean": 42.82599639892578, "std": 48.98682403564453, "min": -97.80610656738281, "p10": -3.4571395874023434, "median": 39.123931884765625, "p90": 98.84653778076176, "max": 181.47128295898438, "pos_frac": 0.84375, "sample": [11.089569091796875, 43.379051208496094, 17.7362117767334, -2.875091552734375, 64.32723999023438, 6.4071044921875, 61.25598907470703, -3.7065887451171875, 51.604827880859375, 12.049970626831055, -2.592376708984375, 34.312110900878906, 40.277076721191406, -6.951324462890625, 48.11827087402344, 52.59436798095703, 11.87563705444336, 38.66560363769531, 44.528099060058594, 17.228759765625, 54.753753662109375, 36.96058654785156, 49.86671447753906, 145.8613739013672, 140.0802001953125, 7.989160537719727, 80.1431884765625, 70.45890045166016, 20.92455291748047, -0.3709259033203125, 13.281829833984375, 1.01641845703125, 23.228309631347656, 43.66143798828125, 119.260009765625, -97.80610656738281, 150.35020446777344, 27.188758850097656, 82.7740478515625, 0.8075523376464844, 13.852184295654297, 62.91923522949219, 84.66474914550781, 24.36248779296875, 181.47128295898438, 51.852684020996094, 174.51226806640625, 87.62651062011719, -31.36016845703125, 39.58226013183594, -7.6930999755859375, 103.65512084960938, 21.067760467529297, 67.66622924804688, 56.74758529663086, -19.388572692871094, 51.948402404785156, 86.34317779541016, 71.59659576416016, 4.078197479248047, 76.59120178222656, 5.978918075561523, -6.921407699584961, 29.95575714111328], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000155.npy"} +{"epoch": 0.2276064610866373, "step": 156, "batch_size": 64, "mean": 27.530269622802734, "std": 55.68471145629883, "min": -138.1324462890625, "p10": -34.13103637695312, "median": 29.372955322265625, "p90": 86.6090919494629, "max": 151.9682159423828, "pos_frac": 0.71875, "sample": [82.37945556640625, 30.475669860839844, 131.0794219970703, 151.80320739746094, 80.35986328125, 44.96315002441406, -92.43753051757812, -3.364227294921875, -13.473114013671875, 14.468193054199219, -59.08317565917969, 90.08979797363281, 43.13592529296875, -8.4239501953125, -20.68832778930664, -86.5804214477539, 39.118438720703125, 85.03494262695312, 145.24830627441406, 33.43206787109375, 9.266891479492188, 68.9327392578125, 12.8016357421875, 36.39957809448242, 12.208395004272461, 3.2208251953125, 12.916984558105469, 30.421104431152344, 151.9682159423828, -2.3103103637695312, 127.28970336914062, 85.21212005615234, 87.20779418945312, 32.456356048583984, -8.785171508789062, 4.666711807250977, 32.283531188964844, -41.660858154296875, -138.1324462890625, -30.033233642578125, 47.740875244140625, -14.6705322265625, 16.825332641601562, 66.13712310791016, 23.320144653320312, -35.887237548828125, 48.20222473144531, 83.41851806640625, 41.38664245605469, 17.9725399017334, -1.8031177520751953, 51.615867614746094, 28.324806213378906, 41.892066955566406, -75.42255401611328, -7.50201416015625, 40.53184509277344, 46.06282043457031, 23.38201904296875, 9.876564025878906, 19.905542373657227, 45.492332458496094, -1.8241462707519531, 73.09141540527344], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000156.npy"} +{"epoch": 0.2290748898678414, "step": 157, "batch_size": 64, "mean": 49.56996536254883, "std": 52.340213775634766, "min": -89.7186279296875, "p10": -8.594980621337884, "median": 45.86173629760742, "p90": 116.72335205078127, "max": 197.2910919189453, "pos_frac": 0.875, "sample": [89.65071105957031, 91.81975555419922, 60.973976135253906, 122.47190856933594, 55.787391662597656, 8.862678527832031, 39.467803955078125, 119.28514099121094, 11.432334899902344, 85.06857299804688, 40.47878646850586, 118.91433715820312, 34.00151062011719, 48.69520568847656, 109.5655517578125, 93.45936584472656, 69.38460540771484, 48.05696105957031, 32.05364990234375, 86.32083129882812, 43.66651153564453, 0.8191375732421875, 51.181640625, -12.260177612304688, 66.01333618164062, 102.63105773925781, -2.4588241577148438, 24.654502868652344, 8.193756103515625, 93.79476928710938, -89.7186279296875, 12.331953048706055, 20.822921752929688, 17.28338623046875, 52.18934631347656, 61.29431915283203, 111.61105346679688, 60.9698486328125, 66.42935180664062, -15.098562240600586, 3.6483802795410156, 185.07730102539062, 22.494428634643555, 17.17998504638672, 3.3710365295410156, 130.21519470214844, 72.43547058105469, -18.669137954711914, 144.35733032226562, 61.555850982666016, 84.9363021850586, 19.52630615234375, -20.331279754638672, 18.367774963378906, 99.02039337158203, 70.91842651367188, 197.2910919189453, 39.90337371826172, 31.618864059448242, -50.124732971191406, 22.41339111328125, -11.224761962890625, 4.988498687744141, 3.406524658203125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000157.npy"} +{"epoch": 0.2305433186490455, "step": 158, "batch_size": 64, "mean": 49.15359115600586, "std": 63.625267028808594, "min": -49.164703369140625, "p10": -9.750825881958008, "median": 34.13654708862305, "p90": 142.25786743164065, "max": 240.41867065429688, "pos_frac": 0.796875, "sample": [16.822914123535156, 7.244178771972656, 37.82984161376953, 33.27030944824219, 37.52904510498047, -0.7304916381835938, 16.66027069091797, 27.52483367919922, 35.20832824707031, 1.4498100280761719, 105.85137939453125, -49.164703369140625, 129.2723846435547, 58.34861755371094, -3.7630767822265625, -17.189716339111328, -5.082895278930664, 74.9078369140625, 2.914651870727539, 6.3573455810546875, 146.21481323242188, -9.486133575439453, 114.01126098632812, -22.45867919921875, 144.85159301757812, 21.77204132080078, 75.26077270507812, 14.04753303527832, 10.895980834960938, 11.984432220458984, 51.635555267333984, 48.814231872558594, 4.601593017578125, 92.99920654296875, -9.864265441894531, -4.4346160888671875, 172.97276306152344, 77.97431945800781, -47.225730895996094, 179.2738494873047, 217.38653564453125, 15.808486938476562, 28.664405822753906, 240.41867065429688, 0.9938812255859375, 18.98324203491211, -16.830577850341797, 62.354949951171875, 66.32200622558594, 35.002784729003906, 211.418701171875, 100.63099670410156, 63.73651123046875, 79.52626037597656, -11.26154899597168, 93.62808227539062, 14.430038452148438, -5.905796051025391, 35.75684356689453, 56.71411895751953, 136.20584106445312, 37.340721130371094, 74.74957275390625, 0.6536102294921875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000158.npy"} +{"epoch": 0.23201174743024963, "step": 159, "batch_size": 64, "mean": 39.663543701171875, "std": 53.95755386352539, "min": -93.00956726074219, "p10": -17.44287891387939, "median": 32.669368743896484, "p90": 112.72923736572267, "max": 201.2825927734375, "pos_frac": 0.8125, "sample": [13.504776000976562, 30.924278259277344, 66.98112487792969, -24.997119903564453, 123.5484619140625, 61.665069580078125, -93.00956726074219, -18.845157623291016, 71.86727905273438, 70.12944030761719, 47.89216613769531, -14.478294372558594, 10.370346069335938, -4.02154541015625, 39.07556915283203, -2.4218292236328125, 0.012237548828125, 47.81828689575195, 108.64509582519531, 95.40504455566406, 201.2825927734375, 7.064338684082031, -23.899322509765625, 9.491409301757812, 10.079662322998047, -12.557235717773438, 14.181236267089844, -20.874359130859375, 27.178749084472656, 34.414459228515625, 4.4649658203125, 115.73274230957031, 22.876953125, 14.250995635986328, 48.26519012451172, -18.713415145874023, 42.15972137451172, -6.7063446044921875, 6.408411026000977, 141.265380859375, 75.78836059570312, 114.47958374023438, 38.64665222167969, 53.6536865234375, 15.300369262695312, 179.03515625, 9.744247436523438, 0.04248809814453125, 35.60235595703125, 49.6278076171875, 83.65047454833984, 23.56861114501953, 0.31932830810546875, -51.82942199707031, 45.79014587402344, 50.53327941894531, 75.0516128540039, 78.00582885742188, 15.014833450317383, 168.1986083984375, 100.86707305908203, 37.932037353515625, 13.522796630859375, 79.48910522460938], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000159.npy"} +{"epoch": 0.23348017621145375, "step": 160, "batch_size": 64, "mean": 45.418617248535156, "std": 61.154991149902344, "min": -46.24053955078125, "p10": -22.844320678710936, "median": 25.919906616210938, "p90": 117.27755355834962, "max": 234.60556030273438, "pos_frac": 0.78125, "sample": [65.82124328613281, 3.3117008209228516, 93.98141479492188, 83.64884948730469, -3.6276512145996094, -18.751693725585938, 48.15587615966797, 105.33656311035156, 103.78277587890625, -0.14874839782714844, -40.72206115722656, 11.802734375, 97.14356994628906, 133.10589599609375, 70.012939453125, -25.727401733398438, 19.140003204345703, 46.880828857421875, -46.24053955078125, 112.73451232910156, 175.5863037109375, -30.393890380859375, 80.16064453125, -22.447708129882812, 71.32752990722656, 20.400909423828125, 20.27179718017578, 6.4256134033203125, 234.60556030273438, 39.07041931152344, 11.269264221191406, 42.607276916503906, 13.314224243164062, 176.54815673828125, -6.4285125732421875, 8.389900207519531, 22.233985900878906, 111.42387390136719, 172.8194580078125, 189.81056213378906, 32.6861572265625, 81.97463989257812, -23.014297485351562, 119.22457122802734, 61.54425811767578, 16.575576782226562, -14.421052932739258, 24.132291793823242, 26.42261505126953, 66.78360748291016, 6.6066131591796875, 25.417198181152344, 29.953323364257812, -35.88166046142578, 24.995391845703125, 54.27405548095703, 2.288656234741211, -1.3245849609375, -39.922332763671875, 72.35881042480469, 11.107864379882812, 110.37725830078125, 1.57855224609375, 56.4178581237793], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000160.npy"} +{"epoch": 0.23494860499265785, "step": 161, "batch_size": 64, "mean": 41.99751663208008, "std": 53.10074234008789, "min": -47.105613708496094, "p10": -20.292691230773926, "median": 34.44633865356445, "p90": 117.07933120727539, "max": 189.48434448242188, "pos_frac": 0.75, "sample": [52.84864807128906, 116.15518188476562, -19.57150650024414, 111.46168518066406, 135.78500366210938, -47.105613708496094, -31.391632080078125, -23.115524291992188, 45.773651123046875, -19.817947387695312, 38.06251525878906, 80.59774780273438, 52.88264465332031, -33.397865295410156, -3.184276580810547, 55.79374694824219, 148.06336975097656, 48.53264617919922, 69.62205505371094, 138.9545135498047, 36.14752960205078, 28.96859359741211, 22.116867065429688, -14.046102523803711, 32.745147705078125, -6.510334014892578, 67.01581573486328, 16.390304565429688, 159.73422241210938, 189.48434448242188, 63.560028076171875, 5.553861618041992, 18.9164981842041, -13.72208023071289, 106.95452880859375, 60.157081604003906, -1.5199432373046875, -30.516565322875977, 11.218719482421875, 57.41025161743164, 23.8502197265625, 12.764778137207031, -16.234939575195312, 26.077301025390625, 40.99183654785156, 26.074996948242188, 115.28433227539062, 59.562904357910156, 8.762481689453125, 50.976524353027344, 101.31689453125, -35.36786651611328, -4.7926177978515625, 121.84510803222656, 21.736434936523438, 17.73895263671875, 21.02655792236328, 56.23674774169922, 71.04637145996094, -20.496152877807617, 21.703256607055664, 117.47539520263672, 74.96098327636719, 48.292747497558594], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000161.npy"} +{"epoch": 0.23641703377386197, "step": 162, "batch_size": 64, "mean": 43.64946746826172, "std": 58.61953353881836, "min": -51.58623504638672, "p10": -13.702000045776368, "median": 34.9099006652832, "p90": 112.90334777832034, "max": 315.4543151855469, "pos_frac": 0.796875, "sample": [-21.84589385986328, 29.963024139404297, 50.340423583984375, 118.17996215820312, 6.1567230224609375, 130.75482177734375, 8.99876594543457, 78.28323364257812, 24.158037185668945, -9.761480331420898, -13.521961212158203, 107.42547607421875, 33.14794921875, -29.765953063964844, -13.779159545898438, -2.5778045654296875, 86.87286376953125, 78.61165618896484, 11.023468017578125, 66.65386962890625, 58.67736053466797, 17.672210693359375, 47.762115478515625, 13.454402923583984, 5.203483581542969, 89.51608276367188, 69.1895751953125, 11.578445434570312, -45.2088623046875, 70.02046203613281, 68.30377960205078, -39.406768798828125, -2.6519527435302734, 16.810731887817383, 89.4617691040039, 45.11030197143555, 117.00125885009766, 315.4543151855469, 50.69804382324219, 48.85386657714844, 44.29198455810547, 48.81842803955078, 84.44569396972656, 100.48469543457031, 90.10386657714844, 6.2367401123046875, 4.2031097412109375, 15.895126342773438, 64.46066284179688, 15.70819091796875, -8.808975219726562, 119.858642578125, 1.4315261840820312, -14.943267822265625, 184.220947265625, 37.245750427246094, -51.58623504638672, 6.6064910888671875, 26.765151977539062, 59.904258728027344, 36.671852111816406, -9.10308837890625, 28.58465576171875, 115.25100708007812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000162.npy"} +{"epoch": 0.23788546255506607, "step": 163, "batch_size": 64, "mean": 41.490657806396484, "std": 48.89918518066406, "min": -44.04656219482422, "p10": -12.222109413146967, "median": 35.5658016204834, "p90": 105.35297470092776, "max": 172.0016326904297, "pos_frac": 0.828125, "sample": [12.272048950195312, 3.2663345336914062, 150.32229614257812, 124.94450378417969, -19.002811431884766, -33.0023193359375, 29.300697326660156, -3.617095947265625, -0.06278228759765625, 42.465423583984375, 54.97051239013672, 8.790313720703125, 69.30419158935547, 172.0016326904297, 48.58221435546875, 60.49427795410156, 73.66770935058594, 61.35406494140625, 12.652259826660156, 12.085762023925781, 26.408706665039062, 63.173667907714844, 53.537841796875, 23.03130340576172, 43.97697448730469, 2.1443214416503906, 82.24078369140625, -38.444854736328125, -14.5511474609375, 44.047889709472656, 86.17500305175781, 9.20926284790039, 68.41114044189453, 16.6655330657959, 41.915069580078125, -6.787687301635742, -23.193206787109375, 147.6002960205078, 76.37007904052734, 17.537063598632812, -15.967864990234375, 30.152603149414062, 17.3077392578125, 98.39894104003906, 39.72246551513672, 66.07770538330078, 8.958335876464844, 157.36395263671875, 67.77719116210938, 7.043367385864258, 31.409137725830078, 49.662994384765625, 29.835845947265625, 108.3332748413086, 59.23899841308594, 3.9668197631835938, 41.00579833984375, 6.715171813964844, -44.04656219482422, 1.573019027709961, 156.81561279296875, 45.287353515625, 89.78665924072266, -1.27374267578125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000163.npy"} +{"epoch": 0.2393538913362702, "step": 164, "batch_size": 64, "mean": 48.205352783203125, "std": 49.44329833984375, "min": -58.76457214355469, "p10": -6.161993598937987, "median": 44.26675796508789, "p90": 103.91645126342776, "max": 223.9791717529297, "pos_frac": 0.875, "sample": [-58.76457214355469, 76.10507202148438, 89.54561614990234, 86.86353302001953, 111.35464477539062, 37.341392517089844, 45.35049819946289, 78.15083312988281, 57.760990142822266, 57.870487213134766, 135.41387939453125, 19.35045623779297, -6.51234245300293, 44.36328887939453, 16.551773071289062, 44.17022705078125, 46.28722381591797, 49.74036407470703, -6.7030487060546875, 32.446922302246094, -12.170686721801758, 56.94397735595703, 46.26092529296875, 71.8958740234375, 34.626564025878906, 52.901458740234375, 3.6058731079101562, -23.715164184570312, 64.9054946899414, 16.164602279663086, 23.775558471679688, -5.344512939453125, 73.77565002441406, 106.19087982177734, 66.79035186767578, 71.41011810302734, 19.92621612548828, 70.97793579101562, 34.48350524902344, 98.60945129394531, 21.73798370361328, 41.23038864135742, 32.22252655029297, 37.18302917480469, 31.275545120239258, 6.651115417480469, 39.61826705932617, 8.433128356933594, -40.31764221191406, 75.59077453613281, 46.72283935546875, 30.02227783203125, 169.17385864257812, 17.696929931640625, 178.79171752929688, 129.10183715820312, 4.4964141845703125, 84.23648071289062, 67.92588806152344, -16.10301971435547, 13.182212829589844, 50.0599365234375, 223.9791717529297, 3.529693603515625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000164.npy"} +{"epoch": 0.24082232011747431, "step": 165, "batch_size": 64, "mean": 32.55958557128906, "std": 43.914371490478516, "min": -84.6120376586914, "p10": -17.9848030090332, "median": 23.77609634399414, "p90": 92.88318786621093, "max": 130.1802978515625, "pos_frac": 0.8125, "sample": [41.44059753417969, 53.829673767089844, -19.391883850097656, 12.914205551147461, -14.241226196289062, 0.6608448028564453, 50.119239807128906, 43.480140686035156, 5.92572021484375, -23.35546875, 38.282310485839844, -84.6120376586914, 2.35601806640625, 28.365554809570312, 22.2042236328125, -34.06389617919922, -5.5552520751953125, 3.0341949462890625, 91.40445709228516, 85.04998779296875, 53.7955436706543, 28.54192352294922, 14.937877655029297, 114.4049072265625, -28.339019775390625, 62.48602294921875, 3.515878677368164, -25.0926513671875, 15.309928894042969, 76.65625, -3.2887649536132812, 18.784040451049805, 92.8193359375, 92.91055297851562, 16.063079833984375, -12.152412414550781, 47.49126434326172, 7.046783447265625, 34.50147247314453, 5.381265640258789, 22.474687576293945, 21.42408561706543, 50.94468688964844, 54.785682678222656, 130.1802978515625, 11.071937561035156, 56.324607849121094, 48.22596740722656, 24.44702911376953, -30.91302490234375, 6.94244384765625, 77.6566162109375, 33.44256591796875, -14.701614379882812, 30.581180572509766, 116.11872863769531, 17.797916412353516, 127.01612854003906, 126.99401092529297, 94.03146362304688, 42.88888931274414, 11.299613952636719, 23.10516357421875, 88.0538101196289], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000165.npy"} +{"epoch": 0.2422907488986784, "step": 166, "batch_size": 64, "mean": 57.041595458984375, "std": 67.72319793701172, "min": -33.60491943359375, "p10": -11.096457672119135, "median": 37.64133644104004, "p90": 152.8443099975586, "max": 282.790771484375, "pos_frac": 0.859375, "sample": [229.4570770263672, 155.8572235107422, 9.602798461914062, 236.99835205078125, -14.224983215332031, 27.598957061767578, 20.781024932861328, 44.855133056640625, 125.43306732177734, -4.6954803466796875, 60.02384948730469, 1.2630805969238281, 6.545734405517578, 83.19647216796875, -4.7635650634765625, 37.23114013671875, 73.34634399414062, 37.486236572265625, 37.79643630981445, 78.43363189697266, 21.87901496887207, 282.790771484375, 17.255237579345703, -14.725444793701172, -33.60491943359375, -21.881885528564453, 22.799972534179688, 54.606422424316406, -13.925689697265625, 109.82540893554688, 163.4610137939453, 215.41806030273438, 71.69103240966797, 68.55891418457031, 14.391281127929688, 208.88510131835938, 17.657119750976562, 75.73551940917969, 46.10211181640625, 64.87838745117188, 14.731578826904297, 47.984004974365234, 10.598983764648438, 28.782596588134766, 131.2702178955078, 62.12655258178711, 145.81417846679688, 19.705524444580078, 50.10826873779297, 70.30917358398438, 21.695709228515625, 66.96697235107422, -19.40358543395996, 24.09130096435547, 3.9149551391601562, -13.810554504394531, 84.65811920166016, 11.399810791015625, 52.929019927978516, 10.038970947265625, 65.49484252929688, 12.711187362670898, 99.45841217041016, 35.066139221191406], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000166.npy"} +{"epoch": 0.24375917767988253, "step": 167, "batch_size": 64, "mean": 60.0372314453125, "std": 55.92289733886719, "min": -70.85662841796875, "p10": 0.6189941406250008, "median": 51.91312599182129, "p90": 134.1592765808106, "max": 198.35836791992188, "pos_frac": 0.90625, "sample": [80.12213134765625, 51.03823471069336, 83.97834777832031, 108.74882507324219, 71.80767822265625, 87.05059814453125, 28.2728271484375, 39.328155517578125, 74.31201171875, 47.5213737487793, 198.35836791992188, 111.9441909790039, 47.778167724609375, -2.287078857421875, 16.7159423828125, 34.64310073852539, 58.14020538330078, -18.943519592285156, 72.17417907714844, 100.54801940917969, 22.256311416625977, 24.262771606445312, 35.149391174316406, 26.291534423828125, 62.48798370361328, -70.85662841796875, 159.8992156982422, 122.35077667236328, 1.4572296142578125, 120.23214721679688, -7.40037727355957, 35.45172119140625, 71.5469970703125, 20.494056701660156, 169.9151611328125, 90.90157318115234, 165.10952758789062, 61.16663360595703, 105.53428649902344, -49.21211242675781, 105.35258483886719, 139.22006225585938, 164.7584228515625, 90.57666778564453, 176.46200561523438, 6.3420257568359375, 97.8856201171875, 36.04115295410156, 108.78927612304688, 101.82377624511719, 61.32647705078125, 22.19586181640625, 23.302133560180664, 27.316394805908203, 15.288688659667969, 85.09403228759766, 52.78801727294922, 40.83454132080078, 10.290496826171875, -30.85216522216797, 33.97380447387695, 6.44340705871582, 8.579544067382812, 0.2597503662109375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000167.npy"} +{"epoch": 0.24522760646108663, "step": 168, "batch_size": 64, "mean": 46.935882568359375, "std": 51.70939636230469, "min": -43.01641845703125, "p10": -9.416509246826172, "median": 40.129024505615234, "p90": 106.14782485961915, "max": 216.760009765625, "pos_frac": 0.828125, "sample": [21.861961364746094, 43.534324645996094, 20.310535430908203, 16.27933120727539, 4.557832717895508, 29.226898193359375, 85.62772369384766, -2.9503936767578125, 26.66183090209961, -9.230705261230469, -40.83061981201172, 93.73184204101562, 200.75225830078125, 55.64409637451172, -9.496139526367188, 100.53681945800781, 106.57581329345703, 105.14918518066406, 96.20828247070312, 40.828468322753906, 118.25981140136719, 153.14572143554688, 104.97537231445312, 86.6477279663086, 1.2410354614257812, -22.0869140625, -3.1989097595214844, 84.31796264648438, 78.76419067382812, 52.74193572998047, 39.42958068847656, 25.58147430419922, -23.084640502929688, 43.79736328125, 27.70155906677246, 78.77476501464844, -10.362907409667969, -3.1651077270507812, 2.297962188720703, 43.28837585449219, 20.55720329284668, 62.900169372558594, 22.536895751953125, 47.06983947753906, 48.89341735839844, 58.60023498535156, 15.050741195678711, -16.206192016601562, 115.37185668945312, 30.412208557128906, 99.3778076171875, 13.728700637817383, 30.73670768737793, 11.198989868164062, 74.24337768554688, 42.4075927734375, 28.70118522644043, -43.01641845703125, 51.33613967895508, 3.1483612060546875, 216.760009765625, 20.267257690429688, 78.57098388671875, 107.20388793945312], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000168.npy"} +{"epoch": 0.24669603524229075, "step": 169, "batch_size": 64, "mean": 41.902549743652344, "std": 65.01738739013672, "min": -91.30522155761719, "p10": -18.591989898681636, "median": 26.246368408203125, "p90": 115.98079528808594, "max": 277.19207763671875, "pos_frac": 0.78125, "sample": [31.48509979248047, 55.071685791015625, -4.430446624755859, 14.531562805175781, 82.98112487792969, 82.07653045654297, 16.632644653320312, 62.71730041503906, 46.093849182128906, 15.516918182373047, -10.799495697021484, 50.13201141357422, -91.30522155761719, -0.8376903533935547, 23.481609344482422, -8.861038208007812, 13.913688659667969, -21.410049438476562, 47.40089416503906, 72.65058135986328, 71.56556701660156, -20.573211669921875, 27.826416015625, -25.675724029541016, 126.57720947265625, 113.80502319335938, 38.01639175415039, 13.727432250976562, 24.017898559570312, 18.99433708190918, 111.56233978271484, 10.603401184082031, 10.856958389282227, 38.30908203125, 54.28581619262695, 75.0344467163086, 116.91326904296875, 10.913909912109375, 63.02872085571289, 112.6790771484375, 17.05690574645996, 257.4034423828125, -26.27465057373047, 6.74664306640625, 37.07146453857422, 33.81104278564453, 24.66632080078125, -13.969139099121094, 196.4881591796875, 5.6411590576171875, 277.19207763671875, -3.8171844482421875, 72.88978576660156, 21.184120178222656, 132.590087890625, 4.01080322265625, 47.59812545776367, -8.225784301757812, 29.679203033447266, 9.956504821777344, -62.170989990234375, 57.9766960144043, 153.46139526367188, -58.713050842285156], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000169.npy"} +{"epoch": 0.24816446402349487, "step": 170, "batch_size": 64, "mean": 52.402130126953125, "std": 53.92622756958008, "min": -79.341552734375, "p10": -3.400487518310547, "median": 44.552337646484375, "p90": 120.4750877380371, "max": 173.5020751953125, "pos_frac": 0.828125, "sample": [109.11226654052734, -49.521751403808594, 47.81629943847656, 54.80933380126953, 112.6866455078125, 43.40488815307617, 93.9140625, 6.007087707519531, 86.95480346679688, -3.3565826416015625, 61.06379699707031, -3.818309783935547, 52.748680114746094, -30.921157836914062, 6.629730224609375, -5.6752471923828125, 61.82917022705078, 165.53009033203125, 31.656936645507812, 144.95989990234375, 3.631113052368164, 5.932527542114258, 93.34521484375, 36.957252502441406, 1.0022811889648438, 30.76837921142578, -79.341552734375, -3.4193038940429688, 37.48480224609375, 120.35086059570312, 83.10336303710938, 83.37692260742188, 45.43827438354492, 61.46669006347656, 120.52832794189453, -2.6777191162109375, 166.22219848632812, 5.560272216796875, -0.2284717559814453, 173.5020751953125, 69.7896728515625, 1.9228515625, 43.66640090942383, 97.02711486816406, 97.59551239013672, -5.629570007324219, 31.91541862487793, 5.114961624145508, 17.380722045898438, 35.63324737548828, 64.06208801269531, 54.6031494140625, 90.0042724609375, 35.71525192260742, 110.13350677490234, 38.70923614501953, 92.72512817382812, 126.20415496826172, -1.2112274169921875, 109.60498809814453, 143.7838134765625, 19.021120071411133, 94.54893493652344, 12.581344604492188], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000170.npy"} +{"epoch": 0.24963289280469897, "step": 171, "batch_size": 64, "mean": 42.92985534667969, "std": 63.7218017578125, "min": -90.75709533691406, "p10": -23.38028564453125, "median": 40.93520545959473, "p90": 127.67419052124028, "max": 297.02374267578125, "pos_frac": 0.78125, "sample": [-24.211044311523438, 132.7562713623047, 44.49732208251953, 33.661834716796875, -30.726181030273438, -5.752265930175781, 9.382240295410156, 43.19956970214844, -32.598724365234375, 81.83428955078125, 9.132057189941406, 90.93338012695312, 58.129356384277344, 52.40034484863281, 40.20407485961914, 4.7682952880859375, 48.882598876953125, -21.441848754882812, 58.14291000366211, -7.506523132324219, 115.81600189208984, 45.44379806518555, 59.229034423828125, 43.57439422607422, 177.44485473632812, 47.92661666870117, -35.16831970214844, 297.02374267578125, -60.280738830566406, 37.94776153564453, 60.908199310302734, -90.75709533691406, 190.9442138671875, 60.967315673828125, 44.74970245361328, -19.8580322265625, 146.15777587890625, 66.64659118652344, 15.453826904296875, 157.7203369140625, 38.882171630859375, 43.396629333496094, 95.34109497070312, 6.683462142944336, -51.80522155761719, 23.11907958984375, 59.79442596435547, 41.66633605957031, -17.820232391357422, 25.442977905273438, -2.2296295166015625, -6.811492919921875, 93.23607635498047, 35.63926696777344, 47.02789306640625, 5.025014877319336, 19.180511474609375, 106.46753692626953, 2.1588134765625, 11.81634521484375, 44.56370544433594, 10.294538497924805, 140.50987243652344, 28.353778839111328], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000171.npy"} +{"epoch": 0.2511013215859031, "step": 172, "batch_size": 64, "mean": 44.155853271484375, "std": 58.185890197753906, "min": -76.8563232421875, "p10": -14.615776443481444, "median": 46.363372802734375, "p90": 104.53092803955079, "max": 252.27098083496094, "pos_frac": 0.71875, "sample": [-4.0047607421875, 73.15318298339844, -76.8563232421875, 54.75041198730469, 12.374542236328125, -11.462173461914062, 92.222900390625, 55.51176452636719, 37.36796569824219, -28.65911293029785, -54.96629333496094, 24.514698028564453, 8.755897521972656, -3.242218017578125, 90.74465942382812, 56.167945861816406, 21.42086410522461, 26.956466674804688, -25.498310089111328, 65.61814880371094, -9.175056457519531, 70.6463623046875, 69.3921890258789, 99.99325561523438, 7.91253662109375, 44.93053436279297, 105.33360290527344, 102.65802001953125, -11.219184875488281, 157.0325927734375, -12.422286987304688, 64.16230773925781, 87.7475814819336, 45.3814697265625, 70.65524291992188, 176.9835205078125, 252.27098083496094, 39.78318786621094, 70.6554946899414, 28.771007537841797, -15.555843353271484, -11.495719909667969, 47.9605712890625, 47.34527587890625, 91.31346893310547, -6.460174560546875, 14.442686080932617, 27.603742599487305, -5.03173828125, 108.78955841064453, 94.62054443359375, 0.321502685546875, -6.136749267578125, 84.28767395019531, 60.463844299316406, -56.706298828125, 61.155033111572266, 62.17933654785156, 143.09315490722656, 50.90711975097656, -4.5715179443359375, -39.59197235107422, 117.53178405761719, 83.1458511352539], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000172.npy"} +{"epoch": 0.2525697503671072, "step": 173, "batch_size": 64, "mean": 44.7443962097168, "std": 50.330284118652344, "min": -73.65020751953125, "p10": -6.9529039382934545, "median": 34.743526458740234, "p90": 109.36931762695313, "max": 194.73204040527344, "pos_frac": 0.859375, "sample": [23.2984619140625, 40.402862548828125, 35.694007873535156, 11.208915710449219, -7.8450469970703125, 53.953338623046875, 6.851947784423828, 107.63653564453125, 80.1434555053711, 5.542488098144531, 20.98291015625, -14.911476135253906, 11.824483871459961, 12.75994873046875, 48.78990936279297, 138.07398986816406, 44.50189971923828, 37.78456497192383, 41.516334533691406, 118.29193115234375, 61.749183654785156, 29.105201721191406, 61.13612365722656, -11.691749572753906, 29.344680786132812, 27.9058837890625, 194.73204040527344, 18.523683547973633, 100.55828094482422, 110.1119384765625, 36.348480224609375, 17.944311141967773, -24.1436767578125, 98.19081115722656, 78.90069580078125, 8.280776977539062, 85.25550079345703, 49.10357666015625, -4.871236801147461, 149.59902954101562, 57.19648361206055, 87.79217529296875, -0.9216232299804688, 23.932960510253906, 14.881195068359375, 12.557544708251953, 155.74049377441406, 73.89912414550781, -25.290267944335938, 13.217803955078125, 47.87727355957031, 33.79304504394531, 63.1470832824707, -28.07855224609375, -73.65020751953125, 168.46395874023438, 27.64671516418457, 0.9168128967285156, 56.46931457519531, 22.319705963134766, 104.27728271484375, 33.513755798339844, 44.03578567504883, 17.318450927734375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000173.npy"} +{"epoch": 0.2540381791483113, "step": 174, "batch_size": 64, "mean": 48.2532958984375, "std": 60.29151153564453, "min": -65.06767272949219, "p10": -17.273013305664062, "median": 37.008684158325195, "p90": 140.29394226074226, "max": 231.52398681640625, "pos_frac": 0.796875, "sample": [9.875221252441406, 16.446929931640625, -25.750152587890625, -18.989089965820312, 0.660247802734375, 16.82623291015625, 3.7543067932128906, 21.646528244018555, 68.31354522705078, 106.02592468261719, 123.39418029785156, -5.5636749267578125, 44.674827575683594, 10.757774353027344, 50.057647705078125, 3.87481689453125, 75.9456787109375, 78.77818298339844, 84.5548095703125, 164.2298583984375, 37.20079040527344, 36.81657791137695, 30.328964233398438, -3.7867965698242188, 148.51559448242188, -65.06767272949219, 74.65746307373047, 154.4569091796875, 0.336334228515625, 66.232177734375, 20.584142684936523, -9.637939453125, 31.934364318847656, -29.68515396118164, 62.569488525390625, -18.22154998779297, 65.21148681640625, 59.12251281738281, -40.262298583984375, 147.5366973876953, 160.09115600585938, 231.52398681640625, 11.222530364990234, 51.86454772949219, -7.001813888549805, -50.45240783691406, -4.536502838134766, 68.32577514648438, 53.93418884277344, 71.53543853759766, 69.43746948242188, 35.35535430908203, 26.244285583496094, 90.04744720458984, -15.059761047363281, 111.83885192871094, 101.34132385253906, 75.72175598144531, 25.49987030029297, 11.900592803955078, 210.89755249023438, 66.58358764648438, 68.95664978027344, 24.583173751831055], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000174.npy"} +{"epoch": 0.2555066079295154, "step": 175, "batch_size": 64, "mean": 68.30412292480469, "std": 62.49799346923828, "min": -30.45825958251953, "p10": 3.0347436904907235, "median": 55.499019622802734, "p90": 153.1569488525391, "max": 267.50872802734375, "pos_frac": 0.90625, "sample": [8.87660026550293, 23.91443634033203, 40.10009002685547, -21.757919311523438, 68.16297912597656, 56.95501708984375, 128.12408447265625, 48.698280334472656, 20.42002296447754, 103.810791015625, 206.9546356201172, -27.467174530029297, 2.65447998046875, 186.56787109375, 43.89113235473633, 20.791183471679688, 137.81753540039062, 116.56796264648438, 98.78865814208984, 115.26708984375, 108.5549545288086, -12.365135192871094, 141.01019287109375, 267.50872802734375, 16.35669708251953, 67.32508087158203, 199.6544189453125, 36.756431579589844, 90.00241088867188, 54.769004821777344, -2.1843414306640625, 31.020095825195312, 166.1651611328125, 21.45626449584961, 23.194351196289062, 30.020605087280273, 13.844844818115234, 158.36270141601562, 9.107637405395508, 39.713897705078125, 88.50375366210938, 79.63832092285156, 86.4354476928711, 117.77251434326172, 3.922025680541992, 119.12883758544922, 118.83657836914062, 129.27987670898438, -7.248111724853516, 85.30013275146484, 116.21710205078125, 59.476318359375, 56.229034423828125, 70.7482681274414, 27.773601531982422, 30.180736541748047, 71.13978576660156, 10.556989669799805, 36.609275817871094, 168.86813354492188, -30.45825958251953, 36.457366943359375, 30.33892822265625, 26.345779418945312], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000175.npy"} +{"epoch": 0.25697503671071953, "step": 176, "batch_size": 64, "mean": 55.23646545410156, "std": 63.42719268798828, "min": -35.61309051513672, "p10": -6.8998886108398425, "median": 43.435462951660156, "p90": 138.49928131103516, "max": 276.81317138671875, "pos_frac": 0.8125, "sample": [25.424530029296875, 17.39166259765625, 46.07666015625, 97.92036437988281, 56.94527053833008, 17.085453033447266, 12.896202087402344, 167.90737915039062, 13.034767150878906, 6.847320556640625, -5.7909393310546875, 223.31558227539062, -11.040262222290039, -5.7178955078125, 139.1968994140625, 49.62101745605469, -5.205535888671875, 166.94776916503906, 88.81459045410156, 5.308631896972656, 8.953826904296875, 62.430564880371094, 73.06779479980469, -26.154991149902344, 136.8715057373047, -7.375152587890625, 18.320552825927734, 108.00224304199219, 93.96029663085938, 4.411735534667969, 62.04884338378906, -15.777557373046875, 66.12635803222656, -35.61309051513672, 24.333526611328125, 86.65863800048828, -12.232715606689453, 30.80126190185547, 81.54175567626953, 76.39274597167969, 87.1243667602539, 0.5908470153808594, -4.855560302734375, -7.498287200927734, 96.00187683105469, 66.87884521484375, 94.13444519042969, 46.46039581298828, 97.79350280761719, 193.57781982421875, 40.79426574707031, -1.4460830688476562, 26.074951171875, 9.752660751342773, 83.14082336425781, 55.65130615234375, 2.4420394897460938, 23.33022689819336, 87.04360961914062, 5.917057037353516, 125.67842102050781, 276.81317138671875, 24.24460220336914, 161.74085998535156], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000176.npy"} +{"epoch": 0.25844346549192365, "step": 177, "batch_size": 64, "mean": 57.55059814453125, "std": 66.4899673461914, "min": -40.419921875, "p10": -11.069068908691403, "median": 38.25611877441406, "p90": 140.3664260864258, "max": 301.1607971191406, "pos_frac": 0.75, "sample": [79.74252319335938, 34.487884521484375, 115.46931457519531, 52.89929962158203, 52.30736541748047, 85.02058410644531, -4.2139434814453125, 47.80784606933594, 127.54296112060547, 93.15093994140625, 134.05885314941406, 17.914833068847656, 63.07447814941406, 143.06967163085938, 76.87650299072266, 54.6948127746582, 15.070075988769531, 201.44247436523438, 12.103927612304688, 129.19769287109375, 19.40855598449707, 121.01979064941406, -33.25251007080078, 102.79319763183594, -1.5994892120361328, 25.06683921813965, 166.849609375, -15.224266052246094, -7.3861541748046875, 25.2027587890625, 29.36859703063965, 174.4719696044922, 96.14619445800781, -0.6708831787109375, 42.02435302734375, 28.918411254882812, 10.7877197265625, 174.99317932128906, -3.029817581176758, -15.642387390136719, -40.419921875, 70.80154418945312, -1.3943023681640625, 34.02375030517578, -34.12617492675781, 90.81941223144531, 105.80702209472656, 2.906696319580078, -1.2657546997070312, -2.947418212890625, 126.94781494140625, 32.744468688964844, 67.91954040527344, -17.69915008544922, 94.46255493164062, 28.468708038330078, 23.89374351501465, -7.308012008666992, 107.75491333007812, 13.445056915283203, 301.1607971191406, 163.8347930908203, 64.0919418334961, -12.6474609375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000177.npy"} +{"epoch": 0.2599118942731278, "step": 178, "batch_size": 64, "mean": 70.92833709716797, "std": 80.305419921875, "min": -90.33758544921875, "p10": -10.561374664306634, "median": 52.46828079223633, "p90": 162.365609741211, "max": 343.4511413574219, "pos_frac": 0.828125, "sample": [41.19040298461914, -4.578834533691406, 102.92972564697266, 1.094268798828125, 89.55768585205078, 53.03101348876953, 85.28135681152344, 24.90833282470703, 18.61392593383789, 185.71482849121094, 9.788955688476562, 90.91766357421875, 138.58966064453125, 60.22731018066406, 33.51188659667969, 96.19139099121094, 235.73934936523438, 51.905548095703125, 343.4511413574219, -13.125320434570312, 127.73814392089844, 41.22620391845703, 20.618988037109375, 49.462318420410156, 22.116701126098633, 128.12918090820312, 168.16473388671875, -24.048782348632812, 36.32354736328125, 3.922454833984375, -21.849777221679688, 67.54962158203125, 138.3157958984375, 19.225223541259766, 69.6310043334961, 134.4978485107422, 49.49932861328125, -29.356849670410156, 44.511077880859375, 134.3756103515625, 17.468902587890625, 77.39021301269531, -90.33758544921875, 265.7431640625, -4.552957534790039, 104.38911437988281, 72.90351867675781, 98.03160858154297, -19.674413681030273, 256.08331298828125, 29.39409828186035, 84.24468231201172, 147.4613037109375, 4.742895126342773, 27.847259521484375, 92.6353759765625, 69.75080871582031, -2.5862884521484375, 30.01318359375, 148.83432006835938, 221.00064086914062, -45.610809326171875, 131.24441528320312, -1.9959487915039062], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000178.npy"} +{"epoch": 0.26138032305433184, "step": 179, "batch_size": 64, "mean": 59.215576171875, "std": 76.84276580810547, "min": -137.9810028076172, "p10": -24.54421806335449, "median": 58.298095703125, "p90": 164.60559234619143, "max": 238.30599975585938, "pos_frac": 0.734375, "sample": [69.93990325927734, 109.56825256347656, 112.62195587158203, 202.52951049804688, 40.86200714111328, -39.162071228027344, 138.73770141601562, 39.803585052490234, 0.37876129150390625, -0.7386760711669922, 81.34104919433594, 40.83653259277344, 18.897857666015625, 58.346649169921875, -13.068016052246094, 89.21826171875, -14.997161865234375, 186.69107055664062, 20.01656723022461, -9.65814208984375, 57.10456848144531, -26.535221099853516, 173.97998046875, 17.520578384399414, 151.9543914794922, 96.37220764160156, 100.28475952148438, 188.55259704589844, 127.77708435058594, 49.68354415893555, 71.63601684570312, 238.30599975585938, 189.8207550048828, 67.177978515625, 66.80309295654297, 71.38653564453125, -18.424346923828125, 8.323848724365234, 1.9523544311523438, 61.65027618408203, 66.28401947021484, -19.898544311523438, 58.249542236328125, -19.837417602539062, 118.33810424804688, 150.3765869140625, -49.52904510498047, 92.88023376464844, 84.50482940673828, -17.14934539794922, -14.232078552246094, 154.47596740722656, 7.3670196533203125, 129.443359375, 167.1180419921875, 158.7432098388672, -88.18359375, 123.00098419189453, -27.294204711914062, -28.781387329101562, -137.9810028076172, 49.30755615234375, 21.31702995300293, -16.215736389160156], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000179.npy"} +{"epoch": 0.26284875183553597, "step": 180, "batch_size": 64, "mean": 64.97667694091797, "std": 67.25618743896484, "min": -67.62396240234375, "p10": -8.613925552368162, "median": 59.7861328125, "p90": 149.175244140625, "max": 260.13372802734375, "pos_frac": 0.828125, "sample": [62.394752502441406, 109.26781463623047, 16.968856811523438, -9.39224624633789, -53.13890838623047, 260.13372802734375, -2.578044891357422, 28.42901611328125, 75.78085327148438, 61.442840576171875, -0.32358360290527344, 127.90955352783203, -67.62396240234375, 92.10433959960938, 86.823974609375, 58.129425048828125, 78.60870361328125, 66.42109680175781, 78.64643859863281, 26.885879516601562, 110.98480987548828, -46.28010559082031, 19.482898712158203, 166.83465576171875, 45.948951721191406, 161.44761657714844, 90.20294952392578, 20.961227416992188, 22.23741340637207, 29.055980682373047, 148.1136016845703, 7.0379791259765625, 139.5150146484375, 92.67349243164062, 80.3687515258789, -2.957094192504883, 78.34864044189453, 124.78467559814453, 8.903915405273438, 126.49564361572266, 50.056846618652344, -6.797843933105469, 182.55935668945312, 27.539695739746094, -46.108489990234375, 41.162044525146484, 18.684722900390625, 137.1842498779297, 149.63023376464844, 32.043670654296875, 65.51251220703125, 144.52748107910156, -10.42652702331543, 202.3877410888672, 47.61796569824219, 0.811431884765625, 103.00904083251953, 52.5887451171875, 26.62322235107422, -26.826393127441406, 178.48350524902344, 129.1743927001953, 17.338768005371094, 120.67947387695312], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000180.npy"} +{"epoch": 0.2643171806167401, "step": 181, "batch_size": 64, "mean": 42.891845703125, "std": 60.580806732177734, "min": -84.01692962646484, "p10": -20.66595802307128, "median": 35.057241439819336, "p90": 127.65270156860352, "max": 220.4949188232422, "pos_frac": 0.84375, "sample": [51.59442138671875, -11.530696868896484, 14.798440933227539, 49.11208724975586, 75.71060180664062, -5.949352264404297, 8.172931671142578, 36.80515670776367, 42.60028839111328, 60.97845458984375, 26.086681365966797, 27.684925079345703, 3.4597854614257812, 5.497764587402344, -66.19720458984375, 126.66844940185547, 49.5664176940918, 1.4771728515625, 58.96437072753906, 3.4899635314941406, 35.21743392944336, -35.40440368652344, -2.524383544921875, 130.40298461914062, -24.581069946289062, 50.74872589111328, 56.317474365234375, 174.53675842285156, 60.912384033203125, 29.689611434936523, 84.25843811035156, 43.274200439453125, -62.72737121582031, 34.89704895019531, 24.39739227294922, 33.9156608581543, 8.148307800292969, 78.67716979980469, 119.82350158691406, -46.43074035644531, 17.338912963867188, -84.01692962646484, 155.24595642089844, 2.6610374450683594, 148.5763702392578, 101.24370574951172, -60.75856018066406, 49.10173034667969, 5.0633392333984375, 71.63350677490234, 128.07452392578125, 220.4949188232422, 74.11212158203125, 33.76742172241211, 85.7210693359375, 1.7643013000488281, 3.197702407836914, 8.936660766601562, 202.4800262451172, 48.72924041748047, 25.31250762939453, 73.60147094726562, 51.55268096923828, 28.70465850830078], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000181.npy"} +{"epoch": 0.2657856093979442, "step": 182, "batch_size": 64, "mean": 68.66345977783203, "std": 74.37088012695312, "min": -131.110107421875, "p10": -16.661261749267574, "median": 54.14805030822754, "p90": 170.45149841308597, "max": 222.8369598388672, "pos_frac": 0.859375, "sample": [173.6524658203125, 51.04750061035156, 189.4356689453125, 92.68740844726562, 59.56086730957031, 106.84864807128906, 136.25289916992188, 158.5975341796875, 222.8369598388672, 56.469303131103516, 49.97377014160156, 36.15753173828125, 21.261938095092773, -4.528247833251953, 140.174560546875, -131.110107421875, 162.98257446289062, 31.465240478515625, 4.98875617980957, 27.793785095214844, 213.0493621826172, 0.7346305847167969, 176.3201904296875, 59.47578430175781, 74.82429504394531, -27.035215377807617, 157.77865600585938, 36.094425201416016, 17.0167236328125, 50.165191650390625, 101.62825012207031, 150.14144897460938, -13.785377502441406, -57.704505920410156, 84.8709945678711, 140.74798583984375, 49.5035400390625, 49.35100555419922, 111.09994506835938, 143.32447814941406, -19.803260803222656, 42.19007873535156, 34.86377716064453, 101.90097045898438, 74.4228515625, 93.90802764892578, 85.05050659179688, 103.89696502685547, -43.80293273925781, 2.4704742431640625, 200.84030151367188, 90.47606658935547, 47.16447448730469, 1.6361331939697266, -17.893783569335938, 6.957460403442383, -71.2118911743164, 22.37291717529297, 154.5001220703125, 24.494041442871094, 191.3583984375, 1.2445640563964844, 51.82679748535156, 111.44760131835938], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000182.npy"} +{"epoch": 0.26725403817914833, "step": 183, "batch_size": 64, "mean": 45.301448822021484, "std": 58.470298767089844, "min": -51.359886169433594, "p10": -32.01211910247802, "median": 37.68613624572754, "p90": 119.60664215087893, "max": 190.54354858398438, "pos_frac": 0.796875, "sample": [51.05110168457031, 52.902183532714844, 9.96800422668457, 38.880767822265625, -0.16158676147460938, 121.74099731445312, 83.76423645019531, 97.00853729248047, 53.59654235839844, 34.18437957763672, -3.955841064453125, 86.94468688964844, 36.49150466918945, 12.48552131652832, 6.838348388671875, 50.31182861328125, 182.74366760253906, 173.74603271484375, 100.97798919677734, 27.093347549438477, 40.229515075683594, 57.088199615478516, 162.7062225341797, 85.13040161132812, 17.091644287109375, 3.782135009765625, 57.81968307495117, 27.68606948852539, 15.923721313476562, -43.86393737792969, 99.90135955810547, 52.79003143310547, -23.710824966430664, 5.727630615234375, 83.1921615600586, 2.6781845092773438, 39.76211166381836, 107.1265640258789, 101.26871490478516, 173.17230224609375, -37.10066223144531, -5.979301452636719, 190.54354858398438, -35.56981658935547, 23.707603454589844, -12.02532958984375, 73.10598754882812, -40.68104553222656, -43.90489196777344, -45.7249755859375, 122.37203216552734, 62.24134826660156, 114.62648010253906, 97.49541473388672, 27.613033294677734, 33.75518035888672, 12.942329406738281, 1.839111328125, 50.06145477294922, 55.39621353149414, -51.359886169433594, -4.273944854736328, 13.409156799316406, 14.6895751953125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000183.npy"} +{"epoch": 0.2687224669603524, "step": 184, "batch_size": 64, "mean": 41.32544708251953, "std": 56.777896881103516, "min": -46.745323181152344, "p10": -15.367185974121085, "median": 28.023112297058105, "p90": 116.9621795654297, "max": 202.81948852539062, "pos_frac": 0.78125, "sample": [12.773895263671875, -37.409690856933594, 39.02855682373047, 58.98759460449219, 102.94984436035156, 71.13092041015625, -1.2115936279296875, 5.245399475097656, 78.86397552490234, -5.934745788574219, 201.16983032226562, 165.55792236328125, 33.203712463378906, 102.8162841796875, 6.5767669677734375, 93.00713348388672, 28.470794677734375, 88.72578430175781, 68.96087646484375, 140.97230529785156, -2.2889442443847656, 155.20645141601562, -24.057403564453125, 9.571479797363281, 15.418575286865234, 35.248321533203125, 27.575429916381836, 5.684822082519531, 75.30892181396484, 9.763021469116211, 39.79151916503906, 32.0474853515625, 97.08372497558594, 117.87886047363281, -42.632022857666016, 132.43460083007812, 8.352073669433594, -19.40966033935547, -3.1020545959472656, 26.533000946044922, 66.1298599243164, -4.365997314453125, 6.812288284301758, 8.387712478637695, 30.042877197265625, 114.82325744628906, 33.59503173828125, 34.056453704833984, -5.181358337402344, -38.599884033203125, -31.372486114501953, 39.56914520263672, 9.408069610595703, 15.379234313964844, -46.745323181152344, 20.432052612304688, 8.730697631835938, 107.33782958984375, 17.286579132080078, 33.89885711669922, 11.688774108886719, -3.559478759765625, 63.96097946166992, 202.81948852539062], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000184.npy"} +{"epoch": 0.2701908957415565, "step": 185, "batch_size": 64, "mean": 34.560455322265625, "std": 58.581912994384766, "min": -89.8758544921875, "p10": -20.70207138061523, "median": 25.35987091064453, "p90": 122.12915420532227, "max": 234.51132202148438, "pos_frac": 0.71875, "sample": [-89.8758544921875, -37.17653274536133, 4.583639144897461, 65.07760620117188, 52.1484260559082, 19.67357063293457, -4.327569961547852, 31.938011169433594, 150.27053833007812, 27.553070068359375, 9.74392318725586, 66.6111068725586, 65.685302734375, 7.2486724853515625, 125.83447265625, 46.55250549316406, -8.75283432006836, 24.507949829101562, 52.35136413574219, -31.474044799804688, -3.3485679626464844, 36.214656829833984, 9.503768920898438, 43.53675842285156, 40.63364791870117, 26.2117919921875, -33.90211486816406, 30.865760803222656, 33.547080993652344, -35.73126220703125, 75.66744995117188, 0.07932662963867188, 81.9054946899414, 69.88252258300781, -0.97882080078125, 185.2622528076172, -22.666648864746094, 57.19562530517578, 123.19287109375, -2.286510467529297, 8.57883071899414, 15.700231552124023, -7.975257873535156, -12.900482177734375, -6.37489128112793, 12.690689086914062, 122.64339447021484, 110.09545135498047, 100.33059692382812, 11.031431198120117, 234.51132202148438, 9.385433197021484, 28.755889892578125, 35.07114791870117, -6.6412200927734375, 135.3046112060547, 18.382415771484375, 74.01953125, -86.55291748046875, 120.92926025390625, -16.118057250976562, 26.68714141845703, 7.43499755859375, -16.078807830810547], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000185.npy"} +{"epoch": 0.27165932452276065, "step": 186, "batch_size": 64, "mean": 61.284664154052734, "std": 60.45098114013672, "min": -65.31378173828125, "p10": 2.5199930191040045, "median": 49.52690315246582, "p90": 141.94999542236334, "max": 240.25340270996094, "pos_frac": 0.90625, "sample": [-65.31378173828125, 88.35111999511719, 164.70010375976562, 80.75938415527344, 85.01419830322266, 21.069915771484375, 56.9520263671875, 20.078296661376953, 151.36972045898438, 115.48519897460938, 30.796127319335938, 126.88522338867188, 8.518867492675781, 42.49562072753906, 33.65689468383789, 4.755496978759766, 54.14697265625, -43.8994140625, 30.160045623779297, 6.426218032836914, 125.97781372070312, 26.714481353759766, 20.06079864501953, 98.20536804199219, 28.63885498046875, 45.78969192504883, 58.701622009277344, 109.09004974365234, 160.07440185546875, 41.918914794921875, 123.7329330444336, 2.2667598724365234, 94.388916015625, 37.150089263916016, 29.663915634155273, 85.06283569335938, 214.88931274414062, 148.4063262939453, 56.14939498901367, -6.952976226806641, 90.66138458251953, 18.3142032623291, 66.62623596191406, 34.311279296875, 9.706741333007812, 240.25340270996094, -2.0658416748046875, 64.72313690185547, -8.375213623046875, 3.110870361328125, 219.70458984375, 111.36638641357422, 41.63710021972656, 102.69003295898438, 21.270751953125, 94.3501968383789, 45.67488479614258, -18.854408264160156, 3.2364959716796875, 66.22355651855469, 71.11976623535156, 63.88111877441406, 17.04995346069336, 53.26411437988281], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000186.npy"} +{"epoch": 0.27312775330396477, "step": 187, "batch_size": 64, "mean": 48.679298400878906, "std": 61.61822509765625, "min": -80.15673065185547, "p10": -12.212477874755857, "median": 40.83677673339844, "p90": 133.52931823730472, "max": 220.56646728515625, "pos_frac": 0.84375, "sample": [-80.15673065185547, 151.65074157714844, -3.3509902954101562, 42.09708786010742, -52.93558883666992, 84.28803253173828, 61.540985107421875, 62.6593017578125, 34.44438171386719, 46.61378860473633, 54.097476959228516, 61.60517120361328, 171.8936309814453, 35.160987854003906, 46.21253967285156, 100.1298828125, 107.22628784179688, 15.469001770019531, 33.22783660888672, -35.5882453918457, 42.846405029296875, 14.635604858398438, 220.56646728515625, 3.3478546142578125, 86.1526107788086, 41.54498291015625, 162.17617797851562, -24.343242645263672, 113.41729736328125, 17.799707412719727, -13.336029052734375, 75.24158477783203, 13.028793334960938, -33.795135498046875, -8.22471809387207, 135.66799926757812, 5.087249755859375, 14.857877731323242, -9.590858459472656, 107.53572845458984, 107.38523864746094, 216.0020294189453, 40.128570556640625, 13.313484191894531, 69.1971435546875, 21.361949920654297, 18.482852935791016, 19.243656158447266, -77.4415283203125, 36.429222106933594, 32.889129638671875, 4.951377868652344, 63.41102600097656, 60.816650390625, 60.782325744628906, 76.61532592773438, 68.0423583984375, 8.750267028808594, 42.24322509765625, 159.08985900878906, 128.5390625, 34.928131103515625, 4.270444869995117, 5.141452789306641], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000187.npy"} +{"epoch": 0.2745961820851689, "step": 188, "batch_size": 64, "mean": 61.76814651489258, "std": 67.51439666748047, "min": -88.1827163696289, "p10": -4.030036163330077, "median": 42.225460052490234, "p90": 151.84251098632814, "max": 254.72694396972656, "pos_frac": 0.859375, "sample": [168.49429321289062, 59.94786834716797, 141.8634490966797, 19.492061614990234, 75.96900177001953, 11.735130310058594, -16.42181396484375, -33.848915100097656, 152.15403747558594, 76.91200256347656, 42.53623580932617, -10.520332336425781, 65.60742950439453, 55.13481903076172, 148.80284118652344, 128.76014709472656, 53.812950134277344, -0.5020294189453125, 43.687896728515625, 5.5294342041015625, 12.121810913085938, 9.958797454833984, 52.3756103515625, 103.25017547607422, 20.842544555664062, 34.920677185058594, 29.114334106445312, 8.93685531616211, 188.73733520507812, 16.63994598388672, 236.0084686279297, -88.1827163696289, 154.24383544921875, 84.25180053710938, 118.7120361328125, 66.19032287597656, 15.129402160644531, 145.13992309570312, -3.110271453857422, 134.78204345703125, 74.06326293945312, 27.90252685546875, -44.259307861328125, 16.70042610168457, 138.23202514648438, 30.445175170898438, 151.11561584472656, 11.955516815185547, 35.99168395996094, 180.9111328125, 23.59502410888672, 106.34001159667969, 41.9146842956543, -8.487724304199219, 102.08192443847656, 41.52601623535156, 74.67190551757812, 254.72694396972656, 69.80899047851562, 33.081336975097656, 39.64935302734375, -4.424221038818359, 15.910026550292969, 10.49960708618164], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000188.npy"} +{"epoch": 0.27606461086637296, "step": 189, "batch_size": 64, "mean": 54.37678527832031, "std": 73.06163024902344, "min": -104.97945404052734, "p10": -20.23615169525146, "median": 39.93207550048828, "p90": 129.8953628540039, "max": 323.9915771484375, "pos_frac": 0.8125, "sample": [22.664344787597656, 127.97898864746094, -45.45184326171875, 188.27020263671875, -8.610855102539062, -35.63409423828125, 29.466249465942383, -1.4743385314941406, -17.713830947875977, 59.17152404785156, 96.89126586914062, 108.53754425048828, 88.78486633300781, 32.72422790527344, 113.6351089477539, 45.416595458984375, 72.53307342529297, 24.08597183227539, -38.26409912109375, 9.994037628173828, -26.36302375793457, 45.19178771972656, 65.66233825683594, 128.58099365234375, 97.55397033691406, 32.85302734375, -21.31714630126953, 39.4658088684082, 20.63680648803711, 151.43191528320312, 15.748554229736328, 31.266508102416992, 75.88963317871094, 234.42327880859375, 13.51348876953125, 11.543071746826172, 7.942499160766602, 170.2534942626953, 60.572113037109375, 26.295578002929688, -72.22572326660156, 81.94330596923828, 6.96173095703125, 46.90288543701172, 22.637737274169922, 222.4275665283203, 126.20277404785156, 323.9915771484375, -4.068408966064453, 94.47761535644531, 130.4586639404297, 23.479080200195312, 40.39834213256836, 90.42280578613281, 41.936927795410156, 81.77460479736328, 101.733642578125, 58.577457427978516, 17.328704833984375, -104.97945404052734, 10.244216918945312, -5.284904479980469, 3.1875152587890625, 87.43583679199219], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000189.npy"} +{"epoch": 0.2775330396475771, "step": 190, "batch_size": 64, "mean": 58.45063400268555, "std": 74.16045379638672, "min": -92.90019226074219, "p10": -20.17230682373047, "median": 47.58584213256836, "p90": 164.44373931884772, "max": 258.8338317871094, "pos_frac": 0.8125, "sample": [-2.2465362548828125, 80.8827896118164, 73.39888000488281, 37.621559143066406, 17.18431854248047, -31.007545471191406, 128.39205932617188, 30.60814666748047, 7.914924621582031, 172.4183349609375, 56.022613525390625, 30.80255126953125, 58.2908935546875, 30.506072998046875, 225.90113830566406, 68.79090881347656, 44.891937255859375, 53.99162292480469, 120.7601089477539, 38.1441650390625, 2.497852325439453, 169.858154296875, -82.89970397949219, 51.99675750732422, 146.96051025390625, 8.641036987304688, 28.1527099609375, 171.0160369873047, 136.5296173095703, 63.491050720214844, 4.5275726318359375, 151.8101043701172, 7.874153137207031, -4.107614517211914, 62.82324981689453, 76.38035583496094, -20.271934509277344, -8.4124755859375, -82.63915252685547, 24.620040893554688, 103.96831512451172, -19.939842224121094, 77.60332489013672, 19.664901733398438, -12.750215530395508, -27.188217163085938, 88.15644836425781, 150.30941772460938, -31.116851806640625, 50.279747009277344, -92.90019226074219, 216.00762939453125, 51.71235656738281, 33.079368591308594, 117.44255065917969, 96.85181427001953, 42.237060546875, 258.8338317871094, 1.501089096069336, 199.022216796875, 62.76551818847656, 146.4702911376953, 25.960865020751953, 30.7518310546875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000190.npy"} +{"epoch": 0.2790014684287812, "step": 191, "batch_size": 64, "mean": 83.02128601074219, "std": 83.24612426757812, "min": -56.31103515625, "p10": 3.3427766799926766, "median": 62.98737716674805, "p90": 197.6971832275391, "max": 347.0641174316406, "pos_frac": 0.9375, "sample": [209.00222778320312, 61.436126708984375, 307.083984375, 10.324134826660156, 162.82235717773438, 57.89628601074219, 97.45515441894531, 184.34039306640625, 2.939023971557617, 201.8046875, 67.13011169433594, 76.86461639404297, 32.235145568847656, 135.49359130859375, 154.70120239257812, 11.823982238769531, 20.932218551635742, 46.20094299316406, 45.44221496582031, 64.53862762451172, 130.54347229003906, 44.86163330078125, 30.33367919921875, 173.29574584960938, 80.42211151123047, 22.243881225585938, 1.040456771850586, 321.8416748046875, 53.34526062011719, 17.57750701904297, 109.73789978027344, -44.763362884521484, 14.587615966796875, -6.045543670654297, 58.29339599609375, 97.8970947265625, 75.57713317871094, 93.40309143066406, 221.22105407714844, 17.95694923400879, 0.1626415252685547, 117.68276977539062, 140.45083618164062, -56.31103515625, 17.22577667236328, 347.0641174316406, 45.028038024902344, -6.828926086425781, 64.8353271484375, 124.23916625976562, 81.68312072753906, 188.11300659179688, 35.19337463378906, 7.8000946044921875, 86.90800476074219, 66.91120910644531, 11.145164489746094, 95.542236328125, 4.2848663330078125, 55.230140686035156, 223.2748565673828, 53.33000183105469, 51.20758819580078, 95.35175323486328], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000191.npy"} +{"epoch": 0.28046989720998533, "step": 192, "batch_size": 64, "mean": 59.959373474121094, "std": 73.8494644165039, "min": -78.76461791992188, "p10": -17.994886016845705, "median": 49.22832107543945, "p90": 179.17081451416018, "max": 258.2056579589844, "pos_frac": 0.8125, "sample": [4.143486022949219, -40.196800231933594, 49.78596496582031, 102.51033020019531, 113.72765350341797, 20.949691772460938, -15.167976379394531, -13.623327255249023, 20.463836669921875, -78.76461791992188, 18.862075805664062, 44.876930236816406, -18.117889404296875, 116.68742370605469, 42.29447937011719, 12.270004272460938, -9.908203125, 53.95549774169922, -19.230316162109375, 89.19073486328125, -38.440773010253906, 214.73367309570312, -28.827190399169922, 55.33270263671875, 3.5655059814453125, 198.28306579589844, 190.97605895996094, 258.2056579589844, 33.41543960571289, 181.14013671875, 50.82684326171875, 60.069087982177734, 257.8740234375, 73.95941162109375, 102.34571838378906, 9.314258575439453, 91.48780822753906, 132.5137176513672, 128.06155395507812, 18.156116485595703, 57.01792907714844, 174.5757293701172, 12.555253982543945, 17.83147430419922, 90.32862091064453, 55.99968719482422, 48.670677185058594, -51.04508972167969, 100.4415283203125, 41.47632598876953, 64.63502502441406, -1.06842041015625, 46.91065216064453, -17.70787811279297, 10.939702987670898, 136.93252563476562, 56.8797607421875, 13.041847229003906, 0.378692626953125, 117.77104187011719, 96.03898620605469, 64.38755798339844, 22.282806396484375, 190.42367553710938], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000192.npy"} +{"epoch": 0.28193832599118945, "step": 193, "batch_size": 64, "mean": 35.77876281738281, "std": 67.36434173583984, "min": -201.79148864746094, "p10": -22.45701332092285, "median": 44.36389923095703, "p90": 99.50141906738286, "max": 218.44395446777344, "pos_frac": 0.71875, "sample": [68.96652221679688, -9.288658142089844, 27.149948120117188, 81.35313415527344, -201.79148864746094, 70.4583511352539, 70.15396881103516, 51.64341735839844, -160.55653381347656, 162.72259521484375, 61.83314514160156, -9.214559555053711, 73.25698852539062, -8.46645736694336, 30.737899780273438, 53.6093864440918, 41.886085510253906, 54.345069885253906, -2.4269027709960938, 73.41424560546875, 2.3053035736083984, 6.122459411621094, 30.544185638427734, 54.86083221435547, 19.25922203063965, 79.11126708984375, -11.983409881591797, 46.841712951660156, 8.4512939453125, -10.235816955566406, -22.052413940429688, 125.96080017089844, -13.855819702148438, -21.743194580078125, 111.13716888427734, -50.35704803466797, 81.51317596435547, -15.472084045410156, 3.456613540649414, 49.36534881591797, 6.581817626953125, 69.95423126220703, 104.69770050048828, 69.4476547241211, 60.080345153808594, -78.481201171875, 186.7400360107422, 82.37364959716797, 59.253082275390625, 72.48808288574219, -34.852699279785156, 56.888336181640625, -22.630413055419922, -14.456344604492188, 24.545989990234375, 87.37676239013672, 218.44395446777344, 83.81645202636719, -32.359615325927734, 20.915199279785156, 144.736572265625, 27.96356201171875, 33.220550537109375, 60.081275939941406], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000193.npy"} +{"epoch": 0.2834067547723935, "step": 194, "batch_size": 64, "mean": 56.68800354003906, "std": 55.03363037109375, "min": -32.45718765258789, "p10": -6.572073364257808, "median": 47.323591232299805, "p90": 146.1473327636719, "max": 173.16807556152344, "pos_frac": 0.84375, "sample": [87.38688659667969, -1.5147247314453125, 25.784887313842773, 34.640785217285156, 48.340606689453125, 131.0367431640625, 25.93360137939453, 111.16719055175781, -2.6620025634765625, 2.478588104248047, 19.492761611938477, 159.28204345703125, -20.398056030273438, 20.642578125, 37.275123596191406, 18.841346740722656, 98.44823455810547, 30.497970581054688, 46.306575775146484, 90.14097595214844, 24.06174087524414, 51.67827606201172, 4.747951507568359, -1.0389442443847656, 81.16594696044922, -32.45718765258789, 1.2864742279052734, 163.44775390625, 54.65300750732422, 80.97370910644531, 168.38385009765625, 94.94831848144531, 149.39085388183594, 98.50228118896484, 18.52094268798828, 173.16807556152344, 51.439327239990234, 14.183675765991211, 88.01175689697266, 32.440391540527344, 130.1387481689453, -11.557662963867188, 12.118623733520508, 2.9749984741210938, 74.224365234375, 113.45555877685547, 6.496479034423828, -11.690414428710938, 33.11328125, -8.247817993164062, 63.518333435058594, -22.421485900878906, 34.71629333496094, 149.7812042236328, 32.6590690612793, 167.58209228515625, 77.76654052734375, 73.45166015625, 138.57911682128906, 65.28620910644531, 112.61746978759766, 68.98391723632812, 54.24696350097656, -10.391502380371094], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000194.npy"} +{"epoch": 0.28487518355359764, "step": 195, "batch_size": 64, "mean": 59.09135437011719, "std": 61.74763488769531, "min": -50.264129638671875, "p10": -15.958446502685543, "median": 55.788265228271484, "p90": 124.30222473144534, "max": 260.6269226074219, "pos_frac": 0.859375, "sample": [60.68133544921875, 29.12579345703125, 140.72654724121094, 42.88442611694336, 54.589599609375, 204.7423095703125, 28.11591339111328, 167.36448669433594, 88.314453125, -2.1010208129882812, -12.666275024414062, 39.15202331542969, 21.357521057128906, -50.264129638671875, 83.420654296875, 69.05101013183594, 70.54878234863281, 126.92826843261719, -24.995811462402344, 242.3383331298828, 0.9611854553222656, 68.96642303466797, -35.42154312133789, 13.995452880859375, 108.45301818847656, 70.65584564208984, 108.16815185546875, 139.48338317871094, 80.87136840820312, 112.4611587524414, 73.9425277709961, -33.53929901123047, 65.38362121582031, 104.81370544433594, 29.289154052734375, 85.05307006835938, 3.0824832916259766, 8.777244567871094, 84.4754638671875, 118.17478942871094, -17.36937713623047, 114.70193481445312, 29.982112884521484, 30.973960876464844, 260.6269226074219, 93.75916290283203, 56.98693084716797, 48.565765380859375, -27.09015655517578, 30.53577995300293, 67.16931915283203, 7.427314758300781, 106.96694946289062, 64.61705017089844, 27.262649536132812, 21.50457763671875, 33.72529983520508, 35.44255828857422, 26.784521102905273, 60.62067413330078, -31.736923217773438, 24.871448516845703, 88.68746948242188, 39.46913146972656], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000195.npy"} +{"epoch": 0.28634361233480177, "step": 196, "batch_size": 64, "mean": 68.53253173828125, "std": 75.67615509033203, "min": -69.24348449707031, "p10": -10.43911323547363, "median": 48.51742935180664, "p90": 175.6976455688477, "max": 251.19139099121094, "pos_frac": 0.859375, "sample": [140.93560791015625, 148.5744171142578, 146.5174560546875, 125.25668334960938, -8.060295104980469, 167.0811004638672, 153.93141174316406, 218.48390197753906, 145.43594360351562, 45.16212463378906, 15.380794525146484, 41.21393585205078, 145.88955688476562, 156.3374481201172, 40.102691650390625, 21.232271194458008, -11.458606719970703, 39.80033874511719, 4.513671875, 62.38341522216797, 251.19139099121094, -7.994873046875, 30.135711669921875, 29.59878158569336, 68.09712219238281, 52.05570983886719, 80.57217407226562, 106.8166275024414, 28.34264373779297, 2.624044418334961, 113.06605529785156, 46.96369934082031, 233.01361083984375, 221.7476043701172, -22.068435668945312, 80.8402099609375, 57.7282829284668, 48.23631286621094, 38.98923110961914, 42.09532165527344, 48.798545837402344, 56.21356201171875, 180.368896484375, 0.10704612731933594, 85.22003936767578, 21.364459991455078, 30.704078674316406, -45.67254638671875, 34.56903839111328, -66.79754638671875, -45.89483642578125, 79.4874038696289, 53.43568420410156, 23.11062240600586, 178.6173858642578, -30.6630859375, -69.24348449707031, 50.41333770751953, 23.254348754882812, 95.4816665649414, 199.7069091796875, 168.88491821289062, 7.29925537109375, 6.551286697387695], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000196.npy"} +{"epoch": 0.2878120411160059, "step": 197, "batch_size": 64, "mean": 54.854408264160156, "std": 61.29789352416992, "min": -70.1456069946289, "p10": -0.07284832000732364, "median": 43.8045711517334, "p90": 130.58452148437502, "max": 244.7076416015625, "pos_frac": 0.890625, "sample": [26.884286880493164, 212.59002685546875, 67.21690368652344, 13.557380676269531, 97.75569152832031, -0.3172340393066406, 30.040809631347656, -23.829513549804688, 42.90142059326172, 213.99716186523438, 244.7076416015625, 117.76863098144531, 52.01165008544922, 58.341644287109375, -35.472816467285156, 4.149143218994141, 39.33406066894531, 175.9606170654297, 47.00470733642578, 44.70772171020508, 83.18753814697266, 114.74528503417969, 48.9799919128418, 9.266342163085938, 42.76292419433594, 8.369447708129883, -25.266517639160156, 64.37480163574219, 97.25786590576172, 81.97110748291016, 175.281005859375, 67.98077392578125, 41.170501708984375, 105.16119384765625, 0.84759521484375, 101.64372253417969, 9.218841552734375, 1.0894775390625, 0.576873779296875, -70.1456069946289, 17.720041275024414, 78.95207977294922, 0.49738502502441406, 26.128761291503906, 10.39521598815918, 34.29571533203125, 123.59652709960938, 90.4720230102539, 39.74195861816406, 133.57937622070312, 1.8899517059326172, 51.28215408325195, 52.825050354003906, 24.10308074951172, 0.7467575073242188, 70.58953094482422, 84.27217102050781, 60.22682189941406, 136.94842529296875, -13.080101013183594, -6.0267791748046875, 26.908035278320312, 12.97735595703125, 63.85765838623047], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000197.npy"} +{"epoch": 0.28928046989721, "step": 198, "batch_size": 64, "mean": 65.08197784423828, "std": 77.35123443603516, "min": -70.98700714111328, "p10": -2.1354936599731436, "median": 50.38179397583008, "p90": 143.06849975585942, "max": 352.4875793457031, "pos_frac": 0.859375, "sample": [-70.98700714111328, 17.988983154296875, 14.365961074829102, 278.0126037597656, 74.58389282226562, 120.23482513427734, 155.92898559570312, 30.086002349853516, 73.27699279785156, 52.129493713378906, 96.3707275390625, 72.90902709960938, -54.76385498046875, 41.68238067626953, 21.74176025390625, 91.32127380371094, 92.5345687866211, 64.41705322265625, 56.628204345703125, 58.84851837158203, -0.9385280609130859, 6.7006072998046875, 27.930177688598633, 47.10575866699219, 11.727888107299805, 36.07971954345703, -21.79723358154297, 130.25692749023438, 68.21324920654297, 93.79287719726562, 63.175201416015625, 59.645591735839844, 37.846466064453125, 26.623092651367188, 70.69160461425781, 9.425968170166016, 40.58342361450195, 100.02875518798828, 71.68167114257812, 148.55917358398438, -43.1571044921875, 318.06524658203125, 42.73839569091797, 44.926673889160156, -1.1066703796386719, 20.879119873046875, 167.02899169921875, -8.389253616333008, 6.059516906738281, 40.14204406738281, -4.324115753173828, 48.63409423828125, 84.10928344726562, 90.26870727539062, 14.96833610534668, 113.96144104003906, 93.29957580566406, 118.86145782470703, -2.576417922973633, 246.29367065429688, 7.415840148925781, 352.4875793457031, 30.954980850219727, 69.06253051757812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000198.npy"} +{"epoch": 0.2907488986784141, "step": 199, "batch_size": 64, "mean": 62.41845703125, "std": 73.17977142333984, "min": -84.23540496826172, "p10": -17.888327026367186, "median": 45.95085906982422, "p90": 159.15824279785159, "max": 253.43829345703125, "pos_frac": 0.828125, "sample": [100.85236358642578, 102.26264953613281, -43.078704833984375, 2.2125701904296875, 146.7047119140625, 59.992164611816406, -22.93883514404297, 135.19601440429688, -84.23540496826172, 95.6661605834961, 26.79279327392578, -17.800933837890625, -42.407867431640625, 101.60000610351562, 142.825927734375, 63.935768127441406, 5.127937316894531, 81.01598358154297, 214.94480895996094, 27.07077980041504, 29.459426879882812, 109.19476318359375, 186.19329833984375, 156.0777130126953, -54.196720123291016, 10.261016845703125, -2.9850692749023438, 55.83983612060547, 132.96575927734375, 24.979873657226562, 21.025894165039062, 160.4784698486328, 247.08815002441406, 41.75213623046875, 25.75403594970703, 134.77810668945312, -17.92578125, 67.95710754394531, 1.949859619140625, 113.06939697265625, 18.541648864746094, 104.81192016601562, 88.24105072021484, 19.34365463256836, 168.20703125, 43.638511657714844, 41.958309173583984, 8.266019821166992, -1.1174278259277344, -67.91431427001953, 122.8696517944336, 36.846649169921875, 38.211669921875, 78.10713958740234, 164.7650604248047, 74.11454772949219, 5.633138656616211, 58.30717468261719, 10.260704040527344, 32.72825622558594, 48.263206481933594, 111.56754302978516, 253.43829345703125, -3.7642364501953125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000199.npy"} +{"epoch": 0.2922173274596182, "step": 200, "batch_size": 64, "mean": 67.68182373046875, "std": 83.56586456298828, "min": -138.7576904296875, "p10": -15.21016502380371, "median": 44.21761894226074, "p90": 186.70709075927735, "max": 344.84088134765625, "pos_frac": 0.8125, "sample": [172.180908203125, 105.05773162841797, 52.22456741333008, 193.3515625, 35.31356430053711, 92.4345474243164, 10.247390747070312, 5.494781494140625, 231.74517822265625, -26.041980743408203, 200.08929443359375, 145.361083984375, 69.65985107421875, -15.91830825805664, 41.92837142944336, 187.44650268554688, 143.0231475830078, 211.3748016357422, 29.808944702148438, 151.4710693359375, 36.77714538574219, 19.2549991607666, 111.8492431640625, 123.54173278808594, -4.636722564697266, 55.033294677734375, 101.74601745605469, -31.167526245117188, 344.84088134765625, 26.407493591308594, 69.09666442871094, -16.334938049316406, 6.221214294433594, 27.12371826171875, 35.58081817626953, 5.711755752563477, -138.7576904296875, 46.506866455078125, 83.88211059570312, 244.14315795898438, 16.07587242126465, 71.30205535888672, 36.11189270019531, 25.752159118652344, -2.0868988037109375, -65.54354858398438, 57.133148193359375, -13.557830810546875, 142.53521728515625, -7.66206169128418, 16.871200561523438, 184.98179626464844, 35.11677551269531, 24.132911682128906, 36.28263854980469, 148.4742889404297, 112.26602172851562, 80.81037139892578, 38.363983154296875, 72.46064758300781, -51.30653762817383, 102.14115905761719, -12.211997985839844, 100.12036895751953], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000200.npy"} +{"epoch": 0.2936857562408223, "step": 201, "batch_size": 64, "mean": 68.60670471191406, "std": 79.2151107788086, "min": -108.25552368164062, "p10": -13.617455291748037, "median": 55.930559158325195, "p90": 159.18893280029297, "max": 316.45849609375, "pos_frac": 0.875, "sample": [121.73454284667969, 37.65607452392578, 120.12677001953125, -96.37493896484375, 37.880516052246094, 2.780181884765625, 16.624231338500977, 12.548444747924805, -2.6230087280273438, 31.258087158203125, 1.178436279296875, 32.276336669921875, 17.7169189453125, 16.51763916015625, 232.98556518554688, 81.77293395996094, 46.525146484375, 67.16448974609375, 22.928611755371094, 160.63934326171875, 224.08566284179688, 128.23831176757812, 22.998870849609375, 48.78851318359375, 86.57125854492188, 208.51959228515625, -33.276668548583984, 101.00233459472656, 119.01839447021484, 12.914670944213867, 60.238521575927734, 99.24130249023438, 107.4495620727539, 102.49339294433594, -18.329360961914062, -24.40230941772461, 139.99783325195312, 37.075599670410156, 34.04906463623047, 316.45849609375, 66.19781494140625, 1.6449813842773438, 66.81957244873047, 41.16413879394531, 22.80537986755371, 130.5143585205078, 51.622596740722656, 134.26129150390625, 208.43716430664062, 62.899147033691406, 68.314697265625, -29.41260528564453, 83.50276184082031, 70.92089080810547, -108.25552368164062, 155.8046417236328, 265.7178649902344, 102.64495086669922, 23.549875259399414, -35.13243103027344, 127.2340087890625, 23.606788635253906, 23.89935302734375, 97.61829376220703], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000201.npy"} +{"epoch": 0.29515418502202645, "step": 202, "batch_size": 64, "mean": 48.29637145996094, "std": 73.00769805908203, "min": -103.914794921875, "p10": -35.35870323181152, "median": 45.9479866027832, "p90": 158.03231506347657, "max": 240.3557891845703, "pos_frac": 0.734375, "sample": [240.3557891845703, -20.928565979003906, 71.63833618164062, 159.6493377685547, 5.13630485534668, 21.44438934326172, 87.14049530029297, -11.077913284301758, 84.43252563476562, 69.41584777832031, 32.21721649169922, -84.98011779785156, 79.4046859741211, 173.71652221679688, 92.56102752685547, 61.071922302246094, -103.914794921875, 19.32037353515625, 59.557289123535156, 206.0647735595703, 32.007545471191406, 154.25926208496094, -43.15131378173828, 124.2025146484375, 11.283729553222656, 1.7503280639648438, 216.33966064453125, -0.6923084259033203, 61.49501037597656, 98.364013671875, 55.057037353515625, 103.98042297363281, 160.11441040039062, 3.6952285766601562, -9.760147094726562, -28.609848022460938, 45.81721496582031, 45.67738342285156, 46.53197479248047, 116.401123046875, 83.26417541503906, -32.42689514160156, 11.469635009765625, 169.4273681640625, 59.96875, 102.29202270507812, 65.4931411743164, -85.43658447265625, 46.078758239746094, 25.98834228515625, 13.86562728881836, 26.835538864135742, -14.376495361328125, -23.48181915283203, 16.95977783203125, -36.61519241333008, 50.67627716064453, -37.000282287597656, -23.020633697509766, 55.51866912841797, 147.00198364257812, -39.8359489440918, 71.34818267822266, -0.015140533447265625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000202.npy"} +{"epoch": 0.2966226138032305, "step": 203, "batch_size": 64, "mean": 57.30005645751953, "std": 61.76263427734375, "min": -53.45024871826172, "p10": -11.64541339874267, "median": 44.251731872558594, "p90": 140.3378463745118, "max": 248.6304931640625, "pos_frac": 0.84375, "sample": [20.57144546508789, 90.10379028320312, 122.97219848632812, 15.392021179199219, -7.113620758056641, 159.91244506835938, -27.40396499633789, 41.96791076660156, 27.608476638793945, 90.250732421875, 12.075050354003906, 74.68549346923828, 84.36376953125, 57.478919982910156, -41.10581588745117, 88.24479675292969, 31.639572143554688, -6.758056640625, 116.71731567382812, 64.46099090576172, 180.43203735351562, 197.38394165039062, 6.517112731933594, 41.6972770690918, 61.521812438964844, 43.481834411621094, 248.6304931640625, 14.645030975341797, -33.021514892578125, 67.19551086425781, -2.0853500366210938, 95.0560073852539, 76.43238830566406, 175.21728515625, 63.37181091308594, 109.68124389648438, 74.07841491699219, 83.37876892089844, 17.447208404541016, 28.584999084472656, 16.1768798828125, 61.512062072753906, 45.021629333496094, 110.06613159179688, 4.052238464355469, -53.45024871826172, -24.672378540039062, 36.72243118286133, -13.587610244750977, 14.353355407714844, 123.71826171875, 31.978919982910156, 42.500797271728516, 33.730247497558594, 30.72576904296875, -26.41619873046875, 87.71387481689453, 45.55741882324219, 112.07870483398438, 188.65478515625, 16.14803123474121, 147.4605255126953, 47.44146728515625, 24.006752014160156], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000203.npy"} +{"epoch": 0.29809104258443464, "step": 204, "batch_size": 64, "mean": 53.61920166015625, "std": 77.09075927734375, "min": -61.266868591308594, "p10": -21.989409255981442, "median": 37.14126014709473, "p90": 130.87333068847656, "max": 310.7162780761719, "pos_frac": 0.75, "sample": [94.66546630859375, 38.66364669799805, 5.357662200927734, 28.996110916137695, 77.9148941040039, 35.618873596191406, 101.78924560546875, -61.266868591308594, 81.53450775146484, 29.385774612426758, -19.771678924560547, 23.85638999938965, 8.237525939941406, 55.26446533203125, 46.94755554199219, -0.5712966918945312, -10.340118408203125, 106.8483657836914, 28.542572021484375, 6.756280899047852, 130.10804748535156, -40.117942810058594, -5.8897552490234375, -23.973384857177734, 15.942840576171875, -5.69639778137207, 45.30084228515625, 112.80027770996094, 131.20130920410156, -11.049583435058594, 288.37066650390625, 10.211723327636719, -9.963607788085938, 14.610296249389648, 1.7278594970703125, 105.88978576660156, 34.82292175292969, 44.179664611816406, 275.87493896484375, -56.29185485839844, 78.19956970214844, 118.99478149414062, 55.472564697265625, 57.07770919799805, 43.880523681640625, 11.032560348510742, 30.54584503173828, 104.08218383789062, 188.3790283203125, -11.05889892578125, 43.712242126464844, 68.57088470458984, -39.4625244140625, 54.456138610839844, 310.7162780761719, 200.37078857421875, -14.340063095092773, 85.88594818115234, 27.543027877807617, 177.31100463867188, -22.939865112304688, 62.92866516113281, 88.69512939453125, -24.912673950195312], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000204.npy"} +{"epoch": 0.29955947136563876, "step": 205, "batch_size": 64, "mean": 66.7889404296875, "std": 76.3533706665039, "min": -109.94151306152344, "p10": -32.400323104858394, "median": 57.07492446899414, "p90": 157.94884643554687, "max": 234.97137451171875, "pos_frac": 0.796875, "sample": [22.557601928710938, 54.450103759765625, 74.40131378173828, 213.73074340820312, 42.594566345214844, 23.881227493286133, -4.73406982421875, 12.727645874023438, 37.42463684082031, 135.53167724609375, 150.15847778320312, -62.25830078125, -29.653907775878906, 74.36945343017578, -7.2479400634765625, 32.93165588378906, 3.4432525634765625, 234.90933227539062, 167.25755310058594, 28.118427276611328, 105.79459381103516, 42.87522888183594, -34.415191650390625, 94.10208129882812, 158.34393310546875, 138.03091430664062, 136.716064453125, 0.19730758666992188, 6.484428405761719, 96.41132354736328, -34.06936264038086, 234.97137451171875, 50.476470947265625, 38.67780303955078, 23.591766357421875, 75.64125061035156, 169.85238647460938, 148.91455078125, 57.29350280761719, 56.56562805175781, -109.94151306152344, 78.35299682617188, 105.79534912109375, 121.48402404785156, 73.72574615478516, -33.57735824584961, 92.94134521484375, 48.81281280517578, 155.86544799804688, 104.9297866821289, 156.0058135986328, 157.0269775390625, 127.31001281738281, 56.856346130371094, 111.09159851074219, 174.90084838867188, 73.95951080322266, -2.8207168579101562, -0.77227783203125, -100.94732666015625, -37.444061279296875, 129.94259643554688, -18.533935546875, 38.47886657714844], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000205.npy"} +{"epoch": 0.3010279001468429, "step": 206, "batch_size": 64, "mean": 73.82017517089844, "std": 77.05616760253906, "min": -92.24810791015625, "p10": -26.47042903900146, "median": 67.93180847167969, "p90": 161.91874694824222, "max": 297.36724853515625, "pos_frac": 0.84375, "sample": [83.46089172363281, 101.1102294921875, 55.55804443359375, 63.69150161743164, 297.36724853515625, 203.77053833007812, 67.1729965209961, 25.755720138549805, 141.4416046142578, 149.61187744140625, 51.337066650390625, 32.83940124511719, 29.56122398376465, 29.582504272460938, -92.24810791015625, 72.44110107421875, 66.23516845703125, 34.79338073730469, 36.389854431152344, -8.485898971557617, 109.42784118652344, -40.53064727783203, 143.02032470703125, 225.508056640625, 76.58621215820312, 107.11068725585938, 32.66233825683594, 143.7681121826172, -61.058441162109375, -61.72688293457031, 127.46720886230469, 165.98670959472656, 130.74758911132812, 111.26817321777344, 20.437089920043945, 146.70028686523438, 14.215751647949219, -43.31484603881836, 139.508056640625, -37.9630126953125, 13.394508361816406, 125.29623413085938, 27.553150177001953, 73.04228210449219, 62.19011688232422, -9.4429931640625, 68.69062042236328, 59.987327575683594, 97.4302978515625, 65.52862548828125, -29.486238479614258, -19.43354034423828, 70.4312744140625, 91.64325714111328, 226.8244171142578, 152.4268341064453, 3.701986312866211, 95.87156677246094, 97.88567352294922, 200.14218139648438, 33.01197052001953, 9.975761413574219, 200.12979125976562, 116.48966979980469], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000206.npy"} +{"epoch": 0.302496328928047, "step": 207, "batch_size": 64, "mean": 83.75286865234375, "std": 87.56193542480469, "min": -53.648582458496094, "p10": 1.1326404571533233, "median": 49.97246551513672, "p90": 201.10044403076174, "max": 339.5751953125, "pos_frac": 0.890625, "sample": [4.493053436279297, -42.97376251220703, 35.24687194824219, 26.27480697631836, 84.11980438232422, 34.67200469970703, 201.4748077392578, 12.116348266601562, 11.306137084960938, -0.149444580078125, 70.84619140625, 52.435142517089844, 154.52978515625, -5.1634368896484375, 56.79803466796875, 95.00672149658203, 48.94636535644531, 115.83922576904297, 18.596771240234375, 23.41602897644043, 35.52992630004883, 112.68529510498047, 21.237747192382812, 169.14625549316406, 13.760650634765625, -14.3763427734375, 175.64881896972656, 39.84046936035156, 18.68604278564453, 12.479320526123047, 200.2269287109375, 143.3034210205078, -24.75164794921875, 4.124172210693359, 45.96888732910156, 170.5897216796875, 27.393997192382812, 121.52285766601562, 47.871150970458984, 83.20063781738281, 161.35739135742188, 100.20091247558594, 44.77308654785156, 128.67916870117188, 136.16366577148438, 150.26953125, 50.65692138671875, 249.68898010253906, 298.037109375, 147.83058166503906, 37.84603500366211, 276.5858154296875, -22.90064811706543, 57.908409118652344, 339.5751953125, 39.1025390625, 49.01520919799805, 62.53569030761719, 49.28800964355469, 272.5830078125, -53.648582458496094, 112.53250885009766, 252.12191772460938, 18.061004638671875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000207.npy"} +{"epoch": 0.3039647577092511, "step": 208, "batch_size": 64, "mean": 67.18119049072266, "std": 77.7541732788086, "min": -122.937255859375, "p10": -2.7397247314453055, "median": 47.19576454162598, "p90": 162.70126495361328, "max": 291.1707763671875, "pos_frac": 0.890625, "sample": [-80.27058410644531, 67.3616943359375, -42.895530700683594, 99.20046997070312, 25.97744369506836, 11.902454376220703, 175.6591033935547, 18.5470027923584, 79.1380844116211, 161.38723754882812, 142.3181610107422, 45.383262634277344, 47.71389389038086, 118.89607238769531, 126.9779281616211, 51.948280334472656, 291.1707763671875, 35.16197204589844, 255.58953857421875, 106.08882141113281, 25.051925659179688, 39.64329528808594, 123.25119018554688, 89.89196014404297, 82.46212005615234, 26.857505798339844, 27.76751708984375, 27.631988525390625, -66.86723327636719, 149.77853393554688, -63.39579391479492, 37.770442962646484, 20.55569839477539, 48.186180114746094, 76.34525299072266, 163.26441955566406, 144.37440490722656, 45.53934860229492, 16.66051483154297, 109.06759643554688, 40.93548583984375, 26.766281127929688, 11.79345703125, 139.4425048828125, 46.677635192871094, 194.66964721679688, 72.74795532226562, -122.937255859375, -16.11969757080078, 55.02708435058594, 4.013519287109375, 32.84678649902344, 193.98443603515625, 61.57597351074219, 39.562103271484375, 5.0567169189453125, 23.942581176757812, 236.98519897460938, 118.59017944335938, 141.03799438476562, 100.7565689086914, 23.154132843017578, 13.628158569335938, -5.63397216796875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000208.npy"} +{"epoch": 0.3054331864904552, "step": 209, "batch_size": 64, "mean": 58.12427520751953, "std": 71.68608856201172, "min": -97.89006805419922, "p10": -16.970952987670895, "median": 51.65634727478027, "p90": 176.4848831176758, "max": 229.69332885742188, "pos_frac": 0.78125, "sample": [203.99134826660156, 50.340267181396484, 109.21406555175781, -1.079376220703125, 204.11265563964844, 212.98077392578125, 181.23410034179688, 35.725189208984375, 72.51705932617188, 53.2432861328125, -14.22652816772461, 55.615753173828125, 102.11367797851562, 24.758949279785156, 114.70536804199219, 77.0967788696289, 125.60105895996094, -58.32017517089844, 1.6595916748046875, 88.08699798583984, 119.11204528808594, 102.17520141601562, 11.627534866333008, -46.862548828125, 51.98388671875, -18.09894561767578, -13.266088485717773, 46.670692443847656, 35.54145050048828, 111.72988891601562, 59.977996826171875, 65.10577392578125, -70.3620834350586, 9.767986297607422, 37.4908447265625, 36.57794189453125, -40.64444351196289, -14.338970184326172, 229.69332885742188, 122.82548522949219, 28.01580810546875, 181.54019165039062, 14.08034896850586, -11.967308044433594, -11.9356689453125, 22.00359344482422, 68.59352111816406, 42.80427932739258, 63.86316680908203, 82.16725158691406, 84.85984802246094, 9.724424362182617, -6.022369384765625, -97.89006805419922, 97.13009643554688, 36.44151306152344, 51.32880783081055, 40.572486877441406, 62.53526306152344, -46.55126190185547, 170.2766876220703, 179.14553833007812, 95.30172729492188, 87.85781860351562], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000209.npy"} +{"epoch": 0.3069016152716593, "step": 210, "batch_size": 64, "mean": 62.687904357910156, "std": 65.9644546508789, "min": -64.02816772460938, "p10": -0.1426170349121087, "median": 42.24260711669922, "p90": 172.03624725341797, "max": 230.91119384765625, "pos_frac": 0.890625, "sample": [0.5110321044921875, 21.062509536743164, 80.66552734375, 53.2783203125, 18.352066040039062, 34.7871208190918, 121.31880950927734, 6.83387565612793, 61.599334716796875, -23.738746643066406, 118.76654052734375, 4.358802795410156, 25.713123321533203, 12.366889953613281, 179.33319091796875, 77.3685531616211, 6.01019287109375, 42.79364013671875, 26.83715057373047, 10.796451568603516, 150.57571411132812, 84.62408447265625, 36.04999542236328, 111.95147705078125, 54.96391296386719, 2.634492874145508, -7.579658508300781, 129.80352783203125, 111.94036865234375, 172.20449829101562, 65.86321258544922, 111.09452819824219, 80.91516876220703, 203.7992401123047, -6.6402587890625, -5.703498840332031, -22.467632293701172, 41.69157409667969, 36.054420471191406, 21.026229858398438, 100.27163696289062, 85.45751953125, 188.7476348876953, 31.170703887939453, 11.747262954711914, 1.43597412109375, 179.2010498046875, 18.94219970703125, 71.54756164550781, 81.11219024658203, 157.72076416015625, 230.91119384765625, 121.0297622680664, 58.610755920410156, 6.890388488769531, 36.654884338378906, 55.179405212402344, 7.435489654541016, 13.923917770385742, 190.58180236816406, -0.42275238037109375, 4.445518493652344, -64.02816772460938, 171.64366149902344], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000210.npy"} +{"epoch": 0.30837004405286345, "step": 211, "batch_size": 64, "mean": 53.358829498291016, "std": 69.30540466308594, "min": -57.46087646484375, "p10": -24.661332702636717, "median": 43.09183311462402, "p90": 137.60346374511718, "max": 257.528564453125, "pos_frac": 0.78125, "sample": [51.78108215332031, -9.999555587768555, 138.4071044921875, 75.5735092163086, 54.5683708190918, 59.93125915527344, 14.44989013671875, 107.52747344970703, 257.528564453125, -29.90301513671875, 27.998939514160156, -1.1370601654052734, 12.443733215332031, -25.491180419921875, 70.90057373046875, 132.67506408691406, 84.0059814453125, 65.28689575195312, 45.65644836425781, 23.83868408203125, 6.299842834472656, 7.4685211181640625, 96.50698852539062, 70.303466796875, -5.353706359863281, 7.9303436279296875, -10.490898132324219, 195.36703491210938, 29.013931274414062, 250.89657592773438, 211.5673828125, -6.399951934814453, -39.48348617553711, 55.048553466796875, 111.32144165039062, 98.02484130859375, -57.46087646484375, 105.226806640625, 30.607528686523438, 27.920654296875, 70.58404541015625, 125.06597900390625, -22.725021362304688, 19.486242294311523, 21.431669235229492, 51.28656005859375, 23.974435806274414, -47.39451599121094, -22.32681655883789, 139.9072265625, -57.25358581542969, 67.47994232177734, 108.27053833007812, 73.42894744873047, 89.09642028808594, 145.41751098632812, 39.618553161621094, 135.72830200195312, 0.7901382446289062, 83.23033142089844, 40.527217864990234, 23.749755859375, 21.153518676757812, -55.9200439453125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000211.npy"} +{"epoch": 0.30983847283406757, "step": 212, "batch_size": 64, "mean": 76.81648254394531, "std": 82.52164459228516, "min": -72.90458679199219, "p10": -9.282021713256835, "median": 70.70221710205078, "p90": 181.73285522460938, "max": 297.3208923339844, "pos_frac": 0.765625, "sample": [-72.90458679199219, -50.161582946777344, 228.40902709960938, 123.0264892578125, 50.056663513183594, 143.47344970703125, 22.743865966796875, -5.033786773681641, 17.496950149536133, 8.581029891967773, 117.5469970703125, 77.92938995361328, -49.386016845703125, -27.230804443359375, -3.545368194580078, 25.26403045654297, 146.78875732421875, 170.334716796875, 243.9099578857422, 110.93992614746094, 54.16326141357422, -53.609169006347656, 40.87039566040039, 74.6998291015625, 82.92010498046875, 88.18087005615234, 152.22021484375, -7.292407989501953, 183.46939086914062, -4.631219863891602, 43.96929931640625, 70.39164733886719, 297.3208923339844, 66.72714233398438, 32.88079833984375, -11.779167175292969, 145.6987762451172, -0.9430694580078125, 98.05965423583984, 276.61529541015625, 145.09463500976562, 71.01278686523438, 48.74275207519531, 87.1403579711914, 19.967628479003906, 24.33685302734375, -9.870372772216797, -0.28928375244140625, 176.65072631835938, 23.062076568603516, 83.43231201171875, 192.01187133789062, 186.98558044433594, 124.00552368164062, 177.68093872070312, -7.909202575683594, 98.83621215820312, 34.488460540771484, 128.07684326171875, 102.43804931640625, 17.932640075683594, 159.4481201171875, -5.51214599609375, 130.31988525390625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000212.npy"} +{"epoch": 0.31130690161527164, "step": 213, "batch_size": 64, "mean": 76.68045806884766, "std": 78.00340270996094, "min": -111.70428466796875, "p10": -10.649186325073241, "median": 70.72335052490234, "p90": 199.17247009277347, "max": 253.15957641601562, "pos_frac": 0.828125, "sample": [201.8419189453125, -18.155258178710938, 24.22887420654297, -28.930299758911133, 83.87403869628906, 10.63092041015625, 34.704532623291016, 94.31028747558594, -111.70428466796875, -18.831634521484375, 3.3171138763427734, 88.41595458984375, 165.1795654296875, -11.730508804321289, 60.537269592285156, 91.02326965332031, 8.662286758422852, 45.38517761230469, 130.4952850341797, 157.84022521972656, 251.6860809326172, 143.57666015625, 39.106117248535156, 215.144775390625, 81.64186096191406, 166.1807861328125, 214.31912231445312, -11.048755645751953, 13.536705017089844, 19.805194854736328, 96.12274932861328, 53.97654724121094, 79.9865951538086, 76.27029418945312, 18.4681396484375, 83.12107849121094, 81.09383392333984, 135.3095703125, 192.94375610351562, 98.13778686523438, 19.691448211669922, 54.76629638671875, 253.15957641601562, -9.378387451171875, 92.45051574707031, 55.65586853027344, 114.74500274658203, 216.64926147460938, -9.71685791015625, 134.54193115234375, 144.23403930664062, 205.09349060058594, 184.43264770507812, 8.600650787353516, -20.814712524414062, -0.7384529113769531, -6.941864013671875, 65.17640686035156, 39.82814025878906, 18.78516387939453, 49.675376892089844, 31.8817138671875, 94.5638427734375, 110.73504638671875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000213.npy"} +{"epoch": 0.31277533039647576, "step": 214, "batch_size": 64, "mean": 62.60892105102539, "std": 77.56240844726562, "min": -114.89823150634766, "p10": -5.110519218444823, "median": 42.888654708862305, "p90": 180.95660247802735, "max": 280.9558410644531, "pos_frac": 0.828125, "sample": [22.77361297607422, 214.71493530273438, 3.788249969482422, 72.43209838867188, 15.527399063110352, -3.6788482666015625, -11.230518341064453, -114.89823150634766, 164.09585571289062, 3.0423927307128906, 106.84710693359375, 62.728057861328125, 56.93189239501953, -1.2037239074707031, 39.768463134765625, 39.67520523071289, 75.50328826904297, -1.4930744171142578, 26.095687866210938, 86.7017822265625, 33.13379669189453, 15.317115783691406, 71.87335968017578, 105.63510131835938, 128.6483154296875, 43.455238342285156, 41.566307067871094, -5.724092483520508, 51.18008804321289, -17.900144577026367, 38.02859115600586, 56.19300079345703, 20.673789978027344, 205.29550170898438, -50.344154357910156, 243.82073974609375, 198.36642456054688, 19.765975952148438, 7.887245178222656, 182.04931640625, 93.383056640625, 81.24091339111328, 178.27561950683594, 178.4069366455078, -84.82495880126953, 280.9558410644531, 15.163084030151367, 91.7796401977539, 133.12051391601562, 64.68516540527344, 1.6206741333007812, 90.98912048339844, -3.1251449584960938, 44.413509368896484, 6.4215850830078125, 99.14590454101562, 34.713809967041016, 187.1748504638672, 36.304283142089844, 48.3189697265625, 42.32207107543945, 163.7650604248047, -26.017070770263672, 1.6941413879394531], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000214.npy"} +{"epoch": 0.3142437591776799, "step": 215, "batch_size": 64, "mean": 67.09068298339844, "std": 73.1328353881836, "min": -81.99464416503906, "p10": -13.638108825683593, "median": 58.22031021118164, "p90": 150.04891357421877, "max": 334.26995849609375, "pos_frac": 0.84375, "sample": [43.32697677612305, 176.1363525390625, 79.52047729492188, 77.6107406616211, -12.57733154296875, -27.95207977294922, -16.647560119628906, 97.26091766357422, 89.85037231445312, 53.99036407470703, 45.98295974731445, 29.010961532592773, 16.850740432739258, 8.731264114379883, 227.84527587890625, 127.5221176147461, 119.43133544921875, 161.73500061035156, 334.26995849609375, 78.22520446777344, 31.70893096923828, 152.51812744140625, 74.2151870727539, -0.488800048828125, 30.136703491210938, 78.77779388427734, -28.644176483154297, 36.398399353027344, 144.28741455078125, 114.88928985595703, 25.202695846557617, 37.79608917236328, 9.325973510742188, 2.573831558227539, 120.64179992675781, 62.45025634765625, -39.24675750732422, 116.26641082763672, 107.85404968261719, 100.35720825195312, 172.44027709960938, 92.7734146118164, 104.86570739746094, 10.747222900390625, 130.30706787109375, 29.968917846679688, 66.43997192382812, -8.514747619628906, 31.325260162353516, 127.14269256591797, -81.99464416503906, 67.10286712646484, 259.7567138671875, 25.18463897705078, 49.491294860839844, 18.614471435546875, 133.75314331054688, -14.092727661132812, 67.45561218261719, 25.708036422729492, 68.71708679199219, 12.392984390258789, -35.31789779663086, 52.39179229736328], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000215.npy"} +{"epoch": 0.315712187958884, "step": 216, "batch_size": 64, "mean": 72.11544036865234, "std": 83.69060516357422, "min": -92.22537231445312, "p10": -25.384266090393066, "median": 50.95331954956055, "p90": 179.53961334228518, "max": 269.0059814453125, "pos_frac": 0.8125, "sample": [143.35372924804688, 161.96728515625, -25.776199340820312, 90.50491333007812, 94.15049743652344, 35.25868225097656, -44.502166748046875, 91.51531219482422, 131.65805053710938, -16.586509704589844, 269.0059814453125, 14.617904663085938, 81.4935302734375, -27.126625061035156, 17.809810638427734, -1.2742919921875, -52.327186584472656, 146.66140747070312, 1.6729850769042969, -92.22537231445312, 9.372238159179688, 177.03228759765625, 6.03515625, 41.82373046875, 105.93878173828125, 79.30509948730469, 13.82696533203125, 101.17909240722656, 113.32064819335938, 40.212615966796875, 72.26129913330078, 96.03217315673828, 53.98030090332031, -6.667858123779297, 215.86465454101562, 145.9551544189453, -34.07585144042969, 209.43365478515625, 20.741012573242188, 36.510459899902344, 16.87030601501465, 153.29837036132812, 23.382295608520508, -91.41950988769531, 47.92633819580078, 101.66692352294922, 146.261962890625, 248.12998962402344, 32.44807434082031, -15.352516174316406, 164.451171875, 35.86541748046875, 30.42371368408203, 54.834556579589844, 45.09355926513672, 180.6141815185547, 130.96694946289062, 161.9527130126953, 225.93890380859375, 214.6143341064453, 21.835479736328125, -24.469755172729492, 24.817489624023438, 167.3035888671875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000216.npy"} +{"epoch": 0.31718061674008813, "step": 217, "batch_size": 64, "mean": 79.8006362915039, "std": 73.16077423095703, "min": -69.77778625488281, "p10": 7.078388595581058, "median": 69.64774322509766, "p90": 200.33450775146488, "max": 266.1133728027344, "pos_frac": 0.921875, "sample": [85.95310974121094, 80.4490966796875, 71.84027099609375, 27.13728141784668, 24.61181640625, 35.573585510253906, 48.79966735839844, 119.49955749511719, 82.74927520751953, 38.69110107421875, 5.7325439453125, 208.91073608398438, 100.3136978149414, 178.46719360351562, -43.40454864501953, 16.272994995117188, 39.82372283935547, -12.596817016601562, 81.59063720703125, 260.9114685058594, 10.218692779541016, 53.82917022705078, 107.2813491821289, 206.6011962890625, 107.12252044677734, 140.39572143554688, 20.26001739501953, 24.416629791259766, -69.77778625488281, 71.43115234375, 266.1133728027344, 119.4852066040039, 12.278913497924805, 59.37074279785156, 121.00521850585938, 26.957000732421875, 88.48614501953125, 153.28839111328125, 10.852310180664062, 204.07127380371094, 205.50502014160156, -31.004234313964844, 173.8358612060547, 146.256103515625, 85.94218444824219, 191.61538696289062, 22.033340454101562, 60.77666473388672, 59.326820373535156, 43.097412109375, 72.22303771972656, 111.6583023071289, 221.192138671875, 39.94190979003906, 106.16845703125, 36.310302734375, -0.04880523681640625, 11.982513427734375, 27.54491424560547, 33.11479187011719, 67.86433410644531, 116.97535705566406, 114.49723815917969, 5.418392181396484], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000217.npy"} +{"epoch": 0.3186490455212922, "step": 218, "batch_size": 64, "mean": 65.4749755859375, "std": 76.97972106933594, "min": -83.49958801269531, "p10": -22.972402954101558, "median": 58.8427677154541, "p90": 152.47944488525394, "max": 341.5948486328125, "pos_frac": 0.828125, "sample": [58.28647232055664, 33.136878967285156, 108.47371673583984, 74.30926513671875, 94.52472686767578, 22.951236724853516, 197.06875610351562, 10.263435363769531, 12.12571907043457, 60.253265380859375, 113.8425064086914, 39.92741394042969, 4.994487762451172, 119.65071105957031, 224.353271484375, 55.39170837402344, 72.72764587402344, 3.7746963500976562, 47.22888946533203, 115.23260498046875, 25.4854736328125, 34.28041076660156, 19.125530242919922, 123.24407958984375, -30.692962646484375, -34.96025085449219, 163.65115356445312, 72.47962951660156, 100.83425903320312, 116.75869750976562, 24.78028106689453, 145.20037841796875, 121.56134033203125, 135.20758056640625, 98.52130126953125, 62.453765869140625, 2.6045475006103516, -15.789140701293945, 80.2018051147461, 9.462963104248047, -11.72821044921875, -43.46314239501953, 90.17157745361328, -6.831279754638672, -16.61693572998047, -25.69617462158203, 341.5948486328125, 169.60113525390625, 126.43528747558594, 3.3037185668945312, -40.0244140625, 139.829345703125, 255.60223388671875, 6.266454696655273, 34.26980972290039, 84.32112884521484, 123.07307434082031, 72.57026672363281, 26.04448127746582, -33.78577423095703, 59.39906311035156, -83.49958801269531, 155.5990447998047, 41.034698486328125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000218.npy"} +{"epoch": 0.3201174743024963, "step": 219, "batch_size": 64, "mean": 55.72784423828125, "std": 72.61895751953125, "min": -137.25607299804688, "p10": -12.630365753173823, "median": 46.50004196166992, "p90": 164.69798278808602, "max": 240.8039093017578, "pos_frac": 0.84375, "sample": [240.8039093017578, 73.0232925415039, 14.156610488891602, 177.24566650390625, 17.607027053833008, 223.8914337158203, -137.25607299804688, 6.380247116088867, 12.15655517578125, 132.6815185546875, 8.303777694702148, 56.367210388183594, 105.7474365234375, 208.2501220703125, 239.18133544921875, 51.674583435058594, 24.935386657714844, 173.53726196289062, 92.33071899414062, 88.46992492675781, 85.96214294433594, 12.837417602539062, -54.828269958496094, 45.893310546875, 42.316307067871094, -3.1686477661132812, 98.37926483154297, 5.61517333984375, 180.46600341796875, 139.85409545898438, 47.106773376464844, 144.072998046875, 22.212799072265625, 22.663190841674805, 18.475006103515625, 94.31681823730469, 75.43273162841797, 6.7361602783203125, 53.87870788574219, 12.47918701171875, 119.14657592773438, 65.95008850097656, 80.86953735351562, -45.266876220703125, 29.07147216796875, 73.57290649414062, -42.455780029296875, 7.907747268676758, 13.095954895019531, 73.45310974121094, 27.8731689453125, -7.636451721191406, 6.233665466308594, -21.63204574584961, -50.26322937011719, 82.56788635253906, 81.25810241699219, 52.65130615234375, 15.236690521240234, -14.770614624023438, 47.65771484375, 44.766998291015625, 74.25321960449219, -7.148212432861328], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000219.npy"} +{"epoch": 0.32158590308370044, "step": 220, "batch_size": 64, "mean": 51.05287170410156, "std": 54.13798141479492, "min": -61.36095428466797, "p10": -13.010181617736814, "median": 56.14223098754883, "p90": 123.25377426147462, "max": 187.47586059570312, "pos_frac": 0.859375, "sample": [128.31845092773438, 20.80388069152832, 71.04401397705078, -6.470939636230469, 42.29889678955078, 187.47586059570312, 183.73748779296875, 3.6239395141601562, 82.68293762207031, 84.43771362304688, 64.52359771728516, 73.12857818603516, 95.40872192382812, 64.52874755859375, -61.36095428466797, 64.28192138671875, 121.29377746582031, 59.62042236328125, 124.0937728881836, 114.57801818847656, 72.78242492675781, 10.581035614013672, 22.40825653076172, -25.31744384765625, -21.01795196533203, 124.68756866455078, 5.797979354858398, 31.401199340820312, 147.16104125976562, 21.697071075439453, 15.14276123046875, 76.99986267089844, 93.74946594238281, 1.207427978515625, 76.52684020996094, 55.04578399658203, 149.29879760742188, 115.58335876464844, 30.40282440185547, 48.727928161621094, 2.8630847930908203, 17.27529525756836, 6.968034744262695, 36.923583984375, -14.318838119506836, 22.40526580810547, 58.215484619140625, 57.31346130371094, -56.45333480834961, 60.689544677734375, 24.833663940429688, 29.622421264648438, 108.51691436767578, -26.30302619934082, 24.486133575439453, 95.26194763183594, 71.0550537109375, 2.7325210571289062, 69.00592041015625, 57.238677978515625, 17.302860260009766, 83.0608901977539, -9.956649780273438, -42.270355224609375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000220.npy"} +{"epoch": 0.32305433186490456, "step": 221, "batch_size": 64, "mean": 94.78233337402344, "std": 77.86430358886719, "min": -65.32173156738281, "p10": 2.2907951354980485, "median": 95.8584098815918, "p90": 194.2901840209961, "max": 271.13323974609375, "pos_frac": 0.9375, "sample": [53.719078063964844, 39.16387939453125, 46.888885498046875, 3.687366485595703, 65.11495208740234, 271.13323974609375, 10.604530334472656, 220.1099090576172, 139.37417602539062, 169.9282989501953, 186.92660522460938, 1.6922645568847656, 185.2943572998047, 74.60015869140625, 82.02762603759766, 125.94757080078125, 5.564666748046875, 193.67703247070312, 1.434286117553711, 125.76900482177734, 127.19324493408203, 86.76556396484375, 183.02828979492188, 238.97503662109375, 249.0882568359375, 163.79632568359375, 130.0203399658203, -65.32173156738281, 32.900550842285156, 32.141822814941406, 100.11978912353516, 122.28924560546875, 6.429695129394531, 46.548255920410156, 41.873329162597656, 98.41401672363281, 60.97876739501953, -59.85772705078125, 120.29512786865234, 194.55296325683594, 0.29099273681640625, 139.97711181640625, 139.77462768554688, 139.52297973632812, 70.14295959472656, 122.86345672607422, 23.69110107421875, 212.61212158203125, 156.35586547851562, 107.759521484375, 25.265239715576172, -22.910207748413086, 54.71178436279297, 135.67422485351562, 17.297012329101562, -10.953414916992188, 30.446741104125977, 112.5866928100586, 50.286441802978516, 93.30280303955078, 186.37637329101562, 29.349485397338867, 219.2645263671875, 119.49232482910156], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000221.npy"} +{"epoch": 0.3245227606461087, "step": 222, "batch_size": 64, "mean": 74.5833740234375, "std": 81.04644775390625, "min": -108.89680480957031, "p10": -14.072964286804197, "median": 68.4288558959961, "p90": 197.65007629394535, "max": 282.45709228515625, "pos_frac": 0.84375, "sample": [-12.857810974121094, 206.39105224609375, 20.070587158203125, -14.593744277954102, 15.788185119628906, 2.365049362182617, -18.368377685546875, 30.177932739257812, 13.4166259765625, 56.989501953125, 71.15926361083984, 92.0110092163086, 277.6988220214844, 39.663265228271484, 99.75226593017578, 49.874977111816406, -108.89680480957031, 98.71939849853516, 140.82754516601562, 87.7481460571289, 121.12601470947266, 65.69844818115234, -44.590667724609375, 190.8553466796875, -42.461097717285156, 127.90937805175781, 282.45709228515625, 54.95183563232422, 62.192344665527344, 9.690376281738281, 95.00824737548828, 13.931427001953125, 113.92214965820312, -33.899200439453125, 124.27047729492188, 103.75990295410156, 28.98664093017578, 106.81373596191406, -10.81607437133789, 99.02630615234375, 0.36586761474609375, 108.07456970214844, -10.132484436035156, 13.226699829101562, 200.56210327148438, 97.41840362548828, 89.86577606201172, 127.5783920288086, 74.75115203857422, 39.680747985839844, 96.56620788574219, 116.11563873291016, 219.2556915283203, 47.96554183959961, 122.23733520507812, 65.28125762939453, 106.10040283203125, -81.49363708496094, 131.74151611328125, 26.62017059326172, 260.583251953125, 34.6566162109375, 216.2197265625, 53.32545471191406], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000222.npy"} +{"epoch": 0.32599118942731276, "step": 223, "batch_size": 64, "mean": 62.964664459228516, "std": 88.54454040527344, "min": -162.86148071289062, "p10": -50.65877304077148, "median": 63.73917007446289, "p90": 179.57464904785158, "max": 292.390380859375, "pos_frac": 0.78125, "sample": [57.55967712402344, -62.254966735839844, 110.78453063964844, 71.69734191894531, 31.205467224121094, 180.34591674804688, 139.60470581054688, 59.956459045410156, 119.93721008300781, 110.51995849609375, -51.77764129638672, 16.904281616210938, 40.982818603515625, 146.85073852539062, 191.2812042236328, 146.22000122070312, 71.36280059814453, 65.22781372070312, 240.28900146484375, -23.450946807861328, 105.50270080566406, -1.4692516326904297, 74.58964538574219, -46.247962951660156, 2.736419677734375, 190.22232055664062, -65.278564453125, 177.7750244140625, 84.5071792602539, -103.8791732788086, 15.82415771484375, 9.053909301757812, 146.63331604003906, -107.30937194824219, 81.53770446777344, 97.59396362304688, -43.72611999511719, 103.7159194946289, 44.6840705871582, 292.390380859375, 172.67556762695312, -52.219390869140625, 75.80559539794922, 124.07710266113281, -48.04808044433594, 23.05398178100586, 127.49179077148438, 146.6258544921875, 54.707862854003906, 7.31903076171875, 42.30992889404297, 4.346382141113281, -37.048797607421875, 198.40560913085938, 29.67137908935547, 41.12310791015625, -6.673391342163086, 20.30360221862793, 89.50112915039062, 112.40274047851562, 62.250526428222656, 98.95378875732422, -162.86148071289062, 183.46188354492188], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000223.npy"} +{"epoch": 0.3274596182085169, "step": 224, "batch_size": 64, "mean": 77.57852172851562, "std": 92.97354125976562, "min": -130.01654052734375, "p10": -13.596275901794433, "median": 64.20595932006836, "p90": 211.6470031738282, "max": 339.1694641113281, "pos_frac": 0.78125, "sample": [118.68556213378906, 5.2724761962890625, 84.74920654296875, 9.492660522460938, -46.46827697753906, 259.2290954589844, 149.1257781982422, 64.95561218261719, 63.45630645751953, -130.01654052734375, 62.71549987792969, 125.18377685546875, 128.0337371826172, 72.87124633789062, 248.65786743164062, 49.62226486206055, 91.11534881591797, 159.71652221679688, 139.0500946044922, 339.1694641113281, 193.836181640625, 69.03636932373047, 155.81594848632812, 143.41607666015625, -9.583938598632812, -27.727397918701172, 153.37551879882812, 75.51779174804688, 128.85269165039062, -15.616584777832031, 27.4996337890625, 23.84130859375, 12.755075454711914, 314.6520080566406, 125.51388549804688, 165.8189697265625, 69.26517486572266, -12.652069091796875, 37.24864959716797, 45.75564956665039, 144.90737915039062, -2.99609375, 18.974136352539062, -13.737970352172852, 45.07649230957031, -47.09016418457031, -6.0568389892578125, -13.265655517578125, -48.02555847167969, 3.5644798278808594, 111.50386047363281, 104.13311767578125, 224.7319793701172, 11.264108657836914, 14.093441009521484, 29.448955535888672, 15.898147583007812, 159.02529907226562, 31.90894317626953, -9.105743408203125, 72.56779479980469, 219.28021240234375, 238.25732421875, -10.57107925415039], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000224.npy"} +{"epoch": 0.328928046989721, "step": 225, "batch_size": 64, "mean": 97.2855224609375, "std": 97.8428955078125, "min": -83.9691162109375, "p10": -23.41173400878906, "median": 95.7482681274414, "p90": 237.88012084960937, "max": 386.9402770996094, "pos_frac": 0.8125, "sample": [167.28555297851562, 96.081787109375, 253.9093475341797, 236.199462890625, 38.64300537109375, 57.946800231933594, 40.69622802734375, -20.17254638671875, -12.607864379882812, 251.33399963378906, 238.60040283203125, 151.455078125, -39.538482666015625, -0.5060806274414062, 191.79286193847656, 15.33328628540039, -75.47968292236328, 25.14644432067871, -28.176082611083984, 31.165481567382812, 136.27894592285156, 95.41474914550781, 110.18244171142578, 111.15050506591797, 1.1381664276123047, 65.72650909423828, 59.127593994140625, 101.08297729492188, 37.03633499145508, 169.5731201171875, 233.6814422607422, 94.94293212890625, 134.64366149902344, 55.54347229003906, -8.904659271240234, 273.5831298828125, 269.90435791015625, -2.1906356811523438, 193.18304443359375, 245.35214233398438, 65.76028442382812, 163.7462615966797, 66.92893981933594, -34.56189727783203, 120.62788391113281, 46.153892517089844, 386.9402770996094, 183.1025390625, 114.72775268554688, 100.88774108886719, 179.2427520751953, 40.531005859375, 144.58261108398438, -24.799957275390625, -54.045902252197266, 198.4521026611328, 149.9713897705078, 99.49288940429688, 128.24307250976562, -83.9691162109375, 93.87714385986328, 100.46014404296875, 11.461158752441406, 32.90118408203125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000225.npy"} +{"epoch": 0.3303964757709251, "step": 226, "batch_size": 64, "mean": 74.54594421386719, "std": 92.88312530517578, "min": -103.13072204589844, "p10": -11.162636566162108, "median": 49.4686164855957, "p90": 203.93233337402344, "max": 370.38262939453125, "pos_frac": 0.84375, "sample": [370.38262939453125, -57.4215087890625, -9.91006851196289, 42.33186340332031, -7.3188934326171875, 127.83236694335938, 23.308181762695312, 69.26370239257812, 84.7551498413086, -14.88473129272461, 92.27272033691406, 22.860084533691406, 6.6689453125, 40.916114807128906, -83.38482666015625, 36.738468170166016, -11.699451446533203, 151.4141845703125, 229.14044189453125, 62.83568572998047, 202.89730834960938, 204.37591552734375, 29.29126739501953, 68.96932983398438, 78.23081970214844, 168.5082244873047, 60.19506072998047, 10.674026489257812, 52.111572265625, 232.67343139648438, 139.98243713378906, -27.440452575683594, 105.40144348144531, 57.42295837402344, 36.085205078125, 263.19818115234375, 131.99990844726562, 318.1342468261719, 175.4827880859375, -0.6652069091796875, 8.938308715820312, 42.939857482910156, -103.13072204589844, 69.64968872070312, 37.685794830322266, 46.825660705566406, 2.2618179321289062, 26.198684692382812, 23.109331130981445, 68.52777862548828, 6.992889404296875, 188.02919006347656, 64.48287963867188, 131.04611206054688, 46.59339141845703, 87.26557922363281, -41.34625244140625, 7.396862030029297, 267.52398681640625, 31.907882690429688, 32.03003692626953, 70.59395599365234, 12.695587158203125, 159.0920867919922], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000226.npy"} +{"epoch": 0.33186490455212925, "step": 227, "batch_size": 64, "mean": 92.38166809082031, "std": 96.87541961669922, "min": -182.25155639648438, "p10": -7.441629219055175, "median": 88.11932373046875, "p90": 233.61342468261725, "max": 333.3125, "pos_frac": 0.84375, "sample": [218.18682861328125, 58.028038024902344, 90.12549591064453, 239.24072265625, 127.57193756103516, 52.544464111328125, 79.87132263183594, 106.1518783569336, 164.51370239257812, -7.9018402099609375, 57.758697509765625, 84.26251220703125, 115.16121673583984, 10.349515914916992, -39.30636978149414, 20.392196655273438, 125.69377899169922, 66.06700134277344, 119.45478820800781, 312.2342529296875, 94.99140930175781, 35.42681884765625, 173.9862060546875, 75.81126403808594, 91.31163787841797, 150.18577575683594, 124.6308364868164, 126.92598724365234, 60.136322021484375, 70.27128601074219, 122.21417236328125, 103.86811828613281, 186.210693359375, 3.505321502685547, -24.300552368164062, -72.84864807128906, 101.7326889038086, 139.01910400390625, 258.64599609375, 86.21365356445312, 109.0931396484375, 259.4829406738281, -80.58973693847656, 333.3125, 18.360036849975586, 29.676795959472656, 108.263427734375, 31.99425506591797, 133.87493896484375, 26.430747985839844, 78.58573150634766, 298.48443603515625, 151.134765625, 26.448959350585938, -2.506418228149414, -6.367803573608398, 220.48306274414062, -182.25155639648438, -2.713939666748047, 71.6678466796875, -10.875066757202148, 24.130151748657227, 90.02499389648438, 277.9443054199219], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000227.npy"} +{"epoch": 0.3333333333333333, "step": 228, "batch_size": 64, "mean": 66.29071044921875, "std": 79.78272247314453, "min": -86.26971435546875, "p10": -40.503799438476555, "median": 66.88349151611328, "p90": 188.8237457275391, "max": 296.56146240234375, "pos_frac": 0.796875, "sample": [-63.87574005126953, -75.29547119140625, -86.26971435546875, 67.528076171875, 61.84546661376953, 190.66880798339844, -66.0310287475586, 83.36338806152344, 100.21097564697266, -50.19951248168945, 15.82452392578125, 126.40768432617188, 198.8524169921875, 122.56639862060547, 215.70188903808594, 218.13414001464844, 209.9720916748047, 106.20501708984375, 76.35738372802734, -5.304664611816406, -20.749923706054688, 192.12835693359375, 71.75763702392578, 141.1630859375, 107.71317291259766, 79.89146423339844, 103.73393249511719, 16.181049346923828, -43.765159606933594, 101.72853088378906, 67.45639038085938, 71.56832885742188, 10.978729248046875, 148.00308227539062, 140.90988159179688, 50.775203704833984, 43.33757019042969, 52.512420654296875, 54.432952880859375, 14.1844482421875, 65.3277816772461, 105.16073608398438, 1.511993408203125, -2.0810546875, 69.81375122070312, 64.52306365966797, 296.56146240234375, 16.26074981689453, 66.31059265136719, 11.906501770019531, 184.5186004638672, 23.820751190185547, -25.50835609436035, 106.10668182373047, 101.23516845703125, 111.69591522216797, 103.21437072753906, 52.251800537109375, -32.893959045410156, 36.707244873046875, -4.521598815917969, 68.63008117675781, -57.5494384765625, 28.999610900878906], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000228.npy"} +{"epoch": 0.33480176211453744, "step": 229, "batch_size": 64, "mean": 80.00627136230469, "std": 89.31297302246094, "min": -185.7428436279297, "p10": -12.218200111389157, "median": 74.98181533813477, "p90": 203.8795593261719, "max": 319.41754150390625, "pos_frac": 0.84375, "sample": [21.448867797851562, 9.940893173217773, 12.85711669921875, 224.92291259765625, 40.66904067993164, -75.77376556396484, 78.80860137939453, 107.262451171875, 62.63435363769531, 134.21340942382812, 204.81588745117188, 201.69479370117188, 51.44268798828125, 265.5909423828125, 102.45539855957031, 77.02217864990234, 113.32781219482422, 74.93830108642578, 35.05234909057617, 214.90151977539062, -23.49842071533203, 73.04421997070312, 11.899223327636719, 139.51800537109375, 69.99320220947266, 41.38823699951172, 319.41754150390625, -41.8485107421875, 19.680160522460938, 124.80242919921875, 142.84793090820312, 174.68923950195312, 142.4959716796875, 197.09620666503906, 221.1282958984375, 46.532676696777344, 148.1376190185547, 31.274681091308594, 75.02532958984375, -9.507095336914062, 82.18204498291016, 77.61302947998047, 42.75695037841797, 118.97169494628906, -13.380102157592773, 56.61903381347656, 125.38165283203125, 138.88478088378906, -8.818679809570312, -61.55435562133789, 91.08724975585938, 274.29400634765625, 25.71059226989746, 61.970829010009766, 89.82238006591797, 113.91456604003906, 8.695831298828125, 114.46920013427734, -2.2162017822265625, 117.02984619140625, -185.7428436279297, 0.30649566650390625, -21.466812133789062, 11.5272216796875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000229.npy"} +{"epoch": 0.33627019089574156, "step": 230, "batch_size": 64, "mean": 97.3758316040039, "std": 95.31262969970703, "min": -169.10723876953125, "p10": -18.065234374999996, "median": 84.89377212524414, "p90": 245.79083557128908, "max": 310.30535888671875, "pos_frac": 0.875, "sample": [82.97222900390625, 238.05734252929688, 5.07750129699707, 185.515380859375, 110.3994369506836, 49.40727996826172, 17.812658309936523, 159.71791076660156, 72.50907897949219, 88.43830871582031, 107.1048583984375, 97.71416473388672, -23.64337921142578, 109.9715576171875, 48.40229797363281, 48.986175537109375, -28.6649169921875, 72.06796264648438, 25.73114013671875, 76.09768676757812, 67.411865234375, -66.0899429321289, -19.076576232910156, 97.27622985839844, 86.81531524658203, 96.68678283691406, 118.52497863769531, 194.03726196289062, 150.33868408203125, 119.20124053955078, 64.5692138671875, 31.36443328857422, 93.55224609375, 284.3711853027344, 92.11622619628906, -22.36290740966797, 50.057334899902344, 120.60750579833984, 152.05520629882812, 297.93304443359375, -28.802528381347656, 10.499237060546875, -169.10723876953125, -15.705436706542969, 9.969001770019531, 29.566940307617188, 251.01712036132812, 77.13182067871094, 248.66140747070312, 267.92071533203125, 143.84976196289062, 66.52055358886719, 74.34461975097656, 33.51354217529297, 220.78033447265625, 143.26113891601562, 310.30535888671875, 62.62934875488281, 239.09283447265625, 125.05644226074219, 162.18211364746094, 52.71513366699219, 66.78474426269531, 296.8021240234375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000230.npy"} +{"epoch": 0.3377386196769457, "step": 231, "batch_size": 64, "mean": 73.71493530273438, "std": 82.30001831054688, "min": -40.07990264892578, "p10": -4.900165557861327, "median": 57.048065185546875, "p90": 161.91689300537112, "max": 354.6712341308594, "pos_frac": 0.84375, "sample": [308.9916687011719, 80.54783630371094, -2.8698196411132812, 131.69281005859375, 21.299072265625, 65.75297546386719, 186.5489501953125, 31.390777587890625, 37.1125602722168, 124.58585357666016, 21.808025360107422, 117.49723815917969, -14.084461212158203, 70.76620483398438, 289.6960754394531, 75.11692810058594, 133.5182342529297, 134.86488342285156, 93.91718292236328, 142.76248168945312, 200.11788940429688, 48.665653228759766, 56.51636505126953, 94.55909729003906, 38.28247833251953, -1.2856216430664062, 165.60845947265625, -31.289661407470703, 117.0848388671875, 354.6712341308594, 277.99163818359375, 142.21676635742188, 82.08314514160156, 18.14698600769043, -28.894943237304688, 31.278411865234375, 50.82554244995117, 126.3558120727539, 23.669570922851562, 0.445587158203125, 4.8862457275390625, 29.808679580688477, 78.38763427734375, 22.250547409057617, -4.256813049316406, 57.57976531982422, -5.1758880615234375, -34.521827697753906, 21.422359466552734, 85.5823974609375, 153.30323791503906, 81.98954772949219, 38.519386291503906, 30.428691864013672, -19.65292739868164, 82.74984741210938, -40.07990264892578, 66.97200012207031, 11.799972534179688, 22.242095947265625, 14.942031860351562, 79.84701538085938, 74.89794921875, 45.86909484863281], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000231.npy"} +{"epoch": 0.3392070484581498, "step": 232, "batch_size": 64, "mean": 58.7244873046875, "std": 86.04212951660156, "min": -85.73152160644531, "p10": -23.505888366699214, "median": 39.665771484375, "p90": 181.38090667724612, "max": 333.0561218261719, "pos_frac": 0.8125, "sample": [97.61029052734375, 76.56575775146484, 3.311412811279297, 59.19646453857422, 7.068870544433594, -20.216232299804688, 201.59848022460938, 110.94558715820312, -11.302715301513672, 173.115478515625, 8.496522903442383, 184.92323303222656, 63.3482666015625, -29.78430938720703, -12.8807373046875, 18.668846130371094, 13.269195556640625, 126.5731430053711, -75.8731460571289, 62.72339630126953, 89.4609375, 298.260986328125, 10.839553833007812, 53.875648498535156, 69.54328155517578, 3.8786048889160156, 281.65283203125, 5.5309600830078125, 9.372344970703125, 153.6212158203125, 85.84806060791016, 97.43648529052734, -24.915740966796875, 36.61039733886719, -44.098297119140625, 57.94477844238281, -72.64949798583984, 13.171737670898438, 18.024925231933594, 333.0561218261719, 16.435821533203125, -8.306022644042969, 26.836875915527344, -85.73152160644531, 3.6797523498535156, 42.72114562988281, 61.48004150390625, -66.85256958007812, 125.10118103027344, 46.8097038269043, 210.9811553955078, 20.440349578857422, 27.494552612304688, 205.55615234375, 150.099853515625, 54.739654541015625, 94.90863800048828, 98.297607421875, -0.6307125091552734, 34.94786834716797, 18.133892059326172, 63.454349517822266, 53.09729766845703, 30.848838806152344], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000232.npy"} +{"epoch": 0.3406754772393539, "step": 233, "batch_size": 64, "mean": 86.78538513183594, "std": 109.06512451171875, "min": -98.62013244628906, "p10": -38.28311729431151, "median": 66.74938201904297, "p90": 240.00167694091797, "max": 433.9913330078125, "pos_frac": 0.78125, "sample": [17.612825393676758, 23.028099060058594, 267.300537109375, 9.901260375976562, 82.9599609375, 131.406982421875, -83.06025695800781, -48.957008361816406, 6.328643798828125, -98.62013244628906, 55.35597229003906, 65.17050170898438, 68.32826232910156, 154.16738891601562, 172.55511474609375, 294.91986083984375, 87.9672622680664, 39.912139892578125, 262.5313415527344, 170.6537628173828, 26.183578491210938, 103.33122253417969, 42.7266845703125, 255.81240844726562, 12.366264343261719, 144.52210998535156, 31.036544799804688, -25.67108154296875, -58.097007751464844, -7.536457061767578, 131.26547241210938, 181.1888427734375, 125.79278564453125, 116.01580810546875, 175.3461456298828, 227.1029052734375, -24.25895118713379, 236.0106964111328, 2.7129364013671875, 47.97590637207031, 107.37825775146484, 272.24481201171875, 13.85036849975586, 139.622802734375, 36.02539825439453, 191.0230712890625, -3.390270233154297, -71.81929779052734, -57.559722900390625, 18.868072509765625, 30.490922927856445, 37.17993927001953, 203.39576721191406, 69.68934631347656, -18.763427734375, 105.82421875, -11.963665008544922, 234.64352416992188, -42.81496047973633, 433.9913330078125, 241.71209716796875, -27.708816528320312, 113.97296142578125, 115.08243560791016], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000233.npy"} +{"epoch": 0.342143906020558, "step": 234, "batch_size": 64, "mean": 90.54090881347656, "std": 83.96233367919922, "min": -90.74676513671875, "p10": -8.941805267333976, "median": 86.22438049316406, "p90": 207.47154846191415, "max": 348.6724853515625, "pos_frac": 0.875, "sample": [-32.40940856933594, 172.3324432373047, 151.01751708984375, 63.60552978515625, 154.92237854003906, 142.33700561523438, 93.96902465820312, 114.02305603027344, 227.27783203125, 45.95487976074219, 146.26828002929688, 63.133026123046875, -90.74676513671875, 43.69023513793945, 32.56200408935547, 116.13182830810547, 32.273223876953125, -26.77173614501953, 50.291717529296875, 155.8392333984375, 348.6724853515625, 133.04461669921875, 39.88865661621094, -12.388008117675781, 63.148582458496094, 16.350082397460938, 136.166259765625, 119.39019012451172, 8.711933135986328, 221.4163055419922, 2.675241470336914, 92.43930053710938, 184.05908203125, 117.22177124023438, 285.2930908203125, 63.41227722167969, 71.38796997070312, 146.35595703125, 74.11834716796875, 119.83689880371094, 137.32870483398438, 98.45338439941406, 145.94088745117188, 40.881683349609375, 17.929290771484375, 53.630489349365234, -0.900665283203125, -35.449180603027344, 121.40597534179688, 97.84468078613281, 54.863197326660156, 225.3509063720703, 271.0389709472656, -40.71350860595703, 99.2846908569336, 71.32002258300781, 0.5286407470703125, 42.76763916015625, 95.55783081054688, 32.97520446777344, 80.00946044921875, -35.68919372558594, 217.50546264648438, 115.85116577148438], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000234.npy"} +{"epoch": 0.3436123348017621, "step": 235, "batch_size": 64, "mean": 62.006805419921875, "std": 85.60012817382812, "min": -127.40009307861328, "p10": -23.659333229064938, "median": 55.96382141113281, "p90": 167.1711395263672, "max": 301.04473876953125, "pos_frac": 0.71875, "sample": [22.802886962890625, 137.48757934570312, -127.40009307861328, 39.0869140625, 125.63770294189453, 301.04473876953125, 77.90325164794922, 218.54104614257812, -33.81351089477539, 86.57845306396484, -78.47526550292969, -25.287429809570312, 70.24111938476562, 158.64613342285156, 32.46539306640625, -15.540550231933594, -21.167587280273438, 15.585777282714844, 74.68386840820312, -81.86735534667969, -12.484859466552734, 19.82135009765625, -15.721237182617188, 7.874427795410156, 179.70556640625, 134.9179229736328, -5.543479919433594, -11.578582763671875, 108.39472198486328, 164.05690002441406, 90.01416015625, 67.82771301269531, 132.89108276367188, -6.776123046875, 66.8082046508789, 107.08497619628906, 32.59043884277344, 127.89058685302734, 50.154144287109375, 21.421783447265625, 24.15323257446289, 33.201324462890625, 251.73739624023438, 61.77349853515625, 168.5058135986328, 114.08067321777344, 112.91045379638672, -19.365928649902344, -51.877906799316406, 35.64634704589844, 287.90618896484375, 140.8486328125, -5.774772644042969, 16.04900360107422, 73.68913269042969, -24.727224349975586, -3.2857208251953125, 71.96842956542969, 66.70895385742188, 8.807373046875, 99.11835479736328, 73.70320129394531, 198.66580200195312, -2.509552001953125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000235.npy"} +{"epoch": 0.34508076358296624, "step": 236, "batch_size": 64, "mean": 88.76435852050781, "std": 79.6263656616211, "min": -31.231216430664062, "p10": 0.9306823730468778, "median": 73.10200119018555, "p90": 201.97659149169928, "max": 298.6544494628906, "pos_frac": 0.890625, "sample": [217.55120849609375, 23.28014373779297, -17.591171264648438, 87.08714294433594, 123.10348510742188, 72.79289245605469, 229.48184204101562, 27.530981063842773, 56.983829498291016, 72.28245544433594, 80.51016235351562, 208.03274536132812, 75.19860076904297, 95.88819122314453, 120.12232208251953, 141.66061401367188, 42.875244140625, 149.78453063964844, -1.2341384887695312, 234.7212677001953, 64.5621337890625, -23.470558166503906, 39.28767395019531, 168.69485473632812, 169.93301391601562, 23.718738555908203, 55.86438751220703, 187.476318359375, 164.00448608398438, 41.35944366455078, 298.6544494628906, 40.72015380859375, -16.141647338867188, 134.29302978515625, 64.46147155761719, 129.17041015625, -0.28691864013671875, 248.30123901367188, 62.952186584472656, 18.910388946533203, 73.4111099243164, 31.544593811035156, 23.41860008239746, 4.914451599121094, 3.7717514038085938, 174.5653839111328, 60.803279876708984, 19.682044982910156, -31.231216430664062, 5.593193054199219, 97.19326782226562, 263.7478332519531, 187.84556579589844, 21.87134552001953, 125.21629333496094, 93.16629028320312, 166.06204223632812, 77.50215911865234, 4.248176574707031, 111.35426330566406, -24.716049194335938, 152.3870391845703, 3.783580780029297, 122.25639343261719], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000236.npy"} +{"epoch": 0.3465491923641703, "step": 237, "batch_size": 64, "mean": 77.66547393798828, "std": 89.26495361328125, "min": -146.18203735351562, "p10": -14.244133377075192, "median": 66.19316482543945, "p90": 207.06165161132816, "max": 342.88287353515625, "pos_frac": 0.796875, "sample": [83.5555419921875, 126.43551635742188, 209.7724609375, 130.0836181640625, -0.2862091064453125, 70.6491928100586, 65.14838409423828, 119.29579162597656, 7.039730072021484, 67.23794555664062, 52.46607208251953, 35.80743408203125, 255.2165985107422, 59.026649475097656, 142.81707763671875, 26.772384643554688, -15.917560577392578, 108.7813720703125, -10.339469909667969, 105.7995376586914, 115.6650390625, -26.13922882080078, 209.68515014648438, 342.88287353515625, -34.3226318359375, 220.9608154296875, 111.19000244140625, 20.216033935546875, 59.31056213378906, 7.493389129638672, 269.319580078125, 89.22895812988281, 140.22828674316406, 153.05291748046875, 31.985145568847656, 103.58238220214844, -8.043922424316406, 42.81970977783203, 142.1972198486328, 9.470130920410156, 156.25686645507812, 7.840385437011719, 121.86490631103516, 127.52127075195312, 231.9953155517578, 17.035675048828125, -7.1525115966796875, 64.46343231201172, -146.18203735351562, 200.94015502929688, -5.936117172241211, 41.052894592285156, -66.5145263671875, -25.619140625, 117.39482879638672, 166.62057495117188, -3.645048141479492, 165.9413604736328, 9.238121032714844, -34.91960144042969, 5.601524353027344, 96.79928588867188, 21.041915893554688, 68.80636596679688], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000237.npy"} +{"epoch": 0.34801762114537443, "step": 238, "batch_size": 64, "mean": 72.58844757080078, "std": 89.50284576416016, "min": -80.39098358154297, "p10": -11.16490783691406, "median": 51.74001693725586, "p90": 165.2431411743164, "max": 369.13037109375, "pos_frac": 0.84375, "sample": [19.439804077148438, -7.9603271484375, 6.627471923828125, 91.18019104003906, 22.369998931884766, 12.744087219238281, 76.05426025390625, 147.35009765625, -34.84459686279297, 202.6356201171875, 66.26940155029297, 124.09266662597656, 51.30027770996094, 24.859241485595703, 30.151037216186523, 94.85552978515625, 155.11349487304688, -32.14793395996094, 26.48408317565918, 45.01576232910156, -0.44898414611816406, 362.6369934082031, 25.866714477539062, 156.32127380371094, -27.880645751953125, 303.9980773925781, 52.17975616455078, 27.82231903076172, 164.47393798828125, -32.6373291015625, 86.01519775390625, 39.6043701171875, 47.79887390136719, 139.38162231445312, 168.83770751953125, 11.30301284790039, 16.65468978881836, 69.99504089355469, 369.13037109375, 69.41968536376953, -7.351581573486328, 1.51910400390625, 39.792789459228516, -80.39098358154297, 15.953292846679688, 61.822410583496094, 71.28224182128906, 116.66424560546875, 58.4737548828125, 237.27725219726562, 142.7550506591797, 4.776519775390625, 99.1517333984375, 165.5727996826172, -66.26881408691406, 161.8431396484375, 102.49114227294922, 32.82542419433594, 0.9458236694335938, -12.538299560546875, 96.59578704833984, 106.6251220703125, 17.157623291015625, 106.62236785888672], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000238.npy"} +{"epoch": 0.34948604992657856, "step": 239, "batch_size": 64, "mean": 83.08114624023438, "std": 84.623046875, "min": -83.31961059570312, "p10": -21.51974296569822, "median": 70.30775451660156, "p90": 218.90325012207032, "max": 263.7237854003906, "pos_frac": 0.875, "sample": [240.11184692382812, 163.87794494628906, 1.916839599609375, 202.73800659179688, 12.44171142578125, 67.76617431640625, 105.77497863769531, 79.92086791992188, -30.05752182006836, -46.272483825683594, 36.51462173461914, 32.83465576171875, 24.904598236083984, 131.14947509765625, 45.025787353515625, 0.8075981140136719, 263.7237854003906, 72.84933471679688, 119.59017181396484, 107.67526245117188, 147.37120056152344, 78.53364562988281, 134.4847412109375, -40.68210220336914, 29.35540008544922, 25.17805290222168, 114.704833984375, 219.87966918945312, 104.27827453613281, 61.28334045410156, 17.944238662719727, 19.1884765625, 221.1149139404297, 222.2277374267578, -48.98774719238281, 59.596763610839844, 27.65180206298828, 216.62493896484375, -33.49803161621094, 188.09078979492188, 40.35904312133789, 200.07630920410156, 129.25872802734375, 116.34003448486328, 8.971359252929688, 42.37702178955078, -83.31961059570312, 11.889938354492188, -44.57393264770508, 104.13841247558594, 92.90636444091797, 46.57122802734375, 181.84959411621094, 43.18590545654297, 253.25086975097656, 143.23477172851562, 234.2839813232422, 38.73848342895508, 86.1114501953125, 90.82682037353516, 38.7593994140625, -1.5982589721679688, 114.29646301269531, 31.624181747436523], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000239.npy"} +{"epoch": 0.3509544787077827, "step": 240, "batch_size": 64, "mean": 77.78254699707031, "std": 89.24916076660156, "min": -94.07518768310547, "p10": -52.44068145751952, "median": 81.66063690185547, "p90": 207.84112396240235, "max": 295.417236328125, "pos_frac": 0.796875, "sample": [105.99120330810547, 121.4871826171875, 97.02971649169922, 27.999755859375, 78.38862609863281, 222.2160186767578, 40.93547821044922, 189.24322509765625, -12.560028076171875, -34.902000427246094, 107.65394592285156, 62.16233825683594, -57.79150390625, 111.66407012939453, 209.79867553710938, 44.667198181152344, -58.88985824584961, 18.54473876953125, 15.5870361328125, 138.97201538085938, 94.74066925048828, 208.6284637451172, 91.10494995117188, 78.87239837646484, 41.747833251953125, 31.55167007446289, -94.07518768310547, 43.64701843261719, 81.18743896484375, 228.79478454589844, 191.3673858642578, 82.13383483886719, 134.2850799560547, 51.881629943847656, -67.29367065429688, 222.52154541015625, 100.49185180664062, 133.0567626953125, -67.39131927490234, 76.51873016357422, -17.432933807373047, 170.9322967529297, -68.32412719726562, 53.4687385559082, 295.417236328125, -39.95542907714844, 159.82635498046875, -28.152055740356445, 176.06497192382812, 100.38478088378906, 67.52326965332031, 231.67434692382812, 30.66869354248047, 88.42073822021484, -26.898300170898438, 107.74620819091797, 102.5924301147461, 9.962434768676758, 117.70657348632812, 103.60771179199219, 29.50360107421875, 104.35108947753906, 206.00399780273438, -88.97959899902344], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000240.npy"} +{"epoch": 0.3524229074889868, "step": 241, "batch_size": 64, "mean": 107.67686462402344, "std": 111.6344223022461, "min": -112.00694274902344, "p10": -9.240324306488034, "median": 94.66849517822266, "p90": 233.40730438232424, "max": 515.350341796875, "pos_frac": 0.875, "sample": [69.31432342529297, 90.16609191894531, -54.661598205566406, -22.44046974182129, -13.313873291015625, 126.87350463867188, 155.02639770507812, 236.32789611816406, 85.83808898925781, 92.23075103759766, 117.92236328125, 7.3821868896484375, 6.985378265380859, 307.63812255859375, -6.192818641662598, 103.10838317871094, 7.469409942626953, 185.83761596679688, 150.97776794433594, 119.7086410522461, 137.48146057128906, 102.82711791992188, 68.63175201416016, 68.46015930175781, 515.350341796875, 190.69281005859375, 99.85406494140625, 97.4175796508789, 377.8269348144531, 12.391151428222656, 440.96368408203125, 173.83302307128906, 226.59259033203125, 67.93914031982422, 63.53962707519531, 47.154144287109375, 89.26470184326172, 88.86067962646484, 24.893035888671875, 90.06156921386719, 98.49034881591797, 43.02228546142578, 45.284393310546875, 15.150909423828125, 147.613525390625, 29.69000244140625, 142.44931030273438, -10.546398162841797, 163.04153442382812, 122.66912841796875, 44.2626953125, 56.358497619628906, 140.2674560546875, 319.0125427246094, 117.90223693847656, 97.10623931884766, -67.9456787109375, 169.1870574951172, 156.55514526367188, 145.11172485351562, -112.00694274902344, 253.08697509765625, 37.957584381103516, -12.634933471679688], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000241.npy"} +{"epoch": 0.35389133627019087, "step": 242, "batch_size": 64, "mean": 74.13463592529297, "std": 88.41925048828125, "min": -116.754150390625, "p10": -10.501987457275387, "median": 58.97693634033203, "p90": 178.90396881103524, "max": 357.9957275390625, "pos_frac": 0.875, "sample": [-31.467376708984375, 67.10475158691406, 59.645713806152344, 131.15382385253906, 27.547115325927734, -116.754150390625, 37.853240966796875, 3.73712158203125, 156.55638122558594, 35.73150634765625, 7.406881332397461, 99.04135131835938, 321.8495788574219, 28.691877365112305, -6.021110534667969, 34.80115509033203, 68.58260345458984, 69.42646789550781, 75.73898315429688, 14.061086654663086, 94.25149536132812, 23.21673583984375, 136.9518280029297, 114.2597885131836, 127.8466796875, -15.737113952636719, 113.3896255493164, 58.30815887451172, 59.764366149902344, -12.42236328125, 96.26073455810547, 114.01512145996094, 41.868377685546875, -58.89813232421875, 152.74261474609375, 65.1016616821289, 188.48150634765625, 259.49615478515625, 47.353416442871094, 4.8115234375, 48.98539733886719, 93.21272277832031, 32.315216064453125, 357.9957275390625, 7.024421691894531, 53.92872619628906, 278.1522521972656, 76.86283874511719, 30.907974243164062, 48.87957763671875, -43.114341735839844, 268.4298095703125, 52.106266021728516, 9.757713317871094, 95.71975708007812, 25.254043579101562, 75.84715270996094, -38.0763053894043, 112.69874572753906, 86.56597900390625, 0.24973487854003906, 230.87216186523438, 134.33609008789062, 9.956108093261719], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000242.npy"} +{"epoch": 0.355359765051395, "step": 243, "batch_size": 64, "mean": 99.05632019042969, "std": 92.74897003173828, "min": -106.85092163085938, "p10": -1.209231567382807, "median": 84.55482482910156, "p90": 221.06571197509766, "max": 390.9772033691406, "pos_frac": 0.890625, "sample": [193.67433166503906, -3.4823646545410156, 53.295753479003906, 238.13490295410156, 162.9691162109375, 94.13152313232422, 58.51394271850586, 139.5455322265625, -46.89694595336914, -44.381858825683594, -43.282012939453125, 35.44917297363281, 159.34048461914062, 58.48509979248047, 11.777725219726562, 58.17109680175781, 39.032073974609375, 204.19326782226562, 60.13453674316406, 90.27545928955078, -106.85092163085938, 28.74248504638672, 91.58399963378906, 35.995811462402344, 33.87908172607422, 170.56336975097656, 36.66682434082031, 81.4114990234375, 86.24376678466797, 191.42947387695312, 154.91622924804688, 203.5867919921875, 18.149179458618164, 97.63302612304688, -4.197601318359375, 206.32467651367188, 109.96759033203125, 262.0814514160156, 153.41819763183594, 126.73638153076172, 217.86131286621094, 48.633026123046875, 283.3270263671875, 390.9772033691406, 256.09228515625, 62.062530517578125, 63.026878356933594, 4.094745635986328, 222.43902587890625, -12.415321350097656, 110.4044418334961, 117.2320327758789, 31.56976318359375, 97.78338623046875, 279.7917785644531, 44.03533172607422, 112.81059265136719, 68.09498596191406, 6.377159118652344, 153.96957397460938, 47.59266662597656, 51.78258514404297, 82.86588287353516, 101.83319854736328], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000243.npy"} +{"epoch": 0.3568281938325991, "step": 244, "batch_size": 64, "mean": 95.97021484375, "std": 85.02116394042969, "min": -77.78265380859375, "p10": -1.631611251831052, "median": 84.88869094848633, "p90": 227.1351028442383, "max": 266.7627868652344, "pos_frac": 0.890625, "sample": [113.79502868652344, 36.023475646972656, 57.30012130737305, 80.34510040283203, 2.9468040466308594, 34.24592208862305, 265.6697082519531, 93.31146240234375, -23.657135009765625, 14.0361328125, 116.52964782714844, 127.91439819335938, -14.87255859375, 177.35939025878906, 32.5867919921875, 110.00213623046875, -2.7908782958984375, 266.7627868652344, 23.85568618774414, 71.15827178955078, -77.78265380859375, 73.83651733398438, 179.9741973876953, -43.8157958984375, 173.64666748046875, 23.67279815673828, 66.0813980102539, 156.61898803710938, 61.59429931640625, 3.4304428100585938, 156.62022399902344, 59.67816925048828, 190.41796875, 211.36068725585938, 261.3785400390625, 220.53123474121094, 127.06513977050781, 238.96353149414062, 50.452823638916016, 233.59715270996094, 129.2065887451172, -9.416572570800781, 229.96533203125, 124.3801040649414, 102.87747192382812, -26.785385131835938, 52.005165100097656, 5.631124496459961, 11.886886596679688, 89.43228149414062, 38.5970458984375, 31.618873596191406, 97.70565032958984, 75.60694885253906, 48.329071044921875, 69.00820922851562, 189.9668426513672, 162.03268432617188, 1.0733451843261719, 114.36007690429688, 93.37190246582031, 241.89401245117188, 115.76701354980469, 203.73410034179688], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000244.npy"} +{"epoch": 0.35829662261380324, "step": 245, "batch_size": 64, "mean": 87.3641357421875, "std": 114.63250732421875, "min": -178.53622436523438, "p10": -12.875754165649413, "median": 70.1639289855957, "p90": 218.73959655761723, "max": 428.05560302734375, "pos_frac": 0.8125, "sample": [46.857627868652344, 262.24932861328125, 103.92001342773438, 88.60548400878906, 100.88833618164062, 100.22599792480469, 43.350624084472656, -13.291244506835938, 145.97622680664062, 82.01496887207031, 42.928253173828125, 75.1170654296875, 152.12852478027344, 38.845645904541016, 30.829797744750977, 137.59515380859375, 162.32901000976562, 193.09461975097656, 55.4979248046875, 15.477062225341797, 361.0281677246094, 82.37722778320312, -15.13336181640625, 230.17674255371094, 14.282938003540039, 2.4243621826171875, 84.64360046386719, -48.660152435302734, -96.16873168945312, 7.502845764160156, 175.4246368408203, 125.94654846191406, 422.006591796875, -53.90504455566406, 130.11050415039062, 11.912742614746094, -8.208930969238281, 64.92218017578125, 197.5022735595703, 106.51455688476562, -2.5633277893066406, 89.52140808105469, 129.5706329345703, -9.051300048828125, 65.6613540649414, -178.53622436523438, 196.74888610839844, 206.37161254882812, 10.950508117675781, 135.7993621826172, 224.0401611328125, 3.273040771484375, 382.25, -11.90627670288086, -6.284008026123047, 428.05560302734375, 74.66650390625, 82.67704010009766, 41.30671691894531, -52.750946044921875, 39.64677047729492, 26.348220825195312, 42.760406494140625, 13.408294677734375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000245.npy"} +{"epoch": 0.35976505139500736, "step": 246, "batch_size": 64, "mean": 80.57941436767578, "std": 102.99098205566406, "min": -103.60826110839844, "p10": -14.814269256591796, "median": 48.21500205993652, "p90": 225.56885070800786, "max": 420.7055358886719, "pos_frac": 0.859375, "sample": [104.129638671875, 183.57183837890625, 73.34219360351562, 26.08710479736328, 209.99737548828125, 148.5278778076172, 420.7055358886719, -87.74480438232422, 21.01363754272461, -103.60826110839844, 29.84252166748047, 61.57347869873047, 38.952701568603516, 9.738264083862305, 129.64501953125, 237.5272674560547, 231.24002075195312, 25.45269775390625, 40.36116027832031, 54.60296630859375, 27.52874755859375, 254.20968627929688, 112.26335144042969, -10.327598571777344, 92.91537475585938, -13.34124755859375, 144.89369201660156, -24.73358154296875, 212.33612060546875, -45.46247100830078, 4.941648483276367, 28.226829528808594, 72.07388305664062, 39.80976867675781, 171.65386962890625, 20.77873992919922, 38.26380920410156, 28.450496673583984, 29.273345947265625, 101.1339340209961, 245.58700561523438, 87.16730499267578, 29.285682678222656, 136.3859100341797, 24.605791091918945, 95.8900146484375, 49.39152145385742, 14.821548461914062, 5.435909271240234, 118.07057189941406, -98.54158020019531, 47.038482666015625, 183.17433166503906, 75.60787963867188, 28.214340209960938, 53.08055877685547, 8.36241340637207, 380.23345947265625, 175.5721435546875, 269.8148193359375, -15.445564270019531, 136.86109924316406, -34.14067077636719, 0.7625770568847656], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000246.npy"} +{"epoch": 0.36123348017621143, "step": 247, "batch_size": 64, "mean": 117.5644302368164, "std": 117.48432922363281, "min": -179.9461669921875, "p10": 0.06613616943359613, "median": 95.57920455932617, "p90": 255.39812011718752, "max": 457.7571716308594, "pos_frac": 0.890625, "sample": [-97.93384552001953, 41.47142028808594, 28.625503540039062, 276.78131103515625, 67.19184875488281, 332.104736328125, 141.89950561523438, 217.4122772216797, 16.240381240844727, 188.99929809570312, 249.18002319335938, 256.8061828613281, 118.269287109375, 247.991943359375, 40.173927307128906, 208.2757568359375, 322.85955810546875, 20.032249450683594, 252.11264038085938, 322.8450012207031, 80.70260620117188, 17.819177627563477, 68.39790344238281, 58.269447326660156, 2.3697509765625, 149.8329620361328, 12.087638854980469, 87.83381652832031, 102.92379760742188, 59.40960693359375, 138.68304443359375, 203.28306579589844, 58.03868865966797, 78.72065734863281, 63.149871826171875, 233.5445556640625, 218.62973022460938, 181.28826904296875, 119.24479675292969, -73.57608795166016, 3.175220489501953, 109.33341979980469, -0.9211273193359375, 95.79220581054688, 38.97343444824219, 20.737319946289062, 71.14019775390625, 194.8775634765625, 352.37811279296875, 95.36620330810547, 29.64832305908203, -40.020904541015625, 63.52037048339844, 457.7571716308594, 114.5059814453125, -1.2040023803710938, 160.2576904296875, 181.2775115966797, 159.0662384033203, -179.9461669921875, 250.47108459472656, -11.735397338867188, 41.92889404296875, 205.75201416015625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000247.npy"} +{"epoch": 0.36270190895741555, "step": 248, "batch_size": 64, "mean": 90.79135131835938, "std": 75.69000244140625, "min": -37.356781005859375, "p10": -0.47164268493652206, "median": 78.91995239257812, "p90": 189.90061950683594, "max": 267.2978515625, "pos_frac": 0.890625, "sample": [158.57675170898438, 166.40243530273438, 31.957536697387695, 122.12071990966797, 108.0546875, 0.8598670959472656, 53.45338439941406, 188.40664672851562, 84.91392517089844, 4.3304443359375, 19.10474395751953, 181.497314453125, 90.5347900390625, 20.1599178314209, 119.81427001953125, 7.301750183105469, 23.35533332824707, -4.39866828918457, 70.88904571533203, 42.35905075073242, 67.73748016357422, 91.0139389038086, 143.18301391601562, 257.4673156738281, 34.31059265136719, -8.758148193359375, 104.43441009521484, 174.80398559570312, 117.26336669921875, 137.53045654296875, 84.55210876464844, -13.540863037109375, 70.66275787353516, 185.57952880859375, 52.40264129638672, 144.3846435546875, 25.1761474609375, 19.256057739257812, 35.03204345703125, 42.0693473815918, 89.41703796386719, 207.52711486816406, 115.84307098388672, 255.92388916015625, 75.36575317382812, 69.75730895996094, 267.2978515625, 190.5408935546875, 34.849609375, 211.28196716308594, -1.0422897338867188, 54.038612365722656, 209.35809326171875, 126.11235046386719, 170.57992553710938, 171.80984497070312, -8.879535675048828, 26.3546142578125, 82.47415161132812, 162.98568725585938, 38.55677032470703, 50.89628982543945, -7.300975799560547, -37.356781005859375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000248.npy"} +{"epoch": 0.3641703377386197, "step": 249, "batch_size": 64, "mean": 65.60900115966797, "std": 100.1478271484375, "min": -154.98512268066406, "p10": -45.949124526977535, "median": 58.6206169128418, "p90": 190.1023666381836, "max": 402.79888916015625, "pos_frac": 0.6875, "sample": [-29.54496955871582, 174.67510986328125, -6.0160980224609375, -38.13908386230469, 47.385414123535156, -36.68304443359375, 34.426025390625, 84.77299499511719, 57.96934509277344, 223.40585327148438, 20.291595458984375, 28.544952392578125, 38.51121520996094, 62.778038024902344, 78.42420959472656, -154.98512268066406, -15.85903549194336, 188.667236328125, 125.81541442871094, -47.54777145385742, 120.17669677734375, 132.66893005371094, -38.39231872558594, 97.11222076416016, -5.8318634033203125, 196.48458862304688, -39.41938781738281, 181.12464904785156, 156.604736328125, 128.2025909423828, -17.399791717529297, 47.68796157836914, 91.89423370361328, 26.156631469726562, 59.271888732910156, 18.053977966308594, 105.36736297607422, -101.32157897949219, 112.43624877929688, 207.74893188476562, 121.48764038085938, -4.94976806640625, 402.79888916015625, -42.21894836425781, -118.07950592041016, -58.20384216308594, 79.7877426147461, -8.966686248779297, 92.63351440429688, 111.79281616210938, 166.32919311523438, 154.23683166503906, -47.58405303955078, 182.36050415039062, 233.38961791992188, 176.8186798095703, 27.13762664794922, -50.4300537109375, -17.036705017089844, 74.51411437988281, 18.55591583251953, 191.380615234375, 190.71742248535156, 6.9853363037109375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000249.npy"} +{"epoch": 0.3656387665198238, "step": 250, "batch_size": 64, "mean": 95.76774597167969, "std": 77.5641860961914, "min": -96.44853973388672, "p10": -3.8761409759521444, "median": 96.36722183227539, "p90": 209.66690521240236, "max": 240.1027069091797, "pos_frac": 0.890625, "sample": [200.96629333496094, 64.3660888671875, 30.127578735351562, -26.34315299987793, -53.9130859375, 140.78599548339844, 87.07331085205078, 220.7347412109375, 96.24095916748047, 6.3809814453125, 56.819557189941406, 233.42269897460938, 93.10581970214844, 105.40318298339844, 96.49348449707031, 93.11509704589844, 20.79645538330078, 96.17950439453125, 228.36236572265625, 173.46572875976562, 101.98916625976562, -34.41996765136719, 0.31613922119140625, 107.26002502441406, 66.16352081298828, 165.55023193359375, 67.78121948242188, 14.316156387329102, 39.39410400390625, 97.22341918945312, 13.761739730834961, 2.1252918243408203, 58.01555633544922, 204.14797973632812, -5.672832489013672, 240.1027069091797, 116.36595153808594, 94.60880279541016, -10.206352233886719, 41.84357452392578, 98.58183288574219, 213.79844665527344, 174.73939514160156, 109.80873107910156, 99.5755615234375, 58.448944091796875, 140.42193603515625, 70.17145538330078, 131.8122100830078, 75.75921630859375, 105.88087463378906, -9.9222412109375, 83.7416763305664, 199.033935546875, -96.44853973388672, 120.30036163330078, 211.66806030273438, 204.99754333496094, 104.92511749267578, 221.31390380859375, 122.30851745605469, 32.984336853027344, 164.5242919921875, 146.46023559570312], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000250.npy"} +{"epoch": 0.3671071953010279, "step": 251, "batch_size": 64, "mean": 80.65234375, "std": 100.45787811279297, "min": -166.6114959716797, "p10": -20.837885284423827, "median": 67.64320755004883, "p90": 215.69212951660157, "max": 339.6930847167969, "pos_frac": 0.796875, "sample": [246.69528198242188, 6.283411026000977, 267.5929260253906, 54.904808044433594, 141.55059814453125, -20.4114990234375, -166.6114959716797, 339.6930847167969, 54.02716827392578, 119.64884948730469, 182.49856567382812, -52.85161590576172, 179.88352966308594, 35.217315673828125, 51.6479377746582, -21.02062225341797, -140.1373291015625, 254.02542114257812, 5.858720779418945, 130.04400634765625, 30.027435302734375, 175.1440887451172, 4.101348876953125, 29.538734436035156, -6.74267578125, 121.80307006835938, -37.60041809082031, -58.83905029296875, 72.60870361328125, 226.07186889648438, 60.227691650390625, -40.909454345703125, 153.77593994140625, 19.359886169433594, 62.677711486816406, 139.39093017578125, 38.36571502685547, 58.85340118408203, 84.0186538696289, 137.22048950195312, 1.5108642578125, 122.52198791503906, -6.916177749633789, 194.8665008544922, 3.1706619262695312, 214.43771362304688, 80.53715515136719, 163.4850616455078, 75.81366729736328, 97.78974151611328, -11.761388778686523, 36.58991241455078, 330.384521484375, 105.5223388671875, 216.229736328125, 146.98947143554688, -7.964048385620117, 137.34112548828125, 73.76632690429688, -1.1134529113769531, 41.340553283691406, 87.108642578125, 39.43513488769531, 83.03123474121094], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000251.npy"} +{"epoch": 0.368575624082232, "step": 252, "batch_size": 64, "mean": 100.01866912841797, "std": 82.44734191894531, "min": -237.30018615722656, "p10": 5.501993560791021, "median": 97.32635879516602, "p90": 202.1216995239258, "max": 259.1549072265625, "pos_frac": 0.921875, "sample": [197.83216857910156, 158.83517456054688, 73.76445007324219, 151.85354614257812, 168.2666778564453, 39.329681396484375, 91.7136001586914, 152.2777099609375, 67.60617065429688, 132.421875, 117.48255920410156, 18.45404815673828, 134.5945587158203, 22.901458740234375, 97.75520324707031, 119.4373779296875, 69.79144287109375, 70.99754333496094, 140.31964111328125, 3.3326034545898438, 198.52662658691406, 117.70890808105469, 43.26841354370117, 203.66244506835938, 168.15994262695312, -36.824703216552734, -10.90257453918457, 72.18582153320312, 10.56390380859375, 79.71012878417969, -2.054370880126953, 83.84439086914062, 109.99125671386719, 146.70599365234375, 93.31494140625, 76.5767822265625, 104.70407104492188, 259.1549072265625, 239.15390014648438, 209.45333862304688, 67.48273468017578, 47.16804504394531, 122.47389221191406, 33.264434814453125, 120.31037902832031, 236.6219940185547, 224.59796142578125, 143.56121826171875, 103.57583618164062, 149.67298889160156, 23.589080810546875, -237.30018615722656, 193.62718200683594, 197.12582397460938, 61.178558349609375, 69.59248352050781, -7.7808380126953125, 85.58101654052734, 244.5989227294922, 1.1534614562988281, 73.76742553710938, 18.792922973632812, 135.7705841064453, 96.89751434326172], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000252.npy"} +{"epoch": 0.3700440528634361, "step": 253, "batch_size": 64, "mean": 89.05569458007812, "std": 103.97410583496094, "min": -240.41571044921875, "p10": -19.37497520446777, "median": 79.93387603759766, "p90": 227.5180877685547, "max": 358.75274658203125, "pos_frac": 0.796875, "sample": [-28.035791397094727, 7.292945861816406, 182.27874755859375, 32.2313232421875, -3.5261077880859375, 65.45972442626953, 177.52008056640625, 67.39900970458984, 190.06568908691406, 50.56373596191406, 85.44686889648438, 358.75274658203125, 122.14585876464844, 18.900169372558594, 85.76376342773438, 118.33460998535156, 227.23117065429688, 60.85591506958008, -3.4250926971435547, 73.12705993652344, -4.98072624206543, -240.41571044921875, 101.17489624023438, 255.66314697265625, 232.385009765625, 7.76667594909668, 312.642822265625, -14.708900451660156, -3.8180084228515625, 10.07351303100586, 12.790153503417969, 2.164308547973633, 74.42088317871094, 137.90243530273438, 206.42819213867188, 174.49722290039062, 105.24740600585938, -55.257728576660156, 235.7076873779297, 91.9542465209961, 126.86614990234375, 208.12864685058594, 118.09273529052734, 184.85696411132812, 227.64105224609375, 34.273406982421875, 112.36293029785156, 221.31979370117188, 61.196434020996094, -85.11542510986328, 150.6011199951172, 120.49187469482422, -51.1231689453125, 23.368391036987305, -13.3968505859375, 70.54684448242188, 86.42689514160156, 136.38502502441406, 155.26673889160156, -26.026031494140625, -21.37472152709961, 27.326812744140625, 235.44973754882812, 65.97917175292969], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000253.npy"} +{"epoch": 0.37151248164464024, "step": 254, "batch_size": 64, "mean": 84.08403015136719, "std": 118.907470703125, "min": -101.49698638916016, "p10": -49.825433731079094, "median": 67.20974349975586, "p90": 233.18675384521487, "max": 416.3049011230469, "pos_frac": 0.75, "sample": [52.21295166015625, -2.491455078125, 114.50946807861328, 72.14823913574219, 82.34247589111328, 106.26591491699219, -29.088333129882812, -39.51987838745117, 67.01243591308594, 18.265296936035156, 100.9398422241211, 137.07371520996094, 137.08389282226562, 80.07857513427734, -53.01935958862305, -71.7503662109375, 131.76071166992188, 7.445835113525391, 78.98750305175781, 110.66401672363281, -32.62600326538086, 174.82427978515625, 324.0970764160156, 227.17974853515625, 227.71722412109375, -76.06329345703125, 235.5308380126953, 226.46701049804688, -42.37294006347656, 24.757837295532227, 30.330102920532227, 150.63157653808594, 67.40705108642578, -96.5875244140625, 14.901729583740234, 164.15924072265625, -19.67937469482422, 282.21478271484375, -53.383846282958984, 39.207157135009766, -5.552497863769531, -83.45912170410156, 29.92860221862793, 284.4752197265625, -4.876501083374023, 143.08761596679688, 3.6785964965820312, 202.19407653808594, 23.95781898498535, 130.59225463867188, 416.3049011230469, 50.95137023925781, 47.85687255859375, 9.475242614746094, -101.49698638916016, 395.74688720703125, 76.42613220214844, 146.73965454101562, 106.76005554199219, 61.45222473144531, -13.971481323242188, 382.48773193359375, 74.505126953125, 34.479583740234375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000254.npy"} +{"epoch": 0.37298091042584436, "step": 255, "batch_size": 64, "mean": 86.84178161621094, "std": 118.63066864013672, "min": -249.89926147460938, "p10": -33.93400573730468, "median": 58.54924774169922, "p90": 242.55034942626955, "max": 369.62445068359375, "pos_frac": 0.8125, "sample": [358.72796630859375, -36.3594970703125, 11.82879638671875, 77.32246398925781, 115.3858642578125, 5.539398193359375, 5.931009292602539, 14.148117065429688, -249.89926147460938, -8.538089752197266, 206.8726348876953, 90.98110961914062, 5.8246002197265625, 21.295902252197266, 175.4832763671875, 136.418701171875, -105.24372100830078, -116.40400695800781, 94.24003601074219, 190.41001892089844, 170.46615600585938, 14.554277420043945, -17.42601776123047, 44.1883430480957, 75.22921752929688, 3.1648635864257812, -60.396484375, 278.5673828125, -23.99203109741211, 159.49948120117188, 28.170669555664062, 122.30838012695312, 92.83576965332031, 264.7785339355469, 286.72259521484375, 193.79534912109375, 198.17958068847656, 234.31671142578125, 236.0579071044922, 369.62445068359375, 21.500831604003906, -26.69460678100586, 200.7745819091797, 101.31731414794922, 20.733089447021484, 100.41325378417969, 49.9114990234375, 145.24203491210938, -68.17979431152344, 239.806884765625, 276.101318359375, 52.649940490722656, 62.77220916748047, 21.81855010986328, -30.104331970214844, 42.940948486328125, 52.76313018798828, 54.32628631591797, 8.850059509277344, 116.88372802734375, -35.575294494628906, 44.07295608520508, 197.21279907226562, 243.7261199951172], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000255.npy"} +{"epoch": 0.3744493392070485, "step": 256, "batch_size": 64, "mean": 97.502197265625, "std": 105.28884887695312, "min": -105.76148986816406, "p10": -34.076559066772454, "median": 93.51844024658203, "p90": 227.92294921875003, "max": 404.97332763671875, "pos_frac": 0.8125, "sample": [93.37908935546875, 404.97332763671875, 92.88640594482422, 127.67176818847656, 158.2247314453125, 185.93972778320312, 25.69132423400879, 125.84058380126953, -41.39879608154297, 32.2291259765625, 159.72422790527344, 183.06280517578125, 107.86166381835938, 168.64016723632812, 192.79896545410156, -44.94746398925781, 57.51272201538086, 160.05581665039062, -79.90269470214844, 35.688411712646484, 44.40800476074219, 60.00599670410156, -52.72559356689453, 176.50889587402344, 54.044979095458984, 184.30374145507812, 1.1829299926757812, -6.058990478515625, -40.96119689941406, 104.15824890136719, 268.4217529296875, 3.1413116455078125, 142.58721923828125, 93.65779113769531, 11.268852233886719, 98.2785873413086, 195.5927734375, 29.810592651367188, 69.02188110351562, 36.223236083984375, -105.76148986816406, -8.436309814453125, 15.085227966308594, 55.331634521484375, 55.99919128417969, -38.35658645629883, 101.38741302490234, 127.0102310180664, -9.52734375, 153.51837158203125, -24.089828491210938, 140.53871154785156, 41.694114685058594, 164.76171875, 235.2678680419922, 284.85589599609375, 138.8283233642578, 49.8652229309082, 343.81585693359375, 324.78472900390625, 135.27597045898438, 230.21493530273438, -13.301107406616211, 222.57498168945312], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000256.npy"} +{"epoch": 0.37591776798825255, "step": 257, "batch_size": 64, "mean": 100.38521575927734, "std": 108.86273193359375, "min": -93.01934814453125, "p10": -20.202583694458006, "median": 91.53260803222656, "p90": 254.25386657714847, "max": 400.9488220214844, "pos_frac": 0.796875, "sample": [258.4322509765625, 5.364292144775391, 63.35420227050781, 91.83370971679688, 244.50430297851562, 22.305816650390625, 168.60409545898438, 208.9602813720703, 174.98846435546875, 109.24732208251953, 4.781915664672852, -21.313682556152344, 61.39344024658203, 201.5874786376953, 130.97833251953125, 23.529098510742188, 129.89105224609375, 172.30401611328125, 25.374629974365234, 284.92218017578125, 72.2650146484375, 297.98309326171875, 211.20802307128906, -66.02537536621094, 99.67606353759766, 98.58841705322266, -26.860698699951172, -2.669921875, -17.61001968383789, 294.9242248535156, 159.44973754882812, 142.59056091308594, 205.774169921875, 400.9488220214844, 91.23150634765625, -12.895782470703125, 80.56155395507812, -4.103355407714844, 114.78704833984375, -49.698570251464844, 70.74063873291016, 243.36001586914062, -6.978496551513672, 175.42152404785156, 32.29039001464844, 133.23666381835938, 50.065086364746094, 15.792427062988281, 345.71563720703125, 159.07369995117188, 120.01686096191406, 180.0620574951172, 13.441551208496094, 97.1577377319336, 41.664306640625, 282.7232360839844, -46.99008560180664, -0.6045303344726562, 72.05828857421875, -93.01934814453125, 10.142379760742188, 2.9705810546875, -31.54376220703125, 106.68934631347656], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000257.npy"} +{"epoch": 0.37738619676945667, "step": 258, "batch_size": 64, "mean": 89.9486083984375, "std": 112.06814575195312, "min": -236.21530151367188, "p10": -22.156257247924806, "median": 81.42008972167969, "p90": 218.0711318969727, "max": 378.385498046875, "pos_frac": 0.8125, "sample": [134.56594848632812, 12.060823440551758, -22.53145980834961, 188.19927978515625, 44.00861358642578, 106.7911148071289, 194.28802490234375, 12.495391845703125, 119.67301940917969, 11.997589111328125, -19.06110191345215, -0.04332733154296875, 7.415506362915039, 74.31407165527344, 69.28130340576172, -7.346466064453125, -21.837059020996094, 61.45098876953125, 170.19151306152344, 40.538230895996094, -2.23822021484375, 195.51773071289062, 341.6693115234375, 136.05064392089844, 258.4568786621094, 174.24258422851562, 55.87744140625, 23.05303192138672, 18.880706787109375, 225.7115478515625, 100.66679382324219, 106.71817016601562, 205.38169860839844, 71.88870239257812, -95.70767211914062, 67.92828369140625, 190.33169555664062, -229.32955932617188, 51.469093322753906, 88.52610778808594, 240.98492431640625, -22.29305648803711, 0.10272216796875, 204.583984375, 38.78860092163086, -236.21530151367188, 378.385498046875, 147.07876586914062, -58.30291748046875, 121.75182342529297, 96.78054809570312, -38.060821533203125, 166.83297729492188, 174.16949462890625, 275.64422607421875, 64.09732818603516, 178.8059539794922, 223.50946044921875, 135.02865600585938, 152.45787048339844, 48.12098693847656, 62.416847229003906, 128.70831298828125, 111.78705596923828], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000258.npy"} +{"epoch": 0.3788546255506608, "step": 259, "batch_size": 64, "mean": 79.83476257324219, "std": 101.83690643310547, "min": -129.94619750976562, "p10": -37.926417541503895, "median": 52.76759147644043, "p90": 219.86225128173837, "max": 310.741943359375, "pos_frac": 0.8125, "sample": [162.9300994873047, -10.920116424560547, 152.55130004882812, 126.25003051757812, 30.159648895263672, -129.94619750976562, 186.02560424804688, 295.71075439453125, 263.0838317871094, 126.40164184570312, 14.824607849121094, -16.137474060058594, 132.3323974609375, -8.976325988769531, 21.558135986328125, -54.3427734375, 146.06454467773438, -15.625091552734375, 268.17626953125, 113.77677154541016, 90.614990234375, 24.23779296875, 30.77374267578125, 39.73093032836914, 163.9091033935547, 12.768749237060547, 54.12432861328125, 58.25103759765625, 310.741943359375, -97.44207763671875, 17.2255859375, 45.572723388671875, 254.16925048828125, 146.33924865722656, 33.82207489013672, 17.115432739257812, 105.45313262939453, 227.93704223632812, -29.72728729248047, 256.18017578125, -62.30313491821289, 12.972549438476562, 189.30564880371094, 7.813251495361328, 57.159088134765625, 21.628082275390625, 194.06027221679688, 179.14797973632812, -95.13319396972656, 69.01092529296875, 6.334321975708008, 201.0210723876953, 195.40487670898438, 37.72059631347656, 164.72616577148438, -41.440330505371094, 51.41085433959961, 153.50509643554688, 87.11243438720703, -60.446990966796875, 36.993587493896484, 32.6756706237793, 75.82235717773438, 29.22791290283203], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000259.npy"} +{"epoch": 0.3803230543318649, "step": 260, "batch_size": 64, "mean": 90.64608764648438, "std": 97.1241226196289, "min": -77.64500427246094, "p10": 2.2764175415039087, "median": 69.10059356689453, "p90": 239.40912170410175, "max": 379.89349365234375, "pos_frac": 0.90625, "sample": [163.45278930664062, -65.19745635986328, -16.930381774902344, -25.38782501220703, 47.971458435058594, 99.55903625488281, 85.3798828125, 95.60191345214844, 158.10626220703125, 40.58821105957031, 29.67428207397461, 279.32861328125, 64.84523010253906, 46.84027099609375, 98.39808654785156, 374.7396240234375, 114.87957000732422, 91.03760528564453, 16.22833251953125, 278.3814697265625, -77.64500427246094, 58.95783233642578, 38.152130126953125, 157.20913696289062, -58.220191955566406, -31.07941436767578, 114.8770751953125, 1.1568145751953125, 148.93907165527344, 88.54153442382812, 48.22019958496094, 193.03939819335938, 7.287895202636719, 136.69862365722656, 72.366455078125, 173.64730834960938, 294.96136474609375, 76.44963836669922, 59.00691604614258, 41.89820861816406, 4.956401824951172, 65.83473205566406, 165.5946044921875, 26.90868377685547, 103.02875518798828, 47.298641204833984, 4.888824462890625, 54.85139465332031, 109.1390151977539, 127.63758850097656, 98.24417114257812, 76.43842315673828, 11.484870910644531, 274.1878967285156, 39.71472930908203, 21.192344665527344, 29.79803466796875, 167.24037170410156, 7.708961486816406, 379.89349365234375, 29.75720977783203, 259.2818603515625, 29.176345825195312, 145.13047790527344], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000260.npy"} +{"epoch": 0.38179148311306904, "step": 261, "batch_size": 64, "mean": 98.16458129882812, "std": 100.16162872314453, "min": -84.3460922241211, "p10": -6.758825874328609, "median": 88.30648803710938, "p90": 211.3804168701172, "max": 517.4583129882812, "pos_frac": 0.859375, "sample": [203.88922119140625, 153.79409790039062, 38.80315399169922, -54.36985778808594, 71.21197509765625, 517.4583129882812, 47.230628967285156, 278.8822021484375, -27.23383331298828, 110.78386688232422, 40.0728759765625, 41.27076721191406, -2.62615966796875, 212.64993286132812, 21.349525451660156, 162.228515625, 115.62191772460938, 186.60781860351562, 160.23550415039062, -24.156890869140625, 67.49343872070312, 55.57489013671875, 104.41814422607422, 119.33289337158203, 132.68719482421875, 227.5055694580078, 207.53350830078125, 44.58683776855469, 242.2879638671875, 102.7935791015625, 6.6408233642578125, 58.559104919433594, 13.294973373413086, 88.63992309570312, 94.95675659179688, 19.725427627563477, 208.418212890625, 201.4258575439453, 148.9147186279297, 59.54560089111328, 186.24642944335938, 64.73078918457031, 188.00796508789062, 220.11660766601562, 38.84064483642578, 164.59600830078125, -44.45180892944336, 123.67732238769531, 32.105125427246094, 100.77455139160156, 115.2788314819336, 268.139892578125, 25.75977325439453, -81.59025573730469, -84.3460922241211, 87.97305297851562, 22.494232177734375, -2.954587936401367, 66.7850570678711, 114.01005554199219, 42.611839294433594, 56.90392303466797, 127.17401123046875, -8.389213562011719], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000261.npy"} +{"epoch": 0.3832599118942731, "step": 262, "batch_size": 64, "mean": 76.17524719238281, "std": 82.68199920654297, "min": -138.5968780517578, "p10": -27.53738441467285, "median": 82.81129837036133, "p90": 165.0045532226563, "max": 301.541748046875, "pos_frac": 0.84375, "sample": [38.030853271484375, 173.88876342773438, 57.61560821533203, 121.27743530273438, -66.23158264160156, 141.62603759765625, 301.541748046875, 83.70634460449219, 81.91625213623047, 117.68616485595703, -62.999549865722656, -138.5968780517578, 57.134368896484375, 281.87371826171875, 34.14678955078125, -85.16107940673828, 138.21078491210938, 115.00216674804688, 26.61736488342285, 91.08456420898438, 99.02755737304688, 43.03395080566406, -36.5466194152832, 136.83859252929688, 100.10655212402344, -9.459640502929688, 4.648380279541016, 228.67254638671875, 105.6475830078125, 175.64642333984375, 32.57196044921875, 49.00536346435547, 52.17401123046875, -27.811199188232422, 115.12187194824219, 87.26567840576172, 32.616943359375, 46.315773010253906, 140.94598388671875, 145.00506591796875, 129.73733520507812, 102.96583557128906, 78.25872039794922, 184.30862426757812, 116.23983001708984, 132.15936279296875, 134.2350616455078, 28.875511169433594, 20.54875946044922, 9.258108139038086, 11.475341796875, 67.76353454589844, -5.4715728759765625, 149.158935546875, -89.31301879882812, 5.143775939941406, 124.51640319824219, 139.75131225585938, 115.87608337402344, 171.7955322265625, 105.38314819335938, -26.898483276367188, 71.42192077636719, 68.82880401611328], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000262.npy"} +{"epoch": 0.38472834067547723, "step": 263, "batch_size": 64, "mean": 108.84306335449219, "std": 97.14408111572266, "min": -103.68922424316406, "p10": -7.994868087768553, "median": 103.66960144042969, "p90": 257.6097320556641, "max": 303.12347412109375, "pos_frac": 0.875, "sample": [199.93991088867188, -103.68922424316406, 104.73612976074219, 69.11473083496094, 183.4113006591797, 22.677024841308594, 184.63223266601562, 51.39024353027344, 23.237930297851562, 131.58837890625, -72.298828125, 147.35455322265625, 43.603843688964844, 146.56491088867188, 157.42955017089844, 112.22605895996094, 14.822052001953125, 188.3764190673828, 88.3437728881836, 70.30963134765625, -9.369729995727539, -12.966049194335938, 60.011268615722656, 255.87380981445312, -6.221073150634766, 48.62577819824219, 177.75192260742188, 62.02043914794922, 1.0529708862304688, 32.89577102661133, 30.006866455078125, 1.471200942993164, 232.22433471679688, 56.60820770263672, 46.77644348144531, 219.9993438720703, 142.62550354003906, 57.40325927734375, 291.9352111816406, 102.011474609375, 34.36609649658203, 208.89569091796875, 135.84201049804688, 164.97634887695312, 258.35369873046875, 104.48774719238281, 280.16033935546875, 44.10346984863281, 153.43246459960938, 164.12203979492188, 151.60171508789062, 263.9927062988281, 303.12347412109375, -78.460693359375, -8.75506591796875, 261.9488830566406, 55.838661193847656, -10.365913391113281, 157.80409240722656, 153.5838623046875, 102.85145568847656, 302.8034362792969, 123.54592895507812, 83.1956787109375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000263.npy"} +{"epoch": 0.38619676945668135, "step": 264, "batch_size": 64, "mean": 87.41431427001953, "std": 91.63023376464844, "min": -97.35459899902344, "p10": -5.8041038513183585, "median": 67.70123291015625, "p90": 233.46177368164066, "max": 335.25079345703125, "pos_frac": 0.859375, "sample": [143.46981811523438, 158.23764038085938, 33.00404739379883, 211.74343872070312, 98.35716247558594, -60.21409606933594, -6.1793670654296875, 236.51864624023438, 24.95895767211914, 257.13104248046875, -4.928489685058594, 37.022743225097656, 61.6444091796875, -30.782554626464844, 78.62836456298828, 114.89730072021484, 177.9669189453125, 126.313720703125, 138.78091430664062, 139.97654724121094, 86.3013916015625, 44.805816650390625, 47.945831298828125, 102.76715087890625, 41.96714401245117, 56.16783142089844, 42.565345764160156, 32.53739929199219, 26.095436096191406, 64.82211303710938, 0.49960899353027344, 8.039787292480469, 57.492347717285156, 265.74359130859375, 3.8397674560546875, 285.5249328613281, 81.94339752197266, 81.28528594970703, 113.47444152832031, 161.12823486328125, 81.62052154541016, 80.41015625, 67.59512329101562, 19.386444091796875, 34.58116912841797, 335.25079345703125, 175.2841796875, -97.35459899902344, 17.53533172607422, -10.394721984863281, 124.456298828125, 16.052265167236328, 67.80734252929688, 262.4449462890625, -2.1330337524414062, 52.26405715942383, 97.02620697021484, 129.10537719726562, -77.12782287597656, 226.32907104492188, 245.51864624023438, 188.4171905517578, -9.096649169921875, 28.044082641601562], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000264.npy"} +{"epoch": 0.3876651982378855, "step": 265, "batch_size": 64, "mean": 89.52692413330078, "std": 109.8080062866211, "min": -150.09471130371094, "p10": -17.61956634521484, "median": 55.7863655090332, "p90": 253.03906555175791, "max": 399.2298583984375, "pos_frac": 0.828125, "sample": [34.59819793701172, 263.398193359375, 228.86776733398438, 131.79502868652344, 41.55341339111328, 10.991071701049805, 127.61013793945312, 7.696144104003906, 363.7722473144531, 46.73516845703125, 178.40310668945312, -63.917335510253906, 37.8426399230957, 23.048961639404297, -7.8434295654296875, 44.910770416259766, 89.86367797851562, -40.348201751708984, 85.95964050292969, 31.547531127929688, 160.91690063476562, 351.9544372558594, -150.09471130371094, 273.6606750488281, -38.294342041015625, 102.96432495117188, 48.6021728515625, 311.5727233886719, -4.948101043701172, 399.2298583984375, -85.15205383300781, 57.83269500732422, 48.19182586669922, 30.018199920654297, 26.915054321289062, -19.402862548828125, 18.65087890625, 45.82478332519531, 170.95028686523438, 15.156646728515625, 206.97686767578125, 53.74003601074219, 45.380348205566406, 200.38861083984375, -13.458541870117188, 78.39192199707031, -1.0259780883789062, 100.12745666503906, 7.2911834716796875, 66.00527954101562, 52.96746063232422, 14.3472900390625, 171.2989501953125, 202.912109375, 80.37539672851562, 103.36239624023438, 107.25758361816406, 85.31916809082031, 109.93069458007812, -30.158222198486328, 278.6290588378906, 83.19961547851562, 163.8450164794922, 161.5853729248047], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000265.npy"} +{"epoch": 0.3891336270190896, "step": 266, "batch_size": 64, "mean": 94.04476928710938, "std": 98.40812683105469, "min": -134.9886016845703, "p10": -15.205927276611325, "median": 69.78214263916016, "p90": 223.81490631103517, "max": 343.4620056152344, "pos_frac": 0.84375, "sample": [54.31611633300781, 318.2132263183594, 83.66773223876953, 69.13401794433594, 182.4737091064453, 118.10102844238281, 70.43026733398438, 200.1578369140625, 107.23794555664062, 253.8775177001953, 125.69197082519531, 178.0731658935547, -11.413803100585938, 188.0355682373047, -20.470870971679688, 38.20738983154297, 177.89332580566406, 53.38372802734375, 152.86099243164062, 25.103492736816406, 59.448970794677734, -134.9886016845703, -45.080299377441406, 310.4148254394531, 37.82770538330078, 225.79592895507812, -1.99560546875, 26.316932678222656, 3.7423324584960938, 205.39129638671875, 260.7604064941406, 132.32568359375, 343.4620056152344, 199.27452087402344, 144.91482543945312, 39.95979309082031, 142.3805694580078, 207.95428466796875, 41.77478790283203, 46.671287536621094, 88.84906005859375, 63.60054016113281, 81.66117858886719, -2.7764835357666016, 143.21925354003906, 26.150226593017578, -42.996337890625, -53.236724853515625, 225.02528381347656, 113.5207290649414, -23.38208770751953, 53.59923553466797, 45.171234130859375, 76.9488525390625, 11.621551513671875, 17.423057556152344, 12.660223007202148, 52.29077911376953, -16.83112335205078, 220.99069213867188, 145.73284912109375, 118.5251235961914, 0.4277229309082031, 49.34398651123047], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000266.npy"} +{"epoch": 0.39060205580029367, "step": 267, "batch_size": 64, "mean": 96.68211364746094, "std": 91.94367980957031, "min": -158.03814697265625, "p10": 2.9537143707275413, "median": 86.73865127563477, "p90": 194.2540084838867, "max": 377.42059326171875, "pos_frac": 0.90625, "sample": [81.76742553710938, 127.48471069335938, 110.99516296386719, 9.017925262451172, 128.54360961914062, -16.19556999206543, 193.72720336914062, 130.9740753173828, -3.182586669921875, 49.25825881958008, 31.979644775390625, 21.1700439453125, 1.9566650390625, 228.21820068359375, 75.37482452392578, 89.62474060058594, 23.324508666992188, 174.88088989257812, 300.4052734375, 377.42059326171875, 73.87431335449219, 79.46668243408203, 133.4036865234375, -4.6147308349609375, -4.1286163330078125, 96.27284240722656, 100.36492919921875, 40.347747802734375, 150.27198791503906, 158.95738220214844, 193.63388061523438, 214.4454345703125, 296.8668212890625, 46.05076599121094, 33.414302825927734, 61.08767318725586, 6.621196746826172, 18.809349060058594, 18.115863800048828, 129.87860107421875, 181.09133911132812, 9.31988525390625, 5.280162811279297, 318.5887756347656, -158.03814697265625, 10.724510192871094, -31.808059692382812, 124.10150909423828, 83.8525619506836, 104.61358642578125, 69.76213073730469, 150.50827026367188, 22.350299835205078, 194.4797821044922, 74.35767364501953, 97.38424682617188, 50.28157043457031, 165.86102294921875, 54.26746368408203, 120.09428405761719, 155.97337341308594, 111.64785766601562, 181.51620483398438, 111.5593490600586], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000267.npy"} +{"epoch": 0.3920704845814978, "step": 268, "batch_size": 64, "mean": 93.34541320800781, "std": 101.3028335571289, "min": -177.52322387695312, "p10": -31.724885559082022, "median": 95.78898620605469, "p90": 182.21241149902346, "max": 371.242919921875, "pos_frac": 0.84375, "sample": [5.449897766113281, 223.3172149658203, 104.73258209228516, -10.268394470214844, 62.19273376464844, 75.09876251220703, 114.35322570800781, 72.0194091796875, 179.640625, 141.4957275390625, 147.48135375976562, -23.929588317871094, 124.7054214477539, 56.86766052246094, 230.88388061523438, -48.23402404785156, 35.337547302246094, 19.60875129699707, -82.099609375, 154.65989685058594, 98.93860626220703, 350.51666259765625, 21.338329315185547, 101.56532287597656, 98.40371704101562, 163.9717254638672, 178.0030517578125, 370.29754638671875, -9.21187973022461, 179.33447265625, -36.84394073486328, 23.0958251953125, 74.81552124023438, 54.67203903198242, 302.8766784667969, 51.56272888183594, 132.27377319335938, 96.74099731445312, 106.77716064453125, 173.58790588378906, 148.9811248779297, 166.761962890625, 46.326942443847656, 10.672300338745117, 117.82368469238281, 49.892364501953125, 183.31460571289062, 135.06399536132812, 17.211715698242188, 104.48790740966797, 94.83697509765625, 91.88285827636719, 371.242919921875, 143.50584411621094, 50.96943664550781, -47.09547424316406, 62.857948303222656, -35.06572723388672, 147.46295166015625, 30.872316360473633, 79.06768798828125, -177.52322387695312, -38.350467681884766, 102.87654113769531], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000268.npy"} +{"epoch": 0.3935389133627019, "step": 269, "batch_size": 64, "mean": 124.49280548095703, "std": 127.92713928222656, "min": -167.38705444335938, "p10": -23.339799499511717, "median": 122.47663879394531, "p90": 288.2470367431641, "max": 423.103759765625, "pos_frac": 0.8125, "sample": [109.20991516113281, 228.251708984375, 135.88467407226562, 97.38853454589844, -19.15332794189453, 162.18734741210938, 262.4309997558594, 151.44973754882812, 26.7862548828125, -24.69060516357422, 209.6704864501953, -20.18791961669922, 240.95938110351562, 85.05519104003906, 253.2199249267578, 137.6810302734375, 12.249900817871094, 82.01679992675781, 69.03465270996094, -167.38705444335938, 184.53372192382812, 37.156314849853516, 159.83645629882812, -106.58332061767578, -19.857837677001953, 220.04957580566406, 10.812675476074219, 344.0508117675781, 396.83203125, 120.68630981445312, 154.35250854492188, -11.71148681640625, 292.27459716796875, 214.00222778320312, 49.55322265625, 168.66758728027344, 199.70733642578125, 113.96057891845703, 223.09771728515625, 11.812225341796875, -28.584815979003906, 157.84837341308594, -52.84019470214844, -1.107147216796875, 124.2669677734375, 278.607177734375, 187.6805419921875, 278.8493957519531, 256.692626953125, 67.30110168457031, 149.12448120117188, 78.8199462890625, 62.42578125, 423.103759765625, -35.3253173828125, 358.7620544433594, 29.838409423828125, 25.73815155029297, 0.7322578430175781, 94.21250915527344, 309.5039978027344, 311.9760437011719, -94.46795654296875, 189.09085083007812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000269.npy"} +{"epoch": 0.39500734214390604, "step": 270, "batch_size": 64, "mean": 91.78181457519531, "std": 117.80372619628906, "min": -161.1959991455078, "p10": -25.439400482177724, "median": 79.43547058105469, "p90": 252.72116088867188, "max": 413.8190612792969, "pos_frac": 0.78125, "sample": [194.77174377441406, 226.21597290039062, 246.42208862304688, -161.1959991455078, -141.51904296875, 24.459213256835938, 24.91742706298828, 188.3045654296875, 93.19403839111328, 182.76412963867188, 298.30267333984375, -9.760467529296875, 59.92742919921875, 102.13209533691406, 5.1269989013671875, 10.98282241821289, 109.65108489990234, 126.77935791015625, -11.897010803222656, 118.21865844726562, 252.85638427734375, 110.24433898925781, -29.933982849121094, 152.153076171875, 269.004638671875, 128.72003173828125, -13.812589645385742, 13.288490295410156, 102.15914916992188, 52.948387145996094, -74.82421875, 310.88104248046875, 394.16925048828125, -68.73086547851562, 5.3473052978515625, 164.48907470703125, 148.6442413330078, 43.577606201171875, 89.07133483886719, 43.635372161865234, 413.8190612792969, -14.952041625976562, 50.499664306640625, -0.9942436218261719, 76.09413146972656, 95.71569061279297, -8.418472290039062, 143.45819091796875, -12.894943237304688, 77.65879821777344, 126.33041381835938, 146.19842529296875, 23.013992309570312, 8.67231559753418, 26.75946044921875, 48.586395263671875, -44.44195556640625, 7.5961456298828125, -48.486412048339844, 287.8578796386719, 137.65736389160156, 219.00265502929688, 81.21214294433594, 252.4056396484375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000270.npy"} +{"epoch": 0.3964757709251101, "step": 271, "batch_size": 64, "mean": 116.81657409667969, "std": 113.19961547851562, "min": -66.73210906982422, "p10": -13.297106552124019, "median": 87.1661148071289, "p90": 280.5080718994141, "max": 409.2940673828125, "pos_frac": 0.875, "sample": [409.2940673828125, 228.80038452148438, 103.28430938720703, 31.83700942993164, 131.53134155273438, 10.59218978881836, 230.21554565429688, 249.40927124023438, 106.49054718017578, 77.9860610961914, 147.96688842773438, -65.07223510742188, 332.89544677734375, -66.73210906982422, 100.2802963256836, 44.15796661376953, 155.00711059570312, 73.12201690673828, 54.99212646484375, -31.596900939941406, -33.23936462402344, 37.85865783691406, 254.1209716796875, 257.46551513671875, 343.3072204589844, 264.056640625, 63.743492126464844, 21.469636917114258, 34.94292449951172, 159.7847900390625, -15.08675765991211, 78.43165588378906, 122.29438781738281, -9.121253967285156, -28.185379028320312, 103.75886535644531, 34.73027038574219, 45.406768798828125, 169.90872192382812, 59.60116195678711, 17.556320190429688, 308.32244873046875, 71.88077545166016, 36.21680450439453, 102.15449523925781, 62.3441047668457, 64.05923461914062, 52.9490966796875, 242.17626953125, 120.29244995117188, 371.1225891113281, 203.37860107421875, 41.03462219238281, 26.307357788085938, 322.47442626953125, -29.56822967529297, 152.58895874023438, 287.5586853027344, 124.73828125, 216.91062927246094, 29.13335418701172, 193.99102783203125, 73.02809143066406, 95.90057373046875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000271.npy"} +{"epoch": 0.39794419970631423, "step": 272, "batch_size": 64, "mean": 133.50997924804688, "std": 128.22683715820312, "min": -163.94586181640625, "p10": -20.920419311523435, "median": 135.7467269897461, "p90": 294.4323852539063, "max": 445.31951904296875, "pos_frac": 0.828125, "sample": [164.57643127441406, 122.93366241455078, 3.2723827362060547, -21.616256713867188, -19.296798706054688, 261.2901611328125, 139.89622497558594, 337.6026611328125, 263.1336975097656, 151.70558166503906, 221.25054931640625, 252.08026123046875, -66.38256072998047, 86.50796508789062, -159.78843688964844, 59.16990661621094, 17.95122528076172, 145.4676055908203, 146.84689331054688, 116.20915985107422, -26.3424072265625, 238.01983642578125, 194.47494506835938, 122.30389404296875, 175.78018188476562, 10.153221130371094, 28.642353057861328, 229.1942596435547, 116.7448501586914, 445.31951904296875, 139.9273223876953, 112.77290344238281, 167.265380859375, 70.34042358398438, 24.280061721801758, 240.6600341796875, 355.12078857421875, -163.94586181640625, 170.13406372070312, 149.68988037109375, 431.4432678222656, 71.79714965820312, 323.46807861328125, 121.35916137695312, 131.59722900390625, -11.241706848144531, 195.15789794921875, 57.98698806762695, 182.1776885986328, 274.3175964355469, -29.961013793945312, 161.77952575683594, -1.5479583740234375, -37.479583740234375, -16.51628303527832, 236.5310821533203, 180.43333435058594, 61.160858154296875, 303.0530090332031, 51.549468994140625, 120.47345733642578, 239.56167602539062, 366.8919372558594, 107.30048370361328], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000272.npy"} +{"epoch": 0.39941262848751835, "step": 273, "batch_size": 64, "mean": 132.7435302734375, "std": 128.73980712890625, "min": -151.6547393798828, "p10": -5.324353027343745, "median": 102.03041076660156, "p90": 319.1155059814455, "max": 445.02728271484375, "pos_frac": 0.875, "sample": [173.31842041015625, 225.69021606445312, 37.68768310546875, 91.30906677246094, 398.6395263671875, 368.9815368652344, 135.38070678710938, -39.363182067871094, 382.89935302734375, -32.06846618652344, 229.96923828125, 86.0900650024414, 39.0157470703125, 59.27141571044922, 20.522066116333008, 140.94796752929688, 16.038249969482422, 104.89053344726562, 117.62301635742188, 222.00167846679688, 167.95204162597656, 25.330137252807617, 44.843017578125, 93.08580780029297, 37.06684875488281, 96.01740264892578, 237.8314666748047, 95.20663452148438, 140.1464385986328, 140.23178100585938, 145.8328094482422, 2.1142349243164062, 244.400634765625, 341.2218933105469, 9.301687240600586, 390.7922668457031, 376.5355224609375, 152.5200958251953, 99.1702880859375, 260.29510498046875, -92.58831024169922, -16.190032958984375, 189.5499725341797, -33.27870178222656, 162.94705200195312, 267.533935546875, 240.22474670410156, 44.12800598144531, 73.25852966308594, -151.6547393798828, 445.02728271484375, 106.7477798461914, 82.16885375976562, -1.0009002685546875, 51.5211181640625, 241.73062133789062, 87.81391906738281, 35.818233489990234, 83.81559753417969, -7.1772613525390625, 252.58099365234375, 30.175758361816406, 258.4879455566406, 265.2044982910156], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000273.npy"} +{"epoch": 0.4008810572687225, "step": 274, "batch_size": 64, "mean": 106.3211669921875, "std": 119.75415802001953, "min": -107.14945983886719, "p10": -35.66765518188476, "median": 96.46549606323242, "p90": 284.30335083007816, "max": 370.1663818359375, "pos_frac": 0.78125, "sample": [65.06607818603516, 47.90996170043945, 18.318252563476562, -81.26731872558594, 155.34710693359375, 115.64239501953125, 149.5186767578125, -14.691719055175781, 44.63367462158203, 71.38946533203125, -30.583677291870117, 278.1575012207031, 189.39378356933594, 169.14993286132812, 195.30282592773438, 110.55189514160156, 94.77249908447266, 348.379150390625, -107.14945983886719, 199.08343505859375, 19.03765869140625, 151.1837158203125, -33.918365478515625, 92.06185913085938, -36.41735076904297, 246.45654296875, 317.9996643066406, -37.896575927734375, 57.111427307128906, 370.1663818359375, -33.369571685791016, 16.724822998046875, 153.304931640625, 88.1539535522461, 35.488426208496094, 176.7845001220703, 301.751953125, -18.439208984375, 187.65113830566406, 53.61970520019531, 218.74533081054688, 59.89863586425781, 130.11611938476562, 141.7704620361328, 210.92874145507812, -29.29619598388672, 98.15849304199219, 13.634710311889648, 105.7071762084961, 340.1240539550781, 247.40234375, 72.17509460449219, 1.723428726196289, 146.00390625, -89.00720977783203, 212.65794372558594, -79.1072998046875, -88.92521667480469, 286.9372863769531, 118.689453125, 83.08489990234375, 289.60772705078125, -14.532257080078125, 201.67697143554688], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000274.npy"} +{"epoch": 0.4023494860499266, "step": 275, "batch_size": 64, "mean": 108.71735382080078, "std": 127.85067749023438, "min": -144.50076293945312, "p10": -49.09644088745117, "median": 79.36230087280273, "p90": 287.69672851562507, "max": 415.666748046875, "pos_frac": 0.8125, "sample": [154.67091369628906, 61.26763916015625, 183.59481811523438, 21.171680450439453, 8.456914901733398, -32.02698516845703, 254.16635131835938, 47.06626892089844, 292.4298400878906, -47.296661376953125, 215.33056640625, 262.6372375488281, 415.666748046875, 67.70499420166016, 145.518798828125, 155.83995056152344, -37.62128829956055, 26.278823852539062, 73.19576263427734, 109.68150329589844, -59.187217712402344, 140.9130859375, 324.59368896484375, -69.93262481689453, 41.73584747314453, -81.4720687866211, 161.75259399414062, 142.37533569335938, 195.76043701171875, 78.64495086669922, -49.867774963378906, 50.09910583496094, 201.8023223876953, 140.44381713867188, 46.621856689453125, 1.6448211669921875, -144.50076293945312, 80.07965087890625, -23.5048885345459, 29.034149169921875, 63.39794921875, 219.1871337890625, -99.47830200195312, 26.59207534790039, 151.81948852539062, 276.6528015136719, 140.30715942382812, 401.618408203125, 195.19125366210938, 324.54766845703125, 305.09832763671875, -78.0303955078125, 209.842041015625, 304.1820068359375, 1.5534133911132812, 202.85693359375, 73.69804382324219, 219.09495544433594, -22.47223663330078, 134.4057159423828, 35.50984191894531, 245.52294921875, 11.292724609375, 30.75042724609375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000275.npy"} +{"epoch": 0.40381791483113066, "step": 276, "batch_size": 64, "mean": 111.8005599975586, "std": 131.83177185058594, "min": -120.97423553466797, "p10": -11.496914291381835, "median": 88.33814239501953, "p90": 302.4560516357422, "max": 497.52081298828125, "pos_frac": 0.828125, "sample": [82.33216094970703, 497.52081298828125, 108.1456298828125, 331.4969482421875, 306.5768127441406, 105.66356658935547, -9.117866516113281, 397.4020080566406, 43.358306884765625, 163.8337860107422, 83.92681884765625, 50.72028350830078, -9.19140625, 153.5146484375, 175.70159912109375, 185.5382080078125, 51.02392578125, -120.97423553466797, 37.66029357910156, 17.135555267333984, 27.173480987548828, 176.7460479736328, 112.54496765136719, 241.85687255859375, 16.355087280273438, -103.32095336914062, 31.248470306396484, 99.82493591308594, 112.25975036621094, 223.47531127929688, -27.484283447265625, 123.92449951171875, 78.93087005615234, 79.11827087402344, 232.85243225097656, 146.78546142578125, -105.31961059570312, -115.77378845214844, 7.308774948120117, 60.560665130615234, 2.2779102325439453, -4.780517578125, -12.484989166259766, 359.5102233886719, 98.85236358642578, 58.69129180908203, 99.71749877929688, 168.69834899902344, 215.8933563232422, 26.57457733154297, 286.10504150390625, -6.028434753417969, 168.70458984375, 181.44482421875, 27.967437744140625, 203.6326141357422, 92.74946594238281, 68.41107177734375, 292.8409423828125, 54.1409912109375, -82.2813720703125, 374.98992919921875, 30.615182876586914, 377.6583251953125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000276.npy"} +{"epoch": 0.4052863436123348, "step": 277, "batch_size": 64, "mean": 108.79533386230469, "std": 112.6235580444336, "min": -78.85542297363281, "p10": -17.206645202636714, "median": 74.36823272705078, "p90": 276.0857330322266, "max": 349.05145263671875, "pos_frac": 0.859375, "sample": [27.28628158569336, 195.16163635253906, -44.75068664550781, -18.979156494140625, 42.52486801147461, 147.52584838867188, 268.829833984375, 143.8704376220703, -13.070785522460938, 179.32565307617188, 118.20819091796875, 39.34886932373047, 202.32310485839844, -12.618947982788086, 227.2222442626953, 166.31617736816406, 7.881460189819336, 66.98735046386719, 141.00851440429688, 139.7506561279297, 17.897727966308594, 193.16848754882812, 39.75416564941406, 44.980613708496094, -60.78339385986328, -40.980201721191406, 349.05145263671875, 93.09130859375, 27.381378173828125, 112.08181762695312, -78.85542297363281, 320.26873779296875, 45.07488250732422, 236.7255859375, 37.708763122558594, -52.407264709472656, 347.88916015625, 179.311279296875, 81.74911499023438, 23.122894287109375, 151.0001220703125, 56.68867492675781, 312.21630859375, 47.53316116333008, 6.392951965332031, 51.46625518798828, 57.76030731201172, 275.7535095214844, 276.11419677734375, 319.89801025390625, -49.81280517578125, 210.48641967773438, 55.724403381347656, 39.172325134277344, 17.064743041992188, 2.045307159423828, 130.0287322998047, 54.676170349121094, 276.0193176269531, 300.33831787109375, 85.08543395996094, 139.9405517578125, 54.00804901123047, 152.91806030273438], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000277.npy"} +{"epoch": 0.4067547723935389, "step": 278, "batch_size": 64, "mean": 123.15806579589844, "std": 132.85069274902344, "min": -95.45172119140625, "p10": -11.226144981384277, "median": 99.85776901245117, "p90": 346.34186401367197, "max": 448.9095458984375, "pos_frac": 0.828125, "sample": [3.5533084869384766, 227.52621459960938, -24.41448211669922, 44.31477737426758, 167.46817016601562, 184.7977752685547, 127.58694458007812, 54.7292366027832, 146.4495086669922, 77.89568328857422, 85.42396545410156, 122.16180419921875, 75.02201080322266, -4.7324981689453125, 137.36471557617188, 175.25843811035156, -30.142730712890625, 215.16677856445312, 375.1600036621094, -1.1786270141601562, 36.23851013183594, 448.9095458984375, 233.5549774169922, 113.01535034179688, 156.43487548828125, 198.58787536621094, 99.09821319580078, 8.789321899414062, 93.45426177978516, 391.7234802246094, 319.5198974609375, -77.01210021972656, 97.46548461914062, -65.08352661132812, 354.657470703125, 353.98785400390625, 1.9880905151367188, 58.36114501953125, 15.389078140258789, 176.08526611328125, 50.76886749267578, 328.501220703125, 100.61732482910156, 20.899948120117188, 32.52293395996094, 325.8907165527344, 84.50714874267578, 161.27174377441406, 18.05245590209961, 355.8428955078125, 108.65403747558594, 134.07838439941406, 16.173095703125, 17.919342041015625, 169.60284423828125, 252.7292022705078, -6.200563430786133, -9.963415145874023, 423.0892333984375, -79.25042724609375, 199.69607543945312, 109.35580444335938, -95.45172119140625, -11.767314910888672], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000278.npy"} +{"epoch": 0.40822320117474303, "step": 279, "batch_size": 64, "mean": 133.2210693359375, "std": 149.61065673828125, "min": -160.80404663085938, "p10": -48.45751609802246, "median": 116.81611633300781, "p90": 350.3849731445313, "max": 532.8648681640625, "pos_frac": 0.796875, "sample": [60.18145751953125, -10.249990463256836, 23.278553009033203, 78.44610595703125, 14.880096435546875, -29.84642791748047, 348.1649169921875, 285.8292541503906, 69.5782699584961, 16.61563491821289, -16.69586944580078, 186.84654235839844, 108.92164611816406, 43.584999084472656, 423.4301452636719, 406.5268249511719, 43.91828536987305, 144.1116943359375, 199.5384063720703, 121.42738342285156, 403.09735107421875, 119.84960174560547, -1.7283573150634766, -160.80404663085938, 2.958253860473633, 59.260162353515625, 88.76333618164062, 197.3271484375, 58.44805908203125, 358.09844970703125, 120.074951171875, 101.74224090576172, 202.04940795898438, 47.73749542236328, 180.93722534179688, 370.74462890625, 172.67550659179688, 532.8648681640625, 351.33642578125, 254.0337677001953, -49.38627624511719, -32.566131591796875, -95.51675415039062, -145.71014404296875, 112.30744171142578, 327.96527099609375, 66.44049072265625, 189.07948303222656, 232.31764221191406, 72.62631225585938, 252.36753845214844, 247.85089111328125, 165.7835693359375, -49.83687973022461, 249.82955932617188, 113.78263092041016, 250.882568359375, 150.53199768066406, -71.62725830078125, -95.04948425292969, 279.33154296875, 222.52996826171875, -46.290409088134766, 200.5504150390625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000279.npy"} +{"epoch": 0.40969162995594716, "step": 280, "batch_size": 64, "mean": 133.7972412109375, "std": 114.7332534790039, "min": -123.84507751464844, "p10": -30.47364349365234, "median": 132.14199829101562, "p90": 283.28838806152345, "max": 374.88995361328125, "pos_frac": 0.84375, "sample": [125.82994079589844, 225.8977508544922, 110.22327423095703, 159.1348114013672, -27.883346557617188, -57.728919982910156, 219.84912109375, 296.1231689453125, 271.34954833984375, 16.881866455078125, 197.4801788330078, -1.2673873901367188, -64.06788635253906, 180.8310546875, 279.7250061035156, 300.13232421875, 98.71512603759766, 222.8509979248047, 129.234130859375, 164.07962036132812, 39.848270416259766, 374.88995361328125, 362.5385437011719, 182.91448974609375, 227.89846801757812, 110.638427734375, 209.96507263183594, 202.5891571044922, 6.653289794921875, 212.883544921875, 148.91098022460938, 133.9727783203125, 331.34625244140625, 95.84063720703125, -31.583770751953125, 144.05979919433594, -47.067901611328125, 284.8155517578125, 132.8899383544922, 133.84478759765625, 158.13233947753906, 62.976707458496094, -123.84507751464844, 340.3944396972656, -60.99711990356445, -39.3021240234375, 94.9612808227539, 214.2510986328125, 127.98294067382812, 127.68282318115234, 131.39405822753906, 271.8746337890625, 74.69535827636719, 97.1959457397461, 16.2626953125, 18.56456756591797, 121.64916229248047, 80.45903015136719, 125.94711303710938, 247.0583038330078, 159.09799194335938, -20.29145050048828, 43.01069641113281, 188.62998962402344], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000280.npy"} +{"epoch": 0.4111600587371512, "step": 281, "batch_size": 64, "mean": 126.9708023071289, "std": 119.35255432128906, "min": -346.8807067871094, "p10": -2.0819944381713755, "median": 127.28911209106445, "p90": 267.59234619140625, "max": 424.18585205078125, "pos_frac": 0.890625, "sample": [65.7808609008789, 234.4986572265625, 112.36367797851562, 184.3805694580078, 101.52149963378906, 66.08334350585938, 92.77247619628906, 153.68145751953125, 136.4475555419922, 58.66779327392578, 125.7367935180664, 268.24969482421875, 202.19129943847656, 42.34062957763672, 181.26625061035156, -61.794586181640625, 157.8559112548828, 357.54180908203125, 69.4837646484375, 128.8414306640625, 57.734954833984375, 181.30650329589844, 210.9903106689453, 294.73565673828125, 125.12039947509766, 136.92788696289062, 161.81631469726562, 173.4695587158203, -11.734882354736328, 424.18585205078125, 254.37615966796875, 95.45440673828125, 147.6881561279297, 290.3653564453125, 156.55227661132812, -346.8807067871094, 105.277587890625, 121.47325134277344, 157.02423095703125, 75.59429931640625, 406.85931396484375, 54.21275329589844, 103.19682312011719, 159.1233673095703, 135.59979248046875, 210.70346069335938, 67.14391326904297, 156.1983642578125, 266.05853271484375, 105.42981719970703, 9.234794616699219, 86.41643524169922, -69.91484069824219, 43.896209716796875, -72.08843994140625, 13.512649536132812, -15.218568801879883, 234.755126953125, 161.98585510253906, 43.26245880126953, 75.38678741455078, 158.15126037597656, -6.932046890258789, 309.7689514160156], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000281.npy"} +{"epoch": 0.41262848751835535, "step": 282, "batch_size": 64, "mean": 97.33328247070312, "std": 129.67420959472656, "min": -97.5494384765625, "p10": -42.36973419189453, "median": 78.56774139404297, "p90": 244.25725708007815, "max": 676.3240356445312, "pos_frac": 0.75, "sample": [309.8575744628906, 74.42658996582031, 78.33528137207031, -58.238441467285156, 163.32781982421875, 197.02297973632812, 75.77578735351562, 37.164825439453125, 103.92855834960938, 231.1801300048828, 247.15106201171875, 119.359130859375, 245.57565307617188, 86.5889663696289, -17.848098754882812, 138.80352783203125, 676.3240356445312, 132.67825317382812, 56.296844482421875, 68.5673828125, 49.554534912109375, 163.96420288085938, 15.27590560913086, -75.70089721679688, 380.8487548828125, 284.99566650390625, 110.45159912109375, 95.26728820800781, 138.23513793945312, 319.0792236328125, 218.2809295654297, 50.80669021606445, 218.9549560546875, -42.624053955078125, 44.23954391479492, 92.37727355957031, -9.052413940429688, -22.226234436035156, 39.2890625, -52.04510498046875, -41.38983154296875, 78.80020141601562, 66.27959442138672, 120.67813110351562, 202.60301208496094, 65.95795440673828, -88.06826782226562, 163.80233764648438, 143.55084228515625, 107.60475158691406, -29.048004150390625, -21.931259155273438, 47.36729431152344, 43.97365951538086, 113.10696411132812, -29.0333251953125, 170.1937255859375, -41.77632141113281, -37.482872009277344, 241.18099975585938, -97.5494384765625, 33.2223014831543, -56.27714538574219, 87.31523132324219], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000282.npy"} +{"epoch": 0.41409691629955947, "step": 283, "batch_size": 64, "mean": 142.69090270996094, "std": 140.4113311767578, "min": -226.0755615234375, "p10": -28.526519966125488, "median": 117.51211929321289, "p90": 319.0350646972656, "max": 497.7466735839844, "pos_frac": 0.828125, "sample": [37.89779281616211, -31.31560516357422, 121.13053894042969, -5.845582962036133, 227.4937744140625, 302.5558166503906, 68.03076171875, 188.0785369873047, 301.43170166015625, 24.94647789001465, -25.134254455566406, 194.70594787597656, -29.69588851928711, 64.64134216308594, 42.97090148925781, 18.592926025390625, -226.0755615234375, 74.57752990722656, 275.6141662597656, -3.7206287384033203, 403.2530517578125, 324.5833740234375, 94.99983215332031, 219.65696716308594, 91.76493835449219, 63.79473876953125, -55.0784912109375, 312.8590087890625, 214.71356201171875, 405.31854248046875, 20.79949951171875, 231.90097045898438, -29.262954711914062, 92.29690551757812, 225.02899169921875, 83.35037231445312, 84.21786499023438, -41.56865310668945, 92.9542007446289, -45.962005615234375, 282.7576904296875, 219.15296936035156, 264.0362243652344, 86.9107666015625, 279.68536376953125, 170.5660400390625, 77.4797592163086, 121.00699615478516, 238.2930908203125, 154.554443359375, 420.1228942871094, 126.78020477294922, 307.917724609375, 114.01724243164062, 34.148704528808594, 229.4236602783203, 358.2550964355469, 172.1231231689453, 77.60494995117188, 497.7466735839844, 321.68194580078125, 66.65119171142578, 127.6078872680664, -26.80817222595215], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000283.npy"} +{"epoch": 0.4155653450807636, "step": 284, "batch_size": 64, "mean": 134.25747680664062, "std": 133.26368713378906, "min": -78.55062866210938, "p10": 6.487386322021487, "median": 107.0523910522461, "p90": 324.6530364990235, "max": 549.12841796875, "pos_frac": 0.90625, "sample": [-30.01116943359375, 84.71415710449219, 192.96156311035156, 32.352333068847656, 110.80091094970703, -25.9212589263916, 219.68873596191406, 28.5086669921875, 202.990478515625, 258.1658630371094, 317.485107421875, 12.462215423583984, 327.7250061035156, 29.161157608032227, 170.39508056640625, 73.17642211914062, 39.355918884277344, 210.40121459960938, 204.22787475585938, 354.3933410644531, 5.484130859375, 459.4620056152344, 126.88694763183594, 220.43438720703125, 96.55499267578125, 43.90806579589844, 96.34059143066406, 47.58517074584961, 200.1042938232422, -20.121749877929688, 78.72102355957031, 516.4234008789062, 126.484619140625, -68.30840301513672, 105.51287078857422, 108.468017578125, 9.831077575683594, 82.21623992919922, 175.91676330566406, 549.12841796875, 105.63676452636719, 159.44522094726562, 97.443603515625, 42.28892517089844, -78.55062866210938, -78.28228759765625, 154.06338500976562, 254.62191772460938, 45.45362854003906, 126.17977142333984, 359.06109619140625, 160.73162841796875, 135.2598876953125, 158.32899475097656, 45.822265625, 9.26590347290039, 8.828315734863281, 35.42084503173828, 192.3911895751953, 221.9593048095703, 46.215492248535156, 338.26947021484375, 60.626014709472656, 217.93060302734375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000284.npy"} +{"epoch": 0.4170337738619677, "step": 285, "batch_size": 64, "mean": 122.10980224609375, "std": 114.34367370605469, "min": -228.00889587402344, "p10": -5.753572463989251, "median": 120.0155258178711, "p90": 268.67498168945315, "max": 390.87994384765625, "pos_frac": 0.890625, "sample": [259.00860595703125, 170.57049560546875, 35.28466796875, 254.82675170898438, 77.9809341430664, 77.09176635742188, 122.08228302001953, 25.520957946777344, 143.8658905029297, 330.8000183105469, 185.34385681152344, 184.6388397216797, 205.87548828125, 84.84003448486328, 110.46894836425781, 390.87994384765625, 74.43045043945312, 20.543487548828125, 144.12326049804688, 172.96820068359375, 270.5996398925781, -58.78069305419922, 49.72216033935547, 126.93841552734375, 264.1841125488281, 55.08203887939453, 12.97372055053711, 108.94540405273438, 60.64642333984375, 17.70038604736328, 197.0366668701172, 66.10728454589844, -41.604896545410156, 1.8434104919433594, 187.2132568359375, 342.6401672363281, 99.88224029541016, -228.00889587402344, 206.24981689453125, 83.25863647460938, 149.982177734375, 4.9030303955078125, 357.167236328125, 144.16107177734375, -25.48230743408203, 187.249755859375, 168.9388427734375, 155.90524291992188, 187.79147338867188, 279.0238342285156, 52.756935119628906, -55.702030181884766, 188.23179626464844, 134.91183471679688, -70.4150390625, 22.370540618896484, 272.43487548828125, 56.17462158203125, 69.19896697998047, 102.87498474121094, -9.009422302246094, 244.42388916015625, 185.41171264648438, 117.94876861572266], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000285.npy"} +{"epoch": 0.4185022026431718, "step": 286, "batch_size": 64, "mean": 132.2296142578125, "std": 131.75633239746094, "min": -171.87704467773438, "p10": -30.568455696105953, "median": 134.11803436279297, "p90": 343.02147521972665, "max": 394.68414306640625, "pos_frac": 0.8125, "sample": [22.656044006347656, 12.023460388183594, 384.2154846191406, 140.50982666015625, 92.12338256835938, 147.87741088867188, 80.24578094482422, 105.79055786132812, 206.10369873046875, -93.42245483398438, 187.4884033203125, 241.24911499023438, -31.669631958007812, -27.662561416625977, 157.62527465820312, 134.127685546875, 357.7335510253906, -11.189163208007812, 145.13214111328125, 154.47645568847656, 350.1907958984375, 100.00003051757812, -171.87704467773438, 323.1225891113281, 198.5803985595703, -48.8492431640625, -8.282894134521484, 249.4348602294922, 233.38726806640625, 350.74658203125, 394.68414306640625, 14.542266845703125, 182.29025268554688, 109.70529174804688, 36.10069274902344, 92.20219421386719, 176.13221740722656, 134.10838317871094, 82.46624755859375, -45.40507507324219, 194.41429138183594, 19.85247039794922, 161.77676391601562, -42.89280700683594, 13.424667358398438, 237.6870574951172, 326.2930603027344, 167.8793487548828, -60.41291046142578, 353.2591857910156, -27.99904441833496, 118.4413833618164, 52.58042907714844, 215.39639282226562, 161.30807495117188, 282.70697021484375, 62.4715576171875, 155.0012969970703, 125.46124267578125, -26.566009521484375, 383.45648193359375, 273.717041015625, 67.55058288574219, 89.17316436767578], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000286.npy"} +{"epoch": 0.4199706314243759, "step": 287, "batch_size": 64, "mean": 114.54322814941406, "std": 132.8385772705078, "min": -197.31790161132812, "p10": -24.65536727905273, "median": 102.05803680419922, "p90": 293.36100769042974, "max": 525.5933837890625, "pos_frac": 0.875, "sample": [126.71537780761719, 102.07774353027344, 6.531759262084961, 50.91350555419922, 108.4105453491211, -42.976409912109375, 201.9852294921875, 424.3818359375, 162.43753051757812, 105.27186584472656, 380.84075927734375, 114.8307876586914, 109.67503356933594, 313.337646484375, -26.780670166015625, -71.25286865234375, 73.84191131591797, 227.3037872314453, 201.77017211914062, 272.95135498046875, 86.06561279296875, 5.023073196411133, 138.56570434570312, -197.31790161132812, 525.5933837890625, 58.43561553955078, 54.34642791748047, 52.1015625, 43.175079345703125, -19.696327209472656, 108.46026611328125, 46.772972106933594, 23.73809814453125, 130.30751037597656, 39.83916091918945, 241.09532165527344, 33.85795593261719, 154.90780639648438, 43.19465637207031, 81.69975280761719, 50.59159851074219, 259.31048583984375, 96.2049789428711, 132.50390625, 3.101409912109375, 105.5491943359375, 64.74749755859375, 35.76475524902344, 255.25315856933594, 102.038330078125, -98.46809387207031, 59.918113708496094, 170.83689880371094, 161.79859924316406, 318.8200378417969, -112.8359603881836, -128.42079162597656, 5.549736022949219, 96.96338653564453, 192.37356567382812, 376.89898681640625, 302.1080017089844, 186.56942749023438, 201.156494140625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000287.npy"} +{"epoch": 0.42143906020558003, "step": 288, "batch_size": 64, "mean": 131.75595092773438, "std": 117.51336669921875, "min": -188.94046020507812, "p10": 5.654684638977055, "median": 118.03561782836914, "p90": 269.1903625488282, "max": 453.15093994140625, "pos_frac": 0.90625, "sample": [206.52676391601562, 97.05902099609375, -52.90473556518555, 224.9591064453125, 220.8482208251953, 205.3822479248047, 9.426694869995117, 453.15093994140625, 75.39505004882812, 223.7265625, 141.79232788085938, 99.23797607421875, 4.038108825683594, 21.247955322265625, 55.4945182800293, 70.93397521972656, 103.59762573242188, 221.17031860351562, 81.36703491210938, 238.975830078125, 125.64341735839844, 188.002685546875, 233.78475952148438, 37.268035888671875, 158.68116760253906, 164.6981201171875, 209.89935302734375, 10.650918960571289, 114.0168685913086, 115.61759948730469, 56.949989318847656, 85.42059326171875, 316.92437744140625, 180.89007568359375, 336.05267333984375, 182.15878295898438, 48.38752746582031, 146.50949096679688, 171.24581909179688, 80.91287994384766, -188.94046020507812, 390.4757995605469, 51.21615982055664, 145.38870239257812, 46.13337707519531, 140.93502807617188, 120.4536361694336, 62.227691650390625, 362.180419921875, 97.95946502685547, 229.5087890625, 256.14068603515625, 102.19800567626953, -135.7617950439453, 86.06497192382812, 276.60186767578125, 166.42861938476562, 41.222137451171875, 42.392578125, 252.46484375, -8.196662902832031, 274.7830810546875, -30.546646118164062, -14.089845657348633], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000288.npy"} +{"epoch": 0.42290748898678415, "step": 289, "batch_size": 64, "mean": 145.47280883789062, "std": 142.8282470703125, "min": -122.05284881591797, "p10": -6.1583169937133775, "median": 111.22484970092773, "p90": 379.8641662597657, "max": 463.03607177734375, "pos_frac": 0.859375, "sample": [90.73046875, 17.1837158203125, 398.98785400390625, 415.9378967285156, -122.05284881591797, 339.8819274902344, 247.6717987060547, 189.71298217773438, 110.58708190917969, 406.9466857910156, 301.4658203125, 254.49241638183594, 243.2411346435547, 388.4229736328125, 214.8063507080078, 229.60079956054688, 67.20631408691406, 454.14788818359375, 32.57282257080078, 223.473388671875, 123.4693603515625, 0.11875724792480469, 111.4686279296875, -6.720577239990234, -4.846376419067383, 132.96566772460938, 224.208984375, 232.95660400390625, 36.79994201660156, 95.18730163574219, 243.38653564453125, 463.03607177734375, 232.92123413085938, -33.56013107299805, 76.79187774658203, 47.92857360839844, 395.9210205078125, 91.98396301269531, -1.6738510131835938, 9.818380355834961, 71.96196746826172, 165.99896240234375, 5.7203369140625, 186.141845703125, 24.414766311645508, 253.3271942138672, 336.3987731933594, 21.00456428527832, 60.41670227050781, -88.83944702148438, -23.36227798461914, 163.4671630859375, 78.87838745117188, 16.406156539916992, 359.89361572265625, 136.058349609375, 43.32537078857422, 69.3232650756836, -15.937164306640625, 158.13253784179688, 85.46728515625, -59.85478973388672, 173.75595092773438, 110.98107147216797], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000289.npy"} +{"epoch": 0.4243759177679883, "step": 290, "batch_size": 64, "mean": 146.79083251953125, "std": 158.4547882080078, "min": -171.3162841796875, "p10": -8.888098144531243, "median": 119.23745727539062, "p90": 381.4948913574219, "max": 602.94970703125, "pos_frac": 0.859375, "sample": [320.6451416015625, 118.81405639648438, 114.67517852783203, 105.21121978759766, -120.67823028564453, 184.48233032226562, 289.95361328125, -11.74945068359375, 1.4008941650390625, 60.92706298828125, -0.7472915649414062, 247.72601318359375, 155.23583984375, -2.21160888671875, 147.27520751953125, 187.23240661621094, 386.856201171875, 55.69757080078125, 107.53384399414062, 57.519962310791016, 36.11445617675781, 46.55665588378906, -43.398101806640625, 120.86569213867188, 100.28201293945312, -171.3162841796875, 102.59696960449219, 165.30270385742188, 566.439453125, 506.3708801269531, 19.920499801635742, 213.39967346191406, 225.4571075439453, 455.3208312988281, 180.26156616210938, 18.74955940246582, 196.24856567382812, 119.66085815429688, 132.9705810546875, 67.6089096069336, 28.765716552734375, 173.89450073242188, 144.1701202392578, 183.81773376464844, 29.644790649414062, 237.94435119628906, 10.020362854003906, -24.059907913208008, 201.98562622070312, -48.234134674072266, 420.5567626953125, 282.99176025390625, 93.3172836303711, 332.0346374511719, 40.92155075073242, 368.98516845703125, 408.55731201171875, 179.0566864013672, 9.32131576538086, 26.35429573059082, 35.26640701293945, 602.94970703125, 214.804443359375, -23.66611099243164], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000290.npy"} +{"epoch": 0.42584434654919234, "step": 291, "batch_size": 64, "mean": 135.4893798828125, "std": 131.27963256835938, "min": -192.06533813476562, "p10": -12.668922424316406, "median": 130.88046264648438, "p90": 285.7816101074219, "max": 451.8714599609375, "pos_frac": 0.859375, "sample": [71.11045837402344, 145.28610229492188, 51.79939270019531, 42.20372772216797, -13.034652709960938, 225.67735290527344, 451.8714599609375, 138.91622924804688, 169.4851837158203, 256.3176574707031, 15.7197265625, 280.3266906738281, 218.78643798828125, 79.8803482055664, 389.4775695800781, 5.9585418701171875, 303.09716796875, 161.7953338623047, 18.53175926208496, 123.21925354003906, 277.8267822265625, 36.60688781738281, 214.95046997070312, 47.711517333984375, 217.52906799316406, 245.6002197265625, 129.612548828125, 265.78289794921875, 284.5450744628906, 13.212947845458984, 123.5222396850586, 135.38296508789062, 222.12998962402344, -20.585647583007812, 80.0845947265625, -117.74142456054688, -15.943351745605469, 104.6338119506836, -192.06533813476562, 271.0829162597656, 190.87130737304688, 280.2076110839844, 247.62075805664062, 170.4078369140625, -87.56092834472656, 63.696441650390625, -19.77864646911621, 240.2100067138672, 7.521146774291992, 105.97838592529297, 9.051040649414062, 58.552001953125, 294.9705810546875, 132.14837646484375, -11.8155517578125, 156.12753295898438, 286.3115539550781, 83.86569213867188, 392.910400390625, 36.844444274902344, -6.165504455566406, 249.56884765625, 324.6945495605469, 4.777551651000977], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000291.npy"} +{"epoch": 0.42731277533039647, "step": 292, "batch_size": 64, "mean": 135.6851806640625, "std": 150.32594299316406, "min": -361.3814697265625, "p10": -31.903346252441388, "median": 142.29865264892578, "p90": 340.7985168457032, "max": 455.63311767578125, "pos_frac": 0.859375, "sample": [161.1265869140625, 119.16969299316406, 27.225082397460938, 144.43353271484375, 136.12533569335938, 145.67926025390625, 109.13238525390625, 72.7295150756836, 195.15997314453125, -361.3814697265625, 438.7284851074219, 56.05425262451172, 369.1202392578125, 89.07492065429688, 177.1126708984375, -6.11693000793457, -137.353759765625, 348.3276672363281, 51.357948303222656, 213.96743774414062, -39.41692352294922, 455.63311767578125, 216.40530395507812, 152.5863037109375, 248.19747924804688, 374.2845153808594, 213.44491577148438, 317.663330078125, 63.330284118652344, 71.3692626953125, -88.81854248046875, 184.37030029296875, 6.640232086181641, 210.164794921875, -14.371665954589844, 109.04132843017578, -72.36962890625, 178.29879760742188, 140.1637725830078, 65.62278747558594, 37.13983154296875, 278.5160217285156, 168.30517578125, 163.06103515625, 158.12860107421875, 204.85397338867188, 260.3267517089844, 217.80001831054688, 443.74127197265625, 62.150299072265625, 84.25846099853516, 109.93115234375, 242.35289001464844, -114.15303039550781, 148.2276611328125, -85.51836395263672, 427.5706787109375, 20.319355010986328, 15.41910171508789, 323.2304992675781, 207.4772491455078, 95.26970672607422, 66.57966613769531, 6.95057487487793], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000292.npy"} +{"epoch": 0.4287812041116006, "step": 293, "batch_size": 64, "mean": 111.73796844482422, "std": 144.4040069580078, "min": -167.55018615722656, "p10": -56.15836029052734, "median": 103.35594177246094, "p90": 294.74034423828124, "max": 464.1791687011719, "pos_frac": 0.734375, "sample": [27.90445327758789, 55.46885681152344, 42.25816345214844, 56.77238464355469, 202.85427856445312, 181.62461853027344, 363.1319274902344, 88.35169982910156, 292.75860595703125, 30.90598487854004, 260.1134033203125, -14.146743774414062, 159.09420776367188, -23.490108489990234, 464.1791687011719, -72.16482543945312, 128.20645141601562, 199.90493774414062, 54.281341552734375, 127.40906524658203, 224.69454956054688, -108.92230987548828, 173.07168579101562, -42.374122619628906, 20.34235191345215, 20.881444931030273, -37.510215759277344, 136.7876739501953, 63.55207824707031, 375.930908203125, 361.5057373046875, 222.20941162109375, 217.8566436767578, 222.4835205078125, 205.0723114013672, -56.58366394042969, 27.690216064453125, 133.41763305664062, 135.59088134765625, -24.58124351501465, -21.910682678222656, 44.96002960205078, -10.34210205078125, 126.53561401367188, 259.3797607421875, 251.009033203125, 150.18951416015625, 227.6902313232422, 118.36018371582031, -148.90090942382812, 59.88960266113281, -167.55018615722656, -55.165985107421875, 295.58966064453125, 414.17047119140625, 53.02716827392578, -54.568572998046875, 173.50767517089844, 52.70042037963867, -23.271469116210938, -58.78009796142578, 274.5147705078125, 370.2239990234375, -76.561767578125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000293.npy"} +{"epoch": 0.4302496328928047, "step": 294, "batch_size": 64, "mean": 122.24322509765625, "std": 149.4528350830078, "min": -139.44906616210938, "p10": -31.33457145690918, "median": 77.52316284179688, "p90": 311.9317138671875, "max": 567.962890625, "pos_frac": 0.859375, "sample": [51.227821350097656, 68.42079162597656, 210.38812255859375, 312.07684326171875, 80.2657470703125, 70.27470397949219, -31.238929748535156, 82.6550064086914, 5.892852783203125, 122.62439727783203, 6.370269775390625, 10.796417236328125, -31.375560760498047, 392.349609375, -16.315452575683594, 190.28659057617188, 150.44288635253906, 257.42822265625, 57.59144973754883, 110.34123229980469, 430.25396728515625, -104.71623229980469, 206.05438232421875, 74.42509460449219, 453.60693359375, 268.87261962890625, 231.21788024902344, 26.415658950805664, 122.7614517211914, -63.996360778808594, 26.170143127441406, 228.80712890625, 188.70223999023438, 567.962890625, -139.44906616210938, -116.20790100097656, 456.21484375, 102.66438293457031, 196.70486450195312, 46.064781188964844, -134.86776733398438, -66.90413665771484, 235.37139892578125, 239.5781707763672, 376.9932861328125, 108.99369812011719, 42.55679702758789, 74.78057861328125, 311.59307861328125, 46.57344055175781, 234.31515502929688, 217.33737182617188, 32.93193817138672, 123.73335266113281, 35.87861251831055, 11.43121337890625, 68.30479431152344, 90.9936294555664, 33.88655471801758, 51.54833984375, 259.89556884765625, 8.10955810546875, 47.535125732421875, 69.96338653564453], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000294.npy"} +{"epoch": 0.43171806167400884, "step": 295, "batch_size": 64, "mean": 125.54583740234375, "std": 135.47486877441406, "min": -137.9392547607422, "p10": -25.22760772705078, "median": 115.67920684814453, "p90": 293.6212829589844, "max": 533.6575927734375, "pos_frac": 0.859375, "sample": [-60.94233703613281, 65.24824523925781, 280.64825439453125, 0.8835010528564453, -110.79609680175781, 185.93402099609375, 19.243091583251953, 212.66018676757812, 177.0855712890625, -137.9392547607422, 207.5706787109375, 20.908065795898438, 104.87113189697266, 285.4483642578125, 272.5070495605469, 220.38668823242188, 178.19837951660156, 115.63258361816406, 17.437042236328125, 78.76309967041016, 243.04745483398438, 29.31308937072754, 55.758056640625, 165.44017028808594, -47.251060485839844, 58.056610107421875, 323.8202819824219, -50.66273880004883, 324.8109130859375, 115.725830078125, 176.02215576171875, 15.584304809570312, -22.15060806274414, -121.79977416992188, 257.268798828125, 413.10784912109375, 20.583961486816406, 50.31175994873047, 41.40653991699219, 117.13139343261719, 116.28146362304688, 177.7156982421875, 270.2791748046875, 14.507431030273438, -25.27050018310547, 9.693161010742188, 144.7416534423828, 79.92899322509766, 162.27993774414062, 383.4111022949219, 152.14178466796875, 181.10833740234375, 52.04510498046875, 533.6575927734375, 297.12396240234375, 269.7037658691406, 124.66889953613281, 151.5448455810547, 107.550537109375, 68.66886901855469, 29.41950035095215, 101.18724060058594, 358.3996887207031, -25.127525329589844], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000295.npy"} +{"epoch": 0.4331864904552129, "step": 296, "batch_size": 64, "mean": 95.79239654541016, "std": 147.8951873779297, "min": -250.43336486816406, "p10": -51.85444374084472, "median": 76.0951156616211, "p90": 262.79142150878914, "max": 481.37127685546875, "pos_frac": 0.734375, "sample": [33.099151611328125, 78.21012115478516, -15.11456298828125, 466.90557861328125, -84.12860107421875, 162.04278564453125, -53.229393005371094, -15.856765747070312, -22.042274475097656, -191.39395141601562, 150.37423706054688, 67.06145477294922, -207.46995544433594, 209.66058349609375, 125.70234680175781, -107.74185180664062, 60.57072830200195, 144.20811462402344, 417.3915100097656, 272.79693603515625, 154.50704956054688, 134.26025390625, -43.08820343017578, 5.147857666015625, 50.3299560546875, 215.19406127929688, 12.621841430664062, -43.08668518066406, -45.252227783203125, -48.6462287902832, 332.86834716796875, 69.98426818847656, 168.88063049316406, 117.69377899169922, 67.65719604492188, 363.8929138183594, 177.7084503173828, 37.72164535522461, -250.43336486816406, 223.711181640625, -80.62582397460938, 237.24734497070312, -40.69800567626953, -14.686067581176758, 295.7520751953125, 122.0877456665039, 73.98011016845703, 161.29490661621094, 83.21622467041016, 65.35022735595703, 72.53852081298828, 21.744117736816406, 15.810615539550781, 481.37127685546875, 213.2762908935547, 232.53707885742188, 112.55615997314453, 153.56784057617188, 239.44522094726562, -44.112335205078125, 131.33140563964844, 161.74951171875, 26.368453979492188, 216.89163208007812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000296.npy"} +{"epoch": 0.434654919236417, "step": 297, "batch_size": 64, "mean": 110.9755859375, "std": 113.45466613769531, "min": -166.47677612304688, "p10": -9.415709877014159, "median": 97.63640213012695, "p90": 275.1850799560547, "max": 408.07452392578125, "pos_frac": 0.875, "sample": [38.249603271484375, 69.95075988769531, 250.31845092773438, 328.76275634765625, 86.66609191894531, -10.124792098999023, 280.3121337890625, 125.8074951171875, -35.913047790527344, -45.988895416259766, 95.85891723632812, 130.91018676757812, 190.19583129882812, 149.2279815673828, 291.1443176269531, 3.921102523803711, 108.7803955078125, 52.82073974609375, 66.69398498535156, -7.7611846923828125, -13.776290893554688, -51.31541442871094, 4.642499923706055, 49.84553527832031, 408.07452392578125, 17.180419921875, 277.3721923828125, 57.605682373046875, 179.21240234375, 144.62118530273438, 99.41388702392578, 109.21347045898438, 232.3152313232422, 135.7663116455078, 161.5984344482422, 69.36553955078125, 120.13206481933594, 78.79865264892578, 319.8489990234375, 131.69931030273438, 67.41578674316406, 181.0059051513672, -130.41033935546875, 131.59400939941406, 34.061439514160156, 45.09033966064453, 143.04095458984375, 59.606689453125, 402.3577880859375, 38.10942077636719, 92.03909301757812, 270.0818176269531, 123.5863037109375, 8.18170166015625, -166.47677612304688, 110.90410614013672, 30.534523010253906, 43.00331115722656, 182.619873046875, 254.20223999023438, 68.95358276367188, 209.95176696777344, 77.42305755615234, 124.11320495605469], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000297.npy"} +{"epoch": 0.43612334801762115, "step": 298, "batch_size": 64, "mean": 138.00003051757812, "std": 158.82704162597656, "min": -105.29180908203125, "p10": -24.55209426879882, "median": 94.76099395751953, "p90": 375.77694396972674, "max": 745.3963012695312, "pos_frac": 0.8125, "sample": [92.8957290649414, -35.56990051269531, 38.94863510131836, -63.57453918457031, 63.885772705078125, -57.88885498046875, -6.45789909362793, 426.4468688964844, 54.45177459716797, 62.030426025390625, 72.0073471069336, 294.26971435546875, 27.603866577148438, 9.895919799804688, -19.00738525390625, 46.83318328857422, 96.62625885009766, 128.48788452148438, 100.76953887939453, 142.2190399169922, 422.9107666015625, -9.656265258789062, 195.9446563720703, -105.29180908203125, 333.18121337890625, 199.19281005859375, 11.198654174804688, 268.6704406738281, 405.3652648925781, 394.0322570800781, -29.471094131469727, -64.10003662109375, -26.92839813232422, 4.107044219970703, 19.335182189941406, 404.79083251953125, 440.526123046875, 167.42369079589844, 263.32586669921875, 241.18569946289062, 52.98976135253906, -7.538604736328125, 745.3963012695312, 145.985595703125, 305.842041015625, 103.75454711914062, 186.56935119628906, 283.8857421875, 86.41680908203125, 62.615455627441406, 0.142486572265625, 250.930419921875, 220.55133056640625, 84.45964813232422, 277.64312744140625, 164.8684539794922, 109.2080078125, 229.320556640625, -3.7928619384765625, 97.36563110351562, 56.196441650390625, 77.48817443847656, 44.864166259765625, 246.22299194335938], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000298.npy"} +{"epoch": 0.43759177679882527, "step": 299, "batch_size": 64, "mean": 130.44891357421875, "std": 156.0242462158203, "min": -187.98361206054688, "p10": -51.84148178100586, "median": 116.50736999511719, "p90": 390.3531341552735, "max": 524.5721435546875, "pos_frac": 0.828125, "sample": [-51.282264709472656, -187.98361206054688, -74.10589599609375, 181.72235107421875, -79.63052368164062, -82.34634399414062, 256.1409912109375, 69.53927612304688, 46.464813232421875, 38.11872100830078, 412.90753173828125, 396.01434326171875, 500.87921142578125, 48.80896759033203, 206.0092010498047, 461.3892822265625, -158.94378662109375, 240.90374755859375, 135.88766479492188, 124.43797302246094, 18.646745681762695, 215.41973876953125, 55.06427764892578, 72.51104736328125, 124.62850189208984, 64.75273132324219, 524.5721435546875, 59.59328079223633, 50.24121856689453, 4.485652923583984, -18.3594913482666, 100.76786804199219, 132.42483520507812, 155.18955993652344, 201.2753448486328, 418.0486145019531, 61.14619445800781, 241.94911193847656, 49.050384521484375, 125.31214141845703, 92.11742401123047, 111.90390014648438, 91.95591735839844, -25.96664810180664, 202.85269165039062, 95.59215545654297, -128.3832550048828, 336.5236511230469, 169.43545532226562, 377.1436462402344, -34.764747619628906, 36.06427764892578, 146.70492553710938, 152.15936279296875, 139.37295532226562, 180.10581970214844, 129.77484130859375, 411.08978271484375, 121.11083984375, 105.71351623535156, -52.081146240234375, 282.9137878417969, 41.886314392089844, 223.85302734375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000299.npy"} +{"epoch": 0.4390602055800294, "step": 300, "batch_size": 64, "mean": 142.15560913085938, "std": 165.4517364501953, "min": -216.46128845214844, "p10": -102.5058006286621, "median": 158.32064819335938, "p90": 329.88146667480476, "max": 552.023193359375, "pos_frac": 0.8125, "sample": [111.3048095703125, 198.71197509765625, 170.38917541503906, 226.4800262451172, 190.2370147705078, -161.78073120117188, 83.77729034423828, 302.9599304199219, 292.61822509765625, 0.5388298034667969, -105.07886505126953, 101.6015853881836, 158.8015899658203, 249.39376831054688, 89.7825698852539, 234.49407958984375, 98.04070281982422, 338.20703125, 20.012929916381836, 70.40030670166016, -15.810821533203125, 191.14407348632812, 160.7362060546875, -15.465354919433594, 219.12979125976562, 314.662841796875, -124.75308227539062, 44.41349792480469, 169.1154327392578, 157.83970642089844, 94.91865539550781, 391.23577880859375, 178.67242431640625, -162.41915893554688, 284.7515869140625, -192.38888549804688, 334.12994384765625, 122.62557220458984, 315.7005615234375, 196.61788940429688, 101.5436782836914, 222.69065856933594, 367.2196044921875, 171.0147705078125, 57.395145416259766, 222.0236358642578, 62.92672348022461, -96.50198364257812, 514.7744140625, 319.9683532714844, 317.69683837890625, -216.46128845214844, 148.50091552734375, -124.83980560302734, 7.800586700439453, 137.5721435546875, 276.18804931640625, 46.827880859375, 431.4758605957031, 552.023193359375, -34.5572509765625, 96.98036193847656, 196.92462158203125, -16.976669311523438], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000300.npy"} +{"epoch": 0.44052863436123346, "step": 301, "batch_size": 64, "mean": 141.4380645751953, "std": 138.804931640625, "min": -104.82315063476562, "p10": 1.5010158538818361, "median": 113.03486633300781, "p90": 321.29707336425787, "max": 546.7921142578125, "pos_frac": 0.90625, "sample": [178.7928009033203, 220.30844116210938, 6.599693298339844, 293.0602722167969, 193.98843383789062, 203.57168579101562, 423.7649841308594, 61.96641540527344, 267.902099609375, -55.831817626953125, 367.26239013671875, 303.2048034667969, 71.76802825927734, 184.7362823486328, -30.21832275390625, 119.984130859375, 69.11505889892578, 301.99517822265625, 5.2133941650390625, 215.1345672607422, 476.2923278808594, 22.39773941040039, -22.47818946838379, 16.368728637695312, 206.16903686523438, 109.90924072265625, 40.59038543701172, 25.331100463867188, 198.839599609375, 205.82220458984375, 254.84803771972656, 103.01182556152344, 62.021663665771484, 211.798828125, 48.577415466308594, 95.44375610351562, 131.97767639160156, 38.48533630371094, 8.322153091430664, 13.328765869140625, 108.15936279296875, 46.918190002441406, 67.35272979736328, 56.00426483154297, 10.736801147460938, 546.7921142578125, 198.0044403076172, 116.16049194335938, 329.0509033203125, 133.2589111328125, 371.83197021484375, 152.913818359375, 237.60964965820312, 250.33401489257812, 247.48956298828125, 81.23025512695312, -104.82315063476562, -20.288467407226562, 395.02618408203125, -70.25967407226562, 41.07667541503906, 1.730316162109375, 1.4027442932128906, 204.94778442382812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000301.npy"} +{"epoch": 0.4419970631424376, "step": 302, "batch_size": 64, "mean": 125.11246490478516, "std": 136.68569946289062, "min": -125.58938598632812, "p10": -18.3678840637207, "median": 111.99885940551758, "p90": 312.2091949462892, "max": 560.7171020507812, "pos_frac": 0.859375, "sample": [133.7847442626953, 254.16513061523438, 377.70440673828125, 3.6714515686035156, 111.62841796875, 61.22430419921875, 228.3883056640625, 2.3756256103515625, 101.26399230957031, 42.5701904296875, 197.42147827148438, 43.150726318359375, 77.74932098388672, 36.092796325683594, 47.277099609375, -19.025596618652344, 238.66542053222656, -1.3187828063964844, 265.6361083984375, 210.26272583007812, 138.73085021972656, 115.72530364990234, 178.947265625, 149.48666381835938, -73.37406921386719, -97.93096160888672, 217.555908203125, 407.5410461425781, 79.98554229736328, -125.58938598632812, 125.35503387451172, 34.945823669433594, 180.2404327392578, -44.11015319824219, 4.874187469482422, 400.6639099121094, 560.7171020507812, 210.32350158691406, 115.6911392211914, -16.833221435546875, 27.893035888671875, 288.13330078125, -124.71392059326172, 204.7373046875, 195.4741668701172, 112.36930084228516, 16.647613525390625, 345.00616455078125, 127.73380279541016, 175.77645874023438, 30.03043556213379, 135.05397033691406, -98.0031509399414, 172.3406219482422, 63.617488861083984, 144.79522705078125, 84.66188049316406, 322.5274353027344, 40.11260986328125, 397.4295654296875, 96.26797485351562, 85.28607177734375, 109.23380279541016, 79.15303802490234], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000302.npy"} +{"epoch": 0.4434654919236417, "step": 303, "batch_size": 64, "mean": 119.92936706542969, "std": 155.2955780029297, "min": -162.84591674804688, "p10": -59.054443740844704, "median": 89.61055374145508, "p90": 345.711605834961, "max": 558.6915283203125, "pos_frac": 0.75, "sample": [143.439208984375, 7.928981781005859, -7.343727111816406, 434.7447509765625, -115.45713806152344, 244.57858276367188, 295.6373596191406, 415.13385009765625, -1.6075363159179688, -34.569244384765625, 255.8482666015625, 121.08045196533203, 269.597900390625, -70.89511108398438, 32.43915557861328, 79.29946899414062, 33.24339294433594, 76.55465698242188, -29.43306541442871, 291.6291809082031, -162.84591674804688, 4.441135406494141, -3.0230026245117188, 197.743408203125, 275.0347595214844, 390.44635009765625, 3.673666000366211, 206.02210998535156, -94.64527130126953, 120.00151062011719, -21.07012939453125, 204.57904052734375, 380.96893310546875, 146.87033081054688, 335.85040283203125, 78.56280517578125, -106.73206329345703, 31.138580322265625, 116.83121490478516, 558.6915283203125, 369.7622985839844, 7.740478515625, 131.81011962890625, 349.9378356933594, 52.71112060546875, 79.650634765625, 158.59225463867188, 271.92950439453125, 93.73616790771484, -72.9183120727539, 130.26004028320312, 40.169944763183594, 141.4442901611328, 33.91267776489258, 125.2864990234375, 323.4427185058594, -37.746150970458984, -68.18656921386719, 85.48493957519531, -26.199825286865234, 143.31544494628906, 35.991058349609375, -2.9490432739257812, 203.91265869140625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000303.npy"} +{"epoch": 0.44493392070484583, "step": 304, "batch_size": 64, "mean": 129.28297424316406, "std": 109.99324035644531, "min": -110.18345642089844, "p10": -11.120149230957022, "median": 124.79442596435547, "p90": 276.5036712646485, "max": 395.4195251464844, "pos_frac": 0.875, "sample": [14.7164306640625, 132.1936798095703, 200.9583282470703, 42.21721649169922, -1.6975250244140625, -82.58680725097656, -15.158416748046875, 51.30433654785156, 132.58544921875, 154.58856201171875, 174.26220703125, 232.4741668701172, 93.26689147949219, 248.60401916503906, 111.16966247558594, 232.66555786132812, 186.29025268554688, 177.57431030273438, 92.60936737060547, 107.01873779296875, 138.67276000976562, 285.65179443359375, 52.78993225097656, 42.9923095703125, 210.50714111328125, 109.27508544921875, 207.97877502441406, 49.34576416015625, 238.58099365234375, 101.99868774414062, 96.73867797851562, 95.11862182617188, 315.99554443359375, 104.56131744384766, -32.960365295410156, -110.18345642089844, 117.39517211914062, 41.66253662109375, 99.84483337402344, 167.61502075195312, -67.00930786132812, 291.0166015625, -77.18382263183594, 18.01405906677246, 288.3682556152344, 80.43473815917969, 6.734535217285156, 65.6813735961914, 218.61578369140625, 143.94166564941406, 220.05996704101562, -87.81646728515625, 108.24568176269531, 395.4195251464844, 301.088623046875, 54.67924499511719, 244.58111572265625, 132.44515991210938, 299.02496337890625, 255.15805053710938, 214.63504028320312, 179.61871337890625, 174.71478271484375, 195.00405883789062], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000304.npy"} +{"epoch": 0.44640234948604995, "step": 305, "batch_size": 64, "mean": 127.37362670898438, "std": 151.46485900878906, "min": -127.67079162597656, "p10": -39.880654907226564, "median": 92.02851867675781, "p90": 315.88135681152346, "max": 734.301025390625, "pos_frac": 0.8125, "sample": [251.76873779296875, 187.83633422851562, 81.47380065917969, 139.37210083007812, 99.23246765136719, 424.802734375, 195.16726684570312, 74.98202514648438, 176.90370178222656, 344.5896301269531, 92.81703186035156, 53.32195281982422, 97.7524642944336, -43.89036560058594, -9.97271728515625, 134.49423217773438, 188.111083984375, 315.4024353027344, 75.21021270751953, 3.2787704467773438, 734.301025390625, 177.18527221679688, 60.4484977722168, 170.03004455566406, 245.675048828125, 55.98974609375, 84.26297760009766, 169.69151306152344, 213.31944274902344, -28.231048583984375, -5.482490539550781, 246.4104461669922, -45.45314025878906, 199.57797241210938, 185.65814208984375, 41.46562194824219, 447.9143981933594, 153.89447021484375, 47.760345458984375, 64.20418548583984, 82.70121002197266, -6.323112487792969, 91.24000549316406, 88.5428695678711, 316.08660888671875, 362.2887268066406, 165.12059020996094, -39.413909912109375, -127.67079162597656, 143.70126342773438, -40.0806884765625, 427.79931640625, 287.2510986328125, 28.05889129638672, -78.44964599609375, 15.469635009765625, -69.26311492919922, 13.445976257324219, 207.83059692382812, -103.11841583251953, 20.805397033691406, 204.55267333984375, 47.19035339355469, 12.870283126831055], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000305.npy"} +{"epoch": 0.447870778267254, "step": 306, "batch_size": 64, "mean": 121.53746795654297, "std": 128.5977783203125, "min": -211.213134765625, "p10": -3.8015148162841785, "median": 111.6677360534668, "p90": 264.7770660400391, "max": 536.3804931640625, "pos_frac": 0.84375, "sample": [42.791263580322266, -2.6791954040527344, 130.902099609375, 32.54969024658203, -48.556427001953125, 33.39482116699219, 251.30691528320312, 93.55799102783203, 63.573333740234375, -10.035865783691406, 116.46458435058594, 261.6575622558594, -1.07989501953125, 210.52870178222656, 82.21392822265625, 104.9114761352539, 19.61355972290039, 276.4004821777344, 116.45955657958984, 46.67057800292969, 26.159996032714844, 39.30341339111328, -19.517181396484375, 245.68539428710938, 174.16748046875, 193.311279296875, 7.862636566162109, -1.8793106079101562, 263.82501220703125, 150.5655059814453, 14.891693115234375, 24.907142639160156, 236.09036254882812, 241.15069580078125, 59.64756774902344, 188.5644073486328, 106.87591552734375, -64.83563995361328, 69.66897583007812, 144.16929626464844, -211.213134765625, 296.85205078125, 128.41363525390625, -58.33580017089844, 19.064210891723633, 217.76080322265625, -4.282508850097656, 183.620361328125, 450.8052062988281, 387.15924072265625, 219.88348388671875, 270.73321533203125, 125.17497253417969, 17.68294334411621, 251.8544921875, 37.014949798583984, 78.25056457519531, 34.20574188232422, 163.18569946289062, 128.620361328125, 536.3804931640625, 146.63809204101562, 172.4843292236328, 265.1850891113281], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000306.npy"} +{"epoch": 0.44933920704845814, "step": 307, "batch_size": 64, "mean": 120.9974594116211, "std": 132.8800506591797, "min": -246.38742065429688, "p10": -30.705348968505856, "median": 116.49665832519531, "p90": 316.55716552734384, "max": 434.53875732421875, "pos_frac": 0.84375, "sample": [250.8319091796875, 125.3322982788086, 43.07061004638672, 118.17259216308594, 145.16683959960938, 114.82072448730469, -46.856170654296875, 66.45122528076172, -24.607742309570312, 392.6720275878906, 60.59539794921875, 22.881752014160156, 121.72581481933594, -66.55632019042969, 288.3162841796875, -246.38742065429688, 359.31549072265625, -110.22959899902344, 160.80587768554688, -32.114402770996094, 150.85250854492188, 139.96322631835938, 91.3000717163086, 141.66233825683594, 393.82080078125, 148.0667266845703, 96.11033630371094, 62.21214294433594, 123.94732666015625, 325.37164306640625, 75.40016174316406, 330.76239013671875, 48.656959533691406, 108.82684326171875, 68.85133361816406, 47.840606689453125, 144.166015625, 82.99934387207031, 276.7235107421875, 285.43206787109375, 54.526588439941406, 210.97576904296875, -21.62274932861328, 215.1883544921875, 207.96302795410156, 38.37678527832031, 128.99566650390625, 203.43980407714844, 23.718271255493164, 94.78048706054688, 157.20030212402344, 5.171131134033203, 295.99005126953125, 326.7428283691406, 214.1025390625, 434.53875732421875, 15.491813659667969, -33.46426773071289, 180.78237915039062, 67.73138427734375, 37.49656677246094, 147.31893920898438, -27.417556762695312, -120.56246948242188], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000307.npy"} +{"epoch": 0.45080763582966227, "step": 308, "batch_size": 64, "mean": 133.1981964111328, "std": 129.76417541503906, "min": -139.01321411132812, "p10": -25.79909229278564, "median": 134.33334350585938, "p90": 324.58787231445314, "max": 454.6454772949219, "pos_frac": 0.828125, "sample": [174.43280029296875, 325.138671875, -113.71240234375, 211.55209350585938, 18.26415252685547, 330.6251220703125, -139.01321411132812, 58.734439849853516, 84.91354370117188, 120.10347747802734, 323.30267333984375, 29.620929718017578, 141.6354217529297, 83.890380859375, 172.778564453125, 338.4798889160156, -116.62144470214844, 160.2406005859375, 117.76399993896484, -9.111305236816406, 145.9700164794922, 358.8515625, 304.45269775390625, 329.29119873046875, 12.330780029296875, 166.0373992919922, 102.87045288085938, 60.534027099609375, 142.31997680664062, 32.53096008300781, 168.44863891601562, 119.80359649658203, 121.22492218017578, 200.8203582763672, 127.03126525878906, 289.0256652832031, 314.2609558105469, 82.94610595703125, -27.354827880859375, 56.735740661621094, 29.078824996948242, 216.3667449951172, 168.99176025390625, 145.46417236328125, 215.54315185546875, 454.6454772949219, 144.236572265625, 101.57420349121094, 34.494239807128906, 195.71087646484375, 389.9372863769531, 210.90960693359375, 85.0951156616211, -39.26383972167969, -22.169042587280273, -4.844118118286133, 232.50384521484375, 246.66854858398438, 156.86830139160156, -38.85045623779297, 207.09780883789062, 27.357627868652344, -42.833431243896484, -11.048805236816406], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000308.npy"} +{"epoch": 0.4522760646108664, "step": 309, "batch_size": 64, "mean": 133.6636199951172, "std": 134.25613403320312, "min": -157.050048828125, "p10": -11.74795932769775, "median": 113.02368927001953, "p90": 310.0714874267578, "max": 496.2375183105469, "pos_frac": 0.84375, "sample": [49.577762603759766, 51.4854736328125, 52.039146423339844, -67.06063842773438, 140.77359008789062, 160.71401977539062, 76.8945541381836, -7.668088912963867, 369.8555908203125, 56.487579345703125, 234.0308837890625, 288.408447265625, 114.48915100097656, 386.1278381347656, 79.6282730102539, 48.03681182861328, -4.880191802978516, 254.3258056640625, 241.72500610351562, 202.51144409179688, 191.90309143066406, -157.050048828125, 203.1641387939453, -4.3909454345703125, 89.64193725585938, 77.03887176513672, 236.16128540039062, 107.91886901855469, -14.019790649414062, 39.610504150390625, -18.14624786376953, 374.369140625, 26.0743408203125, 21.495498657226562, 142.50022888183594, 327.16900634765625, 496.2375183105469, 296.6434631347656, 93.2897720336914, -13.496475219726562, 10.528217315673828, 78.22602844238281, 74.79652404785156, 295.6533508300781, 111.5582275390625, 161.89256286621094, 329.46441650390625, 168.7532501220703, 198.47349548339844, 37.220638275146484, 312.5957336425781, -144.65682983398438, 147.29791259765625, 199.96478271484375, -67.55782318115234, 50.23311233520508, 304.18157958984375, 299.7872009277344, 2.9618072509765625, 124.4994888305664, 140.56861877441406, 66.32102966308594, 286.6786804199219, 121.41334533691406], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000309.npy"} +{"epoch": 0.45374449339207046, "step": 310, "batch_size": 64, "mean": 126.892333984375, "std": 106.34928894042969, "min": -52.14994430541992, "p10": 0.2775733947753962, "median": 127.24983215332031, "p90": 284.77918395996096, "max": 379.4024658203125, "pos_frac": 0.890625, "sample": [-52.14994430541992, 84.80928802490234, 196.28094482421875, -2.1758956909179688, 87.57984924316406, 231.99595642089844, 112.53340148925781, -15.130208969116211, 32.77323913574219, 156.48025512695312, -47.502662658691406, 124.72158813476562, 224.13043212890625, 7.870967864990234, 29.42429542541504, 81.2856674194336, 138.09681701660156, 143.47962951660156, 119.96005249023438, 235.55905151367188, 95.89118957519531, 45.88261413574219, 208.56646728515625, 42.09004211425781, 54.873321533203125, 299.39056396484375, 192.0344696044922, 63.28899002075195, 36.4473876953125, 130.5989990234375, 6.0023345947265625, 344.94525146484375, 46.575740814208984, 91.97122192382812, 379.4024658203125, 153.33444213867188, 216.65399169921875, 26.11729621887207, 40.54833221435547, 9.282722473144531, 40.88531494140625, 142.13330078125, 265.95892333984375, 195.09263610839844, 148.52102661132812, 194.9349822998047, 355.38665771484375, -11.184974670410156, 319.57275390625, 129.778076171875, 157.5858154296875, -50.78742218017578, 139.686767578125, 132.95347595214844, -32.62327575683594, 42.372703552246094, 107.3265380859375, 284.08856201171875, 285.0751647949219, 185.34786987304688, 65.66500854492188, 292.0507507324219, 201.22348022460938, 156.14437866210938], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000310.npy"} +{"epoch": 0.4552129221732746, "step": 311, "batch_size": 64, "mean": 149.58926391601562, "std": 158.60379028320312, "min": -86.05442810058594, "p10": -28.57717323303222, "median": 135.05407333374023, "p90": 354.3935241699219, "max": 594.7139892578125, "pos_frac": 0.796875, "sample": [260.3103332519531, 284.4532775878906, 273.33978271484375, 31.704200744628906, -76.27025604248047, 285.8856201171875, -20.60458755493164, 218.7917022705078, 203.65899658203125, 351.1319274902344, 180.4974365234375, 47.610443115234375, 266.1745910644531, 496.12994384765625, 143.23233032226562, 123.5743179321289, 245.54248046875, 84.53374481201172, 7.253929138183594, 382.7098083496094, 176.34422302246094, 192.06088256835938, 355.7913513183594, 164.7343292236328, 77.33324432373047, 167.8497314453125, 35.30952453613281, -65.14273071289062, 113.7734146118164, 89.65355682373047, -68.75521850585938, 162.09030151367188, 3.4930191040039062, 232.1753387451172, 268.94927978515625, -43.762481689453125, 3.7059249877929688, 594.7139892578125, -17.646575927734375, 190.68429565429688, 88.05402374267578, 22.699073791503906, 198.87619018554688, 42.33479309082031, 437.25543212890625, 90.55868530273438, 5.104022979736328, -3.874021530151367, 96.00650024414062, 91.64434814453125, -2.983001708984375, 490.1893310546875, 171.91769409179688, -31.993995666503906, -86.05442810058594, 324.0045166015625, 165.0482635498047, 126.87581634521484, -9.529182434082031, 153.63902282714844, 326.071044921875, -45.57773971557617, -7.681205749511719, 508.112060546875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000311.npy"} +{"epoch": 0.4566813509544787, "step": 312, "batch_size": 64, "mean": 132.64505004882812, "std": 141.73883056640625, "min": -162.75473022460938, "p10": -28.58611507415771, "median": 118.48138046264648, "p90": 317.30362243652354, "max": 562.7853393554688, "pos_frac": 0.84375, "sample": [75.52412414550781, 562.7853393554688, 357.0321044921875, 116.39826965332031, 76.31629943847656, -31.22801971435547, 75.41704559326172, 65.17323303222656, 120.56449127197266, 290.19207763671875, -162.75473022460938, 343.8481750488281, 471.47216796875, 172.2914276123047, 169.9945068359375, 161.23080444335938, -141.7806396484375, 102.15536499023438, 261.98614501953125, 177.47018432617188, 292.841796875, 16.837017059326172, 89.51808166503906, 156.15695190429688, 129.284423828125, 5.817842483520508, 46.39067840576172, 387.0964660644531, 234.9901123046875, 174.10299682617188, 165.99278259277344, 163.7237548828125, 140.397216796875, 169.9174346923828, -62.89019775390625, 108.10902404785156, 182.17530822753906, 327.7872619628906, 52.36442565917969, 102.77032470703125, 236.99285888671875, 100.91678619384766, -22.42167091369629, 202.09776306152344, 173.08920288085938, 100.50775146484375, -34.70552062988281, 475.9284973144531, 52.64277648925781, 77.29730987548828, 179.31307983398438, 113.65440368652344, -116.47335052490234, 75.26524353027344, -7.744010925292969, 155.6606903076172, 159.33021545410156, -7.7761688232421875, -127.97222137451172, 197.84625244140625, 202.63461303710938, 11.99737548828125, 65.43760681152344, 78.29232788085938], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000312.npy"} +{"epoch": 0.4581497797356828, "step": 313, "batch_size": 64, "mean": 126.86814880371094, "std": 156.14932250976562, "min": -236.06057739257812, "p10": -76.88259506225585, "median": 106.62908172607422, "p90": 326.2564361572266, "max": 590.7507934570312, "pos_frac": 0.828125, "sample": [96.30810546875, 162.49266052246094, 47.464385986328125, 266.05303955078125, -81.08279418945312, 110.30979919433594, 332.74774169921875, 168.18881225585938, 272.93157958984375, 42.0634651184082, 66.60292053222656, -89.32591247558594, 120.837646484375, 87.64497375488281, 15.278322219848633, 45.201690673828125, 180.05905151367188, 169.16143798828125, 97.10811614990234, 80.19989776611328, 53.77545928955078, 316.97698974609375, -184.29931640625, -99.47224426269531, 102.9483642578125, 341.1092529296875, 242.74757385253906, 65.30378723144531, 166.51089477539062, 422.1541748046875, 264.28192138671875, 137.34552001953125, -6.781339645385742, 50.87559509277344, -154.64120483398438, 47.4664306640625, -67.0821304321289, -236.06057739257812, 321.2141418457031, 200.3212127685547, -9.785850524902344, 154.6883544921875, 21.57992935180664, 126.48297119140625, 274.62554931640625, 136.9654083251953, 83.1227798461914, 238.9295654296875, 253.98072814941406, -102.67623901367188, 269.2318115234375, 83.12286376953125, 152.076904296875, 7.1986083984375, 61.494110107421875, 417.49591064453125, 413.0514831542969, 2.1829376220703125, -4.281623840332031, 49.08076477050781, 140.88299560546875, 256.0038757324219, 328.41741943359375, 590.7507934570312], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000313.npy"} +{"epoch": 0.45961820851688695, "step": 314, "batch_size": 64, "mean": 123.65789794921875, "std": 156.8173065185547, "min": -275.69696044921875, "p10": -50.911110305786124, "median": 94.42620086669922, "p90": 292.28956298828126, "max": 550.1939086914062, "pos_frac": 0.78125, "sample": [445.3841247558594, -190.20489501953125, 103.75362396240234, 12.071189880371094, 67.64730834960938, 252.15023803710938, 112.03106689453125, 190.42352294921875, 67.6932373046875, -44.45359802246094, 222.26962280273438, -13.711776733398438, 87.2453842163086, 159.3834991455078, 122.236328125, -275.69696044921875, -88.37608337402344, 89.29446411132812, 476.297607421875, 428.50189208984375, 19.85174560546875, -18.9407901763916, -23.186859130859375, 83.11225891113281, 5.512481689453125, -53.67861557006836, 9.121292114257812, -66.64297485351562, 46.877845764160156, 261.61651611328125, 276.4176025390625, 53.247802734375, 198.08233642578125, 246.96514892578125, 224.71170043945312, 84.80438232421875, -11.941734313964844, 178.31248474121094, 78.30804443359375, 406.43402099609375, 287.6009216308594, 304.98260498046875, 550.1939086914062, 275.11273193359375, -12.087162017822266, 175.35549926757812, -55.160972595214844, 277.2401123046875, -5.102752685546875, 142.622802734375, 207.52284240722656, 54.81524658203125, 215.31781005859375, 242.2794952392578, 294.2989807128906, 225.4359588623047, 99.55793762207031, 47.7022705078125, 147.67860412597656, 40.060081481933594, 10.497024536132812, 17.539569854736328, 208.45098876953125, -60.73295593261719], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000314.npy"} +{"epoch": 0.461086637298091, "step": 315, "batch_size": 64, "mean": 117.0391845703125, "std": 129.14952087402344, "min": -142.9514617919922, "p10": -18.448183059692372, "median": 95.1336669921875, "p90": 280.06748046875, "max": 528.7196655273438, "pos_frac": 0.859375, "sample": [279.58734130859375, 30.80649757385254, 151.4284210205078, 124.69711303710938, -22.777339935302734, 87.00282287597656, 62.40576171875, 108.00563049316406, 119.42384338378906, -7.583221435546875, 184.30250549316406, -29.204374313354492, -126.55946350097656, 153.35238647460938, 104.80657196044922, 321.681884765625, 166.34339904785156, 82.78495788574219, -78.6667251586914, 239.09088134765625, 50.07427978515625, 324.529052734375, 214.24278259277344, 41.24462890625, -32.154083251953125, 119.0097885131836, 280.27325439453125, 119.18787384033203, 311.9118347167969, 70.07447814941406, 346.5287170410156, 99.82695007324219, 35.17811584472656, 82.02262878417969, 181.51832580566406, 231.635009765625, 40.764068603515625, 42.780792236328125, 198.0107879638672, 25.728256225585938, 10.52365493774414, 95.62020874023438, 528.7196655273438, 56.178489685058594, 200.69435119628906, 66.04308319091797, -123.54192352294922, 37.112640380859375, -8.346817016601562, 219.62826538085938, 94.64712524414062, 152.2133331298828, 471.2906799316406, 82.85803985595703, 209.97581481933594, 8.871610641479492, 260.8669738769531, 61.032535552978516, 224.74740600585938, 102.40129089355469, -142.9514617919922, 34.28834533691406, 74.76316833496094, 39.554893493652344], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000315.npy"} +{"epoch": 0.46255506607929514, "step": 316, "batch_size": 64, "mean": 153.7897491455078, "std": 145.36952209472656, "min": -65.98208618164062, "p10": -12.866601562499998, "median": 133.0416259765625, "p90": 384.62697753906264, "max": 570.9765014648438, "pos_frac": 0.875, "sample": [180.85438537597656, 70.91206359863281, 341.2431640625, 69.58845520019531, 121.8658676147461, 472.9017333984375, 115.7131118774414, 425.887451171875, 245.04400634765625, 432.5001220703125, 280.659912109375, 193.45335388183594, 7.7555389404296875, -38.86306381225586, 19.33556365966797, 134.3029327392578, 119.09339904785156, 95.89363098144531, -17.471317291259766, 167.28038024902344, 62.625938415527344, 70.15341186523438, 453.0189208984375, 221.34642028808594, 183.7907257080078, 59.50872039794922, 97.40283203125, 398.47491455078125, 131.7803192138672, 352.31512451171875, 65.58483123779297, 69.26421356201172, 23.714584350585938, 179.00697326660156, 241.14767456054688, 288.95391845703125, 235.0703582763672, 11.895957946777344, 173.66058349609375, 4.5811614990234375, 427.4556884765625, 223.03753662109375, 19.751220703125, 278.8850402832031, 137.95431518554688, 102.4168930053711, -59.75433349609375, -53.80781936645508, 570.9765014648438, 176.99061584472656, 316.6466369628906, -65.98208618164062, 162.55499267578125, 158.40077209472656, 167.95513916015625, -13.523460388183594, -29.83391571044922, 42.63169860839844, -11.333930969238281, 86.2080307006836, 169.575927734375, 42.23352813720703, 73.05764770507812, 158.79891967773438], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000316.npy"} +{"epoch": 0.46402349486049926, "step": 317, "batch_size": 64, "mean": 166.0082244873047, "std": 133.22996520996094, "min": -105.93356323242188, "p10": 5.264222717285157, "median": 166.93192291259766, "p90": 329.43738403320316, "max": 486.88922119140625, "pos_frac": 0.90625, "sample": [221.68524169921875, 89.93183135986328, 19.622962951660156, 320.6334228515625, 389.88458251953125, 6.074462890625, 227.06588745117188, 228.8832550048828, -70.0133056640625, 333.21051025390625, 486.88922119140625, 115.13240051269531, 262.2445068359375, 97.5203857421875, 159.99313354492188, 254.8203125, 113.20027160644531, 40.66870880126953, 154.0819091796875, 257.19830322265625, 19.18133544921875, 226.6459503173828, 214.67019653320312, 346.5531005859375, 34.23112106323242, 385.4009704589844, -105.93356323242188, 64.53905487060547, 420.662841796875, 181.4296875, 290.65264892578125, 268.19061279296875, 120.94581604003906, 300.7834777832031, 252.8940887451172, 37.05455017089844, 164.9291534423828, 95.49388122558594, 248.33338928222656, 119.98837280273438, 67.658203125, 306.7643737792969, -2.7948360443115234, -25.796449661254883, 296.05523681640625, 223.77113342285156, 247.13919067382812, 114.2756118774414, 187.1387939453125, -71.9891586303711, 72.65525817871094, 182.75827026367188, 196.85804748535156, 67.34668731689453, -9.826820373535156, 126.06554412841797, 74.09355163574219, 16.53742218017578, 454.0076599121094, 221.96937561035156, 44.63715362548828, 168.9346923828125, 4.9169769287109375, 265.9752502441406], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000317.npy"} +{"epoch": 0.4654919236417034, "step": 318, "batch_size": 64, "mean": 151.59442138671875, "std": 134.11862182617188, "min": -190.64634704589844, "p10": 3.4206995010376087, "median": 127.80400466918945, "p90": 321.9849395751953, "max": 491.4912109375, "pos_frac": 0.890625, "sample": [52.28125762939453, 86.13813781738281, 491.4912109375, 90.13803100585938, 187.49020385742188, 128.72193908691406, 36.23645782470703, 186.44105529785156, 123.22542572021484, 462.51922607421875, 17.360429763793945, 104.7725830078125, 196.21246337890625, 216.5052947998047, 404.86859130859375, 282.23681640625, 36.755767822265625, 100.7330322265625, -1.3091659545898438, 14.457052230834961, 265.811279296875, 39.91436004638672, -38.3737907409668, 338.65594482421875, 440.2938537597656, 246.79750061035156, -17.50637435913086, 24.893848419189453, 229.06874084472656, 233.33575439453125, 48.887359619140625, 265.1949462890625, 98.42305755615234, 307.5000915527344, 119.30699157714844, 126.88607025146484, 315.8923034667969, 14.91778564453125, -27.162750244140625, 325.9231872558594, 324.5960693359375, 143.51486206054688, 81.06193542480469, 141.76055908203125, 219.12501525878906, 135.85299682617188, 135.38327026367188, 221.049072265625, 92.30070495605469, -43.550628662109375, -190.64634704589844, 113.76335906982422, 254.45925903320312, 200.09552001953125, 281.49005126953125, 279.9300537109375, 160.43801879882812, 44.049930572509766, 34.83258056640625, 106.02590942382812, 48.00000762939453, 263.6908264160156, -8.298759460449219, 87.1834945678711], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000318.npy"} +{"epoch": 0.4669603524229075, "step": 319, "batch_size": 64, "mean": 145.79965209960938, "std": 133.25189208984375, "min": -133.90475463867188, "p10": -31.097027015686017, "median": 148.2676544189453, "p90": 308.6495025634766, "max": 472.6349182128906, "pos_frac": 0.8125, "sample": [214.47433471679688, 129.42295837402344, 161.123291015625, 195.06312561035156, 143.1659698486328, 188.3446044921875, 206.27359008789062, 240.1822509765625, 217.4920654296875, 73.01378631591797, 37.01222229003906, -133.90475463867188, 220.88143920898438, 93.68223571777344, 109.0063247680664, 81.78318786621094, 57.90668487548828, 215.60227966308594, 266.4653015136719, 107.66950225830078, 85.37742614746094, 150.296875, 156.63323974609375, 212.34454345703125, 166.4589385986328, 11.33123779296875, 204.63284301757812, 472.6349182128906, 218.6500244140625, 266.1259765625, -62.035675048828125, 109.35481262207031, 71.56452941894531, 290.3638916015625, -64.52490234375, 152.73687744140625, -38.61627197265625, -46.237735748291016, -8.019134521484375, 257.0601806640625, 390.0625915527344, 357.5760192871094, 424.6333312988281, -13.552122116088867, 161.5843048095703, -3.11077880859375, 306.876708984375, 315.3470153808594, 226.94940185546875, 309.4092712402344, 465.98553466796875, -7.165327072143555, 71.78057861328125, 244.39797973632812, -64.46266174316406, 23.386802673339844, -58.51905059814453, -9.821741104125977, 120.41024780273438, 120.01165771484375, 74.78570556640625, 146.23843383789062, 178.17160034179688, 119.4100341796875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000319.npy"} +{"epoch": 0.4684287812041116, "step": 320, "batch_size": 64, "mean": 163.4992218017578, "std": 154.14967346191406, "min": -117.71685028076172, "p10": -15.138437080383294, "median": 162.95313262939453, "p90": 368.4453033447266, "max": 678.9862060546875, "pos_frac": 0.875, "sample": [33.702301025390625, 160.58816528320312, 459.1624450683594, 89.90263366699219, 259.80487060546875, 110.45142364501953, 210.67984008789062, -117.71685028076172, 509.4923095703125, 176.78379821777344, 293.6092529296875, 195.05532836914062, -17.78746795654297, 189.60113525390625, 181.69741821289062, 372.29583740234375, 270.7818603515625, 202.94839477539062, 181.8876953125, -48.71906280517578, 220.56988525390625, 229.27593994140625, 28.41330909729004, 80.70709991455078, 57.729393005371094, 110.6717758178711, 212.16090393066406, 196.99354553222656, -37.39881896972656, 326.1720886230469, -20.241439819335938, 216.05438232421875, 47.85173797607422, 213.19342041015625, 298.83636474609375, 123.89588928222656, 79.6346435546875, 111.562744140625, 678.9862060546875, 54.78411865234375, 359.4607238769531, 16.71472930908203, 114.96151733398438, 15.810806274414062, 392.52935791015625, 29.614593505859375, 188.7730712890625, 140.9365692138672, -8.957365036010742, 17.34161376953125, 171.31646728515625, 165.31809997558594, 156.91868591308594, 321.1943054199219, 295.4068603515625, 445.5914306640625, 425.87847900390625, 17.069456100463867, -37.322811126708984, 12.848426818847656, 22.4061279296875, 199.63755798339844, 145.7612762451172, -89.33502197265625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000320.npy"} +{"epoch": 0.4698972099853157, "step": 321, "batch_size": 64, "mean": 136.1912384033203, "std": 165.71177673339844, "min": -212.89735412597656, "p10": -41.76095809936522, "median": 132.3379898071289, "p90": 349.1216064453125, "max": 584.367919921875, "pos_frac": 0.8125, "sample": [427.57049560546875, 156.76950073242188, 204.4994354248047, 182.0330810546875, -13.245492935180664, 223.24859619140625, 288.2962951660156, 80.62397766113281, 84.50799560546875, 354.25665283203125, 266.76324462890625, 188.2438507080078, -17.329437255859375, -67.72871398925781, -149.4792022705078, 181.6839141845703, 311.4449462890625, 209.447021484375, 174.12716674804688, 221.2821807861328, 521.3652954101562, 73.4031753540039, -212.89735412597656, 172.7447509765625, 189.6283416748047, 234.94796752929688, 157.5762481689453, -180.1420135498047, -25.539203643798828, 94.68528747558594, 134.93067932128906, 409.543212890625, 15.575897216796875, 245.1227264404297, 103.78812408447266, 337.13983154296875, 38.61503601074219, -122.07366943359375, 584.367919921875, 129.74530029296875, 4.097747802734375, 15.209161758422852, 4.836496353149414, 102.116943359375, -14.171958923339844, 101.79652404785156, 18.9423828125, -48.713138580322266, 115.35305786132812, 212.27493286132812, 491.0916748046875, 424.3922424316406, 16.91738510131836, -145.88525390625, 100.7905502319336, 168.97137451171875, 150.1402587890625, 184.0432891845703, 45.25726318359375, 290.9376525878906, 92.02432250976562, 176.99398803710938, 11.17181396484375, -11.891998291015625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000321.npy"} +{"epoch": 0.4713656387665198, "step": 322, "batch_size": 64, "mean": 149.79257202148438, "std": 144.279296875, "min": -101.10244750976562, "p10": -40.60990524291992, "median": 148.5826644897461, "p90": 343.9173492431641, "max": 466.2591857910156, "pos_frac": 0.828125, "sample": [-38.892295837402344, 391.9601135253906, 163.98773193359375, 466.2591857910156, -31.15546417236328, -80.84176635742188, 367.926513671875, 246.15521240234375, 230.29214477539062, -56.40668869018555, -41.34602355957031, 10.377326965332031, 262.63348388671875, 204.1812744140625, 237.86407470703125, 262.273681640625, 66.08853912353516, 144.48898315429688, -0.06186676025390625, 38.450958251953125, 341.5960998535156, 286.47222900390625, 94.20199584960938, 443.05596923828125, -50.009490966796875, 148.21664428710938, 220.89117431640625, 417.230712890625, -98.64391326904297, 337.93695068359375, 46.30271911621094, 122.15667724609375, 121.76748657226562, 282.70697021484375, 20.2628173828125, 45.05404281616211, 153.3528594970703, 149.5411376953125, -5.494014739990234, 98.078369140625, 213.3025665283203, 344.91217041015625, 218.8672637939453, -64.17049407958984, 340.86328125, 151.0409393310547, 144.59764099121094, 5.775623321533203, 203.382568359375, 115.28819274902344, 70.2121353149414, 418.48406982421875, 206.49972534179688, 226.3323974609375, 177.10308837890625, 148.9486846923828, 41.137786865234375, 98.31890106201172, 27.888078689575195, -101.10244750976562, 45.43560791015625, 87.06019592285156, 290.31829833984375, 157.31591796875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000322.npy"} +{"epoch": 0.47283406754772395, "step": 323, "batch_size": 64, "mean": 145.38555908203125, "std": 145.64553833007812, "min": -215.44027709960938, "p10": -13.374306297302242, "median": 142.22652435302734, "p90": 340.4651336669922, "max": 548.8257446289062, "pos_frac": 0.828125, "sample": [323.11578369140625, 201.21095275878906, 378.478759765625, -215.44027709960938, 338.1273498535156, -92.44408416748047, 375.06182861328125, 548.8257446289062, -38.868160247802734, 91.69133758544922, 259.79632568359375, -9.870208740234375, 105.94195556640625, 168.27500915527344, 3.8429946899414062, 132.38902282714844, 71.06778717041016, 204.68116760253906, 50.36577606201172, 420.4326171875, -0.6045913696289062, 55.20646667480469, 382.239013671875, -14.876062393188477, -20.853679656982422, 255.13987731933594, 35.770729064941406, 138.48680114746094, 221.7032928466797, 79.93923950195312, 315.2781982421875, 177.8765869140625, 168.57655334472656, -8.114439010620117, 205.5774383544922, 10.541587829589844, 77.17268371582031, 107.0616226196289, 246.7455291748047, 231.29776000976562, 230.25213623046875, 76.27633666992188, 141.22052001953125, 318.8223876953125, -104.73895263671875, 143.23252868652344, 233.55905151367188, 341.467041015625, 180.64205932617188, -5.359016418457031, 167.72796630859375, 349.601318359375, 20.52783966064453, 209.89869689941406, -137.3916778564453, 220.83749389648438, 89.40097045898438, 213.4710693359375, 53.482383728027344, 202.6564178466797, 86.51757049560547, 48.36497116088867, 194.12167358398438, 49.2379150390625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000323.npy"} +{"epoch": 0.47430249632892807, "step": 324, "batch_size": 64, "mean": 155.66616821289062, "std": 172.8908233642578, "min": -274.4296875, "p10": -37.45964660644531, "median": 156.89781188964844, "p90": 370.7733398437501, "max": 661.8583984375, "pos_frac": 0.8125, "sample": [-47.64373779296875, 379.3080749511719, 37.39190673828125, 108.79046630859375, 104.6977767944336, 104.931884765625, 194.54742431640625, 200.79684448242188, 178.5990753173828, 185.63302612304688, 556.4579467773438, 39.233978271484375, 271.1395263671875, 74.86738586425781, 49.21625900268555, 466.2334289550781, -77.94206237792969, -150.77857971191406, 210.65866088867188, 172.05712890625, 23.69012451171875, 156.54171752929688, -39.017059326171875, 214.1620330810547, 267.468994140625, 565.3027954101562, 275.185791015625, -155.8406219482422, 134.820556640625, 266.5644226074219, 379.2281799316406, 104.83998107910156, 112.14534759521484, 146.30227661132812, 243.58541870117188, 286.09600830078125, 129.74961853027344, 241.28359985351562, 104.83414459228516, -2.3442935943603516, 264.6961364746094, -274.4296875, 202.064697265625, 270.68402099609375, 320.6402587890625, 215.68057250976562, 55.260440826416016, -161.33106994628906, -33.82568359375, 168.81137084960938, 157.25390625, 188.793701171875, 113.98484802246094, 13.22684097290039, 136.1073760986328, 180.79263305664062, 351.0453796386719, -27.673324584960938, -24.69955825805664, 661.8583984375, 386.1525573730469, 78.02513122558594, -7.9449920654296875, 214.66583251953125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000324.npy"} +{"epoch": 0.47577092511013214, "step": 325, "batch_size": 64, "mean": 131.30233764648438, "std": 137.79505920410156, "min": -221.9534912109375, "p10": -12.587920761108395, "median": 129.87645721435547, "p90": 308.5082763671875, "max": 543.82177734375, "pos_frac": 0.859375, "sample": [-221.9534912109375, -158.47088623046875, 209.50155639648438, -0.16042327880859375, 315.7054138183594, 223.97659301757812, 101.45133209228516, -53.02595520019531, 125.87071990966797, 267.9195556640625, 142.1768341064453, 336.11761474609375, 217.94094848632812, 157.59974670410156, 127.43516540527344, 83.29486846923828, 90.5942153930664, 164.34312438964844, 15.298355102539062, 178.19407653808594, -14.067741394042969, 308.91070556640625, 207.12051391601562, 307.56927490234375, 13.407794952392578, -15.370796203613281, 137.0900421142578, 132.3177490234375, 68.07539367675781, 28.598304748535156, 271.1787109375, 75.42045593261719, 484.1107177734375, -9.135005950927734, 366.3860778808594, 13.966728210449219, 374.41412353515625, 22.005184173583984, 185.9705810546875, 179.73983764648438, 9.696203231811523, 136.7208251953125, 185.12478637695312, -27.254457473754883, 198.46539306640625, 199.70748901367188, 71.22115325927734, 151.28616333007812, 64.59619140625, 164.95953369140625, 199.60772705078125, -121.72845458984375, 13.411827087402344, 543.82177734375, 29.794815063476562, 114.38892364501953, 221.73565673828125, 126.62360382080078, 63.89971923828125, 122.6782455444336, 160.8738250732422, 53.3194465637207, 31.098289489746094, 227.78334045410156], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000325.npy"} +{"epoch": 0.47723935389133626, "step": 326, "batch_size": 64, "mean": 127.40595245361328, "std": 124.18055725097656, "min": -98.55056762695312, "p10": -1.0262710571289022, "median": 103.169189453125, "p90": 289.90169982910163, "max": 501.0677795410156, "pos_frac": 0.890625, "sample": [8.75201416015625, 62.27994155883789, 20.898147583007812, 296.20623779296875, 130.30821228027344, -2.773284912109375, 47.252506256103516, 72.68817138671875, 201.6395263671875, -65.32470703125, 51.928890228271484, -98.55056762695312, 436.76873779296875, 64.97554779052734, -58.49712371826172, 105.75215148925781, 78.20286560058594, 87.36236572265625, 178.72860717773438, 257.6053466796875, 26.437469482421875, 24.01873779296875, 237.32142639160156, 237.42938232421875, 3.0500946044921875, 60.02418518066406, 501.0677795410156, 166.29844665527344, 38.916297912597656, 94.30360412597656, 172.86581420898438, 138.6929931640625, 263.27288818359375, 140.91177368164062, 110.33074188232422, -12.40838623046875, 81.52001953125, 183.671630859375, -43.274967193603516, 186.78353881835938, 111.17201232910156, 72.9608154296875, 149.29287719726562, 73.97918701171875, 67.17860412597656, 83.50996398925781, 204.50076293945312, 209.66693115234375, 100.58622741699219, 170.87525939941406, -73.15834045410156, 337.2449645996094, 412.0450134277344, 122.5055160522461, 296.38958740234375, 156.62924194335938, 264.10076904296875, 245.54052734375, 13.993047714233398, 7.9079132080078125, 294.5032043457031, 279.16485595703125, 20.881229400634766, 45.073875427246094], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000326.npy"} +{"epoch": 0.4787077826725404, "step": 327, "batch_size": 64, "mean": 182.56640625, "std": 162.20574951171875, "min": -203.31509399414062, "p10": -22.439614868164046, "median": 181.77027130126953, "p90": 386.86133422851566, "max": 603.1817016601562, "pos_frac": 0.875, "sample": [185.72909545898438, 138.7436065673828, 387.53704833984375, 215.44729614257812, 49.98774719238281, 294.6163024902344, -28.909751892089844, 265.2472839355469, 192.06692504882812, 107.22041320800781, 255.440185546875, 272.88848876953125, 57.28913879394531, 268.0401611328125, -7.342628479003906, 206.24404907226562, 72.1302490234375, 415.4107971191406, 213.30166625976562, 186.93496704101562, -203.31509399414062, 365.26129150390625, 428.5599365234375, 322.904541015625, 40.98413848876953, 491.90216064453125, 223.72915649414062, -34.4844856262207, 254.0604248046875, 353.13031005859375, 328.5359191894531, -170.6602783203125, 17.961151123046875, 161.09629821777344, 281.10272216796875, 128.1094512939453, 177.8114471435547, 157.4852752685547, 121.2923583984375, 130.75982666015625, 117.44349670410156, 58.686553955078125, 134.75363159179688, 335.6285095214844, 126.25408935546875, -48.830841064453125, 105.01527404785156, 255.0679473876953, 385.28466796875, 33.341522216796875, 334.677978515625, 122.43763732910156, 603.1817016601562, -53.73802185058594, 422.3164367675781, 11.829160690307617, 108.58726501464844, 354.911376953125, 42.64271545410156, 397.40679931640625, 221.76885986328125, 345.2998962402344, 64.55121612548828, -118.51837158203125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000327.npy"} +{"epoch": 0.4801762114537445, "step": 328, "batch_size": 64, "mean": 153.0107421875, "std": 151.1506805419922, "min": -150.24703979492188, "p10": -4.343890190124509, "median": 124.02864456176758, "p90": 362.3960632324219, "max": 522.7840576171875, "pos_frac": 0.859375, "sample": [147.58692932128906, -56.410675048828125, 108.48575592041016, 71.14617919921875, 83.72433471679688, 133.62197875976562, 209.6631317138672, 58.90016174316406, 112.43299865722656, -150.24703979492188, 69.9642105102539, 253.89320373535156, 367.7310791015625, 270.6207580566406, 57.90556335449219, 349.94769287109375, 29.852821350097656, -112.14445495605469, 18.080991744995117, 130.02963256835938, 0.23103713989257812, -6.091339111328125, 249.15252685546875, 522.7840576171875, 162.3668212890625, 327.5336608886719, 198.37136840820312, 102.94216918945312, 443.02752685546875, 320.1573181152344, 31.942899703979492, 129.88558959960938, 223.8404083251953, 41.41703414916992, -2.0853137969970703, 93.06674194335938, 195.54348754882812, 194.9630126953125, 165.82095336914062, 88.0506591796875, -1.920907974243164, 163.60269165039062, -58.13557434082031, 167.35025024414062, -18.224708557128906, 342.0707092285156, 101.12455749511719, 227.73355102539062, 244.48397827148438, 54.999595642089844, 378.3699035644531, 494.1701965332031, 91.21110534667969, 7.687103271484375, 40.37504959106445, 498.45111083984375, 457.91778564453125, 35.059722900390625, 299.068359375, 70.24494934082031, 118.17169952392578, 268.48089599609375, 178.0020294189453, -5.311851501464844], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000328.npy"} +{"epoch": 0.48164464023494863, "step": 329, "batch_size": 64, "mean": 161.3940887451172, "std": 164.27101135253906, "min": -208.1063232421875, "p10": -20.257223510742186, "median": 162.8469467163086, "p90": 386.96117858886726, "max": 636.4505004882812, "pos_frac": 0.859375, "sample": [151.357421875, 279.3310546875, 322.256103515625, 204.50144958496094, 178.95297241210938, -27.97481346130371, 409.8475646972656, 45.562782287597656, 332.6531982421875, 169.6563262939453, 75.60986328125, 172.63442993164062, 29.85578155517578, -18.060317993164062, 120.74925231933594, 172.276611328125, 160.7545928955078, 142.80996704101562, 14.615755081176758, 429.2976989746094, 128.90380859375, 78.56068420410156, 173.3982391357422, 114.76358032226562, -208.1063232421875, 22.15387725830078, 56.82762908935547, 636.4505004882812, 393.1600036621094, 34.75022888183594, 262.43572998046875, -19.385433197021484, 240.84085083007812, 4.635202407836914, 218.8743896484375, 319.2187194824219, 350.03057861328125, 242.41470336914062, 443.93902587890625, 229.7783203125, 65.05630493164062, -20.630847930908203, -132.89952087402344, 372.49725341796875, 330.33563232421875, -33.78158187866211, 336.1018981933594, -72.15293884277344, 166.4542694091797, 80.95233154296875, 164.93930053710938, 30.76304817199707, 175.3634490966797, 233.28956604003906, 59.24402618408203, 48.058128356933594, 206.32447814941406, 259.8062744140625, 495.64605712890625, 36.19805908203125, -146.22946166992188, 139.66677856445312, 411.6087646484375, 32.278419494628906], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000329.npy"} +{"epoch": 0.4831130690161527, "step": 330, "batch_size": 64, "mean": 156.75009155273438, "std": 154.9280242919922, "min": -148.98776245117188, "p10": -56.15992889404294, "median": 146.23468780517578, "p90": 331.063265991211, "max": 598.3441162109375, "pos_frac": 0.828125, "sample": [403.309814453125, 14.5863037109375, -134.97308349609375, -83.30947875976562, 221.48114013671875, 351.71014404296875, 140.6939239501953, 118.99993896484375, -29.563615798950195, 65.10173797607422, -83.51579284667969, 147.4585723876953, 88.63623046875, 262.82904052734375, 298.5469970703125, 80.51923370361328, 171.02328491210938, 166.15090942382812, 173.11358642578125, 303.7656555175781, 39.802215576171875, 81.58792114257812, 220.47311401367188, 116.47108459472656, 140.9981689453125, 239.5315704345703, -108.62152099609375, 145.01080322265625, 217.1353759765625, 120.10759735107422, 224.9208984375, 12.984691619873047, -67.558349609375, 151.19894409179688, 334.49432373046875, 48.10346221923828, 115.64842224121094, 473.2659912109375, -6.303092956542969, 131.64918518066406, -24.164207458496094, 315.28387451171875, 105.58352661132812, 598.3441162109375, -92.95790100097656, 512.5980224609375, 300.9560546875, 201.1234588623047, 291.8597412109375, 102.9460220336914, 196.29864501953125, 233.76966857910156, 87.87445068359375, 407.7486572265625, 86.39669036865234, 284.5818176269531, 162.14340209960938, 233.4242401123047, -148.98776245117188, 323.0574645996094, 132.7569580078125, 196.00685119628906, 229.7811279296875, -11.883880615234375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000330.npy"} +{"epoch": 0.4845814977973568, "step": 331, "batch_size": 64, "mean": 166.2667236328125, "std": 125.61683654785156, "min": -34.41754913330078, "p10": 5.670362854003908, "median": 135.1876449584961, "p90": 319.0373046875, "max": 510.20953369140625, "pos_frac": 0.921875, "sample": [128.6924285888672, 415.61077880859375, 299.3126525878906, 1.417205810546875, 228.77854919433594, 49.355682373046875, 70.14617919921875, -25.477279663085938, 83.47120666503906, 37.799346923828125, 419.4900817871094, 323.0478820800781, 446.0361328125, 220.34249877929688, 184.72630310058594, 153.68002319335938, 510.20953369140625, 88.04869079589844, 238.2816162109375, 145.47341918945312, 309.6792907714844, 292.6753845214844, 330.91937255859375, 4.804340362548828, 106.52247619628906, -4.6858673095703125, 124.12117004394531, 84.0659408569336, 265.1500549316406, 69.66412353515625, 114.45317077636719, 7.691082000732422, 281.82171630859375, 84.52201843261719, 191.70489501953125, 66.6123275756836, -34.41754913330078, 118.32013702392578, 245.22576904296875, 434.1683044433594, 271.8471374511719, 56.36949157714844, 44.309715270996094, 234.4837646484375, 49.58934783935547, 133.28652954101562, 204.51763916015625, 180.01673889160156, 219.84417724609375, 120.46269226074219, -5.38134765625, 75.32333374023438, 111.22172546386719, 122.64973449707031, 239.73495483398438, 137.08876037597656, 252.4419403076172, 90.44837188720703, -16.210220336914062, 286.9788818359375, 213.25779724121094, 189.78585815429688, 221.8957977294922, 95.6461181640625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000331.npy"} +{"epoch": 0.48604992657856094, "step": 332, "batch_size": 64, "mean": 119.06929016113281, "std": 132.80868530273438, "min": -207.90380859375, "p10": -19.24050598144531, "median": 109.0322380065918, "p90": 321.3516510009766, "max": 443.117919921875, "pos_frac": 0.796875, "sample": [384.70501708984375, 144.24343872070312, 113.04063415527344, 186.62152099609375, 62.67182922363281, 1.3355865478515625, 90.82112121582031, 309.1259765625, 324.0137939453125, 117.88423156738281, 125.36967468261719, -4.094743728637695, 24.544158935546875, 347.074462890625, 146.35055541992188, 107.02666473388672, 94.38494873046875, -20.722122192382812, 46.88629913330078, -6.255586624145508, 182.0570068359375, 98.83836364746094, 315.1399841308594, 41.27076721191406, 313.68743896484375, 112.58467102050781, -1.670166015625, 28.23033905029297, 337.04705810546875, 197.0740509033203, 125.158203125, 167.52345275878906, 48.255828857421875, -25.17778778076172, -9.618110656738281, 64.17548370361328, 443.117919921875, 302.9427185058594, -93.67141723632812, 189.2177276611328, 166.24154663085938, -76.13468170166016, 184.51336669921875, 112.78758239746094, 175.48788452148438, 342.630859375, 60.62992858886719, 329.91265869140625, -44.847076416015625, 96.67378234863281, -81.87030029296875, 111.03781127929688, 289.9683837890625, 5.9771728515625, 232.5305633544922, 33.0595588684082, -15.783401489257812, -8.673301696777344, 0.2547149658203125, 82.66326141357422, 185.67547607421875, 78.41268157958984, 135.9784698486328, -207.90380859375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000332.npy"} +{"epoch": 0.48751835535976507, "step": 333, "batch_size": 64, "mean": 163.3745880126953, "std": 176.4287109375, "min": -163.64700317382812, "p10": -63.835654449462886, "median": 155.38597106933594, "p90": 397.40014648437506, "max": 562.8408203125, "pos_frac": 0.78125, "sample": [-65.85035705566406, 150.24911499023438, 562.8408203125, -18.680374145507812, 182.87673950195312, 99.85728454589844, 160.7025604248047, 250.09510803222656, 24.983551025390625, 91.77189636230469, 497.94134521484375, -59.134681701660156, 151.50820922851562, -77.38385772705078, 330.00201416015625, 30.225250244140625, 449.6279296875, -26.068605422973633, 430.6798400878906, 116.80818939208984, -74.82479858398438, 309.26495361328125, -88.55610656738281, 342.69183349609375, 279.46734619140625, 249.07638549804688, 72.46125030517578, 17.061553955078125, 159.26373291015625, 341.52728271484375, 76.36324310302734, 15.197553634643555, -22.986297607421875, 446.3390808105469, 273.2974853515625, 264.9531555175781, -163.64700317382812, 3.5973739624023438, 401.61651611328125, 234.20523071289062, 322.9894714355469, 52.86137390136719, 1.2233505249023438, 232.62911987304688, 313.45159912109375, -27.67285919189453, 150.84056091308594, 211.1742706298828, 387.56195068359375, 225.39846801757812, 375.4205322265625, 288.3191223144531, 514.4014892578125, 79.2774887084961, 237.47470092773438, -74.61189270019531, 326.270751953125, -26.472469329833984, 231.65576171875, -52.45591735839844, 228.50851440429688, -80.2021484375, 40.69139099121094, 77.81669616699219], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000333.npy"} +{"epoch": 0.4889867841409692, "step": 334, "batch_size": 64, "mean": 147.89768981933594, "std": 157.1070556640625, "min": -238.095947265625, "p10": 5.322164535522463, "median": 133.26744079589844, "p90": 351.6571472167969, "max": 494.9858093261719, "pos_frac": 0.90625, "sample": [386.84320068359375, 126.49282836914062, -82.27316284179688, 344.2113037109375, 217.1533203125, 24.926605224609375, 29.167335510253906, 70.8761215209961, 196.74118041992188, 20.878829956054688, 177.36224365234375, 139.83917236328125, 310.90985107421875, 115.59451293945312, 35.13487243652344, 134.34214782714844, 48.32142639160156, 287.6072998046875, -203.60804748535156, 327.83203125, 33.00295639038086, 290.7147521972656, 70.02792358398438, -238.095947265625, 239.11279296875, 111.1588134765625, 350.7154541015625, -123.08699035644531, -17.023235321044922, 72.41514587402344, 13.642471313476562, 4.362262725830078, 58.04493713378906, 54.69146728515625, 352.06072998046875, 421.1036071777344, 288.9765625, 20.200647354125977, 148.40524291992188, 163.90988159179688, 111.9977035522461, 156.43295288085938, 313.86529541015625, 132.19273376464844, 27.450454711914062, 130.0381317138672, 400.08013916015625, 217.07168579101562, 436.3829040527344, 219.02105712890625, 97.23787689208984, -137.3544158935547, 53.238067626953125, 230.26023864746094, 146.87937927246094, 378.26165771484375, 322.5206604003906, 144.1730499267578, 7.5619354248046875, 338.9256286621094, 16.868227005004883, 494.9858093261719, 171.23574829101562, 33.432823181152344], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000334.npy"} +{"epoch": 0.49045521292217326, "step": 335, "batch_size": 64, "mean": 121.41865539550781, "std": 158.9032745361328, "min": -204.26181030273438, "p10": -63.322785949707026, "median": 128.4450569152832, "p90": 320.2089050292969, "max": 612.7890625, "pos_frac": 0.78125, "sample": [149.66571044921875, 245.3757781982422, 265.0352783203125, 101.73928833007812, 130.8773956298828, -59.026832580566406, 56.29499053955078, 275.23272705078125, 62.43901443481445, 5.217569351196289, -122.21366882324219, 80.19975280761719, 139.73544311523438, 177.13755798339844, 100.3158950805664, 368.23590087890625, 126.0127182006836, 158.68222045898438, -154.5074005126953, 148.33419799804688, -32.02458190917969, 80.53053283691406, 156.07359313964844, 142.01345825195312, 313.3617858886719, 303.81097412109375, 313.9550476074219, 179.67308044433594, 53.86540222167969, 444.5540771484375, 612.7890625, -178.81951904296875, 243.16757202148438, -204.26181030273438, 329.9997253417969, 28.17116355895996, 322.8891296386719, 95.24803161621094, 50.84765625, 176.49356079101562, -60.663787841796875, 334.46441650390625, 72.6786880493164, -41.6112060546875, 221.07269287109375, 32.49720001220703, 191.12474060058594, 8.011245727539062, -127.05636596679688, 36.65618896484375, 74.5323486328125, -64.46235656738281, 298.58319091796875, -95.9189453125, -20.883750915527344, 262.57177734375, 147.89137268066406, 357.6474609375, -18.450424194335938, -46.10849380493164, 166.90597534179688, 216.29681396484375, 134.43893432617188, 3.4543800354003906], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000335.npy"} +{"epoch": 0.4919236417033774, "step": 336, "batch_size": 64, "mean": 164.1492919921875, "std": 177.00155639648438, "min": -309.616455078125, "p10": -48.604997253417906, "median": 143.16555786132812, "p90": 366.2436981201172, "max": 658.177978515625, "pos_frac": 0.890625, "sample": [95.27044677734375, 45.04572296142578, 507.8518371582031, 110.62775421142578, 280.5001220703125, 168.63223266601562, 145.57308959960938, -75.13951873779297, -75.01176452636719, 551.1171875, 39.49830627441406, 173.90000915527344, 125.14037322998047, 221.94998168945312, 658.177978515625, 16.657621383666992, 457.9483947753906, 291.6367492675781, 174.80906677246094, 244.77772521972656, -72.77369689941406, 130.0935516357422, 227.61500549316406, 234.11090087890625, 7.78863525390625, 189.97384643554688, 92.6414794921875, 112.99325561523438, 364.1972961425781, 152.62078857421875, 103.9012222290039, 77.2478256225586, 97.62974548339844, 106.4566421508789, 84.21754455566406, 194.92861938476562, -81.16990661621094, 367.1207275390625, 8.831356048583984, 34.160438537597656, 186.67861938476562, 139.29208374023438, 398.5742492675781, 65.04971313476562, 321.19921875, 39.81475830078125, -309.616455078125, 50.80415344238281, 140.75802612304688, 216.03591918945312, 130.153564453125, 82.87926483154297, 235.48265075683594, 119.4197769165039, 283.4347839355469, 314.4984130859375, 252.35308837890625, 603.0369262695312, -175.00083923339844, 217.15406799316406, -201.3575439453125, 237.74789428710938, 279.39410400390625, 286.21893310546875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000336.npy"} +{"epoch": 0.4933920704845815, "step": 337, "batch_size": 64, "mean": 171.725341796875, "std": 156.62600708007812, "min": -147.3826446533203, "p10": -18.347989654541013, "median": 156.77433013916016, "p90": 341.76951599121094, "max": 573.7828369140625, "pos_frac": 0.828125, "sample": [573.7828369140625, 343.2396240234375, -144.37989807128906, -4.445594787597656, 318.53143310546875, 304.8525695800781, -9.522102355957031, 211.3682403564453, 75.06770324707031, 287.88299560546875, 16.18292999267578, 152.00048828125, 146.75265502929688, 270.2647705078125, 338.3392639160156, -28.23084259033203, 221.22125244140625, 225.9861297607422, 85.79341125488281, 309.3276062011719, 111.09907531738281, 173.32965087890625, 41.864891052246094, 328.6181640625, 148.72518920898438, 98.97900390625, 544.3091430664062, 35.165771484375, 16.58562469482422, -56.84918212890625, 141.35116577148438, 170.48106384277344, 228.36407470703125, -45.191627502441406, 177.4730682373047, 145.7104949951172, 329.6982727050781, -15.233627319335938, 157.18995666503906, 197.0042724609375, 26.151714324951172, 455.1966857910156, 156.35870361328125, 27.71533203125, -147.3826446533203, 113.69468688964844, 409.18060302734375, 333.33880615234375, 336.7650451660156, 114.5777359008789, 220.03814697265625, -6.206977844238281, 394.91497802734375, 277.6619873046875, 102.01628112792969, 398.76348876953125, 249.9905548095703, 130.09539794921875, -19.682716369628906, 275.6264953613281, 251.61721801757812, 113.00161743164062, -72.03083801269531, 226.3291473388672], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000337.npy"} +{"epoch": 0.4948604992657856, "step": 338, "batch_size": 64, "mean": 165.9337158203125, "std": 144.62030029296875, "min": -116.71063232421875, "p10": -10.925658798217768, "median": 160.47583770751953, "p90": 366.85985107421874, "max": 510.013916015625, "pos_frac": 0.859375, "sample": [199.41319274902344, 324.6720886230469, -30.565353393554688, -40.590087890625, 164.28094482421875, 99.49778747558594, -93.81466674804688, 166.2056121826172, 453.4568176269531, 189.48483276367188, 96.85243225097656, 47.256107330322266, -4.258241653442383, 324.4543151855469, 82.39263916015625, 160.4813995361328, 158.8138427734375, 192.09518432617188, 244.07861328125, 381.0535583496094, 398.3101806640625, 103.28258514404297, 212.5814208984375, 212.93702697753906, 229.8923797607422, 46.81358337402344, 224.38888549804688, -4.983634948730469, 77.73197174072266, 7.875022888183594, 184.59060668945312, 408.4684753417969, 344.2209777832031, 44.77540969848633, 143.48873901367188, 202.06765747070312, 138.93576049804688, 1.3024024963378906, 299.5948181152344, 109.11138153076172, 353.0942077636719, 28.980955123901367, 134.640869140625, -41.44438171386719, 364.29534912109375, 78.06781005859375, 174.61868286132812, 63.23755645751953, 208.97991943359375, 91.82695007324219, -116.71063232421875, -18.74081039428711, 449.99847412109375, 510.013916015625, 31.14875030517578, 178.6676788330078, 367.95892333984375, 143.03121948242188, 298.83349609375, 160.47027587890625, 333.24090576171875, -13.472240447998047, 53.00318908691406, 285.36962890625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000338.npy"} +{"epoch": 0.49632892804698975, "step": 339, "batch_size": 64, "mean": 159.30018615722656, "std": 143.9568328857422, "min": -108.93875885009766, "p10": -12.591029739379874, "median": 153.76490783691406, "p90": 356.9250549316407, "max": 554.6740112304688, "pos_frac": 0.875, "sample": [149.42507934570312, 48.522239685058594, -2.9151954650878906, -71.48281860351562, 423.70550537109375, 0.4047698974609375, 109.83717346191406, 158.83131408691406, 239.669189453125, 6.389686584472656, 530.6387329101562, 94.51936340332031, 105.87261962890625, 63.45989990234375, 257.727294921875, 246.5277557373047, 7.89280891418457, 79.44597625732422, 41.477317810058594, 434.752197265625, 193.98057556152344, -16.737815856933594, 170.13768005371094, 3.4402999877929688, 103.51507568359375, 164.25735473632812, 368.9440612792969, 26.09479522705078, 259.9879150390625, 554.6740112304688, -82.99629211425781, 153.05303955078125, -62.858055114746094, 67.84053039550781, 394.9352722167969, 189.50059509277344, 125.53665924072266, 78.15848541259766, 154.47677612304688, 247.44622802734375, 79.95402526855469, 144.45745849609375, 193.88941955566406, 275.3301696777344, 59.668190002441406, 328.8807067871094, -108.93875885009766, 402.0268249511719, 217.26181030273438, -40.26167678833008, 226.32260131835938, 223.44305419921875, 210.59901428222656, 186.8856201171875, 126.25483703613281, 101.26310729980469, 101.66851806640625, -24.497703552246094, 216.19305419921875, 203.42864990234375, 292.94097900390625, 270.7572021484375, 280.6163330078125, 208.97979736328125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000339.npy"} +{"epoch": 0.4977973568281938, "step": 340, "batch_size": 64, "mean": 149.84320068359375, "std": 156.49560546875, "min": -253.40936279296875, "p10": -29.640920257568347, "median": 137.8533935546875, "p90": 344.9907653808594, "max": 626.2277221679688, "pos_frac": 0.859375, "sample": [133.9207763671875, 282.9339599609375, 112.23336029052734, 156.8924560546875, 136.7677459716797, 292.2745056152344, 412.3126525878906, -34.25238800048828, 153.4180908203125, 67.8259048461914, 185.67335510253906, 626.2277221679688, 189.7476043701172, -100.71304321289062, 350.523681640625, 9.168159484863281, 337.3412780761719, 113.03742218017578, 35.963768005371094, 33.493892669677734, 165.81031799316406, 13.06024169921875, -253.40936279296875, 327.02783203125, -42.80409240722656, 172.15475463867188, -51.82050323486328, 348.3869323730469, 51.02366638183594, 237.39532470703125, 298.85333251953125, 151.73326110839844, 519.7449340820312, 340.7618408203125, 63.557350158691406, 129.84039306640625, 20.07425308227539, 145.26126098632812, 52.32981872558594, 185.99319458007812, 151.59066772460938, 138.44064331054688, 127.64092254638672, 12.73501205444336, -102.67286682128906, 262.7950744628906, 65.53546142578125, -3.4445724487304688, 346.80316162109375, 27.773178100585938, 59.07787322998047, 356.43463134765625, 297.8579406738281, 295.7803649902344, 160.01348876953125, 201.07073974609375, 61.35036087036133, 110.59800720214844, 274.45391845703125, 137.26614379882812, -98.56011962890625, 299.2660217285156, -18.880828857421875, 57.273399353027344], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000340.npy"} +{"epoch": 0.49926578560939794, "step": 341, "batch_size": 64, "mean": 169.23243713378906, "std": 158.62100219726562, "min": -167.66510009765625, "p10": -48.88282012939452, "median": 171.90189361572266, "p90": 340.1825836181641, "max": 638.4044799804688, "pos_frac": 0.84375, "sample": [638.4044799804688, -62.417510986328125, -12.276809692382812, 5.44709587097168, 183.94406127929688, 297.9249267578125, 302.4270324707031, 387.8516540527344, 174.10733032226562, 111.99852752685547, 65.83154296875, 231.83665466308594, 172.183349609375, 202.5213623046875, 88.88896179199219, -54.10784912109375, -36.69108581542969, 308.72332763671875, 48.34209442138672, 214.7408447265625, -108.46056365966797, 280.34686279296875, 348.6713562011719, 243.57138061523438, 118.5194320678711, 443.34088134765625, 187.96328735351562, 164.85064697265625, 31.249435424804688, -60.32041931152344, 117.0693359375, 267.81817626953125, 387.6592102050781, 255.17318725585938, 275.7640380859375, 270.3068542480469, 135.60585021972656, 4.90672492980957, -3.665262222290039, 171.6204376220703, 318.82769775390625, 217.6243133544922, 136.81405639648438, 90.55976104736328, -116.98854064941406, 268.501953125, -86.70860290527344, 89.93289184570312, -167.66510009765625, 319.8267822265625, 168.81285095214844, 226.12899780273438, 198.200439453125, 57.340049743652344, 529.1031494140625, 70.52012634277344, 324.6678466796875, 337.5649108886719, 312.60821533203125, 61.95706558227539, 130.506591796875, 141.86099243164062, 341.304443359375, 57.90428924560547], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000341.npy"} +{"epoch": 0.5007342143906021, "step": 342, "batch_size": 64, "mean": 161.10507202148438, "std": 162.73526000976562, "min": -178.42681884765625, "p10": -17.88207015991211, "median": 143.15811920166016, "p90": 360.4499633789064, "max": 673.3868408203125, "pos_frac": 0.8125, "sample": [413.89044189453125, -178.42681884765625, -82.90118408203125, 334.6387023925781, 14.911502838134766, 116.34321594238281, 211.25125122070312, 218.1541290283203, 154.85862731933594, 281.80218505859375, 47.89143371582031, 143.9269561767578, 174.53872680664062, 229.89002990722656, 446.0561828613281, 247.258056640625, 110.92985534667969, 288.6981506347656, -42.05841827392578, -17.566436767578125, 225.6353759765625, -43.145164489746094, 172.48263549804688, 10.772132873535156, -18.01734161376953, 185.08258056640625, 673.3868408203125, 262.5533752441406, 140.2405548095703, -104.29563903808594, -11.732555389404297, 595.003173828125, 142.3892822265625, 67.84596252441406, -4.0135040283203125, -15.803825378417969, 409.64453125, 331.69598388671875, 418.80328369140625, 128.5522003173828, -12.037031173706055, 128.87417602539062, 259.8123474121094, 29.198944091796875, 257.2325744628906, 326.228515625, 91.6531982421875, 282.8042907714844, 159.55650329589844, 285.87969970703125, 371.5119323730469, 111.44837188720703, -34.36774444580078, 83.80516052246094, 228.64181518554688, 99.1551284790039, 106.18931579589844, 0.04711341857910156, 50.143775939941406, 125.00152587890625, 196.06663513183594, 266.3099365234375, 158.49679565429688, 57.90557098388672], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000342.npy"} +{"epoch": 0.5022026431718062, "step": 343, "batch_size": 64, "mean": 180.06126403808594, "std": 145.0496826171875, "min": -115.30704498291016, "p10": 9.00086879730225, "median": 163.61162567138672, "p90": 367.5280242919922, "max": 592.0393676757812, "pos_frac": 0.921875, "sample": [288.9686584472656, 106.31945037841797, 158.06492614746094, 97.85591888427734, 202.0233917236328, 2.3172149658203125, 101.48455810546875, 242.9290313720703, 113.14266967773438, 431.9051513671875, 87.42729187011719, 13.39599609375, 160.31951904296875, 75.93228149414062, 130.9615936279297, 54.56252670288086, 70.47705078125, 220.3303985595703, 189.221923828125, 438.47808837890625, 162.6389617919922, 274.3530578613281, 58.38056945800781, 301.14251708984375, 368.49462890625, 320.31097412109375, 57.46873092651367, 64.79208374023438, 469.4832763671875, 196.44834899902344, -88.98986053466797, 60.62057876586914, 258.9672546386719, 38.3370361328125, 209.26669311523438, 156.875, 98.214111328125, 32.25813293457031, 7.117242813110352, 296.42218017578125, -79.96884155273438, 238.05361938476562, 188.47225952148438, 278.2151794433594, 184.8122100830078, 592.0393676757812, 289.50103759765625, 243.93936157226562, 324.8973693847656, 282.0517272949219, 187.6834716796875, 230.76803588867188, 133.79122924804688, -6.906337738037109, 365.2726135253906, 290.6300964355469, 138.96311950683594, 117.34613800048828, 164.58428955078125, 71.8240966796875, -21.40808868408203, 449.3444519042969, 446.6026611328125, -115.30704498291016], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000343.npy"} +{"epoch": 0.5036710719530103, "step": 344, "batch_size": 64, "mean": 153.33029174804688, "std": 162.5640106201172, "min": -160.5049285888672, "p10": -14.786077117919916, "median": 132.8502426147461, "p90": 353.20285644531253, "max": 836.4849853515625, "pos_frac": 0.875, "sample": [28.987586975097656, 7.553400039672852, -160.5049285888672, -9.546539306640625, 112.91847229003906, 155.51242065429688, 137.20155334472656, 166.58489990234375, 22.07562255859375, 131.86476135253906, 71.1551513671875, 504.6689453125, -17.95227813720703, 181.4186248779297, 68.40739440917969, 133.83572387695312, 184.8697052001953, -23.524948120117188, -95.87145233154297, 212.85472106933594, 104.93910217285156, 46.20111083984375, 89.852783203125, 17.847888946533203, 120.61152648925781, 207.34341430664062, 354.60430908203125, 332.8243713378906, 47.11920928955078, 71.57583618164062, 349.93280029296875, 84.378662109375, 204.49667358398438, 30.507108688354492, 210.20785522460938, 130.58773803710938, 270.804443359375, 4.172760009765625, 183.26910400390625, 15.75806999206543, 360.8983154296875, 260.945068359375, 204.288818359375, 128.09786987304688, 60.96649932861328, 836.4849853515625, 341.12017822265625, 262.1436767578125, 63.35032653808594, 398.32293701171875, -17.031593322753906, 176.88931274414062, 194.08912658691406, 372.9436950683594, 96.58976745605469, 107.60264587402344, -116.6791000366211, 274.69329833984375, 385.9272155761719, -151.38955688476562, 241.09071350097656, 248.05320739746094, 151.4833526611328, 242.71438598632812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000344.npy"} +{"epoch": 0.5051395007342144, "step": 345, "batch_size": 64, "mean": 170.38381958007812, "std": 165.67413330078125, "min": -182.4739990234375, "p10": -14.862044143676751, "median": 128.01971435546875, "p90": 406.8660858154298, "max": 656.3436279296875, "pos_frac": 0.875, "sample": [59.94676971435547, 494.4517517089844, 199.9578857421875, 226.68609619140625, 58.310035705566406, -113.95989227294922, 418.935546875, -27.759658813476562, 126.56637573242188, 259.55633544921875, -17.472219467163086, 192.38194274902344, 87.62847900390625, 319.5372619628906, 378.7040100097656, 89.81396484375, -93.90191650390625, 135.15948486328125, 52.084529876708984, 144.98402404785156, 124.55136108398438, 287.12371826171875, 47.04286193847656, 272.3626403808594, 188.87765502929688, 236.1202850341797, 100.33995819091797, 309.52874755859375, 120.0201416015625, 120.24470520019531, 70.74594116210938, 501.7225646972656, 113.79400634765625, 91.53099060058594, 41.694366455078125, 282.6995849609375, 291.9420166015625, 347.775390625, 139.68154907226562, 287.6168212890625, 175.65699768066406, 656.3436279296875, 125.70510864257812, 15.4039306640625, 277.32318115234375, 296.9820556640625, 32.1230354309082, 548.3848266601562, 119.7777099609375, 247.75823974609375, 53.31378936767578, -20.68109893798828, 121.61093139648438, 427.58721923828125, -8.771635055541992, -129.23117065429688, 41.876914978027344, 112.33993530273438, 432.5626220703125, 129.47305297851562, 217.12640380859375, 206.92965698242188, -182.4739990234375, 40.41688537597656], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000345.npy"} +{"epoch": 0.5066079295154186, "step": 346, "batch_size": 64, "mean": 161.12371826171875, "std": 155.08612060546875, "min": -310.7843017578125, "p10": -11.360117530822754, "median": 170.5549545288086, "p90": 368.29050598144534, "max": 508.3390808105469, "pos_frac": 0.828125, "sample": [231.94358825683594, 69.31605529785156, -4.176704406738281, 268.8660888671875, 373.3471374511719, 292.6313171386719, 480.9568786621094, 167.8671875, 179.22669982910156, 508.3390808105469, -310.7843017578125, 77.72831726074219, 237.04116821289062, -3.1709251403808594, 196.06765747070312, 334.8005065917969, 8.21929931640625, 262.56170654296875, 495.739501953125, -11.581571578979492, 223.29515075683594, 91.81045532226562, -46.7779541015625, 345.7545166015625, 133.9400634765625, 196.95892333984375, 199.2476806640625, 75.44485473632812, 216.10494995117188, 139.76687622070312, 64.63424682617188, -13.254447937011719, 145.91505432128906, 83.56306457519531, 402.92242431640625, 151.74061584472656, 224.18614196777344, 156.8724365234375, 221.74191284179688, 21.246231079101562, -10.843391418457031, 356.49169921875, 253.5392608642578, 382.6624755859375, 112.16407775878906, 180.1360626220703, 279.70867919921875, 237.09860229492188, 41.01752853393555, 47.318912506103516, 228.36721801757812, 30.188945770263672, 16.294281005859375, -49.16088104248047, 89.56550598144531, 325.7811279296875, 73.4791259765625, 177.94186401367188, 408.0475158691406, -52.021697998046875, -8.4122314453125, -83.32612609863281, 173.2427215576172, 212.58438110351562], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000346.npy"} +{"epoch": 0.5080763582966226, "step": 347, "batch_size": 64, "mean": 163.08779907226562, "std": 175.5387420654297, "min": -233.5640869140625, "p10": -63.97290573120116, "median": 150.0560760498047, "p90": 351.00710754394544, "max": 625.39013671875, "pos_frac": 0.84375, "sample": [-127.36538696289062, 223.88653564453125, 308.7586669921875, 273.9236755371094, 110.89684295654297, 100.1488265991211, 87.39521026611328, 165.1366729736328, 292.48101806640625, 132.16845703125, 266.7528076171875, 25.044593811035156, 156.95997619628906, -9.280517578125, 235.44427490234375, -187.1039581298828, -233.5640869140625, 2.958953857421875, 595.884765625, 123.97806549072266, 295.63751220703125, 28.359134674072266, 38.698486328125, 243.92440795898438, -68.12228393554688, 385.909912109375, 303.5533142089844, 228.6531982421875, 16.28479766845703, 323.39080810546875, 279.47406005859375, 361.53271484375, 271.31842041015625, 78.33452606201172, 150.22061157226562, -1.4129867553710938, 415.19775390625, 184.40805053710938, 104.58843994140625, -111.05276489257812, 149.89154052734375, 165.55581665039062, 102.59117889404297, 280.12567138671875, 100.5645523071289, 108.7864990234375, 57.89837646484375, 101.0238265991211, 625.39013671875, 52.25714874267578, 242.7920684814453, 404.60540771484375, 208.95140075683594, 617.8560791015625, -54.29102325439453, 129.13751220703125, 177.77279663085938, -92.59397888183594, 326.4473571777344, 81.17875671386719, 317.7049255371094, -71.66398620605469, 61.06626892089844, 271.16668701171875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000347.npy"} +{"epoch": 0.5095447870778267, "step": 348, "batch_size": 64, "mean": 145.63839721679688, "std": 161.53140258789062, "min": -168.15785217285156, "p10": -19.12066535949706, "median": 120.60348510742188, "p90": 333.4512084960938, "max": 681.803955078125, "pos_frac": 0.84375, "sample": [251.43930053710938, 7.047416687011719, 165.15072631835938, 223.0291748046875, 215.68748474121094, -5.587013244628906, 13.597999572753906, 272.7304992675781, 276.30218505859375, -163.4671173095703, 74.77565002441406, 65.17559814453125, 394.8102722167969, 232.50213623046875, 169.11363220214844, 195.18063354492188, 4.167581558227539, 274.3035888671875, 113.44415283203125, 91.09112548828125, 201.93048095703125, 220.18557739257812, 36.576454162597656, 681.803955078125, 337.5187683105469, 196.65087890625, 22.24939727783203, 283.9375305175781, -6.52891731262207, 101.86760711669922, 65.51072692871094, -88.69944763183594, 6.4759521484375, 524.9115600585938, 307.3096008300781, 177.53094482421875, 206.10821533203125, 90.51191711425781, 219.8180694580078, 40.06523132324219, -168.15785217285156, 323.9602355957031, -39.69019317626953, 277.40380859375, 184.098876953125, 102.83704376220703, 45.274085998535156, 24.078596115112305, 379.8586730957031, 183.37648010253906, 60.71973419189453, 85.36045837402344, 410.85833740234375, 13.434196472167969, 319.57672119140625, 144.8248748779297, 28.14765167236328, -11.160213470458984, 19.351547241210938, 448.9781799316406, 127.7628173828125, -22.53228759765625, -65.57569885253906, -48.157413482666016], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000348.npy"} +{"epoch": 0.5110132158590308, "step": 349, "batch_size": 64, "mean": 150.9041748046875, "std": 190.70314025878906, "min": -262.0289001464844, "p10": -40.792142105102535, "median": 106.2527961730957, "p90": 405.06243591308595, "max": 759.7919311523438, "pos_frac": 0.796875, "sample": [4.121696472167969, -262.0289001464844, 269.8824157714844, 105.9455337524414, 95.72682189941406, 46.899871826171875, 347.6978454589844, -38.34519958496094, 192.0582275390625, 358.1400451660156, 131.32034301757812, 242.73556518554688, 204.02487182617188, 72.29425811767578, 24.316877365112305, 29.115102767944336, 46.55730438232422, 26.45720672607422, 190.70748901367188, -47.90454864501953, 291.1943359375, 255.13677978515625, 156.89761352539062, 9.416501998901367, 345.5697021484375, -53.22600173950195, 296.9349670410156, -35.02855682373047, 83.48357391357422, 81.13107299804688, 549.6717529296875, -22.91388702392578, -41.8408317565918, 210.29666137695312, 224.3289031982422, -228.01852416992188, -180.86666870117188, 99.94721221923828, 439.1947021484375, 63.377262115478516, 502.6401062011719, 18.309539794921875, -167.20596313476562, 55.37864685058594, 414.9582824707031, 428.6298828125, 400.7735900878906, 143.20022583007812, 390.56549072265625, -6.613067626953125, 232.24700927734375, 406.9005126953125, -3.6184768676757812, 182.56988525390625, 217.51646423339844, 242.68756103515625, -9.522819519042969, 86.69697570800781, 99.96455383300781, 277.3634948730469, 241.69140625, 106.56005859375, 51.972900390625, 759.7919311523438], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000349.npy"} +{"epoch": 0.5124816446402349, "step": 350, "batch_size": 64, "mean": 158.27227783203125, "std": 171.03952026367188, "min": -302.4638977050781, "p10": -61.51022720336913, "median": 169.8261260986328, "p90": 382.96308898925787, "max": 538.7472534179688, "pos_frac": 0.84375, "sample": [530.21240234375, -179.01498413085938, 94.80474090576172, 256.3157043457031, 252.08078002929688, 155.72410583496094, 275.3441467285156, 30.639862060546875, -45.52064514160156, 181.3785858154297, 90.85968017578125, 302.6615295410156, -183.97833251953125, 179.33770751953125, 176.2161865234375, 0.834930419921875, 538.7472534179688, 271.9577331542969, 427.86761474609375, 282.0638122558594, 204.73287963867188, 144.74147033691406, -56.739036560058594, 311.6875915527344, 73.6043701171875, 453.19305419921875, 83.04791259765625, -125.58172607421875, 81.01441192626953, 93.6016616821289, 307.10638427734375, 82.03311157226562, 161.43478393554688, 242.94638061523438, 265.8696594238281, 172.2551727294922, 87.98155975341797, 208.64170837402344, -118.306884765625, 389.28466796875, 167.39707946777344, 402.3795166015625, 181.75344848632812, -63.555023193359375, 91.5910415649414, 190.59384155273438, 368.2127380371094, 266.7547607421875, 108.63821411132812, 216.78399658203125, -302.4638977050781, 401.0245361328125, 7.9409027099609375, 222.52304077148438, 157.21788024902344, 118.61873626708984, -71.55845642089844, 21.66424560546875, 50.951805114746094, 196.18569946289062, 294.93328857421875, 71.3852310180664, -32.93165588378906, 362.30328369140625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000350.npy"} +{"epoch": 0.5139500734214391, "step": 351, "batch_size": 64, "mean": 184.771728515625, "std": 191.07598876953125, "min": -274.9679870605469, "p10": -27.248312377929686, "median": 181.08673858642578, "p90": 447.2126831054688, "max": 664.0361938476562, "pos_frac": 0.828125, "sample": [50.51129150390625, 172.30111694335938, 350.267578125, 533.8807373046875, 664.0361938476562, 140.03671264648438, 244.26197814941406, 79.05924987792969, 305.5362548828125, 245.31602478027344, 123.4244384765625, 219.09597778320312, 521.4551391601562, 511.48211669921875, 299.583984375, 324.9786682128906, 258.42706298828125, 391.97613525390625, 380.650390625, 88.1666030883789, -274.9679870605469, 148.11184692382812, 151.38711547851562, 172.16188049316406, 153.37982177734375, 206.38751220703125, 245.39251708984375, 262.3824157714844, -216.8985137939453, -13.645706176757812, -85.99874877929688, 195.6735076904297, -127.90959167480469, 232.50204467773438, -2.985504150390625, 629.7481689453125, 25.9208984375, 126.29660034179688, 128.79315185546875, 84.93675994873047, -28.221710205078125, 312.0785827636719, 42.80330276489258, 449.5127258300781, -24.97705078125, 24.902318954467773, 143.53282165527344, -2.552337646484375, 90.78271484375, -149.1111297607422, 27.559560775756836, 354.54876708984375, 89.07286071777344, 209.93035888671875, 464.53839111328125, -35.72355651855469, 255.9247283935547, 441.8459167480469, 294.1933288574219, 206.34329223632812, 189.8723602294922, 246.73204040527344, 227.2598419189453, 49.42682647705078], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000351.npy"} +{"epoch": 0.5154185022026432, "step": 352, "batch_size": 64, "mean": 126.654052734375, "std": 178.97610473632812, "min": -287.9688720703125, "p10": -83.34854202270506, "median": 113.79133605957031, "p90": 330.97838745117195, "max": 706.3220825195312, "pos_frac": 0.765625, "sample": [305.04241943359375, 299.33465576171875, 107.50770568847656, 264.0514221191406, -150.28521728515625, 147.39927673339844, 106.71515655517578, 287.56732177734375, 84.75215148925781, 153.68247985839844, 481.528564453125, 339.7188720703125, 68.51591491699219, 37.65178680419922, 229.50181579589844, -116.58445739746094, -110.26812744140625, 706.3220825195312, 182.79039001464844, -57.95336151123047, 179.39105224609375, 513.185791015625, 107.33988952636719, 311.70074462890625, 31.252593994140625, -14.254280090332031, -35.94680404663086, -201.61441040039062, 66.1934585571289, 52.78952407836914, 132.82864379882812, 408.81634521484375, 17.255733489990234, 20.22411346435547, 339.240234375, 175.58181762695312, -287.9688720703125, 105.98224639892578, 164.90548706054688, 282.0964660644531, 120.07496643066406, 45.54682922363281, 230.9292755126953, -141.90243530273438, -61.0234260559082, -68.78799438476562, -88.2244873046875, 176.22781372070312, 310.39990234375, -35.46971130371094, 176.98216247558594, -48.227989196777344, -71.9713363647461, 299.83123779296875, 220.5247802734375, 162.5008544921875, 264.30322265625, 46.54456329345703, 350.77496337890625, 98.38933563232422, 132.46871948242188, 51.24421691894531, 154.082763671875, 44.650394439697266], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000352.npy"} +{"epoch": 0.5168869309838473, "step": 353, "batch_size": 64, "mean": 141.15277099609375, "std": 170.62257385253906, "min": -286.02374267578125, "p10": -48.84786834716796, "median": 114.19855117797852, "p90": 372.7220794677735, "max": 479.9564208984375, "pos_frac": 0.78125, "sample": [203.82037353515625, 173.98648071289062, 87.39665985107422, 353.00897216796875, 147.95941162109375, 299.9319763183594, 70.88063049316406, -52.570526123046875, 353.542724609375, 64.53314208984375, 148.58132934570312, -21.68573760986328, 98.84437561035156, 358.107421875, 94.59103393554688, 214.7080078125, -71.85014343261719, -10.019317626953125, 476.6233215332031, 378.856689453125, 121.91204833984375, 252.88746643066406, 18.16305160522461, 61.5455322265625, -286.02374267578125, 353.12188720703125, -198.85043334960938, 81.46370697021484, 24.307090759277344, 263.7070617675781, -33.51454162597656, 15.015573501586914, 252.41119384765625, 104.88258361816406, 47.0323486328125, -109.20114135742188, 122.10919189453125, -63.87791442871094, 387.7242736816406, 1.1382389068603516, 83.705078125, 479.9564208984375, 55.48186492919922, 100.4935073852539, 117.14144134521484, 174.4333038330078, 220.81912231445312, 119.89646911621094, -20.4227294921875, 451.3769226074219, 144.12234497070312, -18.81781005859375, 358.4079895019531, 311.1045227050781, 467.814453125, 410.22027587890625, -40.16166687011719, 218.7490234375, 297.95263671875, -57.8658447265625, 16.709556579589844, 111.25566101074219, -32.7449951171875, 278.9491882324219], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000353.npy"} +{"epoch": 0.5183553597650514, "step": 354, "batch_size": 64, "mean": 170.5706787109375, "std": 130.93946838378906, "min": -95.53727722167969, "p10": 3.8201721191406253, "median": 159.34109497070312, "p90": 339.9418548583985, "max": 443.33917236328125, "pos_frac": 0.90625, "sample": [258.59857177734375, 304.0196228027344, 248.02085876464844, 6.45880126953125, 325.7267150878906, 191.997802734375, 407.4104309082031, -18.342395782470703, 211.77377319335938, -95.53727722167969, 64.8620376586914, -22.691856384277344, 53.10026550292969, 428.3138427734375, 283.0821838378906, 158.93661499023438, 57.736427307128906, 168.3936767578125, 265.985107421875, 267.4706115722656, 35.979896545410156, 319.5538635253906, 93.53900909423828, 98.25592041015625, 4.196533203125, 108.02725219726562, 200.52743530273438, 190.91287231445312, 140.87554931640625, 244.50099182128906, 346.0340576171875, -5.38072395324707, 272.8098449707031, 253.38284301757812, 159.74557495117188, 109.09791564941406, 38.32823944091797, 288.21368408203125, 85.45374298095703, -7.1688079833984375, 227.86627197265625, 79.3710708618164, 258.54046630859375, 78.26168823242188, 156.578369140625, 209.6196746826172, 73.72222137451172, 443.33917236328125, 239.50868225097656, 3.65887451171875, 68.55599212646484, 416.4486389160156, 155.6732635498047, 91.95391845703125, -35.565406799316406, 396.34521484375, 47.99535369873047, 311.3314208984375, 287.2894592285156, 118.38108825683594, 390.9892578125, 162.89617919921875, 148.84762573242188, 42.714168548583984], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000354.npy"} +{"epoch": 0.5198237885462555, "step": 355, "batch_size": 64, "mean": 170.591064453125, "std": 139.35092163085938, "min": -100.26362609863281, "p10": 18.830409240722666, "median": 154.20731353759766, "p90": 374.2175079345704, "max": 563.6793212890625, "pos_frac": 0.90625, "sample": [141.08773803710938, 423.9967041015625, 230.16497802734375, 444.4532165527344, 163.34417724609375, 177.21827697753906, 98.3097915649414, 563.6793212890625, -54.18097686767578, 172.2832794189453, 240.07620239257812, 425.523193359375, 53.74797821044922, -3.3283920288085938, 303.504638671875, 284.3712158203125, 88.67416381835938, 382.6278991699219, 80.6207504272461, 126.40067291259766, 295.221923828125, 156.34799194335938, -31.621475219726562, 45.810604095458984, 126.16928100585938, 354.571044921875, 281.968505859375, 162.13824462890625, 110.07502746582031, 13.712661743164062, 77.82461547851562, 112.50332641601562, 152.06663513183594, -72.63348388671875, 132.65518188476562, -100.26362609863281, 230.9441375732422, 409.85784912109375, 119.72146606445312, 444.9940185546875, 87.53199005126953, 224.0576171875, 38.28970718383789, 36.43794250488281, 210.57669067382812, 211.78445434570312, 150.7305450439453, 316.0663146972656, 50.31459426879883, 118.7238540649414, 354.59326171875, 170.19561767578125, 225.2910614013672, 188.34927368164062, 49.76533508300781, 30.771820068359375, 67.26335144042969, -8.678829193115234, 178.24114990234375, 71.61930847167969, 62.53424072265625, 246.21128845214844, 312.5951843261719, 159.92388916015625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000355.npy"} +{"epoch": 0.5212922173274597, "step": 356, "batch_size": 64, "mean": 105.32899475097656, "std": 180.31210327148438, "min": -258.3695373535156, "p10": -94.017276763916, "median": 106.3998031616211, "p90": 333.4492523193359, "max": 530.491455078125, "pos_frac": 0.671875, "sample": [169.42535400390625, -41.764869689941406, 291.9444274902344, 88.2165756225586, 4.267833709716797, -130.90370178222656, -68.63458251953125, 287.6296081542969, -73.18710327148438, 164.28271484375, 319.56549072265625, 8.30935287475586, 530.491455078125, 401.5760803222656, -44.51922607421875, 333.9278259277344, -56.76188278198242, -55.64207458496094, -258.3695373535156, -97.68612670898438, -13.437507629394531, 432.5036315917969, -28.065139770507812, 54.732696533203125, 274.1825866699219, 332.33258056640625, 6.854530334472656, 284.7414855957031, -175.68994140625, 124.5830307006836, -68.96034240722656, 428.6878662109375, 173.43789672851562, -42.79450225830078, 229.2019805908203, 140.65174865722656, 281.6239013671875, 286.60845947265625, -186.7389678955078, 55.02976989746094, 43.07353591918945, -85.45662689208984, 137.65042114257812, 124.85733032226562, 7.851402282714844, 170.70037841796875, 305.259765625, -106.15245056152344, 388.9495849609375, 125.26345825195312, 131.96868896484375, -7.341094970703125, -69.60234069824219, -33.67546844482422, 139.1414794921875, 77.60907745361328, 74.37413024902344, 164.75408935546875, -247.952880859375, 413.3172912597656, 164.04112243652344, 273.29058837890625, 4.688224792480469, 182.79232788085938], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000356.npy"} +{"epoch": 0.5227606461086637, "step": 357, "batch_size": 64, "mean": 160.44131469726562, "std": 141.2873992919922, "min": -189.76220703125, "p10": 0.2093917846679706, "median": 138.59197235107422, "p90": 350.53337402343755, "max": 555.341064453125, "pos_frac": 0.890625, "sample": [276.7364807128906, 316.79681396484375, 136.4082489013672, 397.38818359375, 302.3824157714844, 87.87171936035156, 142.4099884033203, 281.6246337890625, 72.01251220703125, 291.13970947265625, 230.0320587158203, 98.40316009521484, -11.468971252441406, 357.6043701171875, -16.820770263671875, 96.27696990966797, 213.57492065429688, 66.12112426757812, 236.3971710205078, -79.81678771972656, 471.44873046875, 102.5473861694336, 62.683197021484375, 14.243759155273438, -189.76220703125, -0.5823631286621094, 140.77569580078125, 120.52669525146484, 32.05445098876953, 352.99932861328125, 344.77947998046875, 24.93242645263672, 397.1393127441406, 60.95892333984375, 186.0814971923828, 2.0568199157714844, 115.55521392822266, 555.341064453125, 197.11181640625, 307.01812744140625, 104.15811157226562, -26.632043838500977, 229.93350219726562, 237.68418884277344, 116.17310333251953, 40.196842193603516, 200.7814178466797, 207.93643188476562, 43.99831771850586, 148.54913330078125, -25.253568649291992, 255.00343322753906, 147.58441162109375, 38.77171325683594, 285.7526550292969, 131.97235107421875, 95.50323486328125, 253.634521484375, 156.15640258789062, 179.53419494628906, 96.77748107910156, 433.2669982910156, 80.63711547851562, 43.141510009765625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000357.npy"} +{"epoch": 0.5242290748898678, "step": 358, "batch_size": 64, "mean": 137.39901733398438, "std": 146.2180633544922, "min": -162.11837768554688, "p10": -12.504786682128906, "median": 127.90589904785156, "p90": 343.7773529052737, "max": 587.9893188476562, "pos_frac": 0.84375, "sample": [55.136688232421875, 369.6024475097656, 90.74397277832031, 221.71385192871094, 144.18719482421875, 185.13592529296875, 46.5780029296875, 141.30142211914062, 79.76834106445312, 110.7978515625, -69.03885650634766, 84.46249389648438, -20.326080322265625, -13.053512573242188, -6.103410720825195, 482.2638854980469, 15.305763244628906, 149.9469451904297, 39.8131103515625, 38.10440444946289, 62.448509216308594, -4.0874176025390625, 7.313995361328125, 106.70369720458984, 124.94134521484375, -162.11837768554688, 250.57290649414062, -117.8612060546875, 470.1255798339844, 375.9935302734375, 249.6195526123047, 180.72869873046875, 283.518798828125, 17.349327087402344, 420.1885986328125, 145.25894165039062, 110.23798370361328, 94.8697509765625, -73.68791961669922, 130.87045288085938, 207.7220001220703, 211.79080200195312, 25.070831298828125, 185.8428955078125, 265.16864013671875, -13.326324462890625, 258.3882751464844, 418.7502746582031, 83.42149353027344, 77.48439025878906, 165.99432373046875, 159.74105834960938, -11.22442626953125, 134.03924560546875, 275.1177673339844, 587.9893188476562, 6.686868667602539, 205.18133544921875, 171.3385467529297, 19.700439453125, 205.63491821289062, 142.10723876953125, 147.7744598388672, 43.81541442871094], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000358.npy"} +{"epoch": 0.5256975036710719, "step": 359, "batch_size": 64, "mean": 157.00833129882812, "std": 171.7235565185547, "min": -139.8009033203125, "p10": -53.44443969726562, "median": 116.32104873657227, "p90": 367.9344665527344, "max": 687.4493408203125, "pos_frac": 0.875, "sample": [-102.24168395996094, 210.59841918945312, 11.5023193359375, 286.5010070800781, -139.8009033203125, 501.322998046875, -51.82940673828125, 687.4493408203125, 15.993644714355469, -87.58668518066406, -54.27912902832031, -55.38538360595703, 282.544921875, 276.9855651855469, -54.1365966796875, 98.79780578613281, 222.06790161132812, -96.37408447265625, 370.7392578125, 70.93282318115234, 125.2565689086914, 157.01409912109375, 361.38995361328125, 49.33745574951172, 134.3704376220703, 192.25440979003906, 192.12261962890625, 212.75184631347656, 94.67828369140625, 67.55382537841797, 356.41485595703125, 274.2728271484375, 647.8143920898438, 192.72215270996094, 5.2646942138671875, 140.54954528808594, 263.9864196777344, 87.93370819091797, 39.85625457763672, 438.3170471191406, 83.75933837890625, 214.87734985351562, 100.83259582519531, 137.07260131835938, 1.18896484375, 103.51844787597656, 304.8930358886719, 148.5726318359375, 70.88050079345703, 383.141845703125, 17.072959899902344, 83.96842956542969, 10.407485961914062, 216.56222534179688, 103.9737548828125, 301.9716796875, 107.38552856445312, 56.4172477722168, 495.7838439941406, 241.45941162109375, 46.884361267089844, 283.9395446777344, 44.73161315917969, 61.573524475097656], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000359.npy"} +{"epoch": 0.527165932452276, "step": 360, "batch_size": 64, "mean": 138.013427734375, "std": 134.17819213867188, "min": -128.84774780273438, "p10": -12.585195922851556, "median": 105.172607421875, "p90": 308.9957702636719, "max": 483.51318359375, "pos_frac": 0.875, "sample": [24.334869384765625, 232.25323486328125, 7.321922302246094, 111.73736572265625, -52.51634979248047, 329.217529296875, 118.97476959228516, 195.75189208984375, 369.7983703613281, 63.66459655761719, 95.6611099243164, 220.14410400390625, 9.897705078125, 160.28890991210938, 368.1161193847656, 87.98057556152344, 19.652099609375, 13.039413452148438, 310.0374755859375, 125.26211547851562, 256.4734191894531, 185.41156005859375, -28.76494789123535, 254.68565368652344, 84.91900634765625, 237.03028869628906, 298.260498046875, 137.98167419433594, 66.78431701660156, 381.1653747558594, 461.2335510253906, 206.87295532226562, 213.8992462158203, -112.65339660644531, 34.211509704589844, 67.79164123535156, 77.57110595703125, 72.77743530273438, 306.56512451171875, -15.51434326171875, 24.408283233642578, 483.51318359375, -128.84774780273438, 66.638427734375, -5.750518798828125, 98.60784912109375, -44.60430145263672, 59.81475830078125, 212.9310302734375, -25.219402313232422, 35.82643127441406, 244.3211669921875, 73.58675384521484, 261.7198791503906, 198.76287841796875, 233.73902893066406, 283.1605224609375, 61.52696990966797, 141.14312744140625, 195.5759735107422, 39.26033020019531, 32.58550262451172, 59.41173553466797, 233.42825317382812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000360.npy"} +{"epoch": 0.5286343612334802, "step": 361, "batch_size": 64, "mean": 133.99050903320312, "std": 155.7144012451172, "min": -191.24758911132812, "p10": -48.930152893066406, "median": 115.79792404174805, "p90": 364.5839874267579, "max": 513.8407592773438, "pos_frac": 0.828125, "sample": [-22.79648780822754, 119.1434555053711, 43.23603439331055, 154.542724609375, -25.26146697998047, -46.82746124267578, 52.26317596435547, 146.89295959472656, 438.3493347167969, 125.00840759277344, 163.240966796875, 189.64614868164062, -172.83554077148438, 377.99932861328125, -21.986587524414062, -49.83130645751953, 143.97509765625, 111.06761169433594, -140.05699157714844, 112.452392578125, 204.66189575195312, 13.346687316894531, 191.15097045898438, 35.361331939697266, 195.5906219482422, 435.3815002441406, 80.91875457763672, 311.3083801269531, 106.8748779296875, 132.6683349609375, 32.60029602050781, 74.25941467285156, 156.06336975097656, 30.159759521484375, 458.59698486328125, 310.0537109375, -93.25288391113281, 430.71258544921875, 513.8407592773438, 167.9473114013672, 157.3260955810547, 326.8495788574219, 400.8938903808594, 255.52529907226562, 88.21174621582031, 151.46060180664062, 79.86193084716797, 105.71176147460938, 24.989227294921875, -68.0815658569336, 147.674072265625, 101.74465942382812, 333.2815246582031, 81.35868835449219, 57.21086120605469, 293.5684814453125, 172.91131591796875, -61.78340148925781, 207.62860107421875, 33.27809143066406, 4.583251953125, -191.24758911132812, 99.59284973144531, 286.37548828125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000361.npy"} +{"epoch": 0.5301027900146843, "step": 362, "batch_size": 64, "mean": 187.4298553466797, "std": 155.03260803222656, "min": -105.56195831298828, "p10": 16.558513641357422, "median": 172.48998260498047, "p90": 342.78150329589846, "max": 661.310791015625, "pos_frac": 0.90625, "sample": [62.85116958618164, 225.01919555664062, 404.0784912109375, 300.2854919433594, 158.47708129882812, 18.24744415283203, 661.310791015625, 253.0933837890625, 152.88809204101562, 316.1553649902344, 336.6343078613281, 50.272544860839844, 120.46805572509766, 274.3194885253906, 231.0299835205078, 315.3141784667969, 74.96727752685547, -105.56195831298828, 121.77459716796875, 225.38755798339844, 164.55726623535156, 47.844207763671875, 36.477333068847656, 115.54318237304688, -31.856063842773438, -4.630769729614258, 108.05276489257812, 267.8425598144531, -94.94527435302734, 554.2818603515625, 415.9789123535156, 63.179107666015625, 54.355682373046875, 240.27716064453125, 102.92259216308594, 251.5221405029297, 345.416015625, -23.600688934326172, 261.0098876953125, 250.5346221923828, 179.00746154785156, 21.99188804626465, 480.7410888671875, 306.7927551269531, 94.3138198852539, 102.67279815673828, 297.3907775878906, 159.05743408203125, 54.86802673339844, 542.1492919921875, 170.80319213867188, 310.5283203125, 335.0751647949219, 174.17677307128906, 124.7168197631836, 274.5154724121094, 317.69561767578125, -24.050079345703125, 124.4559555053711, 198.01351928710938, 200.26153564453125, 190.88668823242188, 15.834686279296875, 21.836517333984375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000362.npy"} +{"epoch": 0.5315712187958884, "step": 363, "batch_size": 64, "mean": 201.9084014892578, "std": 160.75155639648438, "min": -195.12008666992188, "p10": 31.550781059265137, "median": 204.9633331298828, "p90": 387.2510894775391, "max": 617.147705078125, "pos_frac": 0.9375, "sample": [-195.12008666992188, 229.98858642578125, 364.5995788574219, 234.37054443359375, -78.45249938964844, 171.982421875, 218.994384765625, 201.48712158203125, 130.5358428955078, 137.75575256347656, 280.5572814941406, 93.22233581542969, 314.5433044433594, 87.61869049072266, 208.8784637451172, 564.0972900390625, -177.27845764160156, 315.9375915527344, 106.84915161132812, 507.7616882324219, 310.36016845703125, 178.83346557617188, 202.4908447265625, 285.2003173828125, 126.76367950439453, 251.1470184326172, 617.147705078125, 214.0897979736328, 43.8345947265625, 387.86102294921875, 101.36112976074219, 243.66661071777344, 66.19744873046875, 143.73971557617188, 385.8279113769531, 105.12708282470703, 176.75375366210938, 48.736846923828125, 366.510498046875, 397.6913146972656, 339.95965576171875, 33.444740295410156, 21.403564453125, 379.8846740722656, 30.739084243774414, 285.91412353515625, 354.0389404296875, 19.42742919921875, 288.8511962890625, 404.9838562011719, -10.171028137207031, 133.92269897460938, 64.88269805908203, 264.153076171875, 312.0243225097656, 37.72321319580078, 39.486289978027344, 207.43582153320312, 84.42669677734375, 125.49170684814453, 490.571044921875, 324.2275390625, 103.80297088623047, 213.8438720703125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000363.npy"} +{"epoch": 0.5330396475770925, "step": 364, "batch_size": 64, "mean": 148.92691040039062, "std": 156.99034118652344, "min": -225.24642944335938, "p10": -36.30822372436522, "median": 146.07666015625, "p90": 341.3324615478516, "max": 583.6803588867188, "pos_frac": 0.828125, "sample": [230.6024169921875, 284.5227966308594, -218.8582000732422, 35.853233337402344, -225.24642944335938, 323.3197021484375, -99.84622192382812, 223.36572265625, 279.5823669433594, 583.6803588867188, 144.08290100097656, 349.0522155761719, 295.0351257324219, 158.77528381347656, -93.58988189697266, 34.55329895019531, 22.220672607421875, -19.139739990234375, 221.55917358398438, -17.866729736328125, 126.01409912109375, 224.89132690429688, 157.47018432617188, 231.1648406982422, 165.61492919921875, 42.289703369140625, 42.12563705444336, 384.77630615234375, 211.17247009277344, 223.6527099609375, 79.05641174316406, 393.1748046875, 91.8821792602539, 146.15646362304688, 141.28494262695312, 53.96875, 194.19442749023438, 85.5264892578125, 286.1645812988281, -51.24102783203125, -13.164588928222656, 371.8173828125, 365.9805908203125, 60.29527282714844, 148.0393524169922, -43.66614532470703, 86.03319549560547, 175.93057250976562, 321.18829345703125, 118.34231567382812, 58.85993576049805, 206.658203125, 7.768583297729492, 195.7510986328125, 522.79052734375, 119.19979858398438, 145.99685668945312, 253.76055908203125, -1.255960464477539, 286.24249267578125, 143.38650512695312, 86.28787231445312, 267.0728759765625, -92.99127197265625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000364.npy"} +{"epoch": 0.5345080763582967, "step": 365, "batch_size": 64, "mean": 148.58676147460938, "std": 164.40951538085938, "min": -86.33917236328125, "p10": -29.091649627685545, "median": 106.65069961547852, "p90": 349.59660949707046, "max": 754.14990234375, "pos_frac": 0.859375, "sample": [164.63714599609375, -19.472068786621094, 32.605804443359375, 110.86073303222656, -29.827804565429688, 68.35009765625, 240.33786010742188, -27.37395477294922, 111.55530548095703, 6.839385986328125, 277.99908447265625, 275.706787109375, 2.4674816131591797, 258.7409973144531, -36.00193786621094, 36.940277099609375, 53.36650085449219, 638.0179443359375, 366.3631286621094, 267.1496276855469, 39.38408660888672, 46.36310577392578, 68.58333587646484, 6.44482421875, 83.67967224121094, -86.33917236328125, 104.16067504882812, 116.866943359375, 151.784912109375, 66.5804443359375, -54.99458312988281, -47.78341293334961, 103.20658111572266, 30.998889923095703, 238.99925231933594, 385.58056640625, 52.30298614501953, 101.79667663574219, 90.66873168945312, 136.6265869140625, 187.43751525878906, 5.303504943847656, 236.903564453125, 0.7208652496337891, 109.1407241821289, 309.10382080078125, 264.96502685546875, 166.60055541992188, 214.48477172851562, -52.820892333984375, 94.09851837158203, 255.78366088867188, 18.962976455688477, 242.76300048828125, 189.7745819091797, 207.90866088867188, 754.14990234375, 436.0560607910156, 248.57073974609375, -55.54607009887695, 464.6868896484375, 385.62835693359375, 80.22874450683594, 310.4747314453125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000365.npy"} +{"epoch": 0.5359765051395007, "step": 366, "batch_size": 64, "mean": 154.0113983154297, "std": 152.56243896484375, "min": -103.65695190429688, "p10": -38.7870262145996, "median": 141.99378204345703, "p90": 327.9369750976563, "max": 608.2517700195312, "pos_frac": 0.828125, "sample": [-42.699790954589844, 57.93048095703125, 175.93124389648438, 70.10446166992188, -62.368011474609375, 24.458683013916016, 30.266759872436523, 67.10873413085938, 128.84356689453125, 308.02056884765625, 154.88223266601562, -103.65695190429688, 395.5658874511719, 389.3246154785156, 199.8931884765625, 97.99729919433594, 201.58950805664062, 14.800487518310547, 16.18794822692871, -26.215465545654297, 319.4126281738281, 222.79168701171875, 215.6031494140625, 25.929840087890625, 312.92144775390625, 608.2517700195312, 319.5741882324219, 99.3176040649414, 225.64505004882812, 151.59986877441406, 160.221435546875, -29.657241821289062, -28.612571716308594, 232.50558471679688, 274.10174560546875, 256.53961181640625, 565.8226928710938, 236.65493774414062, 126.14261627197266, 115.61915588378906, 190.96237182617188, 143.685791015625, 67.51263427734375, -82.61123657226562, 348.84466552734375, -44.290504455566406, 232.70831298828125, 39.2784309387207, 262.10772705078125, 331.5210266113281, 260.3819274902344, 131.86099243164062, 248.51168823242188, 129.9320068359375, 439.14019775390625, 221.43508911132812, -53.22538757324219, 100.35520935058594, -1.60430908203125, 90.85781860351562, 40.58733367919922, 140.30177307128906, 213.12374877929688, -102.99842834472656], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000366.npy"} +{"epoch": 0.5374449339207048, "step": 367, "batch_size": 64, "mean": 123.6561050415039, "std": 129.8751220703125, "min": -190.05923461914062, "p10": -21.280994415283196, "median": 103.01998901367188, "p90": 308.2210876464845, "max": 415.1348571777344, "pos_frac": 0.859375, "sample": [12.00518798828125, 58.193939208984375, 338.6497802734375, 136.3279571533203, 415.1348571777344, 77.445068359375, 80.88323974609375, 107.666748046875, 274.72265625, 3.6394424438476562, 35.170135498046875, 15.221467971801758, 31.35174560546875, 257.655517578125, 225.0559539794922, 144.63922119140625, 202.59002685546875, 54.69187927246094, 83.04441833496094, -53.68257141113281, 155.854248046875, 364.03302001953125, -23.57428741455078, 12.191551208496094, 126.95735931396484, 250.8013916015625, 378.5224304199219, 106.58528137207031, 321.92138671875, 81.47479248046875, 346.4818115234375, 39.92445373535156, -1.3747520446777344, 213.74000549316406, 62.72785949707031, 166.17333984375, 177.84605407714844, 270.65679931640625, 276.25372314453125, 105.14140319824219, 212.02613830566406, -93.07516479492188, 19.131214141845703, 40.583099365234375, 42.4290771484375, -89.2216567993164, -15.929977416992188, 150.43045043945312, 225.74673461914062, 25.541595458984375, -38.51909637451172, -190.05923461914062, 135.12680053710938, 22.9993896484375, 173.8180389404297, 222.26414489746094, 224.82496643066406, 87.55879211425781, 273.78045654296875, 28.833938598632812, 100.89857482910156, -31.860191345214844, 365.64593505859375, 88.27218627929688], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000367.npy"} +{"epoch": 0.5389133627019089, "step": 368, "batch_size": 64, "mean": 173.62509155273438, "std": 121.54183197021484, "min": -77.99141693115234, "p10": 23.564829063415527, "median": 159.352294921875, "p90": 304.3258880615234, "max": 562.285400390625, "pos_frac": 0.9375, "sample": [228.94598388671875, 207.75210571289062, 187.27767944335938, 139.30841064453125, 193.80389404296875, 248.3839111328125, 102.68637084960938, -44.41828918457031, 356.8205871582031, 98.2291488647461, 80.21408081054688, 260.07000732421875, 208.92897033691406, 76.34646606445312, 257.2138671875, 252.3095703125, 118.49614715576172, 238.02186584472656, -50.395912170410156, 220.35235595703125, 127.49518585205078, 131.29844665527344, 330.3608703613281, 218.32492065429688, 246.03445434570312, 53.225982666015625, 140.5015106201172, -11.120071411132812, 28.366497039794922, -77.99141693115234, 70.64762878417969, 294.7244873046875, 211.482421875, 142.471435546875, 116.7659912109375, 237.76345825195312, 158.91708374023438, 6.834077835083008, 137.9120635986328, 19.61774444580078, 159.78750610351562, 369.9487609863281, 243.53451538085938, 277.6080627441406, 120.51268005371094, 263.6565856933594, 491.53253173828125, 23.558822631835938, 23.57884407043457, 115.05853271484375, 74.96637725830078, 349.6969909667969, 182.2901153564453, 304.87054443359375, 303.0550231933594, 116.82504272460938, 126.73506927490234, 562.285400390625, 282.0008544921875, 239.11312866210938, 154.54025268554688, 107.31480407714844, 46.17168045043945, 209.38462829589844], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000368.npy"} +{"epoch": 0.540381791483113, "step": 369, "batch_size": 64, "mean": 131.1334228515625, "std": 141.46649169921875, "min": -169.52430725097656, "p10": -70.97810401916502, "median": 118.83595657348633, "p90": 314.276611328125, "max": 491.4420471191406, "pos_frac": 0.8125, "sample": [79.70272827148438, 54.12279510498047, 359.447021484375, -31.314001083374023, -143.0506134033203, 132.0833740234375, 44.45941925048828, 271.7594299316406, 29.73647689819336, 328.8456115722656, 233.60560607910156, 2.2559814453125, 216.80746459960938, 188.226318359375, 197.3211212158203, 491.4420471191406, 243.23486328125, 212.71951293945312, -53.915462493896484, 220.01657104492188, 279.1249084472656, 117.73658752441406, 131.91470336914062, 119.9353256225586, 201.9373779296875, 117.4678955078125, -118.28658294677734, 228.9576416015625, 200.10067749023438, 55.86968231201172, 211.39349365234375, 318.0394287109375, 211.06283569335938, -82.57615661621094, 70.25094604492188, 80.73101043701172, 273.45172119140625, 123.36894989013672, 109.11305236816406, 333.8290100097656, 317.07958984375, -78.29066467285156, 390.7283020019531, 99.82566833496094, 296.47650146484375, 222.037841796875, 197.79000854492188, -105.07400512695312, 3.7521629333496094, 307.736328125, 108.61328125, 78.82902526855469, -113.89801025390625, 59.09900665283203, 73.84380340576172, 42.88006591796875, -169.52430725097656, -16.161048889160156, 112.3243179321289, 240.1455078125, 181.51681518554688, -17.25513458251953, 102.67178344726562, -3.5365524291992188], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000369.npy"} +{"epoch": 0.5418502202643172, "step": 370, "batch_size": 64, "mean": 110.11068725585938, "std": 133.64732360839844, "min": -187.42494201660156, "p10": -55.6012954711914, "median": 121.07450485229492, "p90": 293.59181213378906, "max": 419.61322021484375, "pos_frac": 0.75, "sample": [144.70660400390625, 264.0460510253906, 86.68363952636719, 29.874954223632812, 419.61322021484375, 43.061126708984375, 154.0946044921875, 132.0249786376953, 4.04334831237793, 53.14220428466797, 224.7460174560547, 179.81112670898438, 20.8519287109375, 131.32760620117188, -49.066192626953125, 294.24755859375, 251.30816650390625, -3.9769630432128906, 173.3350067138672, 324.7153625488281, 313.3130798339844, 10.654205322265625, 294.9784240722656, 31.232215881347656, 172.22225952148438, 44.298362731933594, 107.6654052734375, 89.95808410644531, -63.646141052246094, 124.1432113647461, 292.0617370605469, -35.424102783203125, -0.8002471923828125, 128.32737731933594, -15.389312744140625, 157.69650268554688, 212.4422607421875, -9.894561767578125, -95.35151672363281, 20.998106002807617, 127.75167846679688, 143.17984008789062, -187.42494201660156, 138.14718627929688, 263.06988525390625, 118.00579833984375, 44.91046142578125, 304.815185546875, 239.00914001464844, -140.5108642578125, -58.40205383300781, -23.845245361328125, 79.86184692382812, 214.64166259765625, -7.067604064941406, -13.734947204589844, 18.48114013671875, -109.09268951416016, -60.94912338256836, 252.84764099121094, 356.10491943359375, 194.41558837890625, 262.470458984375, 232.32354736328125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000370.npy"} +{"epoch": 0.5433186490455213, "step": 371, "batch_size": 64, "mean": 137.8328857421875, "std": 149.52638244628906, "min": -253.3361053466797, "p10": -43.351550292968746, "median": 139.32108306884766, "p90": 341.61909179687507, "max": 525.6361694335938, "pos_frac": 0.796875, "sample": [255.94039916992188, 235.42742919921875, 206.96327209472656, 30.344655990600586, -253.3361053466797, 128.41973876953125, 412.53131103515625, 148.90115356445312, 349.19940185546875, 211.9379119873047, 108.73468017578125, 13.126150131225586, 212.3636016845703, 182.359130859375, -44.686553955078125, 59.10560607910156, 74.46246337890625, 116.98165893554688, 360.95098876953125, 139.493408203125, -19.68537712097168, -127.97917175292969, 277.144287109375, -67.99581909179688, -62.87652587890625, 222.73130798339844, 373.819580078125, 210.003662109375, 114.23371887207031, 135.98524475097656, 525.6361694335938, -11.037124633789062, 183.81484985351562, 150.3509063720703, -11.670005798339844, -23.693618774414062, 140.035400390625, -24.71491241455078, 88.32620239257812, 360.5, -48.647430419921875, 102.93358612060547, 295.5631408691406, 157.3627166748047, 25.46877670288086, 227.228759765625, -40.236541748046875, 210.72686767578125, 323.93170166015625, 15.524337768554688, 233.09942626953125, 76.19365692138672, 139.1487579345703, 169.94888305664062, 197.4323272705078, 76.27069091796875, 167.08006286621094, 68.04243469238281, -149.28688049316406, 293.4774169921875, 296.5068359375, 438.34295654296875, 118.36231994628906, 44.68096923828125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000371.npy"} +{"epoch": 0.5447870778267254, "step": 372, "batch_size": 64, "mean": 187.437255859375, "std": 185.10679626464844, "min": -158.84194946289062, "p10": -30.800089836120602, "median": 154.94337463378906, "p90": 486.05984802246104, "max": 614.7416381835938, "pos_frac": 0.84375, "sample": [239.14952087402344, 157.21824645996094, 355.5766296386719, -45.13356018066406, 40.10362243652344, 464.1290283203125, 418.88677978515625, 247.1536865234375, 192.712646484375, 2.5973358154296875, 614.7416381835938, 303.81805419921875, 116.24945068359375, 250.88790893554688, -28.746978759765625, 118.37520599365234, 44.36313247680664, 128.18527221679688, 516.6458129882812, -135.3224639892578, 90.13609313964844, 210.07850646972656, 423.9367370605469, -85.3737564086914, 495.4587707519531, 260.8777160644531, 38.46410369873047, 143.55252075195312, 55.54974365234375, 105.29400634765625, 18.62139892578125, 145.34869384765625, 344.74774169921875, -87.72863006591797, 150.8646240234375, -77.1386947631836, 338.76751708984375, 513.9913330078125, 119.3499526977539, 511.8179016113281, 317.5126037597656, 273.7803649902344, 207.5446014404297, -158.84194946289062, 312.7225341796875, 422.4203186035156, 233.75514221191406, 152.6685028076172, -24.436721801757812, 250.55474853515625, 78.33615112304688, 247.3605499267578, 172.23892211914062, 20.634557723999023, 533.2529296875, 179.08436584472656, -31.679994583129883, 128.56939697265625, 25.228883743286133, 516.70947265625, 35.67732620239258, -7.356636047363281, 283.81573486328125, 108.22652435302734], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000372.npy"} +{"epoch": 0.5462555066079295, "step": 373, "batch_size": 64, "mean": 170.8305206298828, "std": 156.42361450195312, "min": -192.38499450683594, "p10": -3.3271141052246036, "median": 161.26331329345703, "p90": 390.68140563964846, "max": 458.2785339355469, "pos_frac": 0.890625, "sample": [2.3929367065429688, -100.78068542480469, 6.98175048828125, 53.06262969970703, 105.49742126464844, 444.9574279785156, 345.02630615234375, 18.573471069335938, 143.401123046875, 112.14093780517578, 458.2785339355469, 166.7331085205078, -15.426116943359375, 39.764305114746094, 270.1260986328125, 128.0023956298828, 15.42577838897705, 425.20086669921875, 186.91476440429688, 79.22405242919922, -123.35440063476562, -5.778564453125, 180.75772094726562, -169.66961669921875, 117.11917114257812, 34.06720733642578, 155.79351806640625, 196.82046508789062, 289.4859313964844, 133.61590576171875, 46.82981872558594, 239.337646484375, 179.10928344726562, 247.85577392578125, 317.7991638183594, -35.16256332397461, 24.07170867919922, 364.75274658203125, 217.4373016357422, 103.31282043457031, 444.781005859375, 57.22441864013672, 269.6964111328125, 330.32073974609375, 348.139892578125, 276.24078369140625, 293.2248229980469, 391.3724365234375, 416.23809814453125, 287.5725402832031, 61.8116455078125, 144.70611572265625, 116.2777099609375, 191.21090698242188, 28.853744506835938, 277.80487060546875, 251.03257751464844, 97.72628784179688, 410.56536865234375, 213.7252960205078, 282.7411193847656, 389.0690002441406, -192.38499450683594, 145.50482177734375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000373.npy"} +{"epoch": 0.5477239353891337, "step": 374, "batch_size": 64, "mean": 157.87384033203125, "std": 172.6921844482422, "min": -117.59501647949219, "p10": -12.126396942138665, "median": 120.13207626342773, "p90": 406.5407958984375, "max": 655.6888427734375, "pos_frac": 0.84375, "sample": [7.288764953613281, 158.57437133789062, 4.1179656982421875, 597.57666015625, 140.49354553222656, 655.6888427734375, 78.6794204711914, 381.2626037597656, 161.0281219482422, 17.598838806152344, 55.59202575683594, 156.39974975585938, 26.912796020507812, -6.479682922363281, 48.300079345703125, -3.4323806762695312, 335.7572021484375, 237.11068725585938, 391.9112548828125, 83.0284423828125, 114.51773071289062, 124.3072509765625, 397.2085876464844, 27.129959106445312, -44.38637161254883, 392.5813903808594, 26.538095474243164, 45.25639724731445, -117.59501647949219, 246.3485107421875, 255.97244262695312, 217.92205810546875, -2.4381103515625, 413.95233154296875, -100.139404296875, 200.5453338623047, -19.959274291992188, 357.06451416015625, 102.1778335571289, -14.546417236328125, 105.10055541992188, 146.32437133789062, 122.27518463134766, 55.783782958984375, 117.98896789550781, 65.87271881103516, 58.86448669433594, 410.5403137207031, 254.8084259033203, 178.79449462890625, 38.68205261230469, 426.388671875, -93.36061096191406, 256.1055908203125, 191.84375, 21.51935577392578, -61.864097595214844, 322.3331604003906, 117.44706726074219, 133.21453857421875, 481.388671875, 9.955810546875, 123.52709197998047, 470.5240173339844], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000374.npy"} +{"epoch": 0.5491923641703378, "step": 375, "batch_size": 64, "mean": 153.18905639648438, "std": 177.14443969726562, "min": -244.07327270507812, "p10": -56.97854042053222, "median": 148.95359802246094, "p90": 395.6269836425782, "max": 612.1226806640625, "pos_frac": 0.828125, "sample": [402.3609619140625, 260.2785949707031, 44.404388427734375, 146.29934692382812, -49.60166931152344, -33.975223541259766, 313.42657470703125, 178.82809448242188, 2.178203582763672, 119.2861328125, 31.41267967224121, 58.84449768066406, 153.1021728515625, -184.75503540039062, 19.928016662597656, 497.2981872558594, 284.9383239746094, 32.029998779296875, 323.3482971191406, 217.4244384765625, 22.502532958984375, 300.328857421875, -153.76992797851562, 147.22323608398438, 164.982177734375, 34.64717102050781, 379.91436767578125, -3.592866897583008, -66.386474609375, 125.0809097290039, 84.42989349365234, -66.50038146972656, 432.4476013183594, 445.9417724609375, -60.14005661010742, 207.05630493164062, 362.90216064453125, 308.84918212890625, 207.05230712890625, 442.1759948730469, 150.6839599609375, 156.14878845214844, 83.46886444091797, -244.07327270507812, 257.4481201171875, 239.59140014648438, 231.04376220703125, -174.01846313476562, 612.1226806640625, 346.6299743652344, 264.8035888671875, 22.355098724365234, 1.8540802001953125, 210.65602111816406, 73.04310607910156, 419.3226623535156, 138.5824737548828, 282.4135437011719, 32.856788635253906, 47.419639587402344, 143.2759552001953, 245.1429443359375, 174.25814819335938, -43.13154220581055], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000375.npy"} +{"epoch": 0.5506607929515418, "step": 376, "batch_size": 64, "mean": 179.0716552734375, "std": 176.3414764404297, "min": -299.30999755859375, "p10": -18.2539710998535, "median": 151.6438980102539, "p90": 424.32076110839853, "max": 567.8046264648438, "pos_frac": 0.859375, "sample": [156.89633178710938, 484.80157470703125, 184.513671875, 104.92485046386719, 567.8046264648438, 71.98167419433594, 262.2672119140625, 18.25886344909668, 59.11351013183594, -42.18162536621094, 210.44464111328125, 2.710651397705078, 431.11700439453125, 73.1832275390625, -25.13574981689453, 285.583251953125, 230.41156005859375, -53.16351318359375, 22.945404052734375, 152.08213806152344, 50.81285095214844, -1.1815357208251953, -119.02023315429688, 477.32928466796875, 438.0224609375, 83.52544403076172, 133.73268127441406, 247.7117462158203, 408.4628601074219, 47.67658615112305, 349.4952697753906, 389.63323974609375, 91.82148742675781, 303.88958740234375, 213.32476806640625, 303.9314880371094, 346.0312805175781, 157.90969848632812, 241.3723907470703, 104.95718383789062, 242.7703094482422, 280.7426452636719, -2.1964874267578125, 230.32382202148438, 358.7682800292969, 90.39070129394531, 542.9390869140625, 125.2371597290039, -56.20005798339844, -35.14910125732422, 18.409255981445312, 109.0396957397461, 476.5162048339844, 370.0552978515625, 233.67471313476562, 378.7066650390625, 90.78260803222656, 151.20565795898438, 142.1307373046875, 62.875709533691406, 35.07337188720703, 52.82123947143555, 392.9812316894531, -299.30999755859375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000376.npy"} +{"epoch": 0.5521292217327459, "step": 377, "batch_size": 64, "mean": 167.6527099609375, "std": 170.15081787109375, "min": -234.9815673828125, "p10": -49.82869338989258, "median": 171.28568267822266, "p90": 384.49409484863287, "max": 515.6636962890625, "pos_frac": 0.8125, "sample": [334.8521728515625, 444.30938720703125, 207.40618896484375, 82.70529174804688, 69.2779312133789, -55.143951416015625, 254.94802856445312, 101.93328857421875, 276.2882385253906, -52.171897888183594, 397.81915283203125, 474.9572448730469, 26.35120391845703, 515.6636962890625, -48.85999298095703, 131.92495727539062, 201.8036651611328, 340.9403991699219, 173.87948608398438, -24.436368942260742, 181.73092651367188, 375.34088134765625, 165.60153198242188, -5.817890167236328, 283.7789611816406, 375.03717041015625, 270.8440856933594, -50.24385070800781, 339.85455322265625, 97.08362579345703, 306.17840576171875, 186.61549377441406, 72.1253890991211, 108.35494995117188, 13.232929229736328, 328.9474792480469, 168.69187927246094, 217.11676025390625, 188.9509735107422, 441.1981201171875, 181.62648010253906, -35.975372314453125, -234.9815673828125, 281.559814453125, 69.23068237304688, 152.3450927734375, 268.4564208984375, 83.63768005371094, -101.88566589355469, 266.8847351074219, 20.251388549804688, 370.82281494140625, -46.12648010253906, -101.12155151367188, 77.9673080444336, 388.4169006347656, 49.13825225830078, 474.6213684082031, 294.63531494140625, 11.643730163574219, 220.54383850097656, 148.98638916015625, 123.3194580078125, -153.29446411132812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000377.npy"} +{"epoch": 0.55359765051395, "step": 378, "batch_size": 64, "mean": 189.63320922851562, "std": 164.38613891601562, "min": -215.4029083251953, "p10": -21.01571369171141, "median": 188.05319213867188, "p90": 369.3300018310547, "max": 724.68310546875, "pos_frac": 0.84375, "sample": [362.5859375, 312.2829284667969, 408.3837585449219, 361.2587890625, 373.57763671875, 129.38487243652344, 158.6644287109375, 292.77545166015625, 226.74365234375, 123.77053833007812, 305.14398193359375, 188.80023193359375, 255.8365478515625, 182.5989532470703, 175.67385864257812, -8.433563232421875, 230.63229370117188, 463.46307373046875, 149.84982299804688, 306.9501037597656, 115.4453353881836, -47.26587677001953, 55.05057144165039, 106.15961456298828, 132.35848999023438, 168.29861450195312, -26.408063888549805, 334.21112060546875, 131.069580078125, 219.24346923828125, 298.4584045410156, 366.3808288574219, 154.81942749023438, 8.886014938354492, -55.02007293701172, 289.3970031738281, 88.74658966064453, 724.68310546875, 28.25800895690918, -4.781005859375, 198.8003692626953, 93.39774322509766, 229.3735809326172, 430.4448547363281, -2.5115966796875, -30.549301147460938, 298.3946838378906, 215.49513244628906, 440.6656188964844, 293.3408203125, 147.5155487060547, 264.4296875, 366.263671875, 187.30615234375, -208.277587890625, 213.55191040039062, 220.78860473632812, 160.53860473632812, -55.966819763183594, 370.59393310546875, 101.83042907714844, -215.4029083251953, 46.33954620361328, 282.2286071777344], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000378.npy"} +{"epoch": 0.5550660792951542, "step": 379, "batch_size": 64, "mean": 183.9088134765625, "std": 189.72230529785156, "min": -182.36007690429688, "p10": -45.623384857177726, "median": 164.54164123535156, "p90": 420.8578186035156, "max": 763.3139038085938, "pos_frac": 0.8125, "sample": [227.97848510742188, 389.20379638671875, 220.07858276367188, 286.37957763671875, -38.10289764404297, 45.300697326660156, 132.42581176757812, 426.52838134765625, 169.52914428710938, 111.70657348632812, 299.2843017578125, 617.0825805664062, 93.00205993652344, 308.7073669433594, 335.17584228515625, 229.99337768554688, -182.36007690429688, 547.0913696289062, 113.22074890136719, 276.35296630859375, 296.16571044921875, 271.5626525878906, 183.0972137451172, 28.970447540283203, 47.6380615234375, -12.703306198120117, -106.17050170898438, 51.05389404296875, 74.82971954345703, 204.1804656982422, 377.8494873046875, 256.85833740234375, -163.889892578125, 85.584716796875, 335.3034973144531, 458.780517578125, 241.7598114013672, 321.853759765625, 421.6336364746094, 55.36309814453125, -32.87078094482422, 458.5216064453125, 135.872802734375, -33.84431076049805, 231.30357360839844, -94.49569702148438, 88.41159057617188, 325.0744323730469, 294.8088073730469, 156.06898498535156, -48.84645080566406, 145.04946899414062, 387.85357666015625, 104.09492492675781, 763.3139038085938, 123.73526763916016, 159.55413818359375, 212.86846923828125, 43.247032165527344, -2.3945579528808594, 38.62837219238281, 419.0475769042969, -70.49496459960938, -72.64402770996094], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000379.npy"} +{"epoch": 0.5565345080763583, "step": 380, "batch_size": 64, "mean": 204.74752807617188, "std": 200.66404724121094, "min": -272.7461242675781, "p10": -71.409056854248, "median": 211.75267791748047, "p90": 443.64109497070314, "max": 666.3380126953125, "pos_frac": 0.859375, "sample": [102.13409423828125, 142.34291076660156, -30.369522094726562, 562.3509521484375, 235.57205200195312, 84.2149429321289, 170.5585479736328, 325.145263671875, 335.6735534667969, 163.90391540527344, 68.61980438232422, 407.4759521484375, -17.381263732910156, 488.80181884765625, 565.8895263671875, 329.4941711425781, 163.6138458251953, 308.18316650390625, 305.20538330078125, 62.77520751953125, 223.13641357421875, -198.57850646972656, 152.58624267578125, 165.597412109375, 145.70169067382812, -88.99742889404297, 445.12261962890625, 304.6628723144531, 192.74725341796875, 440.1842041015625, 246.17007446289062, 364.0379943847656, 263.5282287597656, 352.28765869140625, 392.2749328613281, 347.3545227050781, 58.36503601074219, 23.67132568359375, 210.85414123535156, 89.6479721069336, 666.3380126953125, 4.511615753173828, -102.34123992919922, 85.72036743164062, -141.81817626953125, 212.65121459960938, 348.67822265625, 417.3634338378906, 33.242820739746094, -218.79612731933594, 358.161865234375, 60.651153564453125, 502.21087646484375, 246.1077423095703, 123.34881591796875, -93.21839141845703, 431.4809265136719, 243.3651123046875, 101.9296646118164, 85.876220703125, -272.7461242675781, 300.96868896484375, 491.6238708496094, 313.972412109375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000380.npy"} +{"epoch": 0.5580029368575624, "step": 381, "batch_size": 64, "mean": 166.4598846435547, "std": 208.0777130126953, "min": -303.88323974609375, "p10": -75.88306045532227, "median": 138.3931884765625, "p90": 464.8360137939453, "max": 611.426513671875, "pos_frac": 0.8125, "sample": [169.81185913085938, 5.418670654296875, 18.635194778442383, 467.64654541015625, 239.594970703125, 522.6429443359375, 341.43524169921875, 36.63362121582031, 97.6942367553711, 395.03485107421875, 458.2781066894531, 158.1034393310547, 297.5797119140625, 295.8529052734375, 611.426513671875, -102.53716278076172, 347.23260498046875, 313.20025634765625, 514.2498168945312, 142.96957397460938, 558.7451782226562, 587.4378662109375, -158.437255859375, -175.85121154785156, -71.18001556396484, 176.58090209960938, 450.1258544921875, 78.63459777832031, -88.06836700439453, 32.217918395996094, 144.5219268798828, 228.73681640625, 71.08232116699219, -14.079097747802734, 87.31773376464844, -70.54896545410156, 133.81680297851562, 101.33292388916016, 132.09896850585938, 12.269744873046875, 118.3071517944336, 117.1624755859375, 581.0890502929688, 255.2901153564453, 276.6192626953125, 197.3758544921875, 64.85446166992188, -29.155025482177734, 371.36920166015625, -45.32780838012695, 102.26372528076172, 337.3854064941406, 36.689422607421875, 194.8793182373047, 80.1294174194336, -77.89865112304688, 58.280487060546875, 319.8212890625, 286.39532470703125, 77.27281951904297, 146.99717712402344, -303.88323974609375, -256.7424011230469, 196.599365234375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000381.npy"} +{"epoch": 0.5594713656387665, "step": 382, "batch_size": 64, "mean": 168.00637817382812, "std": 160.6204833984375, "min": -208.40943908691406, "p10": -9.46778564453124, "median": 143.6369171142578, "p90": 394.38841552734385, "max": 585.1895751953125, "pos_frac": 0.890625, "sample": [508.161376953125, 364.2908935546875, 143.00440979003906, 137.83792114257812, 83.7383804321289, 515.13818359375, -41.278465270996094, 40.012596130371094, 167.7264404296875, 38.885231018066406, 141.40513610839844, 240.41751098632812, 290.57208251953125, 125.44520568847656, 249.93885803222656, 254.1871337890625, 142.19740295410156, 13.417892456054688, 332.23382568359375, 62.74882507324219, 253.11172485351562, 9.074813842773438, 1.5856475830078125, 177.97235107421875, 585.1895751953125, -30.602500915527344, 296.6927185058594, -33.434356689453125, 300.5518493652344, 150.97787475585938, 407.287353515625, -67.03719329833984, 25.09976577758789, 248.37136840820312, 239.72482299804688, -208.40943908691406, -36.957122802734375, 277.4376220703125, 8.899429321289062, 334.0984802246094, 33.159202575683594, 82.6322021484375, 152.8562774658203, 24.318750381469727, 227.9234161376953, 74.73667907714844, 187.81687927246094, 289.7231140136719, 104.12080383300781, 144.26942443847656, 433.583251953125, 434.4015808105469, 198.0465087890625, 2.7816104888916016, 160.326416015625, 447.17474365234375, 47.20297622680664, 35.90603256225586, 299.08514404296875, 330.6979064941406, 114.7947006225586, 119.25041961669922, -14.204971313476562, 72.08895874023438], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000382.npy"} +{"epoch": 0.5609397944199707, "step": 383, "batch_size": 64, "mean": 181.7510223388672, "std": 210.09573364257812, "min": -497.926025390625, "p10": -71.76613693237303, "median": 178.03882598876953, "p90": 466.459619140625, "max": 667.52783203125, "pos_frac": 0.828125, "sample": [58.076026916503906, 83.97042846679688, 98.92520141601562, -80.96134185791016, 332.5185241699219, -55.92634582519531, -140.2808837890625, 485.7475280761719, 263.46240234375, 25.902240753173828, 224.67971801757812, 294.7280578613281, 25.934078216552734, 9.053726196289062, -154.50633239746094, 173.17034912109375, 667.52783203125, 10.836463928222656, 203.2601318359375, 170.5092315673828, 532.7847900390625, -160.83541870117188, 73.90362548828125, 454.86236572265625, 192.62359619140625, -61.265052795410156, 496.7887268066406, 100.84535217285156, -76.2666015625, 208.7248077392578, 659.2036743164062, 178.50636291503906, 406.6565856933594, 186.91769409179688, 291.9788818359375, 125.19477844238281, 385.44818115234375, 97.65385437011719, 242.02366638183594, 251.4892578125, 101.62454223632812, 95.61080932617188, 155.86395263671875, 135.90919494628906, 219.73110961914062, 351.4335021972656, -15.17095947265625, 394.27349853515625, 23.689010620117188, 201.78121948242188, 471.42987060546875, 201.19305419921875, -78.23194885253906, 177.5712890625, 306.064208984375, 147.56906127929688, 520.213623046875, 412.9212646484375, -1.3321495056152344, 338.42230224609375, -497.926025390625, 76.17190551757812, 249.32229614257812, 360.0651550292969], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000383.npy"} +{"epoch": 0.5624082232011748, "step": 384, "batch_size": 64, "mean": 146.58035278320312, "std": 145.58290100097656, "min": -185.67733764648438, "p10": -25.57318649291992, "median": 148.48104095458984, "p90": 308.06212158203135, "max": 571.9436645507812, "pos_frac": 0.828125, "sample": [156.20315551757812, 88.32443237304688, 426.54498291015625, -43.44386672973633, 14.595626831054688, -10.327346801757812, -98.70359802246094, -0.7806854248046875, 571.9436645507812, 40.043235778808594, 419.3438415527344, 275.9452209472656, 114.9227294921875, -22.04717254638672, 273.650390625, 54.991939544677734, -48.50593566894531, 32.8431510925293, 210.1951446533203, 235.10829162597656, 120.14407348632812, 59.84307861328125, 111.20793151855469, 169.99876403808594, 189.80978393554688, 91.98863220214844, 36.013511657714844, -69.40137481689453, -89.69762420654297, 203.26983642578125, 208.2333984375, 152.03807067871094, 24.050086975097656, 189.24618530273438, 159.50863647460938, -185.67733764648438, 341.23321533203125, 273.9960021972656, 319.175048828125, 24.757568359375, 141.9456024169922, 188.6981964111328, 219.55300903320312, 279.494384765625, 221.480224609375, 267.93450927734375, 379.29876708984375, 144.92401123046875, 111.21121215820312, 43.95793914794922, 229.6982421875, -17.947349548339844, 159.2112579345703, 462.79638671875, 74.9521484375, -27.084335327148438, 124.0416259765625, 245.32720947265625, 174.7545928955078, 40.5505485534668, 282.1319580078125, 256.5995788574219, 88.48992919921875, 268.5376281738281], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000384.npy"} +{"epoch": 0.5638766519823789, "step": 385, "batch_size": 64, "mean": 176.60423278808594, "std": 146.82525634765625, "min": -146.58218383789062, "p10": -11.930488586425756, "median": 164.6679458618164, "p90": 376.1454071044922, "max": 515.084228515625, "pos_frac": 0.890625, "sample": [45.43255615234375, 18.95386505126953, 39.409263610839844, 270.7931823730469, 368.9409484863281, 193.3203125, 326.4527893066406, 116.48074340820312, 13.197990417480469, 296.90863037109375, 196.91065979003906, 23.174177169799805, 88.7933120727539, 158.97186279296875, 379.2330322265625, -22.69983673095703, 438.8388671875, 224.24314880371094, 87.30939483642578, 112.99552154541016, 291.56878662109375, 116.99601745605469, -85.86082458496094, 263.3074951171875, -89.23541259765625, 219.29345703125, 242.66246032714844, 366.0931701660156, 77.09785461425781, 112.279052734375, 227.93565368652344, 160.6747283935547, 85.54425048828125, 515.084228515625, 458.8990783691406, 201.8861846923828, 114.656494140625, -31.940902709960938, 435.1466979980469, 204.8387908935547, -24.142410278320312, 64.09571838378906, 223.78396606445312, 280.025634765625, 180.39935302734375, 394.8653564453125, 268.3005065917969, 98.1085205078125, 350.5354309082031, 166.4074249267578, -121.66712188720703, 48.80976867675781, 135.63665771484375, -146.58218383789062, 212.83287048339844, 128.022705078125, 267.0672607421875, 151.42852783203125, 191.95835876464844, 455.94317626953125, 152.5382080078125, 254.193359375, 142.59365844726562, 162.928466796875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000385.npy"} +{"epoch": 0.5653450807635829, "step": 386, "batch_size": 64, "mean": 187.22091674804688, "std": 185.82913208007812, "min": -223.06846618652344, "p10": -37.835538482666, "median": 162.44219970703125, "p90": 448.86062011718764, "max": 702.5094604492188, "pos_frac": 0.859375, "sample": [-21.302082061767578, 252.2952423095703, 27.92902374267578, 275.7187805175781, 321.8454895019531, 198.23963928222656, 83.62226104736328, 254.66053771972656, 160.16854858398438, 316.00677490234375, 303.7164611816406, 619.7112426757812, -96.44078063964844, 6.605161666870117, 556.654296875, 204.2391357421875, 228.16995239257812, 274.8363952636719, 1.3862781524658203, -43.30620574951172, 92.68035125732422, 194.90635681152344, 81.81285858154297, 158.54165649414062, 164.71585083007812, 286.59136962890625, 130.03863525390625, 94.08609008789062, 702.5094604492188, 460.7375183105469, -25.070648193359375, 201.5063018798828, 124.51910400390625, -58.618202209472656, 214.6791534423828, 534.7030029296875, 518.4662475585938, 261.44683837890625, 531.3087158203125, 61.96631622314453, 186.44036865234375, 421.1478576660156, -95.64405822753906, 150.02406311035156, 28.7711181640625, 147.05502319335938, 88.83442687988281, 205.9052276611328, -92.65705871582031, 228.81564331054688, -223.06846618652344, 329.79296875, 84.5459976196289, 108.34381866455078, 56.106658935546875, 134.03604125976562, 351.59619140625, 337.1822204589844, -132.5817413330078, 149.70248413085938, 235.4938201904297, 154.2458038330078, 352.16302490234375, 119.60411834716797], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000386.npy"} +{"epoch": 0.566813509544787, "step": 387, "batch_size": 64, "mean": 142.4579315185547, "std": 145.78973388671875, "min": -161.67404174804688, "p10": -26.904286575317382, "median": 150.0173797607422, "p90": 345.59856567382815, "max": 556.2129516601562, "pos_frac": 0.859375, "sample": [190.78509521484375, 29.352882385253906, 167.0369415283203, 180.1514434814453, -25.900447845458984, 556.2129516601562, 128.55709838867188, 36.00254821777344, 284.1546630859375, 398.9657287597656, -161.67404174804688, 248.81109619140625, 165.17950439453125, 190.88856506347656, -108.25125122070312, 194.38375854492188, 163.1917724609375, 366.93701171875, 107.54273223876953, 19.705841064453125, 345.7091979980469, 126.96253967285156, 146.75823974609375, 500.8722229003906, 178.37863159179688, 146.06460571289062, 16.061073303222656, 146.70516967773438, 131.73651123046875, 117.23442077636719, 85.38846588134766, 178.32667541503906, 157.09982299804688, 5.537303924560547, 374.2050476074219, 176.18280029296875, 238.76039123535156, 83.86210632324219, 228.16412353515625, 274.6236877441406, 41.45134735107422, 28.639057159423828, 447.15484619140625, 171.80880737304688, 166.53170776367188, 110.31350708007812, 125.69950103759766, -108.89700317382812, 345.3404235839844, -104.84605407714844, 188.05894470214844, -27.334503173828125, -12.224372863769531, 208.6893768310547, -119.43563842773438, 164.72019958496094, 2.4574203491210938, 206.481689453125, 64.77377319335938, 0.2925872802734375, 208.9119415283203, -43.97071838378906, 153.27651977539062, 108.74772644042969], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000387.npy"} +{"epoch": 0.5682819383259912, "step": 388, "batch_size": 64, "mean": 187.45846557617188, "std": 192.05654907226562, "min": -310.1595458984375, "p10": -10.320598220825193, "median": 182.98216247558594, "p90": 403.3339569091798, "max": 848.4664306640625, "pos_frac": 0.875, "sample": [848.4664306640625, 323.0928039550781, -8.851947784423828, 358.00396728515625, 372.7610168457031, 576.0740356445312, 222.92791748046875, -127.13184356689453, 114.97026824951172, 202.62261962890625, 179.79290771484375, 564.53564453125, 245.7544403076172, 245.44180297851562, -310.1595458984375, 2.0334548950195312, 146.443115234375, 186.17141723632812, 208.22454833984375, 52.54975128173828, 153.76519775390625, 450.382568359375, 88.80573272705078, -99.83679962158203, 201.8203887939453, 267.8175354003906, 3.4047775268554688, 37.70188903808594, 58.054107666015625, 96.54574584960938, -10.950019836425781, 103.73594665527344, 231.80796813964844, 27.300201416015625, 212.5229034423828, 171.19561767578125, 146.55467224121094, 82.47736358642578, 337.37823486328125, 70.29883575439453, 187.16970825195312, 68.42594909667969, 274.4461975097656, 136.937744140625, 126.8236083984375, 162.64645385742188, 324.45965576171875, 219.01942443847656, -95.65982055664062, 303.63262939453125, 292.8575439453125, 676.9839477539062, -112.11640930175781, 278.134033203125, 117.92561340332031, 27.855377197265625, -29.518585205078125, 234.7156219482422, 416.4366455078125, 458.5660400390625, 277.172119140625, 272.6260681152344, 238.6589813232422, 104.63673400878906], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000388.npy"} +{"epoch": 0.5697503671071953, "step": 389, "batch_size": 64, "mean": 180.12693786621094, "std": 188.966552734375, "min": -172.70933532714844, "p10": -65.72170715332027, "median": 173.76023864746094, "p90": 466.90447082519535, "max": 647.8270874023438, "pos_frac": 0.8125, "sample": [132.32308959960938, 473.3575744628906, 222.3125457763672, -172.70933532714844, -107.05894470214844, 257.56634521484375, 175.54161071777344, 280.97833251953125, 198.9373016357422, 99.862548828125, 222.64743041992188, -28.42334747314453, 244.338134765625, -123.28910064697266, 63.07857131958008, 173.22079467773438, 126.0989761352539, 462.76507568359375, 340.0124206542969, -26.478973388671875, -120.82565307617188, 17.53537368774414, 468.6784973144531, -26.900699615478516, 174.2996826171875, 338.9850158691406, -121.35736083984375, 109.96263122558594, 136.65414428710938, 161.5962371826172, 273.6591491699219, 99.00436401367188, 31.130508422851562, 246.69589233398438, -105.42943572998047, 328.3687744140625, 104.13697814941406, 207.86874389648438, 277.0438537597656, 576.8631591796875, 308.54498291015625, -5.60273551940918, 277.2564392089844, 42.31724548339844, 255.28643798828125, 350.111083984375, 106.6898193359375, 157.33811950683594, 393.0677185058594, 175.78909301757812, -9.763603210449219, 478.0160827636719, 29.689163208007812, -81.70671844482422, 245.7473602294922, 499.95208740234375, 53.14654541015625, 647.8270874023438, 445.1639709472656, 47.0322151184082, 143.2188720703125, 1.1502799987792969, 239.3728790283203, 535.42822265625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000389.npy"} +{"epoch": 0.5712187958883994, "step": 390, "batch_size": 64, "mean": 162.02655029296875, "std": 201.16372680664062, "min": -314.1904296875, "p10": -81.1669692993164, "median": 159.78701782226562, "p90": 371.6567993164063, "max": 1079.0438232421875, "pos_frac": 0.828125, "sample": [180.4012908935547, 192.56663513183594, 442.76318359375, 377.01446533203125, -180.24440002441406, -62.154563903808594, -31.55548858642578, 103.16835021972656, 118.6350326538086, 226.14517211914062, 402.6148986816406, 65.75801086425781, 223.74819946289062, 285.1478271484375, 14.816963195800781, 126.30545043945312, -76.92858123779297, 1079.0438232421875, -82.9834213256836, 345.43292236328125, 203.14390563964844, 180.56259155273438, 335.74322509765625, 167.2062225341797, 109.61601257324219, 112.604736328125, 123.68480682373047, 112.832763671875, 157.99415588378906, 121.81147003173828, 64.4932861328125, 86.51739501953125, -170.84967041015625, -104.78412628173828, 161.3372344970703, 382.8792724609375, 411.5787658691406, 380.30010986328125, -264.2945556640625, 224.34890747070312, 351.3410339355469, 253.1642303466797, 165.41732788085938, 140.50839233398438, 293.0984191894531, 158.23680114746094, 135.90362548828125, 64.33598327636719, 357.4071350097656, 359.15557861328125, 244.88870239257812, 166.75645446777344, 98.94601440429688, 240.36090087890625, 308.6281433105469, -29.822683334350586, 34.195526123046875, -129.7333526611328, -314.1904296875, 198.8105010986328, 134.2069091796875, 144.47012329101562, 249.59710693359375, 197.59405517578125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000390.npy"} +{"epoch": 0.5726872246696035, "step": 391, "batch_size": 64, "mean": 114.56845092773438, "std": 167.96388244628906, "min": -265.37261962890625, "p10": -64.75706481933592, "median": 80.10137557983398, "p90": 336.00322570800785, "max": 721.8648071289062, "pos_frac": 0.765625, "sample": [-16.059101104736328, 11.092414855957031, 29.89684295654297, -10.810592651367188, 47.62108612060547, 212.01715087890625, 75.28238677978516, 53.70490264892578, 14.684070587158203, 161.47393798828125, -22.10094451904297, 128.40785217285156, -161.75152587890625, 49.882850646972656, 205.1638641357422, 48.02271270751953, 173.86585998535156, 141.8182373046875, -17.91048240661621, -53.0333251953125, 186.0392303466797, 408.00177001953125, 256.9536437988281, 276.8068542480469, -85.51343536376953, 53.60838317871094, 721.8648071289062, -25.407791137695312, 1.946258544921875, 159.3944091796875, 102.98169708251953, 82.83524322509766, 69.07752990722656, 146.60769653320312, -48.413368225097656, 74.02177429199219, 154.69192504882812, 99.60541534423828, 431.3913879394531, 131.84939575195312, -77.2409439086914, -16.111839294433594, 176.85040283203125, 139.42388916015625, 480.5271911621094, -69.78152465820312, 390.48321533203125, 5.477823257446289, -79.34269714355469, 38.27355194091797, 37.712486267089844, 54.26611328125, -119.93203735351562, 331.47674560546875, 337.9431457519531, 77.36750793457031, 472.6927795410156, 251.8662109375, 198.7797393798828, -265.37261962890625, 166.87899780273438, 248.74822998046875, 122.42416381835938, 159.35877990722656], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000391.npy"} +{"epoch": 0.5741556534508077, "step": 392, "batch_size": 64, "mean": 179.86904907226562, "std": 155.18502807617188, "min": -134.72413635253906, "p10": 16.180538368225097, "median": 157.33472442626953, "p90": 401.3127532958986, "max": 586.041259765625, "pos_frac": 0.953125, "sample": [140.9127197265625, 182.48275756835938, -63.667991638183594, 78.27993774414062, 144.9014434814453, 11.528373718261719, 316.07989501953125, 85.01661682128906, 264.8921203613281, 246.27613830566406, 355.89593505859375, -134.72413635253906, 97.97298431396484, 17.164976119995117, 86.87954711914062, 27.157712936401367, 510.7325439453125, 448.6705322265625, 165.24765014648438, 53.62729263305664, 76.95191192626953, 262.2838134765625, 294.5139465332031, 149.4217987060547, 223.3862762451172, 304.71453857421875, 254.82106018066406, 586.041259765625, 137.27267456054688, 173.68365478515625, 192.4480743408203, 33.522064208984375, 373.7855224609375, 17.548316955566406, 234.51007080078125, 128.638427734375, 205.60801696777344, 179.93612670898438, 115.88162231445312, 95.9515380859375, 169.83421325683594, 266.29351806640625, 15.758636474609375, 10.663246154785156, 413.1101379394531, 474.6706848144531, 178.07823181152344, 346.2525939941406, 255.38906860351562, -100.12312316894531, 68.80908966064453, 65.94010925292969, 85.64224243164062, 141.44873046875, 106.16493225097656, 183.65176391601562, 251.28797912597656, 66.20918273925781, 584.8651123046875, 68.17243957519531, 13.098152160644531, 47.92292785644531, 257.9427795410156, 464.2880859375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000392.npy"} +{"epoch": 0.5756240822320118, "step": 393, "batch_size": 64, "mean": 143.06805419921875, "std": 191.5281982421875, "min": -169.96084594726562, "p10": -80.61171188354493, "median": 108.5123176574707, "p90": 382.91675415039066, "max": 743.0506591796875, "pos_frac": 0.75, "sample": [108.42854309082031, 510.48101806640625, 30.638423919677734, -49.86518096923828, 219.47776794433594, 365.8413391113281, -49.452369689941406, 297.61444091796875, 268.530517578125, 189.3021240234375, 172.83334350585938, 116.83209228515625, 96.802734375, 236.53392028808594, 53.00318908691406, 50.41881561279297, 630.044189453125, 561.8548583984375, 248.08978271484375, -2.78466796875, -1.5249347686767578, 98.61103820800781, -79.88865661621094, -29.508804321289062, 9.958877563476562, 64.14834594726562, -140.3639373779297, 163.33460998535156, 108.5960922241211, 204.96343994140625, 39.22904586791992, 299.94207763671875, 389.216552734375, 172.19937133789062, 254.30882263183594, 138.4097442626953, 266.86224365234375, 97.65921783447266, 335.04315185546875, 19.379592895507812, -87.66868591308594, 182.0880126953125, 433.959716796875, -14.051582336425781, 5.933200836181641, 368.21722412109375, 234.63748168945312, -93.28274536132812, 18.038536071777344, 100.84210968017578, 69.5143814086914, 156.0124053955078, -43.74768829345703, -88.37747955322266, 15.10049819946289, 227.99545288085938, 184.63027954101562, -80.92159271240234, 158.08596801757812, -40.904457092285156, -169.96084594726562, 743.0506591796875, 504.1634216308594, -92.19895935058594], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000393.npy"} +{"epoch": 0.5770925110132159, "step": 394, "batch_size": 64, "mean": 162.78407287597656, "std": 179.14004516601562, "min": -143.56271362304688, "p10": -28.241766357421866, "median": 138.17829132080078, "p90": 406.3679412841797, "max": 714.58056640625, "pos_frac": 0.8125, "sample": [163.97605895996094, 117.84666442871094, -137.3018341064453, 249.88265991210938, 193.80770874023438, 3.38916015625, 233.74806213378906, 113.85450744628906, 397.80023193359375, 54.760284423828125, 419.7804260253906, -5.842720031738281, -122.76522827148438, 41.90635681152344, -17.533798217773438, 38.90668487548828, -114.87561798095703, 394.64031982421875, 296.1852722167969, -136.82228088378906, 445.4666748046875, 25.05908203125, 231.033203125, 243.34957885742188, 137.3853759765625, 220.8870391845703, 226.7705535888672, 120.75675964355469, -143.56271362304688, 401.0943908691406, 344.83355712890625, 454.75213623046875, -19.424240112304688, 149.41082763671875, 161.54029846191406, 41.3599853515625, 100.21526336669922, 56.05260467529297, 58.6939697265625, 403.4051818847656, 3.2319717407226562, 390.5277404785156, 78.17369079589844, 204.09873962402344, -32.02070617675781, 54.6888427734375, 441.38525390625, 407.6376953125, 109.0679931640625, 102.48782348632812, 277.3812255859375, 138.97120666503906, 714.58056640625, 516.6282958984375, 206.98141479492188, 68.21529388427734, 97.53523254394531, -11.03084945678711, -0.49965667724609375, 307.26556396484375, 161.32485961914062, -32.66517639160156, 191.5032958984375, 178.28759765625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000394.npy"} +{"epoch": 0.57856093979442, "step": 395, "batch_size": 64, "mean": 176.14202880859375, "std": 186.7259521484375, "min": -213.79733276367188, "p10": -38.78999786376953, "median": 151.60826110839844, "p90": 441.5172821044923, "max": 662.11767578125, "pos_frac": 0.84375, "sample": [268.7269287109375, 157.26226806640625, 131.99951171875, 41.57365417480469, 117.98466491699219, 312.2801513671875, 313.4857177734375, 172.81362915039062, -213.79733276367188, 233.6947479248047, 86.27867126464844, 248.45330810546875, 247.5269775390625, 58.95366668701172, 188.42575073242188, 73.19020080566406, 378.85235595703125, 269.39715576171875, 491.7951354980469, -41.370819091796875, 404.8100280761719, 84.33499908447266, 358.85467529296875, 157.5310516357422, 27.32276153564453, 88.6982421875, 137.43826293945312, 145.95425415039062, -0.6037826538085938, 57.34050750732422, 554.2965087890625, 585.5621337890625, 186.40684509277344, 119.36976623535156, 204.71298217773438, 10.032440185546875, 134.65667724609375, 98.78096008300781, 472.2695007324219, 180.97808837890625, 321.261474609375, 323.0710144042969, -174.87515258789062, -58.54859924316406, -125.06488037109375, 213.68356323242188, -32.76808166503906, 522.917724609375, 29.6147403717041, -62.27262878417969, 60.747196197509766, -172.37014770507812, 59.21410369873047, 662.11767578125, 457.24896240234375, 256.82763671875, 117.94469451904297, 347.1732177734375, 203.61732482910156, 300.8563537597656, 331.0780029296875, 35.583194732666016, 123.28417205810547, -13.525272369384766], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000395.npy"} +{"epoch": 0.580029368575624, "step": 396, "batch_size": 64, "mean": 157.95355224609375, "std": 148.50491333007812, "min": -184.0894012451172, "p10": -31.895529174804686, "median": 143.19499969482422, "p90": 349.8544036865235, "max": 638.948486328125, "pos_frac": 0.84375, "sample": [144.1486053466797, 339.1496887207031, 57.459312438964844, 154.06893920898438, 115.90313720703125, 400.33740234375, 90.70270538330078, 125.10018920898438, 140.027099609375, 216.76129150390625, 278.19464111328125, 140.92724609375, 142.24139404296875, 257.73663330078125, -42.45395278930664, 160.42955017089844, 72.01113891601562, 66.93614196777344, 63.21937561035156, 176.549560546875, 172.08932495117188, 256.77789306640625, 570.3696899414062, 185.1571044921875, 222.13473510742188, 313.74884033203125, 121.34461975097656, 150.69448852539062, 128.61508178710938, 354.442138671875, -52.30002212524414, 192.04112243652344, -30.38237762451172, 214.66580200195312, 52.213497161865234, 218.8355255126953, 151.5009765625, -86.45203399658203, 270.48223876953125, 367.79498291015625, 157.901611328125, -32.26165771484375, -31.041229248046875, 356.7230529785156, 82.8441162109375, -2.3727760314941406, 311.20635986328125, 33.110023498535156, 130.33584594726562, 210.5263671875, 438.72332763671875, 140.36676025390625, 87.12957763671875, -81.5218505859375, 100.52915954589844, 247.1059112548828, 215.17788696289062, 79.4466781616211, -44.330711364746094, -184.0894012451172, 135.58062744140625, 638.948486328125, 161.96942138671875, 83.7958984375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000396.npy"} +{"epoch": 0.5814977973568282, "step": 397, "batch_size": 64, "mean": 180.58287048339844, "std": 165.93618774414062, "min": -147.86138916015625, "p10": -9.089252853393537, "median": 166.85626983642578, "p90": 389.0917419433594, "max": 621.251953125, "pos_frac": 0.890625, "sample": [29.61203384399414, 360.5906982421875, -16.576141357421875, 248.9617156982422, 73.56523895263672, 32.60599136352539, 74.84030151367188, 159.6865234375, 406.40533447265625, 105.89192199707031, 288.7690124511719, 87.80607604980469, 284.2121887207031, 8.71356201171875, 77.82272338867188, 269.1025085449219, 20.284149169921875, 16.105926513671875, 315.5740966796875, 254.88880920410156, 170.38304138183594, 356.46502685546875, 202.06727600097656, 333.9443054199219, 608.5992431640625, -57.31965637207031, 455.4721374511719, 382.97344970703125, 16.976924896240234, 197.49874877929688, 369.943359375, 80.12995910644531, 68.33135986328125, 447.0672607421875, 282.9322814941406, 280.4839782714844, 138.29034423828125, 44.05011749267578, 122.59307098388672, 185.33448791503906, -147.86138916015625, -80.05525970458984, 91.38028717041016, -141.70803833007812, 241.15786743164062, 133.03663635253906, -35.20671844482422, 188.94805908203125, 8.38015365600586, 279.8114013671875, -28.807788848876953, 391.7138671875, 199.752197265625, 621.251953125, 163.32949829101562, 450.2595520019531, 97.76768493652344, 151.1968994140625, 317.5611267089844, 108.12484741210938, 63.77106475830078, 290.05126953125, 198.79507446289062, 209.57444763183594], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000397.npy"} +{"epoch": 0.5829662261380323, "step": 398, "batch_size": 64, "mean": 176.60476684570312, "std": 180.25575256347656, "min": -209.52273559570312, "p10": -13.663992309570308, "median": 151.0647430419922, "p90": 416.4169586181642, "max": 625.0777587890625, "pos_frac": 0.875, "sample": [84.38323974609375, -17.498397827148438, 360.1271057128906, 389.86004638671875, 24.607772827148438, 131.61944580078125, 169.2512664794922, -103.93852996826172, 61.15617370605469, 23.18726348876953, 309.64178466796875, -158.71824645996094, 431.86260986328125, 341.4013977050781, 308.0401306152344, 625.0777587890625, 183.2198028564453, 79.84716033935547, 427.7984924316406, 263.00018310546875, 325.43060302734375, 140.88973999023438, 459.21453857421875, 237.85659790039062, 437.93402099609375, 19.233245849609375, 93.80923461914062, 68.75559997558594, 347.02508544921875, 161.23974609375, 125.40333557128906, 199.66201782226562, -41.538818359375, -15.639617919921875, 170.44558715820312, 88.32322692871094, 75.2517318725586, 221.02926635742188, 100.71448516845703, 246.59645080566406, 186.78968811035156, 297.8758850097656, 186.76556396484375, 65.33316802978516, 94.36007690429688, 100.79448699951172, 190.96688842773438, 21.027313232421875, 54.66398620605469, 137.19686889648438, 12.276205062866211, 605.606689453125, 383.2330017089844, 567.9928588867188, -209.52273559570312, 342.05963134765625, 297.24603271484375, 332.55523681640625, 313.9849853515625, 38.25677490234375, -9.05419921875, 32.341461181640625, 5.002799987792969, -140.6103515625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000398.npy"} +{"epoch": 0.5844346549192364, "step": 399, "batch_size": 64, "mean": 173.58319091796875, "std": 176.94967651367188, "min": -194.953125, "p10": -59.41816482543944, "median": 165.44649505615234, "p90": 389.37780151367195, "max": 632.576416015625, "pos_frac": 0.8125, "sample": [201.7357177734375, 145.5404510498047, 63.06537628173828, 8.322998046875, 246.54808044433594, -49.54920196533203, 443.9411315917969, 128.14198303222656, 85.11786651611328, 149.77655029296875, 157.75823974609375, -179.24188232421875, 281.3620300292969, -15.57000732421875, 317.9315185546875, 96.36512756347656, 238.92691040039062, -93.51538848876953, 78.3142318725586, 97.73403930664062, 286.5098571777344, 94.7069091796875, 357.9371337890625, 168.39149475097656, 400.08447265625, 449.4779968261719, 124.48064422607422, 292.40277099609375, -140.95042419433594, 216.53060913085938, 38.71034240722656, 315.0697021484375, 271.3037414550781, 141.31211853027344, 313.9605407714844, -130.6505126953125, 632.576416015625, 329.1590270996094, 364.39556884765625, -22.052078247070312, 189.49960327148438, 314.9208679199219, 125.07392883300781, -194.953125, -3.8812828063964844, 315.78192138671875, 71.55140686035156, 448.7926025390625, 174.7403564453125, -37.056148529052734, 272.73175048828125, 563.8084106445312, 246.42633056640625, 123.78178405761719, 233.73329162597656, 255.35194396972656, 86.32485961914062, 268.7621154785156, -80.32156372070312, 267.0638427734375, 446.4290771484375, -63.64772033691406, 162.50149536132812, 15.846197128295898], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000399.npy"} +{"epoch": 0.5859030837004405, "step": 400, "batch_size": 64, "mean": 230.07472229003906, "std": 199.43344116210938, "min": -312.476806640625, "p10": -31.90695114135741, "median": 220.82647705078125, "p90": 469.95984191894536, "max": 805.8753051757812, "pos_frac": 0.859375, "sample": [219.06417846679688, 206.41143798828125, 298.01287841796875, 184.57568359375, 472.1853332519531, 40.60150146484375, 346.8590087890625, 245.28030395507812, -20.941429138183594, 326.35211181640625, 199.28057861328125, 198.66696166992188, 283.3595275878906, 187.1285858154297, -312.476806640625, 428.1763916015625, 446.1393127441406, 93.39617919921875, 94.46879577636719, 223.73995971679688, 534.0946655273438, 384.4010009765625, 135.34982299804688, 272.015869140625, -60.024513244628906, 294.0931396484375, 78.95344543457031, 362.860107421875, 135.51441955566406, 314.564697265625, 190.9534149169922, 113.04486083984375, -52.00963592529297, 145.28392028808594, 229.58242797851562, -80.45515441894531, 38.47777557373047, 474.106201171875, 222.58877563476562, 333.07318115234375, -109.77409362792969, 271.734375, 300.6925354003906, 684.5391845703125, -100.42068481445312, 464.76702880859375, 208.9302978515625, 178.4449920654297, 12.437488555908203, 805.8753051757812, 350.03887939453125, 5.353057861328125, 214.6707763671875, 373.1318359375, 168.94131469726562, 500.81500244140625, -36.60646057128906, 504.9196472167969, 425.173583984375, 259.24749755859375, 171.85256958007812, 398.8893737792969, 446.9486999511719, -2.569122314453125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000400.npy"} +{"epoch": 0.5873715124816447, "step": 401, "batch_size": 64, "mean": 135.82696533203125, "std": 227.90225219726562, "min": -346.75982666015625, "p10": -146.61054763793942, "median": 97.34819030761719, "p90": 503.3798248291016, "max": 618.888916015625, "pos_frac": 0.765625, "sample": [-28.520191192626953, 272.79058837890625, -62.26105880737305, 213.74713134765625, 251.6277313232422, 40.934959411621094, -295.9574279785156, -277.1799011230469, -346.75982666015625, -57.45368576049805, 13.061759948730469, 59.471405029296875, -110.0688247680664, 72.8051986694336, 0.83856201171875, 11.694879531860352, 325.7123718261719, 117.1056900024414, 122.67999267578125, 17.787246704101562, -48.3402099609375, -162.2712860107422, 93.79381561279297, 605.353515625, 87.53119659423828, 368.9188232421875, 491.45404052734375, 207.7294921875, 617.5804443359375, 158.44520568847656, 87.43175506591797, 164.10830688476562, 26.99142837524414, 100.9025650024414, -67.29615020751953, -261.8950500488281, 561.6529541015625, 43.64021682739258, 254.24256896972656, 309.5469665527344, 162.2841033935547, 344.2497863769531, 185.79898071289062, -21.767276763916016, 277.01885986328125, 82.45499420166016, 149.2154083251953, 36.37327575683594, 508.4908752441406, -167.5264129638672, 415.001708984375, 371.7710266113281, 191.33102416992188, 37.325035095214844, 541.50830078125, 618.888916015625, 72.5419692993164, 168.32730102539062, 153.42735290527344, 30.735851287841797, 172.38339233398438, 567.6491088867188, -162.90261840820312, -23.233015060424805], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000401.npy"} +{"epoch": 0.5888399412628488, "step": 402, "batch_size": 64, "mean": 239.6510009765625, "std": 181.82920837402344, "min": -170.92697143554688, "p10": 25.56152439117432, "median": 219.13485717773438, "p90": 471.18769226074227, "max": 742.2738647460938, "pos_frac": 0.953125, "sample": [211.2493133544922, 29.916837692260742, 41.045616149902344, 365.973876953125, 139.07498168945312, 269.4422607421875, 13.022720336914062, 138.39962768554688, 0.9523868560791016, 289.719482421875, 370.35052490234375, 453.6091613769531, 477.3848876953125, 136.1488494873047, 18.326171875, 361.8774719238281, 23.694961547851562, 227.02040100097656, 248.5676727294922, 324.4977111816406, 456.7275695800781, 241.33700561523438, 432.9461364746094, -170.92697143554688, 150.70199584960938, 333.6947326660156, 41.958038330078125, 587.2102661132812, 85.84022521972656, 529.984375, 178.25677490234375, 165.28746032714844, 123.10111999511719, 235.60638427734375, 102.3854751586914, 332.56353759765625, 577.076416015625, 59.88531494140625, 185.32424926757812, 399.45550537109375, 89.43684387207031, -40.80891036987305, 526.6832275390625, 527.4044189453125, 142.959228515625, 273.6859130859375, 423.0225524902344, 193.0745391845703, 256.813232421875, 742.2738647460938, 337.6331787109375, 399.20916748046875, 151.0111083984375, 435.1798095703125, 101.46304321289062, 311.9906005859375, 266.2701416015625, 163.65841674804688, 186.548095703125, -164.19305419921875, 339.4469909667969, 154.5885009765625, 140.49346923828125, 191.1287841796875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000402.npy"} +{"epoch": 0.5903083700440529, "step": 403, "batch_size": 64, "mean": 181.40777587890625, "std": 179.45799255371094, "min": -197.46018981933594, "p10": -43.944038772583006, "median": 179.22061157226562, "p90": 428.47975158691406, "max": 538.9085693359375, "pos_frac": 0.8125, "sample": [50.9334716796875, 179.91104125976562, 64.60062408447266, 159.15402221679688, 325.9114685058594, -24.656291961669922, 186.89303588867188, 303.3707275390625, 9.750600814819336, -88.86369323730469, 80.11260986328125, 30.868988037109375, -141.78684997558594, 433.9306945800781, 167.67588806152344, 431.3401794433594, 171.0595245361328, 68.0552749633789, 365.66925048828125, 325.451416015625, 235.42726135253906, 71.43751525878906, 500.8994445800781, 23.00591278076172, -197.46018981933594, 478.02911376953125, 316.62945556640625, 323.7542724609375, 189.72413635253906, 538.9085693359375, -45.379451751708984, 217.277099609375, 357.77374267578125, 92.70101928710938, 41.55608367919922, 372.2608337402344, -73.25244903564453, 268.3538513183594, 408.25030517578125, 434.95806884765625, -114.28477478027344, 67.4269790649414, 421.805419921875, 11.61781120300293, 332.74346923828125, -40.59474182128906, 411.3061828613281, 354.01226806640625, 161.78070068359375, 448.2021484375, 325.29376220703125, 237.57080078125, 184.8177490234375, 56.35823059082031, -31.012178421020508, -0.1275482177734375, -47.117462158203125, 310.7943420410156, 223.4827117919922, 162.72267150878906, 198.22494506835938, 103.37628936767578, -1.06842041015625, 178.53018188476562], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000403.npy"} +{"epoch": 0.591776798825257, "step": 404, "batch_size": 64, "mean": 189.71990966796875, "std": 183.961181640625, "min": -224.81597900390625, "p10": -72.96236190795895, "median": 185.2892074584961, "p90": 426.6819732666016, "max": 573.5517578125, "pos_frac": 0.84375, "sample": [-89.01716613769531, 173.98013305664062, 38.92205047607422, 334.8595275878906, 291.60186767578125, 128.84051513671875, 130.83961486816406, 525.0494384765625, 248.91668701171875, 524.4140014648438, 512.3465576171875, 155.2564697265625, 159.02076721191406, 392.7103271484375, 118.0520248413086, -116.04878997802734, 425.22149658203125, 166.4697265625, -34.61533737182617, 96.82821655273438, 38.51040267944336, 113.8235855102539, 112.10951232910156, 355.41998291015625, 320.3546142578125, 196.59828186035156, -224.81597900390625, 261.48699951171875, 169.50082397460938, 259.2654113769531, 351.4888916015625, 16.905969619750977, 336.28704833984375, 232.58221435546875, 338.0436096191406, 116.19963836669922, -95.91593933105469, 281.7291259765625, 311.1846008300781, 208.29466247558594, -211.12985229492188, 247.5777587890625, 166.20458984375, 361.38671875, 123.34407043457031, -160.20083618164062, 34.362274169921875, 230.9639892578125, 362.4440002441406, -137.2417755126953, 83.79117584228516, 167.49484252929688, 212.9808349609375, 130.83572387695312, 443.781494140625, 471.11834716796875, 258.423583984375, 53.97535705566406, -35.50115203857422, -7.304338455200195, 573.5517578125, 203.34619140625, 427.3078918457031, 257.8606262207031], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000404.npy"} +{"epoch": 0.593245227606461, "step": 405, "batch_size": 64, "mean": 145.86441040039062, "std": 192.96022033691406, "min": -271.74200439453125, "p10": -78.44425888061522, "median": 119.28273391723633, "p90": 440.96036682128914, "max": 605.2728271484375, "pos_frac": 0.734375, "sample": [356.55328369140625, 422.5621032714844, 267.4791564941406, 76.84529113769531, -63.75117492675781, -2.10479736328125, -271.74200439453125, 62.834991455078125, 363.5257263183594, -219.96929931640625, 83.54285430908203, 468.37762451171875, 352.9367980957031, 57.824951171875, 2.4068679809570312, 495.57464599609375, -40.023651123046875, 68.8763656616211, 191.84793090820312, -62.778560638427734, 122.0209732055664, 227.2928924560547, 69.26487731933594, 379.17919921875, 95.91363525390625, 172.63644409179688, 111.35246276855469, 230.8060302734375, -86.60446166992188, -84.74129486083984, 201.98902893066406, 21.312604904174805, 270.1614074707031, 7.013723373413086, 303.6822509765625, 132.830322265625, 157.05926513671875, -98.04121398925781, -46.490562438964844, -12.462810516357422, 311.04803466796875, 189.78152465820312, 453.0088195800781, 267.0746154785156, -1.2928199768066406, 171.61959838867188, 388.8409729003906, -41.907135009765625, 93.20880126953125, 161.04159545898438, 203.3873748779297, -143.4883270263672, -1.9626293182373047, 498.3030700683594, 83.08191680908203, -179.7755126953125, 157.83514404296875, 84.36917114257812, 116.54449462890625, 605.2728271484375, 456.68243408203125, -13.88916015625, 242.697998046875, 448.8453369140625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000405.npy"} +{"epoch": 0.5947136563876652, "step": 406, "batch_size": 64, "mean": 177.13998413085938, "std": 198.7234649658203, "min": -202.6109161376953, "p10": -47.85411949157714, "median": 159.91261291503906, "p90": 456.4784118652344, "max": 664.135498046875, "pos_frac": 0.78125, "sample": [243.88583374023438, 104.79498291015625, 86.71021270751953, 96.60720825195312, 180.63638305664062, 138.42926025390625, 80.31002807617188, -120.89190673828125, 652.2138671875, 120.80005645751953, 309.5508117675781, 110.47810363769531, 664.135498046875, 322.37091064453125, 220.9928436279297, 33.89603805541992, 217.45574951171875, 263.46148681640625, 287.16717529296875, -68.63516998291016, 513.4403076171875, 309.09307861328125, 211.396728515625, -202.6109161376953, 59.33655548095703, 67.09263610839844, 160.01895141601562, -34.00467300415039, 268.2845458984375, -66.43905639648438, -53.78959655761719, 205.48472595214844, 276.4482727050781, -199.75437927246094, -20.111995697021484, 310.078857421875, 544.205810546875, 77.08856964111328, 199.97828674316406, -4.089742660522461, 159.8062744140625, 182.29803466796875, 359.4728088378906, -21.390525817871094, -9.515411376953125, -3.8363876342773438, 532.8159790039062, 304.5368347167969, 20.8046875, 105.48028564453125, 422.7469787597656, 444.5512390136719, 114.9013442993164, 34.67295837402344, -115.47044372558594, -1.0236320495605469, 461.5900573730469, 231.3461456298828, 647.4291381835938, 244.396484375, 236.98764038085938, 267.38836669921875, 134.89126586914062, 16.562904357910156], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000406.npy"} +{"epoch": 0.5961820851688693, "step": 407, "batch_size": 64, "mean": 146.5013427734375, "std": 184.7505340576172, "min": -424.68988037109375, "p10": -76.76645584106446, "median": 138.4670867919922, "p90": 403.7668731689454, "max": 537.616943359375, "pos_frac": 0.8125, "sample": [-103.7240219116211, 365.4336853027344, 94.89035034179688, -59.837562561035156, -195.7342071533203, 140.50936889648438, 306.20843505859375, -147.18896484375, -3.215181350708008, 20.404632568359375, 380.22760009765625, 143.44810485839844, 199.5832977294922, 519.7945556640625, 452.5579833984375, 21.020524978637695, 126.6585693359375, 353.22552490234375, 115.33667755126953, 80.83206176757812, 181.70098876953125, 14.6417236328125, 142.51612854003906, 160.7872314453125, 219.5498809814453, 259.736083984375, 270.7107238769531, 154.4844970703125, 117.37251281738281, 137.86952209472656, 31.371246337890625, 181.73379516601562, 26.86473846435547, -4.046516418457031, 68.61949920654297, 139.0646514892578, 248.43191528320312, 537.616943359375, 179.16551208496094, -424.68988037109375, 250.0762176513672, 220.90306091308594, 110.50056457519531, 89.50238800048828, 354.01043701171875, 84.82298278808594, 21.57843017578125, 355.20770263671875, 299.8160095214844, 79.3678970336914, 467.92694091796875, -83.64754486083984, -48.820674896240234, 501.6712951660156, -77.29303741455078, 76.12956237792969, 99.78773498535156, 432.13409423828125, 72.95774841308594, -75.53776550292969, 413.8551330566406, 212.66517639160156, -98.58918762207031, 163.12774658203125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000407.npy"} +{"epoch": 0.5976505139500734, "step": 408, "batch_size": 64, "mean": 138.37420654296875, "std": 154.61819458007812, "min": -200.5707244873047, "p10": -45.627986145019534, "median": 131.30612182617188, "p90": 329.6363464355469, "max": 516.6116333007812, "pos_frac": 0.84375, "sample": [153.1796417236328, 404.86810302734375, 506.604248046875, 156.2303466796875, 231.3831787109375, 271.1995544433594, -45.418304443359375, 88.55139923095703, 111.80670166015625, -45.71784973144531, 5.763261795043945, -15.893604278564453, 10.289466857910156, 67.07070922851562, 1.4177627563476562, -122.82292175292969, 113.00575256347656, 249.50927734375, 248.53469848632812, 104.32377624511719, 56.03207778930664, 284.8280029296875, 85.55029296875, 238.66714477539062, -200.5707244873047, 160.515869140625, -57.09394073486328, 64.6213150024414, 63.94422149658203, 325.1296691894531, -78.7994155883789, 135.38531494140625, 331.5677795410156, 21.128883361816406, -164.86561584472656, 40.324554443359375, 217.77310180664062, 243.65467834472656, 38.92491149902344, 127.2269287109375, 203.14883422851562, 210.06973266601562, 185.2692108154297, 199.78802490234375, 236.3485565185547, -81.26866912841797, 427.53668212890625, 70.3296127319336, -28.922100067138672, 3.575653076171875, 58.79029083251953, 146.75631713867188, 17.938026428222656, 62.658905029296875, 154.0889892578125, 233.84262084960938, 81.43939971923828, 349.3261413574219, 200.61231994628906, 218.5266876220703, 468.7933654785156, 516.6116333007812, 294.7507629394531, 198.10841369628906], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000408.npy"} +{"epoch": 0.5991189427312775, "step": 409, "batch_size": 64, "mean": 160.81214904785156, "std": 164.68922424316406, "min": -169.56011962890625, "p10": -46.762805175781246, "median": 161.08330535888672, "p90": 394.8218383789064, "max": 573.041015625, "pos_frac": 0.859375, "sample": [-6.062957763671875, -169.56011962890625, 215.39572143554688, 184.2761688232422, 218.5653839111328, 195.51544189453125, 83.9420394897461, 92.11566162109375, 160.72964477539062, 123.02061462402344, 416.3902893066406, 341.5500793457031, -129.432861328125, 105.24862670898438, 332.32696533203125, 206.95962524414062, 221.02999877929688, 247.68333435058594, 178.63990783691406, 505.7713317871094, 196.77145385742188, -47.223777770996094, 165.87957763671875, 92.78514099121094, 350.30322265625, -93.5788803100586, 36.769737243652344, 22.969940185546875, 82.94314575195312, 172.49891662597656, 73.14506530761719, 565.2841796875, 302.0872802734375, -62.07233428955078, 89.7153091430664, 511.9604187011719, -95.73949432373047, 73.58442687988281, -66.59742736816406, 72.95563507080078, 292.41607666015625, 20.368263244628906, 256.9779052734375, 23.049083709716797, 234.49478149414062, 177.25106811523438, 62.488365173339844, 72.03571319580078, 448.01593017578125, 227.49346923828125, -45.68720245361328, 130.4638214111328, 0.3473987579345703, 161.4369659423828, 285.36248779296875, 413.9012451171875, 257.96368408203125, 169.34872436523438, 210.28887939453125, 66.49369812011719, 573.041015625, 124.30537414550781, 11.72369384765625, 149.85061645507812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000409.npy"} +{"epoch": 0.6005873715124816, "step": 410, "batch_size": 64, "mean": 152.70947265625, "std": 167.44056701660156, "min": -257.7603454589844, "p10": -41.08580932617187, "median": 142.2803497314453, "p90": 419.304672241211, "max": 571.138671875, "pos_frac": 0.84375, "sample": [263.53521728515625, 120.22990417480469, -27.002592086791992, 87.7049331665039, 240.525146484375, 155.73583984375, 255.39199829101562, 437.5727844238281, 241.0340118408203, 16.7608699798584, 102.09449768066406, 44.77770233154297, 14.649276733398438, 215.45321655273438, -49.007240295410156, -212.9405517578125, 439.0491638183594, -257.7603454589844, 7.938350677490234, 225.87171936035156, 256.0771789550781, 440.7273864746094, 141.4785919189453, 143.0821075439453, 334.2936706542969, 434.71356201171875, 61.494815826416016, -23.974552154541016, 219.4665069580078, 571.138671875, 207.35836791992188, 26.533214569091797, 161.294921875, 17.42413330078125, 204.48670959472656, 5.838775634765625, -42.105262756347656, -38.70708465576172, 106.72369384765625, -83.2689208984375, 101.64369201660156, 170.74740600585938, 104.75203704833984, 333.2186279296875, 130.67526245117188, -133.6739501953125, 105.42672729492188, 137.8529815673828, 425.75048828125, 210.50868225097656, 85.24819946289062, 139.4544677734375, 185.39004516601562, 202.6539306640625, 228.7334747314453, 477.54962158203125, 177.22853088378906, 339.53680419921875, -99.35572814941406, 40.45475769042969, 213.93154907226562, 85.75667572021484, 404.2644348144531, 239.9673614501953], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000410.npy"} +{"epoch": 0.6020558002936858, "step": 411, "batch_size": 64, "mean": 165.74508666992188, "std": 179.2735137939453, "min": -157.8436279296875, "p10": -9.693466949462888, "median": 117.88256072998047, "p90": 385.6651153564454, "max": 845.630859375, "pos_frac": 0.859375, "sample": [212.5721893310547, 97.31661224365234, 94.47136688232422, 9.685691833496094, 185.84152221679688, 68.0926742553711, 96.73031616210938, 123.52864837646484, 135.59910583496094, 223.71890258789062, 167.97421264648438, 18.736318588256836, 150.74588012695312, -23.397579193115234, 57.89971160888672, 3.700864791870117, 221.25474548339844, 362.282470703125, 94.883056640625, 235.14816284179688, 541.0999755859375, 293.7664794921875, -7.78900146484375, -13.892297744750977, 95.90604400634766, 95.50227355957031, 83.62226867675781, 136.9275360107422, 171.40899658203125, 610.6190185546875, 32.769989013671875, 207.92813110351562, 419.56793212890625, -10.509666442871094, 329.69757080078125, 45.09551239013672, 139.06788635253906, 291.94805908203125, 395.6862487792969, 107.25408172607422, 112.2364730834961, 330.0992126464844, 221.542724609375, 220.38970947265625, 111.61914825439453, 257.71923828125, 88.92449951171875, -157.8436279296875, 845.630859375, 320.4974060058594, -3.728343963623047, 263.1482849121094, 339.64117431640625, 361.2934875488281, 463.19622802734375, -121.30030059814453, 10.759330749511719, -53.69114685058594, 101.83934783935547, 5.772611618041992, 418.8030090332031, -78.42420196533203, 3.6930694580078125, 43.4062385559082], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000411.npy"} +{"epoch": 0.6035242290748899, "step": 412, "batch_size": 64, "mean": 153.51673889160156, "std": 169.2354278564453, "min": -162.25648498535156, "p10": -67.959464263916, "median": 142.04847717285156, "p90": 360.4129516601563, "max": 557.5191040039062, "pos_frac": 0.8125, "sample": [-90.03905487060547, 219.0791015625, 214.16595458984375, 215.0994110107422, 306.45172119140625, -3.0076522827148438, 426.90130615234375, 19.84593963623047, 82.6597900390625, 238.74624633789062, 98.38388061523438, 51.455596923828125, 56.82037353515625, 277.19677734375, -38.91297912597656, 325.36639404296875, 362.42608642578125, 91.49805450439453, -162.25648498535156, 173.7073974609375, -7.708063125610352, 75.94843292236328, 327.7200927734375, 153.53456115722656, 32.60139083862305, 142.60084533691406, 51.653961181640625, 232.41876220703125, -72.24401092529297, 141.7225341796875, -15.504480361938477, 28.83094024658203, 179.17315673828125, 47.512298583984375, 189.50433349609375, 114.67201232910156, -155.43597412109375, 3.007061004638672, 116.26570892333984, 508.7030944824219, 98.98214721679688, -100.22474670410156, 383.15106201171875, 352.7574462890625, 355.71563720703125, 66.65711975097656, 240.14263916015625, 513.4420166015625, -57.962188720703125, 257.0555114746094, 323.2095642089844, 71.28436279296875, 65.49751281738281, 142.37442016601562, 330.799560546875, 557.5191040039062, -153.3331298828125, 250.54058837890625, 304.10260009765625, -118.0985107421875, 378.76171875, 123.7691650390625, 203.95364379882812, 274.4097595214844], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000412.npy"} +{"epoch": 0.604992657856094, "step": 413, "batch_size": 64, "mean": 149.29052734375, "std": 152.9877471923828, "min": -299.63482666015625, "p10": 8.810533142089847, "median": 145.9926986694336, "p90": 357.4307586669923, "max": 533.3292236328125, "pos_frac": 0.921875, "sample": [31.90704345703125, 24.375629425048828, 28.119657516479492, 196.93270874023438, 180.42221069335938, 19.478073120117188, 41.451995849609375, 144.88720703125, 75.30777740478516, 114.71704864501953, 297.5072937011719, 432.13812255859375, 59.39781951904297, 261.9964904785156, 25.868873596191406, -30.12385368347168, 211.5831298828125, 60.658809661865234, 7.3191070556640625, 417.6534423828125, 22.08604621887207, 382.7737121582031, 181.15921020507812, -299.63482666015625, 28.42840576171875, 254.5137481689453, 222.7981719970703, 129.01263427734375, 246.8727264404297, 533.3292236328125, 135.10145568847656, 147.0981903076172, 386.0331115722656, 315.3968505859375, 12.29052734375, 280.303955078125, 226.84585571289062, 24.6900634765625, 194.40786743164062, 326.31536865234375, 40.43443298339844, 72.44581604003906, 394.4520568847656, 32.046173095703125, 227.41204833984375, 368.75787353515625, 197.31820678710938, -128.30545043945312, -85.34989929199219, -179.01031494140625, 179.82371520996094, 250.56130981445312, 215.66036987304688, 234.49391174316406, 26.967449188232422, 235.4922637939453, 61.293670654296875, 331.0008239746094, 130.9440155029297, 83.45611572265625, 202.24835205078125, 111.99253845214844, 191.79798889160156, 7.24053955078125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000413.npy"} +{"epoch": 0.6064610866372981, "step": 414, "batch_size": 64, "mean": 188.86105346679688, "std": 158.17965698242188, "min": -266.28533935546875, "p10": 1.0246658325195321, "median": 182.4930419921875, "p90": 376.64160461425786, "max": 587.7297973632812, "pos_frac": 0.90625, "sample": [330.6666259765625, 251.4816131591797, 245.40704345703125, 67.35267639160156, 75.13534545898438, 348.0860595703125, 162.0032958984375, -266.28533935546875, 50.41825866699219, 223.5225830078125, 254.05111694335938, 266.92816162109375, 356.1378173828125, 503.81268310546875, 37.78050994873047, -90.81718444824219, 247.3170623779297, 121.07302856445312, -23.274953842163086, 66.69669342041016, 146.94076538085938, 204.49032592773438, 55.9043083190918, 151.64077758789062, 335.20501708984375, 203.77110290527344, 94.32614135742188, 25.548322677612305, 242.2689971923828, 144.71864318847656, 379.8393859863281, 279.44219970703125, -8.564800262451172, 183.93055725097656, 550.585693359375, 152.86187744140625, 228.46424865722656, 1.8739776611328125, -23.480697631835938, 263.0613098144531, 140.2583465576172, 176.44332885742188, 338.95166015625, -14.103544235229492, 30.547090530395508, 400.6094055175781, 32.347198486328125, 405.6125183105469, 47.45728302001953, 121.28721618652344, 323.7962646484375, 211.68406677246094, 134.285400390625, 181.05552673339844, 161.95419311523438, 195.7802276611328, 477.6462707519531, 0.660675048828125, 288.86370849609375, 369.18011474609375, 587.7297973632812, 208.74301147460938, 104.51498413085938, 321.481689453125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000414.npy"} +{"epoch": 0.6079295154185022, "step": 415, "batch_size": 64, "mean": 147.53445434570312, "std": 160.46505737304688, "min": -159.03085327148438, "p10": -25.45377979278564, "median": 121.33483505249023, "p90": 391.71338806152346, "max": 562.421630859375, "pos_frac": 0.84375, "sample": [131.79763793945312, 98.63668823242188, 250.26425170898438, 501.8212890625, -15.251327514648438, 444.7472839355469, 381.83355712890625, 166.66783142089844, -81.81664276123047, 188.46388244628906, 300.8418273925781, 198.82522583007812, 285.299072265625, 294.1876220703125, 81.9445571899414, 57.29545974731445, 195.16091918945312, 75.90609741210938, 207.3522186279297, 43.64704513549805, 7.223091125488281, 159.275146484375, 15.015045166015625, 324.9692687988281, 3.8068313598632812, 71.89395904541016, 437.3636169433594, -15.787635803222656, 193.94354248046875, 123.86273193359375, 45.05609130859375, 181.25723266601562, 129.96157836914062, 37.090789794921875, 562.421630859375, -111.51670837402344, -69.78871154785156, 103.42066955566406, 216.41757202148438, 395.9476013183594, 71.45854187011719, 304.4788513183594, 41.26007080078125, 151.7313232421875, 29.26693344116211, 131.0132293701172, -39.37226104736328, 103.25953674316406, 403.822021484375, -28.313331604003906, 72.3656005859375, 346.7407531738281, 129.70103454589844, 38.96842956542969, 87.61927032470703, -18.781492233276367, 118.80693817138672, 4.756317138671875, 340.9163513183594, -102.6397933959961, 454.0088195800781, 106.94813537597656, -159.03085327148438, 233.76309204101562], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000415.npy"} +{"epoch": 0.6093979441997063, "step": 416, "batch_size": 64, "mean": 210.27268981933594, "std": 168.99057006835938, "min": -114.43475341796875, "p10": -32.95793762207031, "median": 218.5782699584961, "p90": 403.48005676269537, "max": 677.875732421875, "pos_frac": 0.84375, "sample": [386.57623291015625, 282.9317626953125, 62.54132080078125, 280.0386962890625, -52.131126403808594, 482.08477783203125, 143.43984985351562, 166.92745971679688, 281.90008544921875, 582.5740966796875, -19.619606018066406, -33.756011962890625, 65.4034194946289, 221.9353485107422, 231.19296264648438, 273.260009765625, 164.192138671875, 11.295402526855469, 387.23095703125, 144.4664764404297, 234.55471801757812, 169.64376831054688, 77.11946105957031, 240.6771240234375, 379.2151184082031, 257.8676452636719, 193.32101440429688, 381.81109619140625, 9.125030517578125, 426.3380126953125, 181.6760711669922, 307.12158203125, 677.875732421875, -57.134254455566406, -114.43475341796875, 80.8647689819336, 441.7654113769531, 320.3655700683594, -39.57636260986328, 410.0577087402344, 135.924560546875, 388.1322021484375, 163.72100830078125, 263.89666748046875, 225.4371337890625, 317.5718994140625, 205.23765563964844, 347.1496276855469, 149.6632843017578, 215.22119140625, -46.81502151489258, 209.35633850097656, -50.917083740234375, 354.49676513671875, 246.64413452148438, -31.09576416015625, 344.0418701171875, -18.191265106201172, 275.5148620605469, 38.02607727050781, 119.33238983154297, 245.7796630859375, 203.84544372558594, 514.7394409179688], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000416.npy"} +{"epoch": 0.6108663729809104, "step": 417, "batch_size": 64, "mean": 191.36529541015625, "std": 143.2105255126953, "min": -161.16403198242188, "p10": 12.631348609924318, "median": 186.4945831298828, "p90": 385.4268951416017, "max": 546.675537109375, "pos_frac": 0.921875, "sample": [48.292884826660156, 87.02589416503906, 11.951807022094727, 234.0663299560547, 14.21694564819336, 362.646240234375, 146.7792510986328, 77.6925048828125, 4.787878036499023, 493.2577209472656, -4.275966644287109, 311.8929138183594, 186.60594177246094, 105.86003112792969, 234.0579376220703, 81.89497375488281, -28.20274543762207, 418.8687744140625, 421.399658203125, 47.383697509765625, 500.7648010253906, 131.42276000976562, 359.08837890625, 124.8963623046875, 340.5079650878906, 546.675537109375, 214.75372314453125, 134.81002807617188, 269.5242919921875, 284.62860107421875, 115.0840835571289, 115.65730285644531, -27.205175399780273, 117.08075714111328, 222.13916015625, 400.6634521484375, 184.57781982421875, 106.5710220336914, 281.7886047363281, 252.8199005126953, 121.90640258789062, 193.37095642089844, 186.3832244873047, 260.8487548828125, -161.16403198242188, 196.32867431640625, 237.7646484375, 67.56395721435547, 239.78175354003906, 164.10052490234375, 320.7193603515625, -33.538856506347656, 288.4878234863281, 139.36221313476562, 178.07843017578125, 34.628623962402344, 110.6543197631836, 237.61102294921875, 395.1900329589844, 320.7805480957031, 319.57940673828125, 261.72674560546875, 196.01148986816406, 38.75041961669922], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000417.npy"} +{"epoch": 0.6123348017621145, "step": 418, "batch_size": 64, "mean": 176.95785522460938, "std": 204.2498016357422, "min": -224.57456970214844, "p10": -49.55918731689452, "median": 153.42801666259766, "p90": 456.1124206542969, "max": 826.9228515625, "pos_frac": 0.8125, "sample": [216.38385009765625, -144.070068359375, 216.837646484375, -68.36843872070312, 133.45399475097656, 619.775634765625, 31.116256713867188, 65.88299560546875, 468.79364013671875, 258.4574279785156, 58.250732421875, -144.74636840820312, 154.9014892578125, 180.30654907226562, 635.9252319335938, 212.81581115722656, 325.89276123046875, -37.09040832519531, 180.44375610351562, 128.89602661132812, -53.11004638671875, 344.27484130859375, -5.7330322265625, 213.20132446289062, 112.10228729248047, 49.77183532714844, 176.01611328125, 352.53204345703125, 148.41094970703125, 185.86599731445312, 376.3178405761719, 38.52366638183594, 41.61890411376953, 444.9127197265625, -157.13717651367188, -1.6432418823242188, -70.3476333618164, 184.7900390625, 200.5732879638672, 362.58245849609375, 826.9228515625, -25.58092498779297, 150.97625732421875, 84.91378784179688, 232.08782958984375, 151.9545440673828, 106.81328582763672, 135.33413696289062, -224.57456970214844, -41.27384948730469, 314.34625244140625, 148.56739807128906, 331.5347595214844, 26.077896118164062, 460.91229248046875, 177.0611572265625, 242.9508819580078, 650.3419189453125, 67.41825103759766, 31.330730438232422, 203.58441162109375, 512.1875610351562, 230.478271484375, 93.55670166015625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000418.npy"} +{"epoch": 0.6138032305433186, "step": 419, "batch_size": 64, "mean": 188.58261108398438, "std": 184.1721649169922, "min": -220.8070068359375, "p10": -27.214999771118148, "median": 167.4314193725586, "p90": 449.2344940185547, "max": 611.2547607421875, "pos_frac": 0.859375, "sample": [564.2182006835938, -220.8070068359375, 152.27822875976562, 266.7369384765625, 168.5628204345703, 245.9510040283203, -73.90225219726562, -43.77791213989258, 5.1805419921875, 453.8653564453125, 355.9744567871094, 94.95010375976562, 200.9468994140625, 368.8669738769531, 188.21527099609375, 235.6702117919922, 104.15005493164062, 96.50621032714844, -168.06703186035156, 77.1820297241211, 86.2092056274414, 450.8141174316406, -34.58747482299805, 259.2316589355469, 167.18411254882812, -172.55038452148438, 108.99899291992188, 71.5053482055664, 396.3623352050781, 590.61083984375, 220.58175659179688, 83.61773681640625, 268.9625244140625, 241.9356689453125, 114.74484252929688, 221.70895385742188, 128.48033142089844, 135.1209716796875, 427.9692687988281, 217.0790557861328, 462.192626953125, 135.0430908203125, 256.1910095214844, 118.91807556152344, -0.6831531524658203, 167.67872619628906, 60.377471923828125, 25.055206298828125, 555.09912109375, 328.412353515625, 360.4468688964844, 209.65762329101562, -84.29351806640625, 611.2547607421875, 278.51025390625, 339.4159851074219, 321.7962646484375, 53.91374969482422, 117.41928100585938, 445.5487060546875, -10.012557983398438, 46.09808349609375, 89.20088958740234, 125.36581420898438], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000419.npy"} +{"epoch": 0.6152716593245228, "step": 420, "batch_size": 64, "mean": 155.41830444335938, "std": 197.66421508789062, "min": -363.8829345703125, "p10": -53.13401107788085, "median": 131.4396629333496, "p90": 403.6452117919922, "max": 655.4364624023438, "pos_frac": 0.796875, "sample": [-83.20804595947266, 170.51455688476562, 206.3763427734375, 142.1861572265625, 279.495849609375, -42.39153289794922, 25.31342315673828, 2.153289794921875, 286.53057861328125, 407.31719970703125, -59.32173156738281, 98.60913848876953, 8.968639373779297, 54.83030319213867, -21.36993408203125, 155.1981201171875, 257.1263427734375, 184.438232421875, 56.54478073120117, 137.84182739257812, -3.485902786254883, 310.6724548339844, -39.4364013671875, 357.2057189941406, 73.83953857421875, 166.36590576171875, 164.7804412841797, 310.05828857421875, -12.096443176269531, 75.45004272460938, 424.704833984375, 308.3902587890625, 331.33935546875, -90.65885925292969, 125.0374984741211, 261.7687072753906, -57.73793029785156, 19.41284942626953, 385.3932800292969, 429.4560241699219, -10.329641342163086, 34.1317138671875, 473.455322265625, 160.32321166992188, 315.4417724609375, 108.23440551757812, 395.0772399902344, 392.9316711425781, 290.66314697265625, 215.88739013671875, 33.181556701660156, -236.8343505859375, 33.78538513183594, 70.51022338867188, 42.49058151245117, 655.4364624023438, -363.8829345703125, 74.64002990722656, 178.9858856201172, 108.02338409423828, 633.594482421875, 70.49567413330078, -128.92990112304688, 591.8460693359375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000420.npy"} +{"epoch": 0.6167400881057269, "step": 421, "batch_size": 64, "mean": 140.71896362304688, "std": 168.3171844482422, "min": -189.92135620117188, "p10": -91.61617813110351, "median": 170.39398956298828, "p90": 332.7249176025391, "max": 477.6711730957031, "pos_frac": 0.78125, "sample": [297.38604736328125, 60.77703857421875, 170.1197052001953, -93.05353546142578, 217.51092529296875, 254.57540893554688, 477.6711730957031, -31.178985595703125, 0.2965526580810547, 47.87434768676758, 173.6416015625, -2.7055091857910156, 319.4137878417969, 395.91937255859375, 299.5960693359375, 307.3320617675781, 211.5262908935547, -189.76808166503906, 163.62237548828125, 75.43112182617188, 469.6532897949219, 338.4296875, 177.33709716796875, -119.10198974609375, 192.67481994628906, 15.105819702148438, 177.8512420654297, 161.58984375, 80.46332550048828, 130.359375, 305.3619079589844, 267.85540771484375, 438.73724365234375, 195.40130615234375, 178.89385986328125, -111.68572998046875, 223.16259765625, 177.50914001464844, -143.00625610351562, 427.1139221191406, 62.29450225830078, -15.340852737426758, 219.2870330810547, 454.887939453125, 283.5237121582031, 253.5185089111328, 79.75100708007812, 6.80567741394043, 6.992166519165039, 228.3080596923828, 97.79411315917969, -72.13317108154297, 250.3239288330078, -189.92135620117188, -127.13417053222656, 97.44465637207031, 203.6607666015625, -36.102237701416016, -67.31829833984375, 170.66827392578125, 0.37847328186035156, -88.26234436035156, 147.32833862304688, 299.56512451171875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000421.npy"} +{"epoch": 0.618208516886931, "step": 422, "batch_size": 64, "mean": 150.78555297851562, "std": 183.1394500732422, "min": -302.80548095703125, "p10": -62.35778465270994, "median": 167.0867919921875, "p90": 367.14184570312506, "max": 626.7039184570312, "pos_frac": 0.8125, "sample": [170.2075653076172, 398.4121398925781, 440.17864990234375, 256.99298095703125, 211.38136291503906, -71.24256896972656, -302.80548095703125, 329.4208984375, 124.05956268310547, 626.7039184570312, -39.64009094238281, 214.317138671875, 59.80503845214844, 33.062896728515625, 339.0677490234375, 84.73764038085938, 72.62950897216797, -70.7993392944336, 37.259159088134766, 221.92311096191406, 84.91194152832031, 430.51751708984375, 185.7006378173828, -0.1669292449951172, 357.8084411621094, 284.1431579589844, 173.265869140625, 371.1418762207031, 64.99877166748047, 276.2317199707031, 524.8133544921875, 232.6751251220703, 71.20797729492188, 126.40495300292969, 10.51108169555664, 172.3231658935547, 295.8613586425781, -249.10903930664062, 132.23526000976562, -31.356536865234375, -11.968372344970703, 268.8899230957031, 73.78853607177734, -261.2554626464844, -42.660823822021484, 389.5514831542969, 303.09814453125, 228.2942352294922, 346.01470947265625, 17.129045486450195, 282.7777404785156, 163.9660186767578, 0.0509033203125, 30.623584747314453, 219.69151306152344, 228.06094360351562, 142.1279296875, -124.73176574707031, 226.79745483398438, 274.9521484375, -176.64451599121094, 125.00041198730469, 185.54083251953125, 111.38981628417969], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000422.npy"} +{"epoch": 0.6196769456681351, "step": 423, "batch_size": 64, "mean": 160.45176696777344, "std": 160.43505859375, "min": -175.99859619140625, "p10": -28.134346008300767, "median": 154.43988037109375, "p90": 388.4417053222657, "max": 542.0096435546875, "pos_frac": 0.828125, "sample": [37.141700744628906, 145.81600952148438, -7.1153411865234375, 103.07925415039062, -13.981658935546875, -119.16474151611328, 475.3830261230469, 31.508209228515625, 267.82232666015625, 244.8563232421875, 363.0003967285156, -51.81572723388672, 261.3568420410156, 194.44668579101562, 258.7970275878906, -54.478397369384766, 449.2530517578125, 463.99420166015625, 399.3451232910156, 431.4179992675781, 5.2149658203125, 10.68035888671875, 287.68902587890625, 101.98503875732422, 228.3458709716797, 181.16830444335938, 111.72533416748047, 150.97677612304688, -175.99859619140625, -140.51101684570312, 211.49542236328125, 68.70039367675781, 77.9211196899414, 273.4376525878906, 347.95953369140625, 211.0398406982422, 43.76734924316406, 226.1197509765625, -3.7003345489501953, 39.834869384765625, 156.19500732421875, 356.4522705078125, 113.80026245117188, 152.68475341796875, 542.0096435546875, 199.57679748535156, 189.59779357910156, -34.19978332519531, 226.4636993408203, 140.96751403808594, 3.881053924560547, 141.0911102294922, 208.0161590576172, 251.44244384765625, 171.59170532226562, 214.05076599121094, -131.4115447998047, 90.1895523071289, 97.17459106445312, -4.769561767578125, 349.3880920410156, 91.4487075805664, 199.62820434570312, 405.1294860839844], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000423.npy"} +{"epoch": 0.6211453744493393, "step": 424, "batch_size": 64, "mean": 165.27096557617188, "std": 179.28770446777344, "min": -213.877685546875, "p10": -30.98098621368408, "median": 147.9356460571289, "p90": 372.8422241210938, "max": 720.9833374023438, "pos_frac": 0.84375, "sample": [-70.61347961425781, 299.2029724121094, 63.11048889160156, -31.160045623779297, 29.7208251953125, 253.409912109375, 233.32318115234375, 243.289306640625, 67.90702056884766, -12.5797119140625, 257.78314208984375, 136.13104248046875, 137.4884033203125, -73.35320281982422, 205.5973663330078, 514.6052856445312, 221.0277099609375, 383.6693115234375, 87.66871643066406, 289.3297119140625, 379.86322021484375, -1.8474559783935547, 230.5098876953125, 356.45989990234375, 268.03570556640625, -73.2551498413086, 76.81658935546875, 253.14266967773438, -93.28579711914062, 35.15450668334961, -30.563180923461914, 720.9833374023438, 31.79207992553711, 196.71360778808594, 204.13404846191406, 30.108856201171875, 286.6750183105469, -116.62596130371094, 45.357444763183594, 219.1024169921875, 8.152105331420898, 86.45044708251953, -213.877685546875, 56.23923110961914, 138.23284912109375, 169.2423095703125, 32.84252166748047, 68.23914337158203, 104.88510131835938, 646.3352661132812, 82.53778839111328, 286.46343994140625, 247.44906616210938, 201.81483459472656, 240.86659240722656, 58.600399017333984, 440.2328186035156, 71.81969451904297, 157.63844299316406, 630.474853515625, 262.3948974609375, 244.62355041503906, 180.75770568847656, 120.09591674804688], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000424.npy"} +{"epoch": 0.6226138032305433, "step": 425, "batch_size": 64, "mean": 142.17543029785156, "std": 147.49411010742188, "min": -168.81912231445312, "p10": -39.419227600097656, "median": 141.5770263671875, "p90": 327.5818237304688, "max": 432.48443603515625, "pos_frac": 0.8125, "sample": [-40.641845703125, -13.48723030090332, 260.3017883300781, -168.81912231445312, -152.9691925048828, 218.03201293945312, 64.45805358886719, 208.33102416992188, 42.792327880859375, 189.204833984375, 97.12833404541016, 336.47235107421875, -135.02896118164062, 3.7366905212402344, 307.15692138671875, 107.44955444335938, 54.554298400878906, -30.846710205078125, 186.2125701904297, -11.294967651367188, 256.6438293457031, 147.27386474609375, 78.87934875488281, 281.02978515625, 325.74176025390625, 30.24768829345703, 226.92918395996094, 12.016616821289062, 148.92674255371094, 432.48443603515625, 140.78408813476562, 318.07568359375, 225.14248657226562, -36.29978942871094, 226.15354919433594, 115.38192749023438, 247.89756774902344, 109.65872192382812, 93.30265808105469, 385.40960693359375, -82.81056213378906, 222.53025817871094, 111.13493347167969, -142.58746337890625, 305.44537353515625, 253.3933868408203, 143.19358825683594, 142.36996459960938, 20.442108154296875, 355.8686218261719, 264.8345947265625, 234.44244384765625, -36.56645202636719, 52.68658447265625, 121.87715911865234, 138.59095764160156, 328.37042236328125, 417.6842041015625, 369.16363525390625, 305.22674560546875, 121.34738159179688, -49.53057098388672, 170.1129150390625, 43.58415985107422], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000425.npy"} +{"epoch": 0.6240822320117474, "step": 426, "batch_size": 64, "mean": 174.22409057617188, "std": 169.6143341064453, "min": -266.0386962890625, "p10": -10.003599548339832, "median": 144.29157257080078, "p90": 396.53720092773443, "max": 503.9173583984375, "pos_frac": 0.890625, "sample": [186.05723571777344, 14.491004943847656, 43.98261260986328, 127.04736328125, 137.84735107421875, 115.22969055175781, 90.21673583984375, 198.81723022460938, 374.47149658203125, 173.7389373779297, 132.0449981689453, 80.03642272949219, 153.01300048828125, -121.2351303100586, 371.4634094238281, 24.749080657958984, 102.41357421875, 226.63259887695312, 327.8488464355469, 142.030517578125, -67.52567291259766, 494.3905029296875, -66.3801498413086, 153.67959594726562, 90.83384704589844, 106.7071533203125, 64.96342468261719, -35.413307189941406, 503.9173583984375, 318.380859375, 152.6055450439453, 431.767333984375, 66.9341049194336, 106.28240966796875, 142.20811462402344, 441.2005615234375, 84.03076171875, -202.20008850097656, 125.85570526123047, 295.84783935546875, 13.82281494140625, 225.89002990722656, 503.45819091796875, 261.57757568359375, 286.6004638671875, 64.0451889038086, 189.27725219726562, 373.24127197265625, -15.236419677734375, 146.37503051757812, 14.609870910644531, 371.6371765136719, 2.2063140869140625, 235.03839111328125, 375.824951171875, 268.99224853515625, 59.121437072753906, 347.58734130859375, 420.716796875, -266.0386962890625, 405.41387939453125, 358.0435791015625, 69.52713012695312, 329.6273193359375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000426.npy"} +{"epoch": 0.6255506607929515, "step": 427, "batch_size": 64, "mean": 124.92091369628906, "std": 144.9481201171875, "min": -180.93838500976562, "p10": -34.09647235870361, "median": 107.92219924926758, "p90": 325.65366210937503, "max": 526.6851196289062, "pos_frac": 0.859375, "sample": [145.21539306640625, 29.876781463623047, 174.11358642578125, 343.1189880371094, 446.9770202636719, 159.36550903320312, 99.0271224975586, -36.69123077392578, 231.45895385742188, 212.96368408203125, -11.514068603515625, 118.69845581054688, -79.77464294433594, 34.70796585083008, 142.0988311767578, 168.57179260253906, 354.4049987792969, 478.5216064453125, 90.14337921142578, -180.93838500976562, 10.660942077636719, 38.2227783203125, 162.656005859375, 146.96792602539062, 22.62566375732422, 212.21499633789062, 126.1532211303711, 526.6851196289062, 60.8267822265625, 160.40972900390625, 39.760032653808594, 301.3805847167969, 327.988525390625, -50.695594787597656, 13.811641693115234, 209.18885803222656, 96.24585723876953, 5.519840240478516, 299.0428466796875, 71.72882843017578, 273.10223388671875, 116.81727600097656, 12.508247375488281, 163.03773498535156, -93.97321319580078, 38.79541778564453, 292.2901916503906, 233.62356567382812, -93.319580078125, 31.714378356933594, 326.640380859375, 15.20794677734375, 161.72186279296875, 24.567893981933594, 323.351318359375, 7.34991455078125, 24.038742065429688, 165.98886108398438, 229.73617553710938, 41.2713623046875, 4.316810607910156, -28.042036056518555, 70.13880157470703, -47.68605041503906], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000427.npy"} +{"epoch": 0.6270190895741556, "step": 428, "batch_size": 64, "mean": 159.47756958007812, "std": 182.626708984375, "min": -321.8248291015625, "p10": -46.914250183105466, "median": 158.9441680908203, "p90": 403.74703979492193, "max": 561.62353515625, "pos_frac": 0.828125, "sample": [-19.286693572998047, -41.71181869506836, 122.33751678466797, 301.3602600097656, 561.62353515625, 164.27613830566406, 108.19770812988281, 224.62551879882812, 320.61761474609375, 482.4840087890625, -30.386566162109375, 36.79541015625, 324.6842346191406, 33.447998046875, 410.164794921875, -96.44180297851562, -321.8248291015625, 504.2625427246094, 10.853775024414062, 334.3498840332031, 289.16595458984375, 109.3568115234375, 13.401948928833008, 87.6697998046875, 297.0430908203125, 418.1199951171875, 250.134521484375, 156.59658813476562, -49.143863677978516, 187.4053192138672, 177.92324829101562, 123.02899169921875, 353.8213806152344, -50.96514129638672, 290.6942443847656, 38.85591125488281, 235.43780517578125, 237.1725616455078, 21.366230010986328, 19.125553131103516, 92.51248168945312, 87.53305053710938, 41.58556365966797, 5.851219177246094, -111.7694091796875, 158.13125610351562, -19.764907836914062, 183.34600830078125, 468.51983642578125, 159.757080078125, 258.97808837890625, 229.7998046875, 285.8059387207031, -285.98046875, 283.8187255859375, 174.9845733642578, 374.3064880371094, 449.98626708984375, 234.9663848876953, 65.66691589355469, 388.77227783203125, -58.96394348144531, 89.14292907714844, 12.937522888183594], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000428.npy"} +{"epoch": 0.6284875183553598, "step": 429, "batch_size": 64, "mean": 146.47256469726562, "std": 178.04672241210938, "min": -392.6571960449219, "p10": -53.71096305847167, "median": 129.52052688598633, "p90": 378.1133941650391, "max": 517.8726806640625, "pos_frac": 0.765625, "sample": [127.31993865966797, 103.62033081054688, -46.96397018432617, 268.5794982910156, 8.457389831542969, 257.91070556640625, -22.70494842529297, 70.55878448486328, 109.41009521484375, 172.94020080566406, -82.20628356933594, 234.84042358398438, -7.595802307128906, 112.9738998413086, -74.73063659667969, 204.0723876953125, 82.52352905273438, -15.114761352539062, 414.0240173339844, -40.25627136230469, 437.00213623046875, 517.8726806640625, -26.5712890625, 277.5672912597656, 183.85948181152344, 20.55313491821289, 311.07208251953125, 23.897563934326172, 450.9698181152344, 296.89630126953125, -56.60253143310547, 131.7211151123047, 369.7921142578125, -129.85617065429688, 77.6136703491211, -100.83412170410156, -44.501991271972656, 150.7025146484375, 216.5234832763672, 102.11331939697266, 223.70431518554688, 290.546875, 148.2791748046875, 27.81695556640625, 515.7842407226562, 110.65686798095703, 318.6770935058594, 74.70587158203125, 125.00321960449219, -157.11343383789062, 279.8336486816406, 123.82475280761719, 364.3113708496094, 248.46014404296875, -17.225046157836914, -392.6571960449219, 3.227020263671875, 167.69638061523438, 202.72076416015625, 429.3709411621094, 232.465576171875, 381.6796569824219, 326.5723876953125, 258.45428466796875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000429.npy"} +{"epoch": 0.6299559471365639, "step": 430, "batch_size": 64, "mean": 146.6612548828125, "std": 157.09303283691406, "min": -163.6243438720703, "p10": -31.043233489990225, "median": 120.81015014648438, "p90": 333.10679931640624, "max": 581.4852905273438, "pos_frac": 0.84375, "sample": [252.9869384765625, 79.94291687011719, 120.2787857055664, -84.38494873046875, 365.9410705566406, 233.27365112304688, -34.313385009765625, 130.68673706054688, 83.53469848632812, 244.0421142578125, -1.6177024841308594, 7.768218994140625, 22.208412170410156, 164.95599365234375, 190.7253875732422, 299.35125732421875, -101.6497802734375, 301.7487487792969, 55.03395080566406, 108.63421630859375, 88.12481689453125, 16.00012969970703, 125.96913146972656, 121.34151458740234, 32.845802307128906, 40.77449035644531, 301.0526123046875, 227.97869873046875, -20.344451904296875, 167.44032287597656, 103.21430969238281, -163.6243438720703, 573.0466918945312, 107.05998229980469, 172.0758056640625, 226.68626403808594, 109.92522430419922, -23.412879943847656, 16.89124870300293, 170.72865295410156, 334.1749267578125, 391.7728576660156, 279.3160095214844, 110.0445327758789, 136.40322875976562, 447.57958984375, 107.97535705566406, 39.746429443359375, 290.0560607910156, -67.46766662597656, 152.2733917236328, 45.04280471801758, 330.614501953125, 130.4611053466797, -77.6416015625, 134.59698486328125, 265.0679931640625, -107.18415069580078, 280.7693176269531, 483.2628173828125, 581.4852905273438, 109.90386962890625, 79.5777359008789, 75.56773376464844], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000430.npy"} +{"epoch": 0.631424375917768, "step": 431, "batch_size": 64, "mean": 207.43408203125, "std": 174.67648315429688, "min": -118.67967224121094, "p10": -29.63300132751462, "median": 224.34233856201172, "p90": 436.83195800781255, "max": 673.7188720703125, "pos_frac": 0.875, "sample": [-69.01972961425781, 202.51744079589844, -40.56502151489258, 341.64447021484375, 250.74618530273438, 121.19385528564453, 535.7132568359375, 81.66921997070312, 103.87035369873047, 311.18096923828125, 154.299560546875, 278.6794738769531, 44.96624755859375, 10.038978576660156, -75.19489288330078, 51.13212585449219, 76.3511962890625, -92.67253875732422, 266.7757568359375, -118.67967224121094, 267.48211669921875, 465.43914794921875, 542.328369140625, 409.41558837890625, 110.19224548339844, 427.0032043457031, 382.57916259765625, 227.50225830078125, 221.1824188232422, 339.76422119140625, 457.468505859375, 205.57803344726562, 247.62109375, 240.1258087158203, 354.799072265625, -91.68196868896484, 673.7188720703125, -4.1249542236328125, 241.42269897460938, 29.984962463378906, 350.6651306152344, 262.3028564453125, 407.3939208984375, -99.79702758789062, 86.46217346191406, 72.25608825683594, 292.8924255371094, 28.12607765197754, 151.66815185546875, 263.0174560546875, 441.0442810058594, 309.16888427734375, 291.4325866699219, 292.754150390625, 190.10357666015625, 334.87506103515625, 140.45150756835938, 67.32495880126953, 60.054107666015625, 127.68717956542969, 445.81927490234375, 206.448486328125, 137.1410369873047, 234.0402374267578], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000431.npy"} +{"epoch": 0.6328928046989721, "step": 432, "batch_size": 64, "mean": 182.02349853515625, "std": 162.6422119140625, "min": -230.13002014160156, "p10": 8.060308837890625, "median": 147.5904541015625, "p90": 377.409652709961, "max": 739.9205322265625, "pos_frac": 0.921875, "sample": [78.89924621582031, 59.867496490478516, 99.75035095214844, 223.76495361328125, 95.70481872558594, 29.287582397460938, 133.6630859375, 92.92054748535156, 227.21875, 47.41181945800781, 85.05764770507812, -13.685436248779297, 367.0200500488281, 326.05120849609375, 68.99891662597656, 739.9205322265625, 343.37322998046875, 223.08401489257812, 59.37770080566406, 153.6229248046875, 89.82818603515625, 113.07870483398438, 412.8277587890625, -13.252410888671875, 250.74725341796875, 304.5481262207031, 240.8945770263672, 42.2970085144043, 192.12551879882812, 129.9327392578125, 211.064453125, 8.903030395507812, 59.25578689575195, 102.14572143554688, -36.877105712890625, 432.648681640625, 176.5268096923828, -230.13002014160156, 264.5737609863281, 87.14103698730469, 156.43035888671875, -89.07992553710938, 344.754150390625, 173.49586486816406, 231.87615966796875, 306.3372802734375, 271.5999755859375, 367.6956787109375, 299.3953857421875, 7.6991424560546875, 88.19075012207031, 340.7920837402344, 98.92817687988281, 224.08177185058594, 79.84571075439453, 105.300537109375, 549.1336059570312, 113.38471221923828, 293.1243896484375, 381.5727844238281, 4.190025329589844, 383.38397216796875, 141.5579833984375, 496.2240905761719], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000432.npy"} +{"epoch": 0.6343612334801763, "step": 433, "batch_size": 64, "mean": 133.83766174316406, "std": 170.11981201171875, "min": -330.2420959472656, "p10": -103.5698028564453, "median": 142.6808090209961, "p90": 328.3500854492188, "max": 433.33111572265625, "pos_frac": 0.75, "sample": [-1.987274169921875, 323.6702880859375, 347.9206848144531, 234.5625762939453, 308.6314697265625, -129.1983184814453, 117.22879791259766, -76.28709411621094, 304.2217712402344, 300.07354736328125, -330.2420959472656, 10.986114501953125, 390.91644287109375, 146.86367797851562, 30.472381591796875, 154.0847930908203, 278.293701171875, 279.48504638671875, 423.2611083984375, -179.08380126953125, -13.36233139038086, 265.5506286621094, 290.8881530761719, 293.3843688964844, 170.971435546875, 206.84381103515625, 38.02796936035156, 112.50479888916016, 138.49794006347656, 298.4659729003906, -0.463714599609375, 281.2807922363281, -105.92324829101562, 45.549041748046875, 62.94532775878906, 344.8411560058594, 303.1326599121094, -98.07843017578125, 264.74310302734375, 74.19098663330078, 433.33111572265625, 330.355712890625, 100.03617095947266, -139.8990478515625, -136.08087158203125, 99.53662872314453, 192.03952026367188, 61.77863311767578, -44.7292594909668, -124.26364135742188, 406.64105224609375, 107.64163208007812, 130.77142333984375, -66.37919616699219, 153.76885986328125, -66.74156188964844, 86.26825714111328, 261.4110412597656, -25.356260299682617, 166.02694702148438, 256.0467529296875, 251.53428649902344, 150.63851928710938, 73.36920166015625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000433.npy"} +{"epoch": 0.6358296622613803, "step": 434, "batch_size": 64, "mean": 150.973388671875, "std": 161.44374084472656, "min": -194.27525329589844, "p10": 1.1983947753906317, "median": 116.11874771118164, "p90": 339.381787109375, "max": 800.7782592773438, "pos_frac": 0.890625, "sample": [342.39837646484375, 220.116943359375, 800.7782592773438, 24.532360076904297, -82.28926086425781, 11.037567138671875, 100.06812286376953, 251.437255859375, -62.22942352294922, 32.40483093261719, 264.77337646484375, 44.12043762207031, 117.99324798583984, 318.5998840332031, -1.5713615417480469, 97.88418579101562, 51.55445098876953, -74.34619140625, 220.73178100585938, 440.67218017578125, 205.50439453125, 94.13793182373047, 202.4600372314453, 83.33235931396484, 268.35223388671875, 207.41189575195312, 76.03629302978516, 27.325559616088867, 114.24424743652344, 118.43456268310547, 88.47000122070312, 250.93252563476562, 211.62503051757812, -194.27525329589844, 93.08302307128906, 332.34307861328125, 7.661159515380859, 77.98104858398438, 135.509521484375, 221.11541748046875, 9.43377685546875, 376.4037170410156, 356.7633972167969, 79.51942443847656, 147.73159790039062, 255.43438720703125, 197.68551635742188, 77.69261169433594, 172.29922485351562, 329.5103759765625, 168.5647735595703, -20.452028274536133, 447.6390380859375, 88.92802429199219, 74.97232055664062, 35.993255615234375, 168.30271911621094, 178.06771850585938, 95.60943603515625, 19.07842254638672, 526.7910766601562, 217.43560791015625, 9.745376586914062, -91.20578002929688], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000434.npy"} +{"epoch": 0.6372980910425844, "step": 435, "batch_size": 64, "mean": 171.58197021484375, "std": 173.33937072753906, "min": -287.943359375, "p10": -21.861203956604, "median": 182.24565887451172, "p90": 396.24039916992206, "max": 558.136962890625, "pos_frac": 0.875, "sample": [555.230224609375, 52.404869079589844, -170.2275390625, 238.84869384765625, 434.05010986328125, 237.75311279296875, 182.91513061523438, 347.9554443359375, 282.7210388183594, 181.57618713378906, 330.71966552734375, 70.94364929199219, 17.42755126953125, 325.3974304199219, 55.80812072753906, 417.5914611816406, 351.39385986328125, 357.684814453125, 219.37213134765625, 30.573501586914062, -85.1063232421875, 18.421051025390625, -22.550262451171875, 71.55345153808594, -97.83026123046875, 89.72630310058594, 31.931163787841797, 189.18081665039062, -112.24024200439453, 131.12698364257812, 184.80450439453125, -20.253400802612305, 136.43780517578125, 125.68850708007812, 302.36907958984375, 249.74282836914062, 276.0126037597656, -287.943359375, 187.52975463867188, -162.367919921875, 80.48703002929688, 190.8482208251953, 448.7823791503906, 35.45918273925781, 214.17897033691406, 259.25579833984375, 412.76422119140625, 296.30810546875, 99.77743530273438, 59.33294677734375, 558.136962890625, 34.6646728515625, 230.9007568359375, 77.76229858398438, 308.3835144042969, 499.75048828125, 284.04425048828125, 102.306396484375, 99.91172790527344, 282.46099853515625, 126.13369750976562, 108.33454895019531, 264.84967041015625, 180.03939819335938], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000435.npy"} +{"epoch": 0.6387665198237885, "step": 436, "batch_size": 64, "mean": 170.72122192382812, "std": 153.1236572265625, "min": -128.30258178710938, "p10": -25.415382385253885, "median": 156.32225799560547, "p90": 393.4248168945314, "max": 560.4927978515625, "pos_frac": 0.875, "sample": [47.23704528808594, 73.26878356933594, -53.93403625488281, 123.8561019897461, 368.83740234375, 197.9795379638672, 183.71945190429688, 486.783935546875, 294.0952453613281, 31.370460510253906, 162.2161102294922, 556.5714111328125, 49.447174072265625, -34.94048309326172, 205.9584197998047, 184.00949096679688, -65.37690734863281, 55.937957763671875, 461.1886291503906, -95.90730285644531, 213.6311798095703, 180.27981567382812, 234.2693634033203, 196.69979858398438, 424.2708435058594, 110.33428955078125, 310.2089538574219, 403.9622802734375, 96.92695617675781, 132.77635192871094, 560.4927978515625, 125.22974395751953, 358.3540954589844, 213.91392517089844, 246.22613525390625, 150.3878173828125, 154.06455993652344, -3.1901473999023438, 225.17062377929688, 74.66567993164062, 271.1172790527344, 42.905948638916016, 70.68020629882812, 77.01265716552734, 155.36134338378906, 251.65072631835938, 108.62751770019531, -57.53325653076172, 157.4646759033203, 107.85955810546875, 205.87860107421875, 191.43359375, 157.28317260742188, 320.7524719238281, 464.0484619140625, 31.552536010742188, 125.1796646118164, 139.1082305908203, -128.30258178710938, 120.91294860839844, -55.956321716308594, 228.16998291015625, 270.7444152832031, 29.212554931640625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000436.npy"} +{"epoch": 0.6402349486049926, "step": 437, "batch_size": 64, "mean": 157.938720703125, "std": 150.06642150878906, "min": -213.10594177246094, "p10": -36.697378921508786, "median": 152.58318328857422, "p90": 356.84221801757815, "max": 540.4963989257812, "pos_frac": 0.859375, "sample": [138.4252471923828, -110.05734252929688, 140.11988830566406, 245.740478515625, 162.0032958984375, -213.10594177246094, 324.09210205078125, 115.11468505859375, 196.2924041748047, 38.2076416015625, -33.652286529541016, 165.3571319580078, -38.002418518066406, 349.03955078125, 61.70391845703125, 183.12232971191406, 151.71102905273438, -103.62974548339844, 18.805877685546875, 365.684814453125, 28.74657440185547, 303.85162353515625, 123.46067810058594, 147.91751098632812, -98.15357971191406, 408.2194519042969, 65.5858383178711, 73.07166290283203, 274.9747314453125, 204.76625061035156, 453.67718505859375, 274.5775146484375, 447.3061218261719, -84.91868591308594, 540.4963989257812, -30.265823364257812, 68.37177276611328, -69.25732421875, 202.49923706054688, 244.45217895507812, 112.09566497802734, 149.99514770507812, 243.49412536621094, 179.8952178955078, 303.5279541015625, 133.43386840820312, 380.71868896484375, 241.4976806640625, 360.18621826171875, 199.5799560546875, 71.88673400878906, 177.19039916992188, 212.1941375732422, 62.266998291015625, 46.46472930908203, 331.44036865234375, 252.068115234375, 153.45533752441406, 88.76133728027344, 146.7245330810547, 143.88137817382812, 174.96067810058594, 198.02313232421875, 7.983406066894531], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000437.npy"} +{"epoch": 0.6417033773861968, "step": 438, "batch_size": 64, "mean": 180.56558227539062, "std": 152.6371307373047, "min": -115.31472778320312, "p10": 11.623118019104004, "median": 150.7465057373047, "p90": 437.7142730712891, "max": 486.1976318359375, "pos_frac": 0.9375, "sample": [7.897432327270508, 159.31373596191406, 322.0951843261719, 11.537796020507812, 104.36553955078125, 80.81320190429688, 486.1976318359375, -59.41685485839844, 71.7457046508789, 66.06515502929688, 202.3310546875, 75.8694839477539, 52.995113372802734, 97.24846649169922, 235.0704803466797, 334.73406982421875, 235.4229278564453, 480.45892333984375, 130.7652130126953, 153.2061767578125, 80.88202667236328, 440.1038818359375, 282.3336181640625, 464.27923583984375, 87.30184936523438, 79.1114501953125, 104.02741241455078, 283.43231201171875, -115.31472778320312, 24.512664794921875, 11.822202682495117, 22.772706985473633, 185.47093200683594, 351.8857421875, 449.8368225097656, 148.28683471679688, 197.87078857421875, 64.58728790283203, 367.47186279296875, 230.80670166015625, 432.1385192871094, 222.67596435546875, 44.876163482666016, 481.42108154296875, -51.672576904296875, 119.26815795898438, 304.4817810058594, 309.67578125, 197.1484832763672, 218.99237060546875, 70.01214599609375, 233.18704223632812, 367.07098388671875, 103.75794982910156, 58.102874755859375, 117.29190063476562, 169.447509765625, -48.34580993652344, 5.848320007324219, 289.4945983886719, 88.46528625488281, 468.4577941894531, 90.24668884277344, 253.95657348632812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000438.npy"} +{"epoch": 0.6431718061674009, "step": 439, "batch_size": 64, "mean": 143.0908203125, "std": 166.68077087402344, "min": -112.92741394042969, "p10": -72.66128540039061, "median": 113.39131164550781, "p90": 394.3713287353516, "max": 546.6541137695312, "pos_frac": 0.78125, "sample": [190.15277099609375, 442.03515625, 50.62727355957031, 269.72662353515625, -34.96534729003906, -55.24267578125, 168.8282470703125, 125.30873107910156, 128.04330444335938, 88.5832748413086, -80.65631103515625, 399.9556884765625, 18.309555053710938, -6.679435729980469, 87.8050308227539, 277.75042724609375, 242.01768493652344, 279.43975830078125, 280.1168518066406, 109.06292724609375, 94.54277038574219, 528.9074096679688, 133.99496459960938, -104.36892700195312, 90.11421966552734, -80.12640380859375, 96.82843017578125, 102.86357116699219, -1.9708309173583984, 38.67565155029297, 301.5853576660156, -30.263954162597656, 262.5212097167969, 227.09266662597656, 504.7096862792969, 233.78091430664062, 381.3411560058594, 272.5983581542969, 118.62052917480469, 59.82825469970703, 57.819358825683594, -109.4461898803711, -112.92741394042969, 2.138427734375, 269.4068908691406, 77.41584014892578, 138.78053283691406, 218.20298767089844, -86.28762817382812, 285.32305908203125, 546.6541137695312, 94.39903259277344, 57.03562927246094, 413.9988708496094, 132.3715362548828, 43.55216979980469, 302.42535400390625, 197.48770141601562, -40.062625885009766, 34.898162841796875, 441.44610595703125, 117.71969604492188, -100.05834197998047, -35.97561264038086], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000439.npy"} +{"epoch": 0.644640234948605, "step": 440, "batch_size": 64, "mean": 133.9937744140625, "std": 173.3215789794922, "min": -245.32266235351562, "p10": -33.05714111328125, "median": 93.13737106323242, "p90": 389.8146118164063, "max": 570.054931640625, "pos_frac": 0.828125, "sample": [280.7992248535156, 102.47290802001953, 403.8160705566406, 177.9054718017578, 20.6943359375, 28.75438690185547, 307.9112548828125, 39.29126739501953, 95.91188049316406, 478.60540771484375, 49.14342498779297, 58.630218505859375, 363.39434814453125, 93.81072998046875, 323.31890869140625, -55.24201965332031, 114.55758666992188, -159.66139221191406, 74.77317810058594, 14.427238464355469, 56.975860595703125, 430.2170715332031, 59.94944763183594, 32.370216369628906, 116.47969055175781, 114.93464660644531, -26.3199405670166, 233.6815185546875, 179.19200134277344, 48.938602447509766, 130.38973999023438, 390.608154296875, -91.12802124023438, 142.4976348876953, -41.2174072265625, -21.407745361328125, 481.17578125, -245.32266235351562, 92.4640121459961, 16.18085289001465, 377.5692443847656, 98.9267349243164, 570.054931640625, 282.5828857421875, 95.91351318359375, 149.2813720703125, 47.41607666015625, -33.76820373535156, 243.92202758789062, 272.836669921875, 76.90228271484375, 516.1580810546875, 2.5253524780273438, 49.696434020996094, 12.713436126708984, 387.9630126953125, 3.971454620361328, -31.397994995117188, 37.39257049560547, -9.988739013671875, 342.4102478027344, 239.32278442382812, -111.56957244873047, 40.792572021484375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000440.npy"} +{"epoch": 0.6461086637298091, "step": 441, "batch_size": 64, "mean": 168.28616333007812, "std": 200.389404296875, "min": -351.8939208984375, "p10": -18.09986114501952, "median": 117.12174224853516, "p90": 430.98809814453125, "max": 871.9410400390625, "pos_frac": 0.859375, "sample": [157.6676788330078, 76.27995300292969, 69.09088134765625, 320.94268798828125, 233.18057250976562, 506.1134033203125, 13.742340087890625, 36.020545959472656, 277.0888366699219, -181.14248657226562, -1.5517082214355469, 51.27259063720703, 412.6616516113281, 462.0674133300781, 315.72308349609375, 107.65196990966797, 83.69480895996094, 361.24713134765625, 72.10464477539062, 61.182098388671875, 17.352407455444336, 57.67530822753906, 119.09170532226562, 428.3522033691406, 469.2542724609375, 40.401763916015625, 74.52540588378906, 325.720947265625, 202.58505249023438, 871.9410400390625, 21.82892417907715, -224.0372772216797, 115.15177917480469, 369.61370849609375, -104.70072937011719, 92.80633544921875, 77.8864517211914, 311.77984619140625, 479.0257568359375, 27.101806640625, -6.86248779296875, 181.97857666015625, -22.915878295898438, -351.8939208984375, 250.95738220214844, 196.9846954345703, 223.5082550048828, 432.1177673339844, 220.83547973632812, 28.901432037353516, 21.141921997070312, -55.212806701660156, 47.930572509765625, 256.70709228515625, 169.26771545410156, 65.24551391601562, 381.3799133300781, 172.2364044189453, 314.5465393066406, 467.85699462890625, 88.23895263671875, -30.20623016357422, 194.90847778320312, 314.26849365234375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000441.npy"} +{"epoch": 0.6475770925110133, "step": 442, "batch_size": 64, "mean": 196.18905639648438, "std": 181.53651428222656, "min": -153.0502166748047, "p10": 8.41243743896485, "median": 155.03395080566406, "p90": 467.8856506347656, "max": 692.4766235351562, "pos_frac": 0.921875, "sample": [20.709144592285156, 108.72346496582031, 410.91943359375, 74.10430908203125, 248.82147216796875, 85.19808959960938, 41.87702178955078, 144.83558654785156, 48.49853515625, 154.63763427734375, 145.4883270263672, 281.5601806640625, 72.48101043701172, 228.7115936279297, 221.35079956054688, 416.591796875, 461.0014343261719, 64.52845764160156, 26.9622802734375, 130.57325744628906, 93.67865753173828, 368.0657958984375, 41.2763671875, 241.54415893554688, -1.255483627319336, 181.39120483398438, 470.8360290527344, 157.36009216308594, 478.24993896484375, 250.55255126953125, 111.11129760742188, -153.0502166748047, 190.62686157226562, 37.48491287231445, 28.476417541503906, 305.10504150390625, -20.163997650146484, 390.1665954589844, 194.78335571289062, 567.8920288085938, -101.53131103515625, 264.6443176269531, 410.86578369140625, 34.91796112060547, 43.98101806640625, 594.1199951171875, 482.22705078125, 221.01190185546875, 439.7704772949219, 146.589111328125, 53.006103515625, 5.338897705078125, 4.881591796875, 137.0028839111328, 15.584030151367188, -60.79991149902344, 360.71978759765625, 285.18927001953125, 692.4766235351562, 191.73703002929688, 490.0207824707031, 155.43026733398438, 111.04060363769531, 256.1702880859375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000442.npy"} +{"epoch": 0.6490455212922174, "step": 443, "batch_size": 64, "mean": 137.07171630859375, "std": 189.19361877441406, "min": -313.8285217285156, "p10": -93.07962341308593, "median": 112.79312896728516, "p90": 335.79843444824223, "max": 723.9342041015625, "pos_frac": 0.828125, "sample": [723.9342041015625, 73.78706359863281, 501.426025390625, 11.103652954101562, 194.19326782226562, 57.23243713378906, 81.8155746459961, -52.12696838378906, 70.20118713378906, 121.25115203857422, -63.24693298339844, 59.73274230957031, 108.32987213134766, 281.10382080078125, 29.18212890625, -313.8285217285156, 225.0883331298828, -97.19096374511719, 211.10549926757812, 104.2908935546875, 651.1637573242188, 321.508056640625, 250.9498291015625, 248.88546752929688, -128.2354736328125, -300.85430908203125, 206.53738403320312, 148.33187866210938, 351.7383117675781, -221.13397216796875, 221.97071838378906, 265.7200012207031, 60.60487365722656, 97.05204772949219, 77.41453552246094, -37.10492706298828, 91.46790313720703, 296.82080078125, 183.101318359375, 114.44499206542969, -126.65577697753906, -83.48649597167969, 246.10195922851562, 74.70193481445312, 48.450687408447266, 65.56544494628906, 235.99755859375, 111.14126586914062, 126.44470977783203, 161.22613525390625, 297.5758361816406, 101.62408447265625, 121.39356994628906, 187.2035675048828, -122.98329162597656, 212.60797119140625, 99.38127136230469, 341.9228820800781, 170.94332885742188, 106.831787109375, 399.5694274902344, 488.6937255859375, 20.011817932128906, 260.5591125488281], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000443.npy"} +{"epoch": 0.6505139500734214, "step": 444, "batch_size": 64, "mean": 194.06898498535156, "std": 199.19985961914062, "min": -256.13262939453125, "p10": -23.280255508422847, "median": 166.9800262451172, "p90": 425.08591918945314, "max": 799.2061767578125, "pos_frac": 0.84375, "sample": [197.6481170654297, -33.5712890625, 46.551124572753906, 331.6311340332031, 5.116079330444336, 205.04266357421875, 44.6171760559082, 761.8529052734375, 185.15682983398438, -186.7970733642578, 107.1300048828125, 264.6514587402344, 323.4317626953125, 373.81512451171875, 432.66424560546875, 170.828125, 327.6373291015625, 226.8981170654297, -41.420166015625, 19.947647094726562, -24.3721866607666, -20.7324161529541, 340.6774597167969, 510.729248046875, -53.86079406738281, 271.04840087890625, 201.825927734375, -15.283935546875, 125.50321960449219, -64.29180908203125, 101.3312759399414, 799.2061767578125, 270.0072937011719, 142.63604736328125, 71.08628845214844, 426.1226806640625, 242.7529296875, 385.8424072265625, 279.665771484375, 142.62161254882812, 50.19097900390625, 555.593017578125, 422.66680908203125, 73.84362030029297, 299.72137451171875, 383.7492980957031, 235.70338439941406, 104.30121612548828, 150.46803283691406, 99.90213012695312, 406.056884765625, 106.51106262207031, 118.59282684326172, 94.3903579711914, 313.76953125, 163.13192749023438, 40.48858642578125, 87.98423767089844, -20.278528213500977, 34.382904052734375, 505.942138671875, -256.13262939453125, 211.37310791015625, 342.71600341796875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000444.npy"} +{"epoch": 0.6519823788546255, "step": 445, "batch_size": 64, "mean": 190.07302856445312, "std": 180.88926696777344, "min": -99.13471984863281, "p10": -9.463719367980955, "median": 162.99077606201172, "p90": 403.45192260742186, "max": 637.5178833007812, "pos_frac": 0.84375, "sample": [172.18771362304688, 15.662473678588867, 121.14171600341797, 561.2530517578125, 114.15870666503906, 366.15313720703125, 162.80645751953125, 47.112815856933594, 168.47769165039062, 175.7710723876953, 32.711299896240234, 378.5329284667969, 87.69496154785156, 163.1750946044922, 70.47782135009766, -99.13471984863281, 179.24880981445312, 339.3236389160156, 21.385263442993164, 92.40684509277344, 134.7537841796875, 190.30511474609375, 354.2875061035156, 486.18963623046875, 10.347034454345703, 82.1744613647461, 637.5178833007812, 389.2634582519531, 177.51937866210938, 513.9105834960938, 404.31597900390625, 401.435791015625, 229.9530487060547, 400.3469543457031, 139.36497497558594, -10.250457763671875, 215.47256469726562, 269.6394348144531, -43.81578063964844, 397.23309326171875, -28.991188049316406, 636.6988525390625, 312.46942138671875, -2.2194976806640625, 252.521240234375, 304.8211669921875, 93.92804718017578, -79.19336700439453, 32.178138732910156, 106.06797790527344, 182.0489501953125, 84.19954681396484, 121.14669799804688, 321.07891845703125, 287.26739501953125, 162.7276153564453, -2.78338623046875, -7.627996444702148, 569.2276611328125, -58.07124328613281, 156.73272705078125, 182.43157958984375, 59.794921875, -72.29144287109375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000445.npy"} +{"epoch": 0.6534508076358296, "step": 446, "batch_size": 64, "mean": 192.0048828125, "std": 167.7165069580078, "min": -130.7541961669922, "p10": 0.9644021987915172, "median": 177.93719482421875, "p90": 419.7940063476563, "max": 620.9629516601562, "pos_frac": 0.890625, "sample": [30.87543487548828, -12.473373413085938, 60.89601135253906, 391.1022644042969, 617.6729125976562, 293.5547790527344, 128.4704132080078, -130.7541961669922, 87.11592102050781, 404.500732421875, 620.9629516601562, 173.84942626953125, 34.173622131347656, 329.72711181640625, 15.018333435058594, 400.8085021972656, 352.25537109375, 203.29669189453125, 118.635009765625, 287.4654541015625, 26.56201171875, 64.0645751953125, 197.67747497558594, 115.63139343261719, 92.31752014160156, -79.06721496582031, 161.6471405029297, 233.19735717773438, 228.2984619140625, 220.7855224609375, 266.1186828613281, -76.85844421386719, 169.93978881835938, 148.7366485595703, -4.578340530395508, 245.74462890625, -14.41973876953125, 334.417724609375, 83.3594970703125, 426.3482666015625, 302.22760009765625, 27.92756462097168, 291.6366882324219, 471.214111328125, 182.02496337890625, 13.897468566894531, 507.9732971191406, 321.4538879394531, 110.25028991699219, 334.9109802246094, 224.99099731445312, 25.322494506835938, -16.609954833984375, 98.60470581054688, 32.62767791748047, 143.4908447265625, 85.46017456054688, 234.5286407470703, 456.90673828125, 202.96966552734375, 49.809165954589844, 441.09228515625, 206.61138916015625, 291.9146423339844], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000446.npy"} +{"epoch": 0.6549192364170338, "step": 447, "batch_size": 64, "mean": 194.4075927734375, "std": 200.45513916015625, "min": -148.04669189453125, "p10": -28.63780746459961, "median": 172.1294937133789, "p90": 423.324658203125, "max": 1079.6866455078125, "pos_frac": 0.859375, "sample": [89.19161987304688, 171.8599395751953, 210.592529296875, 211.0774383544922, 36.426910400390625, -27.221601486206055, 357.4529724121094, 212.94869995117188, 210.5319061279297, -28.687408447265625, -86.03089904785156, 310.06494140625, 182.42898559570312, 166.50746154785156, 415.6918640136719, 404.19659423828125, 559.232177734375, 110.88623046875, 1079.6866455078125, 533.843994140625, 144.8028564453125, 431.6236267089844, 274.31829833984375, 251.755859375, -70.05110168457031, 102.72138214111328, 235.349365234375, -115.03811645507812, 114.3639907836914, 86.04371643066406, -31.64361572265625, 83.04859924316406, 84.48279571533203, -52.762359619140625, 319.42401123046875, 113.78617858886719, 127.30328369140625, 377.848388671875, 26.679931640625, 19.31914520263672, 382.86871337890625, -28.522071838378906, 165.6723175048828, -148.04669189453125, 471.7344055175781, 173.461181640625, 229.3905029296875, 20.719026565551758, 157.00143432617188, 29.57834815979004, 286.137451171875, 204.806396484375, 223.86158752441406, 172.3990478515625, 23.86199951171875, 355.09259033203125, 248.7122344970703, 88.72428894042969, 531.0457153320312, 426.5958557128906, 86.84734344482422, 256.0366516113281, 396.1766662597656, 43.874053955078125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000447.npy"} +{"epoch": 0.6563876651982379, "step": 448, "batch_size": 64, "mean": 213.71690368652344, "std": 195.1259307861328, "min": -152.6901092529297, "p10": -12.613234710693357, "median": 186.00963592529297, "p90": 501.46840515136716, "max": 702.3887939453125, "pos_frac": 0.84375, "sample": [-9.988410949707031, 79.91461181640625, 50.279754638671875, 118.53860473632812, 578.8604736328125, 51.001953125, 177.23915100097656, -48.44544219970703, 268.7855224609375, -13.7381591796875, 295.58209228515625, 99.8793716430664, 527.87890625, 79.40542602539062, 245.02764892578125, 319.96331787109375, 244.16297912597656, 105.55903625488281, 418.02264404296875, -50.043731689453125, -152.6901092529297, 90.22021484375, 500.9080505371094, 260.4515380859375, 159.23570251464844, 146.75387573242188, 143.97592163085938, 142.24676513671875, 307.35992431640625, 552.709716796875, 272.52301025390625, 501.70855712890625, 58.031455993652344, 325.1495666503906, 702.3887939453125, 41.00672912597656, 152.87245178222656, 84.39738464355469, -2.9980697631835938, 210.158935546875, 297.26513671875, 433.20416259765625, 51.7327995300293, 12.235160827636719, 656.7680053710938, 249.7369842529297, 103.86639404296875, -28.294069290161133, 194.78012084960938, 266.89453125, 235.22332763671875, -24.685195922851562, 644.2841796875, -8.950370788574219, 438.52862548828125, 447.9448547363281, -52.82843017578125, 232.01412963867188, 110.50214385986328, 129.96847534179688, 439.2981872558594, 273.1465759277344, 255.57757568359375, 285.37176513671875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000448.npy"} +{"epoch": 0.657856093979442, "step": 449, "batch_size": 64, "mean": 189.0150146484375, "std": 237.73338317871094, "min": -533.3011474609375, "p10": -30.361655044555658, "median": 148.49334716796875, "p90": 476.57139587402355, "max": 1022.371826171875, "pos_frac": 0.796875, "sample": [186.5131072998047, -16.955711364746094, 197.51197814941406, 123.78190612792969, 509.84539794921875, 430.5707092285156, 90.57611083984375, -33.15296173095703, 152.54962158203125, 180.76919555664062, 455.3044738769531, 0.6898689270019531, -3.6177749633789062, 96.47077178955078, 441.4679870605469, 54.05100631713867, 323.5574951171875, 144.43707275390625, 67.41209411621094, -47.59599304199219, 324.1219787597656, 180.46949768066406, 411.42510986328125, 405.2969055175781, 91.07752990722656, 485.685791015625, 61.978485107421875, 31.671951293945312, 206.62078857421875, 186.17837524414062, 171.35899353027344, 72.77145385742188, -50.078857421875, 345.359375, 334.30419921875, 542.4076538085938, -36.565574645996094, 80.88969421386719, 15.25030517578125, -85.74839782714844, 1022.371826171875, 425.6134948730469, -172.56260681152344, 252.905517578125, 239.53424072265625, 130.45770263671875, -8.3604736328125, -533.3011474609375, -23.84860610961914, 660.6071166992188, 209.50930786132812, 559.0300903320312, 138.31993103027344, -2.4609851837158203, 173.44480895996094, 132.88336181640625, 326.5855712890625, 784.7911376953125, 243.338134765625, 70.23454284667969, 46.08958435058594, -13.12835693359375, 97.26383972167969, 208.98089599609375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000449.npy"} +{"epoch": 0.6593245227606461, "step": 450, "batch_size": 64, "mean": 211.74847412109375, "std": 190.30068969726562, "min": -435.5663757324219, "p10": 27.394324111938484, "median": 213.73578643798828, "p90": 428.4889617919922, "max": 698.6475830078125, "pos_frac": 0.921875, "sample": [65.41959381103516, 290.73968505859375, 431.5873718261719, 385.7952575683594, 54.151123046875, 525.8512573242188, 694.5758056640625, -80.17707824707031, 162.29769897460938, -59.708518981933594, 358.6595458984375, 463.03973388671875, 362.7852783203125, 95.54427337646484, 254.09735107421875, 421.25933837890625, 116.73992919921875, 177.9871826171875, 202.13739013671875, 91.49545288085938, 138.5310821533203, 218.38531494140625, 228.618408203125, 372.2243957519531, 276.864013671875, 74.91815185546875, 139.69627380371094, 577.9442138671875, 240.85169982910156, 311.591796875, 145.95555114746094, 360.15057373046875, 236.88487243652344, 71.78469848632812, 281.31829833984375, 23.597003936767578, 66.71299743652344, 409.22332763671875, 238.025146484375, 90.70455932617188, 572.4627685546875, -103.22969055175781, 204.78683471679688, 289.3708801269531, 179.10452270507812, 270.15325927734375, 8.891716003417969, 269.56610107421875, 216.110107421875, 68.03086853027344, 316.3887634277344, 97.54733276367188, 247.115234375, 179.37538146972656, 36.254737854003906, -435.5663757324219, 211.36146545410156, 243.28208923339844, -58.30290985107422, 135.38087463378906, 104.60779571533203, 698.6475830078125, 51.99750518798828, 230.30519104003906], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000450.npy"} +{"epoch": 0.6607929515418502, "step": 451, "batch_size": 64, "mean": 179.7703399658203, "std": 207.10409545898438, "min": -573.5909423828125, "p10": -30.209365844726555, "median": 177.71124267578125, "p90": 450.13643188476584, "max": 585.7369384765625, "pos_frac": 0.859375, "sample": [95.37124633789062, 113.47552490234375, 131.43734741210938, 129.919921875, 291.4880065917969, 164.06369018554688, 220.53004455566406, 385.9455261230469, 92.4418716430664, 293.54693603515625, 383.1892395019531, 554.0406494140625, 390.0594787597656, 498.2874755859375, 179.92474365234375, 177.72171020507812, 47.25884246826172, -573.5909423828125, 5.989013671875, -73.507568359375, 106.22063446044922, 136.04379272460938, 292.1950988769531, 398.698486328125, 280.8722229003906, 182.20811462402344, 203.47772216796875, 168.9230194091797, 177.70077514648438, 470.4375915527344, 78.32827758789062, 585.7369384765625, 557.6493530273438, 94.90545654296875, 320.68109130859375, 281.6326599121094, 237.33590698242188, -23.531723022460938, 265.41143798828125, 101.03651428222656, -33.07121276855469, 16.12728500366211, 402.7670593261719, -389.17510986328125, 169.8977508544922, 148.088623046875, 184.17453002929688, 141.77017211914062, 202.0961151123047, 46.818580627441406, 250.97137451171875, -86.43778991699219, -133.81082153320312, 493.89630126953125, 366.10797119140625, 94.61282348632812, 349.2620849609375, -142.48434448242188, 105.92153930664062, 10.547983169555664, 472.6391296386719, -1.5492744445800781, 208.42303466796875, 204.1517333984375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000451.npy"} +{"epoch": 0.6622613803230544, "step": 452, "batch_size": 64, "mean": 198.1666717529297, "std": 172.41441345214844, "min": -137.65011596679688, "p10": 10.363523292541505, "median": 196.70263671875, "p90": 411.28611450195314, "max": 720.6454467773438, "pos_frac": 0.90625, "sample": [288.0047302246094, -30.557289123535156, 331.7499084472656, 244.51730346679688, 33.332801818847656, 330.84423828125, 287.3601989746094, 306.3830261230469, 267.19293212890625, 23.39301300048828, 584.8952026367188, 224.51202392578125, 54.430389404296875, 89.02435302734375, 322.295166015625, 147.0830078125, 345.08148193359375, 262.0606689453125, 353.66497802734375, 10.48115348815918, 228.74691772460938, 47.905006408691406, 285.91290283203125, 190.22613525390625, 229.81785583496094, 528.9066772460938, 36.25484085083008, -137.65011596679688, 310.47381591796875, 80.3917236328125, -6.280120849609375, 351.89971923828125, 59.50906753540039, 203.17913818359375, 532.2604370117188, 132.7943572998047, 218.4247283935547, 290.46917724609375, 125.74730682373047, 105.93883514404297, 73.22549438476562, 10.3131103515625, 93.7281494140625, 513.5943603515625, 92.41175079345703, 255.89666748046875, 66.0009765625, 403.3971862792969, 720.6454467773438, 117.94963073730469, 133.70384216308594, 207.01971435546875, 273.50341796875, 42.617671966552734, -110.60336303710938, 132.00396728515625, -18.190643310546875, 414.6670837402344, 158.13760375976562, 462.5234375, 63.992469787597656, 259.7178039550781, -17.078414916992188, 42.81212615966797], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000452.npy"} +{"epoch": 0.6637298091042585, "step": 453, "batch_size": 64, "mean": 178.55931091308594, "std": 199.5140380859375, "min": -197.6995849609375, "p10": -43.82433357238769, "median": 141.61947631835938, "p90": 427.7690185546876, "max": 1013.6514282226562, "pos_frac": 0.828125, "sample": [300.71881103515625, 409.04119873046875, 11.657600402832031, 110.8843002319336, 264.30548095703125, 159.35401916503906, 154.798095703125, 195.4018096923828, 100.00822448730469, 222.05340576171875, 227.10311889648438, 247.22695922851562, 298.64892578125, 391.1809387207031, 156.69952392578125, 2.3524322509765625, 307.560302734375, 207.2918701171875, -36.55281066894531, 22.692276000976562, 268.56256103515625, -102.37712097167969, 106.5615005493164, 96.34117889404297, 463.9990539550781, -120.02332305908203, 105.92304992675781, 305.5858154296875, -197.6995849609375, 251.69189453125, 6.0849761962890625, 105.77536010742188, 270.26129150390625, 510.3563537597656, 487.44677734375, 58.76784133911133, -55.93669128417969, 102.34747314453125, 1013.6514282226562, 398.2777099609375, -2.599700927734375, -90.55562591552734, 505.28216552734375, 54.57549285888672, 115.25151062011719, 40.50532531738281, 346.1927490234375, 57.18152618408203, 314.76470947265625, 300.6126708984375, -16.255874633789062, 258.00518798828125, -46.94070053100586, 325.314208984375, 148.01235961914062, -51.62671661376953, 15.050460815429688, 122.19768524169922, 472.82611083984375, 435.79522705078125, -9.818897247314453, 135.22659301757812, 55.110443115234375, 115.6652603149414], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000453.npy"} +{"epoch": 0.6651982378854625, "step": 454, "batch_size": 64, "mean": 207.1181182861328, "std": 214.07447814941406, "min": -297.88037109375, "p10": -13.792335319519033, "median": 176.29754638671875, "p90": 446.30006713867203, "max": 847.3021850585938, "pos_frac": 0.875, "sample": [151.7718505859375, -18.177978515625, 553.3685913085938, 143.5554656982422, 365.0036926269531, 73.32118225097656, 370.7490539550781, 214.53050231933594, 61.48807144165039, 232.7450714111328, -297.88037109375, 239.1831512451172, 237.89096069335938, 236.04888916015625, 145.67864990234375, 73.187744140625, 164.82249450683594, 252.0331573486328, -86.25884246826172, -202.82009887695312, 45.36045837402344, 387.26312255859375, -19.135467529296875, 647.2417602539062, -18.667396545410156, 115.3431396484375, 406.93621826171875, -3.5591678619384766, 211.46690368652344, 226.71087646484375, 346.85382080078125, 93.24690246582031, 189.462890625, 144.29083251953125, 391.8184814453125, 118.09664154052734, 175.71334838867188, 105.23626708984375, 305.5648193359375, 210.67124938964844, 306.40863037109375, 385.0947570800781, 37.64802551269531, 254.69796752929688, 29.165313720703125, 463.1702880859375, 847.3021850585938, 176.88174438476562, 129.11460876464844, 145.37588500976562, 699.7149047851562, 837.8787841796875, 36.994606018066406, 291.86944580078125, 110.60518646240234, 390.2252502441406, 259.9045715332031, -111.64584350585938, 240.76422119140625, 72.23463439941406, 131.4861297607422, 28.804363250732422, 478.67254638671875, 23.034229278564453], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000454.npy"} +{"epoch": 0.6666666666666666, "step": 455, "batch_size": 64, "mean": 224.86865234375, "std": 213.12152099609375, "min": -173.48387145996094, "p10": -46.37418365478514, "median": 204.05608367919922, "p90": 545.7261657714844, "max": 728.8853149414062, "pos_frac": 0.84375, "sample": [-5.203052520751953, 101.08489990234375, 348.3406677246094, 503.2904052734375, -124.72215270996094, 249.29696655273438, 471.2523193359375, -50.991119384765625, 539.51806640625, -85.4223403930664, 607.3569946289062, 157.01979064941406, 500.339599609375, -57.573944091796875, 613.3450927734375, -78.8430404663086, 223.83053588867188, 597.9225463867188, 39.078269958496094, -35.60133361816406, -2.2869873046875, 323.7249755859375, 161.75218200683594, 55.95612335205078, 281.2347412109375, 728.8853149414062, 249.36631774902344, 35.18451690673828, 203.1341094970703, 164.83958435058594, 292.71502685546875, 413.3763427734375, 195.8524169921875, 204.97805786132812, 101.3697509765625, 136.20941162109375, 150.51153564453125, 247.8885498046875, 66.06588745117188, -173.48387145996094, 240.94174194335938, 40.86898422241211, 299.67327880859375, 413.54254150390625, 18.893997192382812, 83.54923248291016, -55.18408203125, 278.142578125, 490.5272216796875, 286.3362121582031, 469.0683288574219, 65.1954345703125, 227.7296905517578, 353.9836120605469, 120.20502471923828, 77.97135925292969, 11.446735382080078, 553.7671508789062, 298.48431396484375, 218.6201171875, 548.3867797851562, 178.81594848632812, 196.27757263183594, 623.7571411132812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000455.npy"} +{"epoch": 0.6681350954478708, "step": 456, "batch_size": 64, "mean": 226.49349975585938, "std": 180.54428100585938, "min": -11.492057800292969, "p10": 31.54075164794922, "median": 175.43710327148438, "p90": 508.6656402587891, "max": 710.1380615234375, "pos_frac": 0.96875, "sample": [51.70040512084961, 105.60722351074219, 500.90106201171875, 243.95071411132812, 29.56756591796875, 216.200439453125, 45.195980072021484, 142.74099731445312, 155.98733520507812, 421.62652587890625, 306.1802978515625, 16.91588020324707, 137.1795196533203, 469.8684997558594, 152.33819580078125, 179.83416748046875, 62.74327087402344, 129.78494262695312, 216.7133026123047, 226.48388671875, 94.43629455566406, 39.48869323730469, 79.369384765625, 567.7109375, -5.980747222900391, 331.4661560058594, 528.4524536132812, 215.80126953125, 432.17071533203125, 98.04611206054688, 171.0400390625, 339.1953430175781, 393.8377380371094, 201.57745361328125, 102.34004974365234, 62.435890197753906, 36.14485168457031, 270.2103271484375, 580.729248046875, 38.471466064453125, 480.108154296875, 196.64495849609375, 243.2875213623047, 108.84545135498047, 8.44021987915039, 66.4075927734375, 16.75617027282715, 82.00997924804688, 531.6608276367188, 164.03054809570312, 710.1380615234375, 488.02056884765625, 225.06295776367188, 2.5065689086914062, 155.93325805664062, 162.43238830566406, 376.17291259765625, 114.07400512695312, 273.11322021484375, -11.492057800292969, 434.9853820800781, 511.9933166503906, 536.3172607421875, 229.671875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000456.npy"} +{"epoch": 0.6696035242290749, "step": 457, "batch_size": 64, "mean": 202.02642822265625, "std": 180.33663940429688, "min": -236.7521209716797, "p10": -48.637748718261676, "median": 222.5390167236328, "p90": 435.6202728271485, "max": 668.7362060546875, "pos_frac": 0.875, "sample": [96.62686920166016, 256.5220642089844, 263.781982421875, 133.48751831054688, 240.68080139160156, 294.5308532714844, 285.3931884765625, 439.5437927246094, 184.70697021484375, 44.62214660644531, 668.7362060546875, 358.7887878417969, 54.23538589477539, -98.76387023925781, 183.4056854248047, 235.60353088378906, 306.81488037109375, 216.25247192382812, 461.2536315917969, 231.95884704589844, 502.90789794921875, -2.4278221130371094, 177.26332092285156, 256.21746826171875, 306.8473205566406, 426.46539306640625, -117.21095275878906, 15.563762664794922, 391.7053527832031, 396.0208740234375, 97.84577941894531, 149.1243896484375, 100.55229187011719, 270.6108703613281, 228.3530731201172, -157.8844757080078, 320.1311340332031, 293.1592712402344, 74.85549926757812, 53.96044921875, 211.47242736816406, 464.02178955078125, 216.72496032714844, -68.44200134277344, 32.980960845947266, -72.58201599121094, 347.34686279296875, 84.84015655517578, 332.90020751953125, 301.78485107421875, 15.600896835327148, 318.7688293457031, 229.41360473632812, 543.2159423828125, -236.7521209716797, 454.52703857421875, 128.0685272216797, -76.0687484741211, 165.3560791015625, 65.01481628417969, 46.34589385986328, 256.8381042480469, 139.338134765625, 386.7333679199219], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000457.npy"} +{"epoch": 0.671071953010279, "step": 458, "batch_size": 64, "mean": 219.8893585205078, "std": 213.11241149902344, "min": -152.7190704345703, "p10": -35.4969554901123, "median": 222.78020477294922, "p90": 417.5255584716797, "max": 1059.0955810546875, "pos_frac": 0.796875, "sample": [-93.52735900878906, 179.91815185546875, 321.850830078125, 181.0861358642578, 208.5675811767578, -38.68748474121094, 1059.0955810546875, -152.7190704345703, 304.151123046875, 291.3101501464844, 74.37688446044922, 248.65234375, 419.3243408203125, 257.1470642089844, 527.7327880859375, -117.3779296875, -28.052387237548828, -1.3876895904541016, 166.0142364501953, 239.13392639160156, 374.9708557128906, 413.3283996582031, 219.766357421875, 378.768798828125, 210.0472869873047, 169.5498504638672, 258.1395568847656, -16.52655029296875, 273.0946044921875, 214.98080444335938, 252.799072265625, 583.2293701171875, 354.2288818359375, 456.0495300292969, 219.86785888671875, -20.80352783203125, 244.08560180664062, 46.44508361816406, 454.60400390625, 329.705810546875, 355.7037048339844, 230.58767700195312, -1.5791549682617188, -102.2908706665039, 204.2532958984375, -104.33074188232422, 295.1870422363281, -27.65343475341797, 147.96205139160156, 840.8890380859375, 197.84385681152344, 240.25454711914062, 225.6925506591797, 131.44773864746094, 85.99392700195312, 338.0124816894531, 356.61505126953125, 241.158447265625, 136.86634826660156, 330.8206787109375, 106.1891098022461, 322.3741760253906, -151.57943725585938, 209.56044006347656], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000458.npy"} +{"epoch": 0.6725403817914831, "step": 459, "batch_size": 64, "mean": 197.2677459716797, "std": 180.34136962890625, "min": -257.146728515625, "p10": -0.479778289794919, "median": 179.1401138305664, "p90": 436.7090759277345, "max": 603.4029541015625, "pos_frac": 0.890625, "sample": [42.096771240234375, 121.3052978515625, 249.8787841796875, 243.1942138671875, 2.3682174682617188, 176.40283203125, -35.24753189086914, 50.90287780761719, 299.06610107421875, 213.20265197753906, 295.1786804199219, 150.8634033203125, 282.7183837890625, 379.9508056640625, -3.5007801055908203, 264.75244140625, 97.2275390625, 232.57815551757812, 29.20867156982422, 548.7767944335938, -1.700347900390625, 576.9785766601562, 451.3857421875, 12.364910125732422, 187.22097778320312, 73.68318176269531, 355.00177001953125, 113.04570007324219, 136.8999481201172, 232.31085205078125, 160.52308654785156, 193.0103759765625, 471.8278503417969, -257.146728515625, 601.9158325195312, 177.0044708251953, 603.4029541015625, 194.49261474609375, 130.87432861328125, -58.698333740234375, 118.97032928466797, 95.94576263427734, 194.07440185546875, 98.296630859375, 4.78814697265625, 181.2757568359375, -175.52114868164062, 234.1324005126953, 111.85304260253906, 408.5284423828125, 386.0465087890625, 111.00302124023438, 119.61389923095703, 361.3743896484375, 166.22866821289062, 254.68136596679688, 362.7279968261719, -91.08246612548828, 445.8976745605469, 398.5314025878906, 415.2690124511719, 115.82147216796875, 17.23532485961914, 294.12115478515625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000459.npy"} +{"epoch": 0.6740088105726872, "step": 460, "batch_size": 64, "mean": 217.61569213867188, "std": 188.01373291015625, "min": -166.32125854492188, "p10": -19.216245841979962, "median": 210.1498565673828, "p90": 469.58610229492194, "max": 724.5175170898438, "pos_frac": 0.875, "sample": [544.4096069335938, 262.9449462890625, 180.199951171875, 46.570831298828125, 61.19017028808594, 438.41241455078125, 495.3105773925781, 251.1958465576172, -96.19840240478516, 150.4080047607422, 33.89533233642578, 441.49468994140625, -36.08308029174805, 393.4939880371094, 74.70069885253906, 165.05877685546875, 118.67259979248047, 85.02690124511719, -68.26707458496094, 724.5175170898438, 415.7953186035156, 192.4237060546875, 404.1648254394531, 14.9505615234375, 508.8132629394531, 290.8116149902344, 624.501708984375, 218.71726989746094, 185.75863647460938, 25.1240234375, 322.5765380859375, -166.32125854492188, 452.51800537109375, -28.663650512695312, 216.60614013671875, 495.93988037109375, -27.0416316986084, 269.0765075683594, 203.69357299804688, 156.65496826171875, 392.0112609863281, 308.3583679199219, 240.4482879638672, 274.1897888183594, 307.7799072265625, 37.934326171875, 183.8280792236328, 199.4017791748047, 232.19891357421875, 197.7386932373047, -0.9570121765136719, 3.0396957397460938, 47.93547821044922, 136.23007202148438, 11.169540405273438, 178.8758544921875, 322.5728759765625, 229.88494873046875, -102.9302749633789, 337.70611572265625, 307.93505859375, 314.856201171875, 247.24139404296875, 476.9010009765625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000460.npy"} +{"epoch": 0.6754772393538914, "step": 461, "batch_size": 64, "mean": 162.4608154296875, "std": 200.34022521972656, "min": -232.4873046875, "p10": -67.36269683837891, "median": 135.77301025390625, "p90": 421.9129730224611, "max": 686.2587890625, "pos_frac": 0.796875, "sample": [75.997802734375, -208.7497100830078, 210.08175659179688, 130.19424438476562, 136.72943115234375, 363.59228515625, -232.4873046875, 438.21539306640625, 36.761077880859375, 194.2855987548828, 53.757591247558594, -70.75647735595703, 179.7030029296875, 289.7627868652344, 621.582763671875, 241.5246124267578, 548.0975952148438, -46.43109130859375, 26.623306274414062, 98.00599670410156, 90.79151916503906, 323.9621276855469, 359.68505859375, 625.1912231445312, 51.63587951660156, -26.79803466796875, 588.4248046875, 686.2587890625, 172.19970703125, -65.2239990234375, 253.03860473632812, 132.1769561767578, 333.6546630859375, 28.178817749023438, 256.05584716796875, 109.87147521972656, -121.91706085205078, 18.049652099609375, -64.31317138671875, 111.18685913085938, 448.4063720703125, -89.03736114501953, 253.20877075195312, 217.24337768554688, -124.18971252441406, 320.0014343261719, 162.32469177246094, 356.47052001953125, 20.897186279296875, 173.4810791015625, 383.8739929199219, 58.710655212402344, 82.282470703125, 134.81658935546875, 149.6825408935547, -40.35895538330078, -12.161407470703125, 163.14447021484375, 184.4259033203125, 134.02349853515625, 25.673425674438477, -68.27928161621094, 239.2358856201172, 275.0153503417969], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000461.npy"} +{"epoch": 0.6769456681350955, "step": 462, "batch_size": 64, "mean": 159.60598754882812, "std": 174.8517303466797, "min": -143.09434509277344, "p10": -31.727632522582997, "median": 159.97157287597656, "p90": 368.2053924560547, "max": 680.1500854492188, "pos_frac": 0.78125, "sample": [129.588134765625, -21.00670623779297, -6.740501403808594, 44.65653991699219, 425.9406433105469, 372.4906921386719, 37.343170166015625, 175.76132202148438, 345.54083251953125, 240.5588836669922, 149.7391357421875, -118.27149200439453, -36.32231521606445, -143.09434509277344, 107.28118133544922, 288.92620849609375, 358.20635986328125, 168.87295532226562, -1.77923583984375, 148.45309448242188, -11.007301330566406, 218.15032958984375, 218.0526885986328, -113.37481689453125, 25.610095977783203, 40.218475341796875, 300.9476318359375, 44.08757019042969, 31.354976654052734, 46.0328254699707, 680.1500854492188, 199.52676391601562, 75.92054748535156, 180.7506561279297, 418.295166015625, 628.1763916015625, -1.262472152709961, -58.188392639160156, 276.9693298339844, 260.2274169921875, -4.36766242980957, -10.887069702148438, 267.366455078125, 158.8248291015625, 73.07089233398438, 21.488021850585938, 208.39303588867188, 416.4186096191406, -58.017601013183594, 161.11831665039062, 7.973175048828125, 217.4371337890625, 584.867919921875, 266.94842529296875, 317.1620788574219, 200.2823486328125, 276.6795654296875, 207.94139099121094, 88.73822021484375, 313.924072265625, 62.31172180175781, 173.50408935546875, -36.38315963745117, 173.20614624023438], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000462.npy"} +{"epoch": 0.6784140969162996, "step": 463, "batch_size": 64, "mean": 172.01210021972656, "std": 207.21437072753906, "min": -189.68069458007812, "p10": -74.20376358032225, "median": 191.3749771118164, "p90": 417.1373809814453, "max": 663.5060424804688, "pos_frac": 0.765625, "sample": [20.966575622558594, 344.1039733886719, -177.13111877441406, 88.97537231445312, 107.28076934814453, 276.9215087890625, 3.0548019409179688, -40.06580352783203, 224.586181640625, -17.188796997070312, 343.9152526855469, -81.05941772460938, 202.8401336669922, 416.2850036621094, 77.67218017578125, 216.28182983398438, 235.10604858398438, 196.4951629638672, 237.43260192871094, 164.9037628173828, 149.22042846679688, 663.5060424804688, 130.62301635742188, 137.90347290039062, 417.502685546875, 331.3893127441406, -1.1394920349121094, -58.207237243652344, -9.248016357421875, 257.5635070800781, -189.68069458007812, -2.0316543579101562, -189.21902465820312, -99.41613006591797, 558.3121337890625, -185.30270385742188, 36.06614685058594, 24.266571044921875, 232.70303344726562, 9.646533966064453, 282.58306884765625, 273.2240905761719, 440.106689453125, 201.91995239257812, 409.7098388671875, 544.0153198242188, 79.01156616210938, 202.3817138671875, 352.8028869628906, -47.94648742675781, 300.2002258300781, 642.4052734375, 198.5665740966797, 317.5634765625, -159.76336669921875, -49.0198974609375, 186.25479125976562, 128.8150634765625, 12.073745727539062, 275.7175598144531, 245.41940307617188, 392.8002014160156, 622.0045776367188, 102.09371948242188], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000463.npy"} +{"epoch": 0.6798825256975036, "step": 464, "batch_size": 64, "mean": 245.82643127441406, "std": 217.58596801757812, "min": -224.944091796875, "p10": 11.478487586975099, "median": 227.26953125, "p90": 488.54268188476567, "max": 981.2280883789062, "pos_frac": 0.90625, "sample": [-122.94520568847656, 465.4309997558594, 240.62823486328125, 79.8419189453125, 182.6328125, 444.8285827636719, 261.44647216796875, 10.959096908569336, 62.606292724609375, 227.36737060546875, 423.95770263671875, 415.2826232910156, 37.745697021484375, 418.5865783691406, 264.0387878417969, 357.43658447265625, 342.65045166015625, 97.89961242675781, 79.578125, 490.93072509765625, 395.265380859375, 227.17169189453125, 513.0964965820312, -100.28703308105469, 499.236572265625, 116.8779067993164, 223.94436645507812, 452.57647705078125, -16.724876403808594, 89.82910919189453, 164.13104248046875, 281.5270080566406, 186.9243621826172, 40.27109146118164, 67.38997650146484, 335.7755432128906, 438.836669921875, 286.5552978515625, 44.3828125, 145.9288330078125, 76.7613525390625, 169.80136108398438, 412.24908447265625, 58.68645477294922, 84.30104064941406, 370.3430480957031, -197.5013427734375, 215.20980834960938, 12.690399169921875, 443.3387145996094, 529.3638305664062, 677.9581298828125, 981.2280883789062, 386.5057373046875, 155.37872314453125, 445.25494384765625, 498.89898681640625, -28.360736846923828, 482.9705810546875, 400.75274658203125, 360.3410949707031, -224.944091796875, 214.96083068847656, 33.090614318847656], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000464.npy"} +{"epoch": 0.6813509544787077, "step": 465, "batch_size": 64, "mean": 233.79568481445312, "std": 187.0790557861328, "min": -139.49642944335938, "p10": 1.4830329895019556, "median": 216.17455291748047, "p90": 478.8786682128908, "max": 754.4774169921875, "pos_frac": 0.90625, "sample": [401.29376220703125, 434.10589599609375, 186.06011962890625, 129.9882354736328, -88.89615631103516, 87.07925415039062, 347.82098388671875, 94.1968994140625, 183.81304931640625, 244.13360595703125, 300.6265563964844, 142.03375244140625, 240.00741577148438, 62.20787811279297, 493.1154479980469, 553.9782104492188, 282.984375, 388.7868957519531, 298.1922607421875, 141.26800537109375, -27.344566345214844, 445.6595153808594, 496.60797119140625, 181.32861328125, 754.4774169921875, -19.332015991210938, 365.4598388671875, 63.00750732421875, 78.60704803466797, 233.2059326171875, 358.3880920410156, 314.3184814453125, 88.0031509399414, 180.03988647460938, 175.99766540527344, 425.0848388671875, 0.5009536743164062, 153.65283203125, 67.43075561523438, 299.2400207519531, 396.6731872558594, 191.64566040039062, 195.85488891601562, -135.946044921875, 217.68063354492188, 48.04167175292969, 113.03517150878906, 336.78631591796875, 68.55445861816406, 304.0162353515625, 571.024169921875, -62.77545166015625, 562.5277099609375, -139.49642944335938, 530.800537109375, 214.66847229003906, 3.7745513916015625, 210.450927734375, 230.27565002441406, 383.05340576171875, 38.53092956542969, 365.9171142578125, 437.25634765625, 323.44403076171875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000465.npy"} +{"epoch": 0.6828193832599119, "step": 466, "batch_size": 64, "mean": 199.3838348388672, "std": 207.38125610351562, "min": -315.7829895019531, "p10": -33.507534027099595, "median": 211.75064086914062, "p90": 480.7665161132815, "max": 684.7955932617188, "pos_frac": 0.828125, "sample": [305.093505859375, 30.061935424804688, 176.8133544921875, -80.03112030029297, 343.05938720703125, 273.23834228515625, 270.1510925292969, 285.68939208984375, 322.0526123046875, -39.67396545410156, 73.78510284423828, 223.58731079101562, -132.98696899414062, 608.4063720703125, -2.9298973083496094, 273.21612548828125, 219.98948669433594, 35.03627395629883, 315.6721496582031, 216.2191162109375, 234.56234741210938, 189.88421630859375, 215.7392120361328, 286.1783447265625, 566.7505493164062, 186.41339111328125, 121.82160186767578, 413.8915710449219, 207.76206970214844, 374.4196472167969, 82.5804443359375, 199.7413330078125, 241.05389404296875, 546.4823608398438, -303.4472961425781, 684.7955932617188, 268.7365417480469, 576.90234375, 356.1783447265625, -19.11919403076172, 308.7884826660156, 6.454746246337891, 164.482666015625, 183.03665161132812, 418.87957763671875, 620.7033081054688, -11.34323501586914, -111.30768585205078, -2.6802444458007812, 138.087646484375, 326.3826904296875, 127.21638488769531, 130.7569580078125, 52.000083923339844, 312.7172546386719, 345.15130615234375, -315.7829895019531, 24.13519287109375, 507.28948974609375, 23.761154174804688, 249.157958984375, -58.55571746826172, 159.68936157226562, 13.767791748046875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000466.npy"} +{"epoch": 0.684287812041116, "step": 467, "batch_size": 64, "mean": 220.30905151367188, "std": 247.68508911132812, "min": -287.94500732421875, "p10": -79.4729766845703, "median": 215.39796447753906, "p90": 538.1967346191407, "max": 797.3414306640625, "pos_frac": 0.796875, "sample": [428.61993408203125, 489.6570739746094, 431.02459716796875, 545.1657104492188, -184.8079376220703, -287.94500732421875, 258.8238525390625, 228.86264038085938, 78.641357421875, 190.92501831054688, 188.94619750976562, 342.14361572265625, 596.2340087890625, 25.51606559753418, 308.78765869140625, 384.36053466796875, 116.39813995361328, 238.19444274902344, 713.1083374023438, 439.49969482421875, 324.517578125, 65.34104919433594, 103.61138916015625, 352.8636779785156, -89.59352111816406, -68.38583374023438, -33.45549011230469, 255.48336791992188, 662.7537841796875, 252.0650177001953, 475.4610290527344, 121.27897644042969, 261.0961608886719, -33.99658966064453, 521.935791015625, 63.027313232421875, 691.391845703125, 49.459224700927734, 137.8114013671875, 405.36492919921875, 169.51528930664062, -236.6595458984375, 201.56524658203125, -25.095062255859375, 197.1024932861328, 221.87640380859375, -148.61395263671875, -84.224609375, -40.843994140625, 464.1281433105469, 338.0572509765625, 312.03759765625, 255.44400024414062, 62.85396957397461, 797.3414306640625, 208.8599853515625, 425.3260498046875, 208.91952514648438, -64.50093078613281, 674.461181640625, -246.56723022460938, 38.70858383178711, 40.65569305419922, 279.24383544921875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000467.npy"} +{"epoch": 0.6857562408223201, "step": 468, "batch_size": 64, "mean": 222.0347900390625, "std": 219.2637176513672, "min": -237.18484497070312, "p10": -21.60180130004882, "median": 216.81076049804688, "p90": 484.99548950195316, "max": 782.5230712890625, "pos_frac": 0.84375, "sample": [353.55615234375, -103.59723663330078, 711.8015747070312, 254.339599609375, 212.71078491210938, 436.6739501953125, 301.71710205078125, 438.0295715332031, 409.9670104980469, 415.0943298339844, 320.2381896972656, 95.38533020019531, 85.70339965820312, 144.84841918945312, -29.862760543823242, 87.25482177734375, 94.78059387207031, 80.65503692626953, 342.96246337890625, 152.08316040039062, 125.45122528076172, 34.15925216674805, -25.23255157470703, 220.91073608398438, 293.7357177734375, 285.0889587402344, 237.4871826171875, 649.5466918945312, 570.9906616210938, 145.96731567382812, -198.52333068847656, -13.130050659179688, 430.0888671875, 228.88125610351562, 319.6607971191406, 268.94403076171875, 202.81094360351562, 180.2322998046875, 470.0733642578125, -208.13662719726562, 671.531005859375, 59.78706359863281, 73.91133117675781, 292.9310607910156, 266.60284423828125, 284.7759094238281, -1.5516357421875, 91.15499877929688, 5.9092254638671875, 782.5230712890625, 175.13088989257812, 571.9894409179688, 198.40805053710938, 248.60037231445312, 96.47858428955078, 431.7669372558594, 368.6437683105469, 491.39068603515625, 300.8111877441406, 40.2471923828125, 71.76008605957031, -237.18484497070312, -87.96214294433594, -10.776435852050781], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000468.npy"} +{"epoch": 0.6872246696035242, "step": 469, "batch_size": 64, "mean": 228.04100036621094, "std": 213.21047973632812, "min": -195.20668029785156, "p10": -0.8166275024414025, "median": 170.4022445678711, "p90": 515.3675842285157, "max": 736.1522216796875, "pos_frac": 0.890625, "sample": [414.03424072265625, 240.14004516601562, 88.01912689208984, 167.1030731201172, 421.58453369140625, 46.261924743652344, 473.1951904296875, 145.2135009765625, 86.55232238769531, 150.90997314453125, 318.5885314941406, 72.73829650878906, 250.92491149902344, 64.32168579101562, 719.6549072265625, 152.60198974609375, 427.0988464355469, 370.0284729003906, 290.94183349609375, 369.5900573730469, 717.219970703125, 260.4248962402344, 143.18711853027344, 223.00218200683594, 736.1522216796875, -2.369659423828125, 423.0240478515625, 134.62265014648438, 523.1778564453125, -178.39031982421875, 160.60018920898438, 143.06871032714844, 286.8645324707031, 25.18398094177246, 230.1233673095703, 40.08635711669922, 138.88839721679688, -26.388092041015625, 2.8071136474609375, 12.925674438476562, 270.9479675292969, 3.5295791625976562, 591.34912109375, -46.82158279418945, -74.42054748535156, 181.09295654296875, 338.8468017578125, 420.51739501953125, 238.5415496826172, -84.21977996826172, 273.27252197265625, 497.14361572265625, 157.62686157226562, 637.4635009765625, 621.5118408203125, 389.8320617675781, 61.709556579589844, 310.62896728515625, 146.33799743652344, 173.701416015625, 160.53115844726562, 157.70201110839844, 99.29148864746094, -195.20668029785156], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000469.npy"} +{"epoch": 0.6886930983847284, "step": 470, "batch_size": 64, "mean": 231.9114990234375, "std": 215.24620056152344, "min": -509.2549133300781, "p10": -18.699761199951162, "median": 185.7377471923828, "p90": 516.285223388672, "max": 690.5436401367188, "pos_frac": 0.875, "sample": [123.10514068603516, 297.7614440917969, 136.81814575195312, 68.20459747314453, 185.89773559570312, 121.22515106201172, 241.82876586914062, 86.66646575927734, 79.31588745117188, -61.37867736816406, 285.593994140625, 107.23503112792969, 394.2846984863281, 178.08180236816406, 345.5799865722656, 540.7740478515625, 11.51753044128418, 528.510498046875, -509.2549133300781, 399.33502197265625, 171.55599975585938, 201.01336669921875, 148.540283203125, -92.49297332763672, 268.31488037109375, 153.35000610351562, 26.441043853759766, 275.4842834472656, 175.67547607421875, 111.0268783569336, 385.002685546875, -93.23396301269531, 552.1493530273438, 124.40040588378906, 31.87346839904785, 395.420166015625, 171.99078369140625, 487.75958251953125, 620.0655517578125, 330.66802978515625, 406.7095947265625, 149.96792602539062, 429.2869873046875, 452.7469787597656, -29.134384155273438, 664.0601806640625, 259.1851806640625, 173.52720642089844, 482.7796936035156, 450.876953125, 141.127197265625, -10.307456970214844, 185.5777587890625, 213.17141723632812, 375.3680419921875, 591.4820556640625, -39.32386779785156, 690.5436401367188, 370.5793762207031, -22.296463012695312, 452.85308837890625, 257.839111328125, 51.93993377685547, 137.66873168945312], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000470.npy"} +{"epoch": 0.6901615271659325, "step": 471, "batch_size": 64, "mean": 194.92843627929688, "std": 194.93287658691406, "min": -214.9939422607422, "p10": -27.46202049255371, "median": 215.5313720703125, "p90": 411.9381561279297, "max": 793.2521362304688, "pos_frac": 0.796875, "sample": [407.36309814453125, -11.055068969726562, 216.897705078125, 451.3854675292969, -11.3704833984375, 328.10479736328125, 350.556396484375, 763.2439575195312, 374.9627685546875, 69.928466796875, 183.68382263183594, 216.58404541015625, 247.5763702392578, 245.78402709960938, 309.1307067871094, -19.08133316040039, -58.00381088256836, -84.82695007324219, 159.49981689453125, 338.4159240722656, 793.2521362304688, 93.97877502441406, 109.41976165771484, 346.3930358886719, 323.0606994628906, -18.54962730407715, -69.68484497070312, -1.3185710906982422, 11.445259094238281, 252.91754150390625, -26.08235740661621, 497.04022216796875, 167.35488891601562, 413.8988952636719, 198.13929748535156, 422.1449279785156, 245.611572265625, 152.83045959472656, 310.3829345703125, 464.6788330078125, 115.7842788696289, 64.0998764038086, 223.17066955566406, 92.44938659667969, 340.10406494140625, 349.8909912109375, 214.47869873046875, -180.40066528320312, 339.667724609375, 30.086349487304688, -214.9939422607422, 123.30289459228516, 50.817237854003906, 305.3660888671875, 42.50982666015625, 238.34088134765625, -28.05330467224121, -33.97291564941406, 252.34222412109375, 238.83718872070312, 239.73301696777344, 291.1462707519531, 1.3701152801513672, 213.64939880371094], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000471.npy"} +{"epoch": 0.6916299559471366, "step": 472, "batch_size": 64, "mean": 245.75267028808594, "std": 246.41600036621094, "min": -113.07247924804688, "p10": -15.030651092529272, "median": 170.50826263427734, "p90": 628.6029724121095, "max": 923.8145751953125, "pos_frac": 0.890625, "sample": [-110.25559997558594, -51.06111145019531, 411.2989807128906, -97.23568725585938, 70.66282653808594, 191.55392456054688, 223.84628295898438, -73.65176391601562, 124.56204223632812, -113.07247924804688, 143.36927795410156, 169.09762573242188, 89.00682067871094, 298.0721740722656, 75.6392822265625, 653.168212890625, 97.16807556152344, 193.65948486328125, 82.53680419921875, 165.6234130859375, 633.5327758789062, 105.72059631347656, 231.016357421875, 38.260711669921875, 163.4168701171875, 43.69367980957031, 119.56829833984375, 141.98147583007812, 164.2117919921875, 533.6415405273438, 27.968765258789062, 472.9873352050781, 282.9320983886719, 443.5023498535156, 923.8145751953125, 502.87457275390625, 186.9835662841797, 787.551025390625, 246.92808532714844, 118.22256469726562, 457.39971923828125, 525.9580078125, 77.1033935546875, 199.0035858154297, -55.24220275878906, 705.0247802734375, 874.5489501953125, 177.1500701904297, 617.10009765625, 84.03242492675781, 735.6514892578125, 9.941482543945312, 81.03208923339844, -25.732994079589844, 279.736083984375, 362.4664001464844, 458.54937744140625, 424.7052307128906, 72.74407958984375, 300.93402099609375, 171.9188995361328, 64.03507995605469, 286.6076354980469, 130.7057342529297], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000472.npy"} +{"epoch": 0.6930983847283406, "step": 473, "batch_size": 64, "mean": 171.7787322998047, "std": 257.2103271484375, "min": -557.5667724609375, "p10": -119.39866790771484, "median": 181.66602325439453, "p90": 394.3253845214844, "max": 1175.252685546875, "pos_frac": 0.765625, "sample": [-28.520057678222656, 174.505859375, 290.0416259765625, 371.3575439453125, -14.80111312866211, 283.54168701171875, 65.43618774414062, 137.11798095703125, 379.422119140625, 79.86131286621094, 388.97613525390625, 485.7353210449219, 162.7545623779297, 136.2320556640625, -65.353271484375, -71.84949493408203, 305.7230224609375, 20.839431762695312, 239.69895935058594, 316.059814453125, -109.73491668701172, 226.6932830810547, 345.9797668457031, 183.02000427246094, 218.8157196044922, 64.14749908447266, 334.976806640625, 396.617919921875, 172.77316284179688, 186.77774047851562, -279.0030822753906, 178.6749267578125, -175.3741455078125, 121.18531799316406, -557.5667724609375, -217.14097595214844, 1175.252685546875, 319.28497314453125, 278.9308776855469, 677.5392456054688, 240.1459197998047, 247.197265625, 384.430908203125, 326.6485595703125, 153.19825744628906, 171.29830932617188, -64.9049301147461, -398.2160949707031, 38.52912902832031, 306.358642578125, 402.7672119140625, 361.06927490234375, -123.54027557373047, -31.778038024902344, 205.15066528320312, 6.508811950683594, 474.5475769042969, -182.88038635253906, 280.72711181640625, 180.31204223632812, -56.78166198730469, 294.1307678222656, 42.300682067871094, 537.9892578125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000473.npy"} +{"epoch": 0.6945668135095447, "step": 474, "batch_size": 64, "mean": 176.81515502929688, "std": 227.57882690429688, "min": -436.58642578125, "p10": -85.71146011352538, "median": 184.5677032470703, "p90": 441.3908447265627, "max": 771.0702514648438, "pos_frac": 0.796875, "sample": [-217.18959045410156, 214.44964599609375, 68.02182006835938, 350.7601013183594, 310.0132141113281, 63.668617248535156, 520.3884887695312, -6.99748420715332, 213.97415161132812, 370.3933410644531, 349.0927429199219, 604.015869140625, 17.567785263061523, -1.8648910522460938, 99.7851791381836, 310.2358703613281, 104.82510375976562, 3.9169845581054688, -163.47161865234375, 156.50704956054688, 70.25447082519531, 338.5838928222656, -77.19043731689453, 13.23824691772461, -10.578405380249023, 321.417724609375, 205.036865234375, 166.56671142578125, 319.2972412109375, 212.96705627441406, 114.2177734375, -51.665828704833984, 185.75433349609375, 38.436729431152344, 315.02685546875, 178.1985321044922, 391.6110534667969, 107.11248779296875, 38.17854309082031, -193.05364990234375, 228.85069274902344, 183.38107299804688, 367.22125244140625, 40.69293212890625, 107.5539779663086, 132.38015747070312, 461.49981689453125, 669.5146484375, -89.36332702636719, -341.7632751464844, 280.8583984375, 394.46990966796875, 203.01394653320312, 369.86004638671875, 485.7419128417969, 255.6050567626953, 270.90032958984375, -101.90662384033203, 336.8468933105469, -51.22314453125, 200.46188354492188, 525.5877075195312, -436.58642578125, 771.0702514648438], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000474.npy"} +{"epoch": 0.6960352422907489, "step": 475, "batch_size": 64, "mean": 240.01370239257812, "std": 215.73182678222656, "min": -201.08900451660156, "p10": -30.63566265106201, "median": 252.78646087646484, "p90": 505.5507446289063, "max": 853.9320068359375, "pos_frac": 0.8125, "sample": [58.32769775390625, -27.584789276123047, 305.9667053222656, 175.40989685058594, 52.733642578125, 391.32171630859375, 714.6082763671875, 156.60247802734375, 505.77545166015625, 439.3761291503906, 205.29202270507812, 211.34823608398438, 283.9925537109375, 66.78643798828125, 256.5731201171875, -56.548858642578125, 487.8789978027344, 154.3296661376953, 288.1866455078125, -29.746912002563477, 248.9998016357422, 203.0205841064453, 335.8316345214844, 122.50704956054688, -5.341892242431641, 266.5718078613281, 268.7113342285156, 296.01312255859375, 42.6189079284668, 322.531494140625, 108.8655776977539, 386.4356994628906, -201.08900451660156, 458.3860778808594, 376.1508483886719, 505.02642822265625, 260.8340148925781, 332.6156005859375, 144.73512268066406, 683.6857299804688, 237.3523406982422, -3.5515613555908203, 186.44747924804688, -112.32426452636719, 370.0172119140625, 179.26600646972656, 52.10851287841797, -31.016555786132812, 293.58953857421875, 369.663330078125, -31.303985595703125, 265.6091003417969, 404.81011962890625, -28.816131591796875, 583.0791015625, -53.16502380371094, 158.15916442871094, 853.9320068359375, 400.18328857421875, 595.2694702148438, 134.03579711914062, 508.48114013671875, -125.94670104980469, 357.2588195800781], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000475.npy"} +{"epoch": 0.697503671071953, "step": 476, "batch_size": 64, "mean": 204.19937133789062, "std": 200.898193359375, "min": -216.547119140625, "p10": -12.487621688842767, "median": 176.41761016845703, "p90": 520.9554321289064, "max": 743.28369140625, "pos_frac": 0.859375, "sample": [457.7282409667969, -2.217611312866211, 22.599218368530273, 180.79396057128906, 283.5120544433594, 270.70196533203125, 564.7551879882812, 337.3611755371094, 103.35781860351562, -119.72187042236328, 550.9896240234375, 24.161109924316406, 627.4251098632812, 430.99761962890625, 428.44189453125, 301.75054931640625, 357.7235412597656, 29.787620544433594, 224.365234375, 210.15518188476562, 212.9813232421875, 297.14251708984375, 2.5897674560546875, 308.4888916015625, 119.5025634765625, 281.3343505859375, -216.547119140625, -35.77661895751953, 108.03347778320312, 51.518882751464844, 14.310722351074219, 92.76206970214844, -5.233882904052734, -15.596366882324219, 354.718994140625, 56.36430740356445, 91.66449737548828, -30.025835037231445, 152.786865234375, 172.041259765625, 34.005558013916016, 299.1043395996094, 165.45822143554688, -79.67161560058594, 52.88121795654297, 157.61293029785156, 489.65948486328125, 534.3679809570312, 743.28369140625, 622.7210083007812, 186.53663635253906, 107.81233215332031, 191.82809448242188, 251.79608154296875, 122.75605773925781, 82.33272552490234, 251.53497314453125, 352.08740234375, 68.04280090332031, 260.9679870605469, 560.9412841796875, 222.34375, 137.15809631347656, -44.52988815307617], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000476.npy"} +{"epoch": 0.6989720998531571, "step": 477, "batch_size": 64, "mean": 194.8673858642578, "std": 225.46160888671875, "min": -528.9474487304688, "p10": -45.87134399414061, "median": 169.32174682617188, "p90": 481.5807189941406, "max": 798.0501708984375, "pos_frac": 0.859375, "sample": [-51.235107421875, 295.88238525390625, 312.77764892578125, -59.08721160888672, 410.2170715332031, 302.29046630859375, 170.77029418945312, 245.5711669921875, 196.55291748046875, 274.5196228027344, 112.86762237548828, 120.83731842041016, 212.9100799560547, 476.9472351074219, 573.5015869140625, 13.631235122680664, 288.44158935546875, 278.01513671875, 684.959228515625, 148.45611572265625, 202.16539001464844, 79.07113647460938, 18.174819946289062, 177.90476989746094, -80.13363647460938, 328.6650390625, 160.2027587890625, 154.94186401367188, 3.211641311645508, -33.35589599609375, -266.6617736816406, 258.7687072753906, -528.9474487304688, -147.51358032226562, 573.9659423828125, 95.76773071289062, 87.6614990234375, 266.0136413574219, 447.1719970703125, 143.41372680664062, 362.755126953125, 49.60226821899414, 342.6961669921875, 483.5664978027344, 76.17361450195312, 108.1819076538086, 132.05630493164062, 494.28240966796875, 333.7782287597656, 164.71755981445312, 464.5980529785156, -144.06710815429688, 66.97599029541016, 202.1619873046875, 2.8127899169921875, 513.682373046875, 167.87319946289062, -1.662506103515625, 57.027008056640625, 20.059261322021484, 66.7169418334961, 292.20989990234375, 798.0501708984375, 467.949951171875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000477.npy"} +{"epoch": 0.7004405286343612, "step": 478, "batch_size": 64, "mean": 222.39512634277344, "std": 209.9766387939453, "min": -112.31918334960938, "p10": 6.150836753845221, "median": 182.15145111083984, "p90": 486.7217803955079, "max": 908.4031372070312, "pos_frac": 0.90625, "sample": [11.992509841918945, 115.97291564941406, 81.30974578857422, 151.03900146484375, 103.0338363647461, 188.3699493408203, 237.75253295898438, 3.6472625732421875, -0.7252273559570312, 64.81417846679688, 74.33247375488281, 247.70352172851562, -26.58017921447754, 157.62721252441406, 58.59833526611328, 463.96240234375, 277.5025939941406, 662.3779907226562, 103.80363464355469, 202.03836059570312, -50.30405807495117, 27.190460205078125, -112.31918334960938, 281.4485168457031, 273.7748718261719, 641.3804931640625, 378.037109375, 108.12937927246094, -37.612091064453125, 164.6716766357422, 908.4031372070312, 420.2188415527344, 232.7085418701172, 199.17919921875, 297.595703125, 410.7496032714844, 496.4757995605469, 259.92156982421875, 306.7189025878906, 81.67042541503906, 361.9372863769531, 33.23871612548828, 46.410125732421875, -34.091514587402344, 622.711181640625, 239.665771484375, 191.4315185546875, 720.9212646484375, 106.41398620605469, 229.89646911621094, 385.958740234375, 110.27046966552734, 260.93695068359375, 315.0400085449219, 71.19743347167969, 175.93295288085938, 414.3706359863281, 41.42323303222656, 721.7131958007812, 87.69025421142578, 153.19583129882812, 158.0780792236328, 331.8101501464844, 20.523178100585938], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000478.npy"} +{"epoch": 0.7019089574155654, "step": 479, "batch_size": 64, "mean": 181.93515014648438, "std": 203.56866455078125, "min": -525.4420166015625, "p10": -58.45975723266598, "median": 150.37899780273438, "p90": 450.4197418212891, "max": 706.5718383789062, "pos_frac": 0.859375, "sample": [114.0625991821289, 129.51992797851562, 460.4744567871094, 273.0066223144531, 158.80963134765625, 381.2098388671875, 76.66804504394531, -102.96959686279297, 380.87042236328125, 14.164121627807617, 276.3592529296875, 43.153499603271484, 32.161781311035156, 326.62078857421875, 503.2903137207031, 175.9415283203125, 177.9889678955078, -6.34130859375, 15.306663513183594, 53.63089370727539, 56.70223617553711, 386.43927001953125, 108.37937927246094, 92.858154296875, 470.17791748046875, 518.1680908203125, 455.0278015136719, 324.7201232910156, 382.07769775390625, 35.04217529296875, 231.28155517578125, 210.2735595703125, 284.6927490234375, 439.6676025390625, -182.22256469726562, 424.274169921875, -116.90878295898438, 141.9483642578125, 113.97989654541016, 102.52015686035156, 221.2593231201172, 487.10821533203125, 273.78125, 290.39495849609375, 77.23640441894531, -525.4420166015625, 188.11572265625, 102.010986328125, -17.05315399169922, 95.2108154296875, 112.1878662109375, -80.9604263305664, -76.2054443359375, -90.0027847290039, 259.4971923828125, 74.20673370361328, 283.11181640625, 119.39630126953125, 114.15322875976562, 706.5718383789062, 275.32080078125, 439.053466796875, 290.7420654296875, 61.12689971923828], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000479.npy"} +{"epoch": 0.7033773861967695, "step": 480, "batch_size": 64, "mean": 160.836669921875, "std": 215.56460571289062, "min": -323.3348693847656, "p10": -103.43091125488282, "median": 130.41851043701172, "p90": 440.4656982421876, "max": 732.8080444335938, "pos_frac": 0.75, "sample": [-88.00788879394531, 130.2864227294922, -96.92630004882812, 130.55059814453125, 38.4083251953125, 295.2554016113281, 249.4592742919922, 400.0313720703125, -1.4432601928710938, 394.648193359375, 183.57806396484375, 570.9048461914062, -9.657363891601562, -120.48918151855469, 120.95343017578125, -123.83688354492188, 71.39732360839844, 260.4508361816406, 194.4764404296875, 555.1725463867188, 121.31636810302734, -148.33262634277344, 13.788162231445312, 262.58740234375, 116.70928192138672, 85.39964294433594, 270.3026123046875, 37.23277282714844, 376.9348449707031, -149.33778381347656, 459.434814453125, 507.01849365234375, 377.29986572265625, -44.06560134887695, -323.3348693847656, 157.7183837890625, 315.71246337890625, -67.72019958496094, 243.42364501953125, 94.8394775390625, -103.60978698730469, -306.4996337890625, 379.2283630371094, 272.65008544921875, 732.8080444335938, 111.8366470336914, 272.29248046875, 408.22955322265625, -32.682003021240234, 55.157135009765625, -0.12890625, -103.01353454589844, 112.2640151977539, 179.84817504882812, 125.66970825195312, 133.43748474121094, 214.92446899414062, 72.53436279296875, 355.26153564453125, 454.28118896484375, 365.26251220703125, 18.019460678100586, 234.8798065185547, 478.7566223144531], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000480.npy"} +{"epoch": 0.7048458149779736, "step": 481, "batch_size": 64, "mean": 203.39276123046875, "std": 205.274658203125, "min": -250.3341064453125, "p10": -23.511239242553707, "median": 174.77808380126953, "p90": 478.0523376464844, "max": 680.5797119140625, "pos_frac": 0.84375, "sample": [425.4696350097656, -243.87513732910156, 444.738037109375, 96.66219329833984, 143.73118591308594, 114.66133117675781, -25.390609741210938, 223.87057495117188, 244.4488067626953, 592.6151733398438, -4.610542297363281, -116.69204711914062, 40.30757141113281, -19.126041412353516, 244.6562957763672, 138.85623168945312, 56.64262390136719, 307.525390625, 626.6923828125, 36.62049865722656, 146.09075927734375, 324.02374267578125, 203.41146850585938, 372.8591613769531, 193.09329223632812, 124.20870208740234, 137.82437133789062, 318.83575439453125, 78.01223754882812, -141.14736938476562, -93.55253601074219, 331.85687255859375, 100.6768798828125, -250.3341064453125, 399.3703308105469, 246.68353271484375, 72.3333511352539, 183.8480987548828, 395.2924499511719, 267.75823974609375, 106.4730224609375, 165.38099670410156, 322.25140380859375, 480.376953125, -9.234580993652344, 142.00799560546875, 516.0693359375, 110.90824127197266, 680.5797119140625, 537.255859375, 252.59820556640625, 17.69744873046875, 82.96023559570312, 297.9177551269531, 97.95873260498047, 165.70806884765625, -87.98291015625, 233.34970092773438, 452.6274108886719, 319.2756042480469, 602.1741943359375, 122.60038757324219, 196.60536193847656, 472.62823486328125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000481.npy"} +{"epoch": 0.7063142437591777, "step": 482, "batch_size": 64, "mean": 152.8795166015625, "std": 204.54066467285156, "min": -537.4500122070312, "p10": -63.500842285156196, "median": 131.9051742553711, "p90": 397.9768493652344, "max": 765.797607421875, "pos_frac": 0.875, "sample": [309.305419921875, -130.7832489013672, 282.4169616699219, 48.866241455078125, 393.81793212890625, 451.8810729980469, -84.937255859375, 441.1495361328125, 232.10728454589844, 60.37750244140625, 151.46746826171875, 129.83328247070312, 5.788055419921875, 96.0225830078125, 466.78924560546875, 395.55389404296875, 206.849365234375, 209.15159606933594, 167.94810485839844, 26.053722381591797, 17.458080291748047, 21.445526123046875, 176.1149444580078, 399.0152587890625, -189.21983337402344, 94.91091918945312, 4.168632507324219, 375.1807861328125, 269.2716369628906, 48.86368942260742, 191.27566528320312, 284.9295654296875, 386.8037109375, 178.71963500976562, -98.47354125976562, 175.46493530273438, 270.3799133300781, 55.961456298828125, 549.8329467773438, -128.36370849609375, 383.2334289550781, 62.032386779785156, 28.401403427124023, 17.103187561035156, 269.234619140625, 47.3485107421875, -537.4500122070312, 209.67633056640625, 765.797607421875, 46.951683044433594, 28.68328857421875, 24.278060913085938, 42.7606201171875, 146.0282440185547, -167.27392578125, 248.43838500976562, 117.46788024902344, 418.83148193359375, -13.4825439453125, 133.97706604003906, 14.571983337402344, 50.981040954589844, 385.02435302734375, 118.27490234375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000482.npy"} +{"epoch": 0.7077826725403817, "step": 483, "batch_size": 64, "mean": 234.26724243164062, "std": 221.85745239257812, "min": -138.60787963867188, "p10": -31.015588569641107, "median": 209.96392822265625, "p90": 528.8091369628908, "max": 1038.27587890625, "pos_frac": 0.875, "sample": [-50.01142883300781, 72.13851928710938, 485.41217041015625, 414.7188415527344, -47.05375671386719, 385.45391845703125, 211.124267578125, 113.79022216796875, 425.69873046875, 429.8382568359375, 426.86322021484375, 361.5052490234375, 286.21710205078125, 29.613174438476562, 60.303306579589844, 161.25286865234375, 606.8912353515625, 201.3899688720703, -123.6646957397461, 274.88671875, 65.57585906982422, -33.80840301513672, 547.4078369140625, 634.811279296875, 108.1463623046875, 211.88961791992188, 672.981201171875, 288.8323059082031, 69.16578674316406, 464.3782958984375, 6.656085968017578, 588.431396484375, 52.359764099121094, 605.9852905273438, 73.42341613769531, 382.33392333984375, 295.8070373535156, 265.0624084472656, 94.11990356445312, 452.79364013671875, 253.2806396484375, 336.20391845703125, 176.03639221191406, 5.6381072998046875, 107.25553131103516, 74.16675567626953, 130.02102661132812, 141.1010284423828, 308.4382019042969, -41.16487121582031, 202.901611328125, -138.60787963867188, 115.54940795898438, 1038.27587890625, 45.81471252441406, 249.19468688964844, 217.6863250732422, 286.9264831542969, 248.66635131835938, -24.499021530151367, 193.83023071289062, 364.072265625, 208.8035888671875, -79.2081298828125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000483.npy"} +{"epoch": 0.7092511013215859, "step": 484, "batch_size": 64, "mean": 208.47601318359375, "std": 189.2229766845703, "min": -82.94915771484375, "p10": -8.522444152832024, "median": 168.1836166381836, "p90": 472.666259765625, "max": 804.7877197265625, "pos_frac": 0.875, "sample": [278.25323486328125, 112.84919738769531, 250.870849609375, 252.77882385253906, -23.615615844726562, 58.20997619628906, 63.52537536621094, 125.49633026123047, 482.799072265625, 157.44967651367188, 41.54605484008789, 139.43792724609375, 243.28311157226562, 305.4289245605469, 334.55511474609375, 163.6009979248047, 182.04818725585938, 116.14935302734375, -77.25971221923828, -25.764389038085938, 108.49307250976562, 140.8051300048828, 205.30355834960938, 220.53457641601562, 804.7877197265625, 16.90441131591797, 224.3970947265625, 412.15380859375, 11.622142791748047, -24.796707153320312, -82.94915771484375, 172.7662353515625, 473.19195556640625, 579.855712890625, -1.3399887084960938, 212.78768920898438, 103.96356201171875, 132.65650939941406, 141.8304901123047, 117.52210235595703, 103.69861602783203, 66.36165618896484, 434.90191650390625, 248.369384765625, 436.2059326171875, 70.8533935546875, 311.0373840332031, 384.6609191894531, 28.869394302368164, -11.600639343261719, 14.618120193481445, 186.10076904296875, 551.4873046875, 300.1654968261719, 211.23960876464844, 385.67816162109375, 562.38037109375, 212.93096923828125, 448.82916259765625, -34.281150817871094, 615.3173828125, 108.46983337402344, 82.59871673583984, 471.43963623046875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000484.npy"} +{"epoch": 0.71071953010279, "step": 485, "batch_size": 64, "mean": 217.1118621826172, "std": 201.75173950195312, "min": -300.3165588378906, "p10": -14.732728576660152, "median": 224.5414810180664, "p90": 457.9211578369141, "max": 667.8858642578125, "pos_frac": 0.875, "sample": [312.4242858886719, 359.40411376953125, 217.9706573486328, 22.03167724609375, 348.863525390625, 59.938331604003906, 391.4731140136719, -81.3211669921875, 247.6107940673828, 335.1835021972656, 179.74606323242188, 12.585929870605469, 462.0719299316406, 321.33294677734375, 167.807861328125, 640.7903442382812, -164.67420959472656, 82.0689926147461, 429.5142517089844, 383.6461486816406, 409.8385314941406, 364.0584411621094, 113.15460205078125, 304.4648742675781, 667.8858642578125, 620.3289794921875, -16.69084930419922, 21.11795997619629, 283.0500793457031, -97.0459213256836, -46.093788146972656, 104.79852294921875, 97.90796661376953, -21.206703186035156, 231.1123046875, 279.17633056640625, 448.23602294921875, 366.1243896484375, 80.15046691894531, 347.958984375, 168.36322021484375, 93.08596801757812, 466.0157165527344, 265.87445068359375, -10.163780212402344, 286.6485290527344, 31.95296287536621, 166.8349609375, 217.14573669433594, -300.3165588378906, 564.2952270507812, 384.3572082519531, 56.82309341430664, 64.38043212890625, 88.98497009277344, 19.718948364257812, 519.56103515625, 47.89784240722656, 22.175804138183594, 184.08041381835938, 251.38800048828125, 236.0967254638672, 431.65948486328125, 351.5024108886719], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000485.npy"} +{"epoch": 0.7121879588839941, "step": 486, "batch_size": 64, "mean": 215.93392944335938, "std": 235.28794860839844, "min": -309.8589172363281, "p10": -110.2894622802734, "median": 220.2700653076172, "p90": 516.6001922607422, "max": 810.5906982421875, "pos_frac": 0.8125, "sample": [508.1765441894531, 203.58750915527344, 97.06446838378906, 137.29293823242188, 225.18572998046875, 108.83119201660156, 129.61923217773438, -28.417320251464844, 552.0821533203125, 13.59197998046875, 103.06448364257812, 395.1943359375, 377.56719970703125, 95.29668426513672, -197.2149658203125, 670.8529052734375, 427.2476806640625, 298.5594482421875, -162.72850036621094, 534.42724609375, 550.8302612304688, -133.5969696044922, 381.26483154296875, -127.69732666015625, -59.1679573059082, 171.75161743164062, 311.8526306152344, 339.38775634765625, 69.12741088867188, 411.3902587890625, 291.3665771484375, 171.9078369140625, -227.95379638671875, -236.8388671875, 150.20069885253906, 322.7508544921875, 77.39801025390625, 520.2103271484375, 372.6107177734375, 281.3581237792969, 158.76971435546875, 352.6518249511719, 299.914306640625, -69.67111206054688, 327.52978515625, 655.190185546875, 487.1047668457031, 336.4486999511719, 335.5657653808594, 474.35687255859375, -58.696998596191406, 232.50904846191406, 137.40713500976562, 215.35440063476562, 306.71875, 129.06788635253906, 53.57825469970703, 328.2352600097656, -309.8589172363281, 205.96315002441406, 259.1775817871094, -5.996124267578125, 30.42699432373047, 810.5906982421875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000486.npy"} +{"epoch": 0.7136563876651982, "step": 487, "batch_size": 64, "mean": 195.3286895751953, "std": 157.20809936523438, "min": -247.84201049804688, "p10": -12.044841766357422, "median": 204.79669952392578, "p90": 376.5369476318359, "max": 557.9483032226562, "pos_frac": 0.875, "sample": [197.63755798339844, 290.1729431152344, 109.69340515136719, 204.72665405273438, 557.9483032226562, 293.3371887207031, 393.85845947265625, 209.61209106445312, 375.79779052734375, 48.78125762939453, 79.9427490234375, -24.07231903076172, 376.8537292480469, -148.35232543945312, 82.32030487060547, 276.4725341796875, 435.99627685546875, -14.69839096069336, 322.0618591308594, 474.10968017578125, 308.90447998046875, 266.3026123046875, -12.026069641113281, 219.571044921875, 147.9134521484375, 239.48809814453125, 87.70690155029297, 244.90420532226562, 329.4609375, 49.99095916748047, 204.8667449951172, 27.333921432495117, 162.01422119140625, 305.01593017578125, 180.58169555664062, 66.18165588378906, 217.2542724609375, -247.84201049804688, 348.3886413574219, 272.25823974609375, 218.4951629638672, 275.034423828125, 133.6785430908203, 24.863882064819336, 196.52142333984375, 308.5888366699219, 49.04689025878906, 164.38461303710938, 305.8681640625, 128.18496704101562, 160.8418731689453, 188.21810913085938, 283.5787658691406, 316.454833984375, 182.7267608642578, -90.95028686523438, 104.92526245117188, 481.4981384277344, 435.41094970703125, 347.6821594238281, -92.73297119140625, 261.1650085449219, -12.052886962890625, 169.13357543945312], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000487.npy"} +{"epoch": 0.7151248164464024, "step": 488, "batch_size": 64, "mean": 194.02163696289062, "std": 157.0642852783203, "min": -105.20577239990234, "p10": 0.45303401947021715, "median": 191.26122283935547, "p90": 413.15304260253913, "max": 594.043212890625, "pos_frac": 0.890625, "sample": [-1.3490982055664062, 117.24346923828125, 420.19964599609375, 106.42500305175781, 527.559326171875, 27.149215698242188, 337.4900207519531, 26.17670440673828, -0.505889892578125, 199.02008056640625, 168.62344360351562, 154.70103454589844, 183.37625122070312, 167.125732421875, 283.566162109375, 396.7109680175781, 73.54592895507812, 18.39069175720215, -95.43901062011719, 67.28762817382812, 130.4121856689453, -105.20577239990234, 355.4287109375, 145.25665283203125, 34.62956237792969, 144.48687744140625, 184.02926635742188, 288.6602783203125, 594.043212890625, 198.49317932128906, 219.39205932617188, 265.53704833984375, -64.1103286743164, 227.74285888671875, 365.306884765625, 135.54795837402344, 449.4128112792969, 338.6856689453125, 279.8265075683594, 220.42156982421875, 229.14590454101562, 313.72418212890625, 297.3153991699219, 96.07945251464844, 237.52854919433594, 108.12261962890625, 269.60186767578125, 306.0259704589844, -74.04070281982422, 94.2953109741211, 204.67762756347656, 237.13970947265625, 301.2251281738281, 178.2890625, 258.31207275390625, 438.5107727050781, 2.690523147583008, 256.5339050292969, 21.616222381591797, 17.566368103027344, -33.27408981323242, 437.3145446777344, 132.13153076171875, 501.5585021972656], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000488.npy"} +{"epoch": 0.7165932452276065, "step": 489, "batch_size": 64, "mean": 165.90423583984375, "std": 180.9261474609375, "min": -111.59112548828125, "p10": -22.930049514770506, "median": 113.66617202758789, "p90": 414.8149780273438, "max": 660.6674194335938, "pos_frac": 0.875, "sample": [295.60833740234375, -51.39866638183594, 48.86973571777344, 70.62565612792969, 39.133995056152344, 109.873291015625, 45.474361419677734, 370.3742980957031, 56.781463623046875, 107.03387451171875, 53.98805618286133, -21.68313980102539, 224.01394653320312, 117.45905303955078, 240.69601440429688, 7.9579925537109375, 97.12178039550781, 306.7580871582031, 131.78509521484375, 410.8077087402344, 120.64064025878906, 11.107240676879883, 8.669265747070312, 178.51510620117188, 102.86627197265625, 45.37989807128906, 69.14274597167969, 20.54654312133789, 149.3227081298828, 32.03965759277344, 135.9688720703125, 54.54872131347656, 4.667051315307617, 44.642852783203125, 531.7562255859375, 304.61383056640625, 197.5235137939453, 59.14045715332031, 97.19567108154297, 150.94064331054688, 316.9510498046875, -23.464439392089844, 576.6296997070312, -65.49911499023438, 416.5323791503906, 315.28765869140625, 280.85302734375, 483.13458251953125, 131.22532653808594, -34.720489501953125, 640.2001342773438, 262.7815246582031, 212.1165313720703, 660.6674194335938, -106.83284759521484, 40.572959899902344, 381.8904724121094, 234.19265747070312, 370.4249572753906, -45.483123779296875, 13.234085083007812, 417.28033447265625, 270.9788513183594, -111.59112548828125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000489.npy"} +{"epoch": 0.7180616740088106, "step": 490, "batch_size": 64, "mean": 188.4940643310547, "std": 212.68284606933594, "min": -277.67138671875, "p10": -34.141625213623044, "median": 141.40260314941406, "p90": 508.27969665527354, "max": 632.0840454101562, "pos_frac": 0.8125, "sample": [45.47357177734375, -277.67138671875, 419.3055419921875, -2.528390884399414, 82.24627685546875, 40.29016876220703, 237.04742431640625, 200.43405151367188, 379.1888732910156, -151.2664337158203, 22.009262084960938, 60.259613037109375, 430.2144470214844, 125.95101928710938, 417.220947265625, 15.166847229003906, -128.11558532714844, 218.50299072265625, 238.10821533203125, -97.25436401367188, -28.425464630126953, 73.3531494140625, 61.94597625732422, 27.964555740356445, -161.58192443847656, 518.3939208984375, 144.17483520507812, 109.42823791503906, 67.29534149169922, 315.1297302246094, 120.54737091064453, 447.2972412109375, 123.78802490234375, 371.3045959472656, 134.21673583984375, 172.62155151367188, 536.9847412109375, 213.19525146484375, 594.98828125, -13.289653778076172, 318.9638977050781, 211.11285400390625, 554.5198974609375, 535.769775390625, 211.26300048828125, 570.2404174804688, -170.60601806640625, 380.5447692871094, 23.5821533203125, 311.12835693359375, 238.69053649902344, -35.525978088378906, 632.0840454101562, 138.63037109375, 286.5730285644531, 437.259765625, -30.911468505859375, -3.5552406311035156, 210.09420776367188, 484.6798400878906, 43.08564376831055, 382.6088562011719, 122.68511199951172, 106.78704833984375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000490.npy"} +{"epoch": 0.7195301027900147, "step": 491, "batch_size": 64, "mean": 153.2397918701172, "std": 190.87603759765625, "min": -241.13922119140625, "p10": -68.26938400268554, "median": 153.71981811523438, "p90": 380.25315856933594, "max": 713.3764038085938, "pos_frac": 0.78125, "sample": [269.48199462890625, -69.77549743652344, 374.844970703125, 246.191650390625, 399.09210205078125, 157.4333038330078, 87.21649169921875, -72.91822814941406, 154.83949279785156, 21.90802001953125, 181.44419860839844, 189.77040100097656, 25.35669708251953, 79.12541198730469, 187.25473022460938, 13.630725860595703, 152.6001434326172, 237.44967651367188, 75.04743957519531, 249.85037231445312, 126.75640106201172, -64.75511932373047, 382.5709533691406, -40.726409912109375, 169.02780151367188, 135.11497497558594, 399.7670593261719, -165.50469970703125, 370.408203125, 62.48937225341797, 462.03411865234375, 251.70333862304688, -241.13922119140625, 49.90630340576172, 92.23094177246094, 247.40457153320312, -25.063819885253906, 547.3908081054688, 191.46803283691406, 319.89361572265625, 129.24330139160156, -74.57970428466797, 688.2947998046875, -39.007781982421875, -12.034194946289062, 184.26022338867188, 18.379913330078125, 219.98377990722656, -15.603538513183594, 183.26895141601562, 77.91868591308594, -130.1652069091797, 326.3646240234375, 177.6690216064453, 86.81157684326172, 228.87449645996094, 363.7248229980469, 345.9772033691406, 66.56990051269531, -9.685989379882812, 713.3764038085938, 16.09581756591797, 220.9882049560547, -190.19992065429688], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000491.npy"} +{"epoch": 0.7209985315712188, "step": 492, "batch_size": 64, "mean": 130.72726440429688, "std": 228.6665802001953, "min": -409.76324462890625, "p10": -139.39808654785153, "median": 119.59025955200195, "p90": 431.3792541503907, "max": 613.9810180664062, "pos_frac": 0.734375, "sample": [69.9686508178711, 157.78868103027344, 372.4557800292969, -87.42481231689453, 210.33709716796875, -52.46458435058594, 8.76873779296875, 438.158203125, -9.857646942138672, -69.72465515136719, 4.702537536621094, 216.57347106933594, -287.0782470703125, 1.6390151977539062, -301.7931213378906, 492.46319580078125, 124.80416107177734, 264.25, 18.723787307739258, 498.9321594238281, -190.4469451904297, 61.458717346191406, 203.26693725585938, 6.327117919921875, 391.28863525390625, 197.51629638671875, 300.7085266113281, -30.879287719726562, 163.77659606933594, 114.37635803222656, -87.29412841796875, 82.90300750732422, 55.419883728027344, -234.9648895263672, -99.82357788085938, 344.2246398925781, 601.31640625, 2.1334266662597656, -409.76324462890625, 102.25408172607422, 613.9810180664062, -101.80096435546875, 337.73358154296875, -4.288871765136719, -61.0271110534668, 415.56170654296875, 195.39059448242188, 126.63882446289062, 549.1499633789062, 156.3131103515625, 264.1823425292969, 84.38261413574219, 292.72723388671875, -155.51113891601562, 403.285400390625, 239.28932189941406, 293.8341064453125, 87.46600341796875, 20.370716094970703, 259.7096252441406, -252.36767578125, 540.4742431640625, 146.85855102539062, 269.171142578125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000492.npy"} +{"epoch": 0.7224669603524229, "step": 493, "batch_size": 64, "mean": 160.08062744140625, "std": 170.4226531982422, "min": -207.72894287109375, "p10": -39.66382446289062, "median": 160.5846176147461, "p90": 378.0985168457032, "max": 564.0595092773438, "pos_frac": 0.796875, "sample": [197.0274200439453, 82.53234100341797, 314.9853820800781, 223.07138061523438, -28.692535400390625, 564.0595092773438, 414.1458740234375, 104.02462768554688, 219.39425659179688, -207.72894287109375, 300.6153869628906, 82.45808410644531, 440.82696533203125, 160.91204833984375, 124.51060485839844, 251.03219604492188, 127.28961181640625, 243.07290649414062, 158.96017456054688, 33.906227111816406, 175.3505096435547, 190.35983276367188, 271.0088195800781, 281.94659423828125, 281.64459228515625, 152.5428466796875, 337.619140625, -42.013648986816406, 10.387823104858398, -26.23096466064453, 56.98602294921875, -187.60752868652344, 281.2229919433594, 14.446945190429688, 360.7112121582031, 21.192712783813477, 385.4316711425781, 362.10186767578125, 142.4647216796875, -63.85191345214844, 93.51242065429688, 191.17929077148438, 331.7931213378906, 340.763916015625, 265.4996643066406, 500.071533203125, -68.76703643798828, 220.35296630859375, 445.78564453125, 384.9542236328125, 182.81393432617188, 140.58966064453125, -183.3469696044922, 195.050048828125, -4.199462890625, -107.9771499633789, -34.18090057373047, -32.49094772338867, 222.04220581054688, 160.25718688964844, 19.269386291503906, 67.53457641601562, -5.431243896484375, 107.96626281738281], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000493.npy"} +{"epoch": 0.723935389133627, "step": 494, "batch_size": 64, "mean": 225.864501953125, "std": 187.33641052246094, "min": -279.95806884765625, "p10": 0.6952018737792998, "median": 224.66558837890625, "p90": 420.82911987304686, "max": 661.543212890625, "pos_frac": 0.890625, "sample": [-1.384267807006836, 249.59315490722656, 190.4219207763672, 334.0862121582031, 661.543212890625, 429.5477294921875, -279.95806884765625, 419.07958984375, 129.77688598632812, 48.137054443359375, 344.91204833984375, 223.19505310058594, 418.9661560058594, 220.6358184814453, 113.48326873779297, 581.3944091796875, 338.82879638671875, -96.6435546875, 340.9268493652344, 404.14349365234375, 59.93267059326172, -270.15643310546875, 384.01177978515625, 355.36199951171875, 397.9149169921875, 86.91463470458984, 322.4905700683594, 526.623046875, 206.82180786132812, 123.39248657226562, 3.5971908569335938, -0.5485076904296875, 226.13612365722656, 40.45421600341797, 373.29644775390625, 246.59402465820312, 336.380859375, 71.93453216552734, 143.3404083251953, 565.4212646484375, 565.5311889648438, 158.3040771484375, -3.9295196533203125, 54.284812927246094, 172.12725830078125, 279.3370666503906, 421.57891845703125, 226.98049926757812, 137.78775024414062, 237.5544891357422, 197.1820831298828, -22.169845581054688, 336.4012451171875, 221.22372436523438, 248.69668579101562, 105.92010498046875, 410.2908935546875, 321.79681396484375, 149.65924072265625, 370.42950439453125, 275.7162170410156, 182.1024627685547, 123.50143432617188, 14.42144775390625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000494.npy"} +{"epoch": 0.7254038179148311, "step": 495, "batch_size": 64, "mean": 203.76222229003906, "std": 197.69436645507812, "min": -171.32618713378906, "p10": 2.8763000488281305, "median": 167.72028350830078, "p90": 449.2148071289063, "max": 814.5677490234375, "pos_frac": 0.90625, "sample": [269.6905517578125, -62.66105651855469, 202.6234130859375, 74.84182739257812, 39.428466796875, 163.91799926757812, -104.34611511230469, 85.54376220703125, 198.08172607421875, 26.090072631835938, 193.1245574951172, 427.98870849609375, 358.23431396484375, -171.32618713378906, 272.21075439453125, 71.61165618896484, 38.83258056640625, 430.1786804199219, 70.45849609375, 370.6651916503906, 68.36410522460938, -164.01962280273438, 217.70106506347656, 81.3753662109375, 222.15499877929688, 453.9815368652344, 447.5858459472656, 214.42227172851562, 110.53075408935547, 303.53509521484375, 167.23085021972656, 268.64556884765625, 173.03729248046875, 124.52166748046875, 194.14955139160156, 517.1951293945312, 435.5495300292969, -81.83644104003906, 625.0423583984375, 385.063232421875, 228.46633911132812, 111.64097595214844, 92.3636245727539, 143.30194091796875, 115.11546325683594, 45.421875, 8.436630249023438, 158.46383666992188, 106.90560150146484, -97.91334533691406, 168.209716796875, 0.4933013916015625, 351.91644287109375, 328.82818603515625, 449.9129333496094, 109.66429138183594, 127.62063598632812, 344.01275634765625, 62.75398254394531, 656.5823974609375, 572.5691528320312, 258.4913330078125, 814.5677490234375, 163.53585815429688], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000495.npy"} +{"epoch": 0.7268722466960352, "step": 496, "batch_size": 64, "mean": 157.27761840820312, "std": 170.4345703125, "min": -189.9241485595703, "p10": -49.47971954345701, "median": 139.7317352294922, "p90": 423.874887084961, "max": 561.851318359375, "pos_frac": 0.828125, "sample": [20.113250732421875, -81.78914642333984, 454.2846374511719, 469.8570861816406, -68.84994506835938, 489.25799560546875, 149.322509765625, 237.54974365234375, 355.6224060058594, 172.09036254882812, 195.20274353027344, 61.06433868408203, 146.12551879882812, 113.0028076171875, 150.5758056640625, 112.2935791015625, 63.1202392578125, 187.03872680664062, -27.89116668701172, 176.8722381591797, 264.8409423828125, 334.7580871582031, 561.851318359375, 190.833740234375, 486.08612060546875, 133.33795166015625, 7.0442657470703125, 326.50592041015625, 60.7701416015625, 146.85037231445312, 64.13741302490234, -19.85877799987793, 228.52023315429688, 278.5948791503906, 12.953422546386719, -5.9059600830078125, 81.33111572265625, 46.096168518066406, -189.9241485595703, 199.3114471435547, -59.496944427490234, -10.874082565307617, 347.71612548828125, -58.731956481933594, 58.34413146972656, 462.6844787597656, 11.553085327148438, 403.9353332519531, 432.42041015625, 322.9529724121094, 26.261322021484375, -80.05677795410156, 341.71136474609375, 130.5315704345703, 35.499786376953125, 61.73060607910156, 183.38893127441406, 264.3709716796875, 49.00579071044922, 19.42493438720703, 307.3603820800781, 219.60874938964844, 85.0261459350586, -71.59765625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000496.npy"} +{"epoch": 0.7283406754772394, "step": 497, "batch_size": 64, "mean": 149.05459594726562, "std": 175.15858459472656, "min": -246.45762634277344, "p10": -57.71114654541015, "median": 133.83070373535156, "p90": 357.3957733154297, "max": 568.4569091796875, "pos_frac": 0.796875, "sample": [32.8355827331543, 360.0040283203125, 310.34417724609375, 470.61236572265625, 9.155462265014648, 150.12249755859375, 341.0667724609375, -23.051315307617188, 64.62559509277344, -35.84895324707031, 321.9232177734375, 335.52978515625, 492.68890380859375, 176.83937072753906, 27.565284729003906, -52.98504638671875, 295.3146667480469, 126.09153747558594, 303.59942626953125, 75.76522827148438, 119.08317565917969, 92.41486358642578, 351.3098449707031, 230.82171630859375, -75.22148132324219, 42.894805908203125, 209.03707885742188, -5.2450103759765625, 65.71275329589844, -6.7309722900390625, 260.42376708984375, 156.60618591308594, -0.9811210632324219, 334.0635986328125, 125.61029052734375, 552.2174072265625, 185.8768310546875, 568.4569091796875, 172.72450256347656, 23.7396240234375, 107.74108123779297, 165.0966339111328, 0.4199047088623047, 331.44915771484375, -107.61302185058594, 139.1339111328125, 185.5240478515625, 86.80172729492188, 249.34625244140625, 267.67529296875, 205.60757446289062, 62.77507019042969, 384.2274475097656, 160.6985626220703, -75.94945526123047, 128.52749633789062, -59.73661804199219, 223.65087890625, -208.7882843017578, 18.4542236328125, 368.3140869140625, -246.45762634277344, 120.95060729980469, -153.36875915527344], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000497.npy"} +{"epoch": 0.7298091042584435, "step": 498, "batch_size": 64, "mean": 166.9569091796875, "std": 206.3229217529297, "min": -294.66070556640625, "p10": -98.4926383972168, "median": 150.94210052490234, "p90": 433.4763854980469, "max": 694.7138061523438, "pos_frac": 0.796875, "sample": [-99.76737976074219, 115.81356048583984, 156.13812255859375, 107.19882202148438, 82.85641479492188, 206.45001220703125, 436.7869873046875, 417.88330078125, 247.57131958007812, 110.046142578125, 193.43756103515625, 476.499267578125, 116.61822509765625, 262.75860595703125, -294.66070556640625, 338.673828125, 281.71044921875, 456.2908935546875, 83.47138214111328, 425.75164794921875, 27.859275817871094, 251.75515747070312, -72.99807739257812, 202.13211059570312, -44.92144775390625, 141.37411499023438, 97.12876892089844, 36.27722930908203, -145.16439819335938, 105.47791290283203, 145.74607849121094, 250.2091522216797, -265.8083801269531, 694.7138061523438, 170.1168212890625, 33.43975830078125, -50.55382537841797, 179.9605712890625, 417.8775634765625, 450.77178955078125, 591.0845947265625, -95.51824188232422, -35.7822265625, 111.38020324707031, 407.4595031738281, 138.5640869140625, -79.04641723632812, 328.9029541015625, 390.8145446777344, 237.30816650390625, 172.23667907714844, -148.0587158203125, 199.40921020507812, 117.5455551147461, -106.41535949707031, 0.981842041015625, 381.4229736328125, 74.3325424194336, 384.0409240722656, 207.9060516357422, 24.46067237854004, 264.43121337890625, 503.86297607421875, -133.00393676757812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000498.npy"} +{"epoch": 0.7312775330396476, "step": 499, "batch_size": 64, "mean": 216.70993041992188, "std": 203.493408203125, "min": -245.57374572753906, "p10": -33.59198226928708, "median": 235.80724334716797, "p90": 475.6135681152344, "max": 641.4676513671875, "pos_frac": 0.875, "sample": [478.9920654296875, 153.79273986816406, 212.0970458984375, -149.54898071289062, 112.65796661376953, 281.1139221191406, 235.7515869140625, 208.70477294921875, 38.99022674560547, -45.902069091796875, 200.82850646972656, 258.947998046875, 84.16707611083984, 411.4991455078125, 246.24383544921875, 388.9918518066406, 272.87322998046875, 0.6074371337890625, 641.4676513671875, 175.36305236816406, 170.47265625, 19.787342071533203, 448.6728820800781, 87.52275848388672, -245.57374572753906, 333.3855285644531, 254.35861206054688, 235.86289978027344, -195.1784210205078, 249.18038940429688, 201.98275756835938, 26.01043701171875, 307.3492126464844, 232.92111206054688, 175.46519470214844, 145.10292053222656, 366.36572265625, 494.8060302734375, 310.9215087890625, -183.74745178222656, 16.49464225769043, 333.7978820800781, 451.5887145996094, 261.8524169921875, 543.5732421875, 240.82229614257812, 481.4530029296875, 559.5006103515625, 233.64239501953125, 96.18280792236328, 259.0503234863281, 359.5793762207031, 228.7564697265625, 270.8391418457031, 10.345218658447266, 354.1409912109375, 466.59716796875, 251.11514282226562, -197.37075805664062, 18.808027267456055, -111.08024597167969, 467.73040771484375, -4.868448257446289, 633.5772094726562], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000499.npy"} +{"epoch": 0.7327459618208517, "step": 500, "batch_size": 64, "mean": 185.16128540039062, "std": 209.10366821289062, "min": -263.84356689453125, "p10": -57.57970428466796, "median": 149.89288330078125, "p90": 488.9345031738282, "max": 791.8225708007812, "pos_frac": 0.84375, "sample": [51.391937255859375, -47.18119430541992, 44.67688751220703, 791.8225708007812, -121.33853912353516, 251.95022583007812, 232.82745361328125, 386.9512023925781, 228.3075714111328, 213.703369140625, 495.065185546875, 349.5238037109375, 204.0989227294922, 119.96041870117188, -61.13124084472656, 173.69387817382812, 37.19854736328125, -49.29278564453125, 146.98570251464844, 405.96356201171875, 534.725341796875, -69.36332702636719, 275.61334228515625, 512.8460693359375, 70.57784271240234, 626.2255249023438, -242.612060546875, 214.66331481933594, 474.62957763671875, -263.84356689453125, 143.8280487060547, 394.4229736328125, 328.09967041015625, 20.16725730895996, 84.52893829345703, 319.6575012207031, -201.7728729248047, 95.06803131103516, 239.69961547851562, -158.26341247558594, 112.59807586669922, 181.66285705566406, 119.69647979736328, 328.0812072753906, 112.38993072509766, 96.06487274169922, 283.1611022949219, 50.972557067871094, 166.99169921875, 148.60781860351562, 565.8016357421875, 151.17794799804688, 382.4624938964844, 101.2553939819336, 138.87017822265625, 296.8536376953125, 304.0443115234375, 94.622802734375, 127.58580780029297, 222.10031127929688, 33.14225769042969, 526.6304321289062, -11.982572555541992, 63.45570373535156], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000500.npy"} +{"epoch": 0.7342143906020558, "step": 501, "batch_size": 64, "mean": 202.94744873046875, "std": 209.478515625, "min": -206.67752075195312, "p10": -60.27877197265623, "median": 173.30213165283203, "p90": 475.91625061035165, "max": 715.0003051757812, "pos_frac": 0.84375, "sample": [-84.97916412353516, 242.9419708251953, 240.7333984375, -69.59464263916016, 42.009883880615234, 522.4105224609375, 65.70277404785156, 433.2640686035156, 114.06303405761719, 68.02969360351562, 220.0166015625, 82.48768615722656, 330.4731750488281, 135.4821319580078, 692.9758911132812, 414.5072326660156, 544.3837890625, 160.91790771484375, 157.68736267089844, 327.87457275390625, 354.28997802734375, 241.36122131347656, 25.923019409179688, 118.05493927001953, 383.8871154785156, -1.498565673828125, -94.95732116699219, 114.2314453125, 484.7401123046875, 152.65866088867188, -162.0028533935547, 153.89547729492188, 185.18687438964844, 133.01080322265625, 351.33441162109375, 715.0003051757812, -32.75669860839844, 276.3454895019531, 214.49087524414062, 382.613525390625, 418.0997314453125, 47.305442810058594, 172.15040588378906, 63.724090576171875, 455.3272399902344, 185.41317749023438, -77.3297119140625, 88.15779113769531, 36.62191390991211, -206.67752075195312, 174.453857421875, 197.3623046875, -173.9347381591797, 50.260406494140625, 27.78958511352539, -38.54174041748047, 290.5479736328125, 412.49725341796875, 490.66424560546875, 294.5425720214844, 694.8375244140625, 354.3771057128906, 311.80401611328125, 81.98696899414062], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000501.npy"} +{"epoch": 0.73568281938326, "step": 502, "batch_size": 64, "mean": 215.4339599609375, "std": 150.2885284423828, "min": -57.66993713378906, "p10": 18.45761013031006, "median": 218.08038330078125, "p90": 414.65295715332036, "max": 620.3895263671875, "pos_frac": 0.921875, "sample": [340.2093505859375, -55.76695251464844, 112.69925689697266, 307.3097839355469, 33.458984375, 247.99325561523438, -54.674827575683594, 69.9420166015625, 442.559814453125, 284.92718505859375, 234.47991943359375, 196.34780883789062, 240.4632110595703, 309.6790771484375, 342.7583923339844, 137.6236572265625, 103.91180419921875, 79.28697204589844, -45.10699462890625, 179.87149047851562, 356.6357116699219, 317.933837890625, 200.83993530273438, 186.84219360351562, 18.172637939453125, 227.1954803466797, 103.44920349121094, 146.25904846191406, 340.82061767578125, 254.2813720703125, 232.05921936035156, 450.79443359375, -13.466316223144531, 198.80401611328125, 389.89111328125, 398.6429138183594, 130.06507873535156, 522.307861328125, -57.66993713378906, 19.12254524230957, 427.20001220703125, 223.5795135498047, 167.14529418945312, 143.7470703125, 213.52737426757812, 421.514404296875, 620.3895263671875, 34.68988800048828, 382.80816650390625, 16.45587158203125, 291.95306396484375, 93.65458679199219, 270.824462890625, 55.286590576171875, 477.9288635253906, 118.7530517578125, 239.89149475097656, 276.7055969238281, 381.027587890625, 156.9127960205078, 222.63339233398438, 175.9700469970703, 269.10491943359375, 175.1163787841797], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000502.npy"} +{"epoch": 0.737151248164464, "step": 503, "batch_size": 64, "mean": 181.41030883789062, "std": 161.6510009765625, "min": -129.4680633544922, "p10": -16.093994903564447, "median": 161.5581283569336, "p90": 357.47160949707035, "max": 637.1380004882812, "pos_frac": 0.875, "sample": [105.33982849121094, -30.808462142944336, 486.8479309082031, 89.17228698730469, -129.4680633544922, -79.5172119140625, 396.0987548828125, 155.69996643066406, -10.366264343261719, 7.574623107910156, -46.13612365722656, 226.0718231201172, 132.77731323242188, 152.3343963623047, 637.1380004882812, 272.8116455078125, 253.5009307861328, 215.96981811523438, 174.85791015625, 303.5843811035156, 74.42189025878906, 152.0749969482422, 143.76658630371094, 71.07508087158203, 348.13482666015625, 292.24871826171875, -125.00188446044922, 359.4603271484375, 72.53292083740234, 260.15521240234375, 162.58517456054688, -18.548736572265625, 54.971778869628906, 239.23977661132812, 14.862983703613281, 128.330810546875, 416.9609680175781, 589.3263549804688, 241.77853393554688, 322.743896484375, 76.3055419921875, 160.5310821533203, 180.4540557861328, 150.7754364013672, -92.09317016601562, 252.029052734375, 85.4891128540039, 352.8312683105469, 116.94942474365234, 265.7827453613281, 56.019020080566406, 290.7942199707031, 278.359619140625, 190.55397033691406, 574.9031372070312, 294.89898681640625, 88.0768051147461, 232.27822875976562, 86.6593246459961, 63.6260986328125, 104.65482330322266, 198.8936004638672, 300.66302490234375, 186.22018432617188], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000503.npy"} +{"epoch": 0.7386196769456681, "step": 504, "batch_size": 64, "mean": 231.68557739257812, "std": 204.697509765625, "min": -213.34176635742188, "p10": -45.36184844970701, "median": 207.98920440673828, "p90": 469.7932281494141, "max": 832.9796142578125, "pos_frac": 0.84375, "sample": [166.89608764648438, 218.24542236328125, 473.7651062011719, 155.99046325683594, -5.7888641357421875, 186.80551147460938, 349.52850341796875, -99.3953628540039, 832.9796142578125, 658.341552734375, 710.5601196289062, 13.403129577636719, 284.2991638183594, 293.958984375, -5.8237762451171875, 84.88249206542969, 366.58154296875, 109.37542724609375, 98.43846893310547, 124.11691284179688, 337.8469543457031, 325.652099609375, 207.2311248779297, 129.60496520996094, 233.18629455566406, 143.1908416748047, -63.50818634033203, 339.63043212890625, 90.3525390625, 224.9342803955078, -213.34176635742188, 419.3770751953125, 177.5419464111328, 501.6495361328125, 460.5255126953125, 400.15966796875, 327.7358093261719, 371.31768798828125, 186.03741455078125, 112.66378021240234, 249.2039794921875, 629.7132568359375, 372.7695007324219, 198.39625549316406, 343.8784484863281, -64.92922973632812, -74.62329864501953, 93.33695983886719, 93.46604919433594, 315.0262145996094, 130.9051513671875, 321.572509765625, 386.2818908691406, 332.985595703125, 126.51203918457031, 146.844482421875, -55.57676696777344, 208.74728393554688, 546.5856323242188, 193.8851318359375, -71.79721069335938, 266.09539794921875, 431.17572021484375, -21.52703857421875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000504.npy"} +{"epoch": 0.7400881057268722, "step": 505, "batch_size": 64, "mean": 144.8794708251953, "std": 220.18846130371094, "min": -301.1656188964844, "p10": -123.5467628479004, "median": 107.6971435546875, "p90": 422.03383178710953, "max": 736.2578125, "pos_frac": 0.71875, "sample": [25.386070251464844, 380.7498779296875, -136.73699951171875, 365.65948486328125, -20.884933471679688, 477.0021057128906, -150.58836364746094, 13.724746704101562, 96.05670166015625, 85.86293029785156, 511.9270324707031, 364.7567443847656, 736.2578125, -4.1017608642578125, 256.74200439453125, 336.690185546875, 108.94935607910156, 38.8216552734375, 244.67156982421875, -36.74550247192383, 149.69166564941406, 438.8645935058594, 150.59722900390625, 55.13140869140625, -223.541015625, -122.77078247070312, 69.93153381347656, 306.6399841308594, 356.5140380859375, 327.723388671875, -97.87696838378906, 227.76806640625, 343.3883056640625, -69.41537475585938, 70.3935775756836, 382.7620544433594, 199.44107055664062, 123.72746276855469, -21.69841766357422, 196.24058532714844, -165.74285888671875, -301.1656188964844, 134.00315856933594, 171.05238342285156, 20.016143798828125, 255.8831787109375, 596.6058959960938, 278.5349426269531, 87.13600158691406, 630.123779296875, -55.63722610473633, -123.87932586669922, -80.89012908935547, 537.9341430664062, 106.44493103027344, -55.55815124511719, 174.66412353515625, -167.82179260253906, 105.6507568359375, 175.3050079345703, 326.117431640625, -67.12484741210938, 100.83468627929688, 32.08620071411133], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000505.npy"} +{"epoch": 0.7415565345080763, "step": 506, "batch_size": 64, "mean": 207.0786895751953, "std": 203.5921173095703, "min": -239.42198181152344, "p10": -47.088960266113276, "median": 173.0141143798828, "p90": 458.38736572265634, "max": 778.6141357421875, "pos_frac": 0.84375, "sample": [-239.42198181152344, 34.2589111328125, 368.22027587890625, 91.83616638183594, 107.60639953613281, 252.28785705566406, -30.726551055908203, 331.785888671875, 430.80438232421875, 33.854793548583984, 278.3924560546875, 77.06541442871094, 85.09172058105469, 263.125732421875, -49.50086212158203, 503.0615234375, 283.31109619140625, 58.70957946777344, 174.42196655273438, 331.36846923828125, 165.515869140625, 431.6636962890625, 357.2093505859375, 171.60626220703125, 778.6141357421875, -73.94290161132812, -120.46880340576172, 323.0302429199219, 306.63616943359375, 143.77769470214844, 131.97242736816406, 429.55059814453125, 387.2560729980469, 83.37208557128906, 400.02191162109375, -116.0916976928711, 417.2413635253906, 219.04486083984375, 61.0841064453125, 563.1715087890625, 467.8070068359375, 283.6150207519531, 26.444168090820312, 239.12828063964844, 72.95121002197266, -81.74474334716797, 165.06954956054688, 137.12564086914062, 529.50634765625, 386.8017272949219, 327.57037353515625, 1.0069999694824219, 153.09149169921875, 436.408203125, 518.1403198242188, 514.4826049804688, -91.27439880371094, 209.30320739746094, 90.41539001464844, 26.73542022705078, -38.297119140625, 151.57083129882812, 322.8218688964844, -41.46118927001953], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000506.npy"} +{"epoch": 0.7430249632892805, "step": 507, "batch_size": 64, "mean": 183.6204376220703, "std": 198.4649658203125, "min": -175.4586639404297, "p10": -14.783483886718745, "median": 147.49353790283203, "p90": 448.33509216308596, "max": 938.032958984375, "pos_frac": 0.84375, "sample": [37.785484313964844, 110.23930358886719, 39.20738220214844, -175.4586639404297, 640.0728759765625, 228.91098022460938, 151.98081970214844, 143.00625610351562, 92.60965728759766, 226.324951171875, 68.60267639160156, 257.1958923339844, -112.59819030761719, 270.3014831542969, 331.19952392578125, 96.494873046875, -133.24473571777344, 126.4527587890625, 169.8300323486328, 938.032958984375, 142.79908752441406, 205.38421630859375, -6.571735382080078, 267.9382019042969, 248.14968872070312, 320.32867431640625, 225.75674438476562, 300.19903564453125, 15.742488861083984, 446.9681396484375, 179.1858367919922, 91.15750122070312, -65.49590301513672, 121.29804992675781, 54.27252960205078, 214.67221069335938, 122.89593505859375, 280.1197814941406, 163.49746704101562, 12.853324890136719, 261.28472900390625, 16.145774841308594, 34.9205322265625, 39.349998474121094, 574.1138305664062, 469.64532470703125, 139.0361328125, 309.59844970703125, -8.542724609375, 307.56549072265625, 353.70599365234375, 15.873647689819336, -17.288970947265625, 403.9119873046875, 456.7291259765625, 396.107421875, 496.35162353515625, 448.9209289550781, 189.52764892578125, -28.59502601623535, 20.012496948242188, -8.937347412109375, -51.15159606933594, 85.3248519897461], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000507.npy"} +{"epoch": 0.7444933920704846, "step": 508, "batch_size": 64, "mean": 169.13079833984375, "std": 216.04246520996094, "min": -388.6576232910156, "p10": -50.42977752685547, "median": 143.12342834472656, "p90": 435.02205200195317, "max": 790.70166015625, "pos_frac": 0.75, "sample": [143.30398559570312, 60.35260009765625, -177.7343292236328, 201.41064453125, 240.0201873779297, -69.7486572265625, -20.70486831665039, 790.70166015625, 412.2051696777344, 415.2491455078125, 242.19100952148438, 421.90606689453125, 236.94500732421875, 678.4915771484375, 396.49761962890625, 224.39686584472656, 212.18756103515625, 12.95556640625, 507.4872131347656, 413.75244140625, 80.88119506835938, 135.9261016845703, 122.02604675292969, -3.09716796875, 501.35107421875, 234.58767700195312, 195.4984130859375, -50.568992614746094, 14.000139236450195, 166.11041259765625, 24.074630737304688, 176.06103515625, 65.70643615722656, -16.062206268310547, 629.0326538085938, -388.6576232910156, -45.907325744628906, 103.97948455810547, 84.66146850585938, 289.0154724121094, 23.565704345703125, 249.1871337890625, 43.36932373046875, 142.94287109375, 314.92852783203125, 161.27777099609375, 398.1512756347656, 72.23295593261719, 342.4701843261719, 446.8330993652344, 270.6477966308594, -33.498687744140625, -72.24275207519531, 440.6431884765625, 367.55743408203125, -77.97724151611328, -28.337066650390625, 25.158905029296875, -55.74275207519531, 170.26132202148438, -43.42787170410156, -42.03528594970703, -50.104942321777344, 98.02388763427734], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000508.npy"} +{"epoch": 0.7459618208516887, "step": 509, "batch_size": 64, "mean": 173.21424865722656, "std": 196.40260314941406, "min": -218.198974609375, "p10": -25.5002326965332, "median": 142.77076721191406, "p90": 474.56458129882816, "max": 879.91015625, "pos_frac": 0.84375, "sample": [93.24019622802734, -9.741592407226562, -23.481361389160156, 391.4448547363281, 49.355712890625, 265.11114501953125, 96.72211456298828, 22.161109924316406, 287.87933349609375, 175.72525024414062, 269.49957275390625, 296.71435546875, 148.31304931640625, 27.509004592895508, 49.11427307128906, -26.365463256835938, 623.1849975585938, -105.94792175292969, -218.198974609375, 166.19888305664062, 483.23358154296875, 259.28338623046875, 103.22642517089844, 99.9566421508789, 294.32598876953125, 102.64308166503906, 149.0225830078125, 139.47598266601562, 255.82095336914062, 223.75753784179688, 212.6309051513672, -134.37623596191406, 19.003986358642578, -80.45576477050781, 106.80459594726562, 571.9251098632812, 205.64268493652344, 133.1635284423828, 293.0377502441406, 89.96199035644531, 119.93060302734375, 143.3258514404297, -14.184127807617188, 97.4493408203125, 29.425758361816406, -111.27528381347656, 491.86669921875, 475.9908447265625, 142.21568298339844, 879.91015625, -66.30428314208984, 176.71560668945312, 8.206707000732422, 50.091339111328125, 324.7113037109375, 96.74205017089844, 154.2393798828125, 100.52914428710938, 169.97146606445312, 240.6092987060547, 471.23663330078125, 565.6937866210938, 201.19598388671875, 230.8950653076172], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000509.npy"} +{"epoch": 0.7474302496328928, "step": 510, "batch_size": 64, "mean": 213.26223754882812, "std": 184.8616180419922, "min": -153.42202758789062, "p10": -29.550827217102047, "median": 220.07587432861328, "p90": 466.2154296875001, "max": 582.5009765625, "pos_frac": 0.84375, "sample": [-36.60195541381836, 205.5162353515625, 531.6664428710938, 340.5759582519531, -46.776214599609375, 216.8608856201172, 473.1446533203125, 336.37646484375, 65.39134216308594, 320.047607421875, 263.8538513183594, 288.3125305175781, -26.655656814575195, 169.60089111328125, 450.0472412109375, 334.6602478027344, 327.0988464355469, -153.42202758789062, -93.53545379638672, 421.4945983886719, 137.65609741210938, 159.07850646972656, 311.97869873046875, 33.187713623046875, 158.54666137695312, 296.1339416503906, 370.2266845703125, 582.5009765625, 427.6164855957031, -125.3214111328125, 477.4416809082031, 387.6100158691406, -3.2326087951660156, 39.46953582763672, 345.72076416015625, 223.29086303710938, 135.04159545898438, 168.7622833251953, -30.791614532470703, 284.14862060546875, 120.63545227050781, 289.8271179199219, 332.4429626464844, 210.43968200683594, 5.9643402099609375, -7.005104064941406, 144.84669494628906, 224.35031127929688, 146.8707733154297, 246.11892700195312, 559.5249633789062, 4.934026718139648, 248.89361572265625, 504.15386962890625, 45.27662658691406, 386.8562927246094, 267.362060546875, 409.6610412597656, -55.71477508544922, 58.618797302246094, 109.30731964111328, 539.0445556640625, 65.05673217773438, 24.59518051147461], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000510.npy"} +{"epoch": 0.748898678414097, "step": 511, "batch_size": 64, "mean": 201.6386260986328, "std": 232.25221252441406, "min": -388.73101806640625, "p10": -50.379957580566405, "median": 184.10678100585938, "p90": 512.2473815917969, "max": 806.7557373046875, "pos_frac": 0.8125, "sample": [23.56397247314453, 633.5645751953125, 100.13070678710938, 231.6013946533203, 199.2928466796875, 185.95640563964844, 306.1337890625, 494.2781677246094, 168.73487854003906, 195.5350799560547, 408.7846374511719, 319.0684814453125, 512.63671875, 503.0310974121094, -22.51820182800293, 620.6366577148438, -388.73101806640625, 515.4363403320312, 53.7840576171875, 377.37115478515625, 309.7404479980469, 373.7587585449219, 54.43653869628906, 301.45916748046875, 161.08343505859375, 114.9134292602539, -300.189697265625, 343.79638671875, -38.889129638671875, 394.6488952636719, 511.33892822265625, 59.90956115722656, -29.184282302856445, 167.1884307861328, 300.57763671875, 271.7259216308594, 370.7619323730469, 150.66510009765625, -45.989288330078125, 131.86077880859375, 384.0435485839844, 199.25601196289062, 90.28138732910156, -112.1997299194336, -36.443870544433594, 583.0587158203125, 140.84695434570312, -52.26167297363281, -96.57978057861328, 70.94490051269531, -120.60184478759766, 125.21004486083984, 316.12933349609375, 21.085866928100586, 102.09609985351562, 182.2571563720703, 219.26950073242188, 13.911476135253906, 806.7557373046875, 85.19830322265625, 213.88912963867188, -141.12408447265625, 214.7142333984375, 653.2296142578125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000511.npy"} +{"epoch": 0.750367107195301, "step": 512, "batch_size": 64, "mean": 220.6194305419922, "std": 181.20030212402344, "min": -106.9072265625, "p10": 1.9589384078979497, "median": 229.52136993408203, "p90": 495.0150482177736, "max": 648.6357421875, "pos_frac": 0.90625, "sample": [509.4455261230469, 559.578857421875, 267.9620361328125, -61.032073974609375, -30.514328002929688, -37.472991943359375, 461.34393310546875, 38.27619171142578, -106.9072265625, 90.68560028076172, 125.74894714355469, 278.6636047363281, 1.7333621978759766, 212.0657958984375, -49.3900146484375, 127.40017700195312, 326.83282470703125, 298.59906005859375, 131.86590576171875, 2.6310176849365234, 2.4852828979492188, 333.7345886230469, 36.237548828125, 289.804931640625, 648.6357421875, 297.51580810546875, 210.57530212402344, 147.6976318359375, 86.66442108154297, 359.7190246582031, 220.24066162109375, 252.8658447265625, 203.34075927734375, 307.1627197265625, 174.5721893310547, 219.4441680908203, 218.0066375732422, 277.59466552734375, 14.103080749511719, 418.66229248046875, 100.03890991210938, 317.31976318359375, 332.3843688964844, 528.407958984375, 10.705718994140625, 7.7207489013671875, 339.298828125, 28.347843170166016, 76.35722351074219, 263.5967102050781, 574.4246826171875, 22.951087951660156, 242.45155334472656, 428.6224670410156, 253.5303955078125, 274.1263122558594, 317.2181396484375, 148.86331176757812, 238.8020782470703, 257.41815185546875, -23.087085723876953, 514.1270141601562, 573.9265747070312, 425.51141357421875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000512.npy"} +{"epoch": 0.7518355359765051, "step": 513, "batch_size": 64, "mean": 164.82261657714844, "std": 208.74606323242188, "min": -375.91094970703125, "p10": -85.43922958374023, "median": 153.17645263671875, "p90": 463.8481689453125, "max": 651.4669189453125, "pos_frac": 0.78125, "sample": [465.4164123535156, 247.23681640625, 242.7216339111328, 13.006263732910156, 133.55877685546875, 241.03550720214844, 54.52446746826172, 140.7863311767578, -257.1754150390625, 460.1889343261719, 293.1801452636719, 119.220458984375, 368.91552734375, 133.583740234375, 4.201013565063477, 373.21173095703125, 56.85308837890625, -89.266845703125, 147.96043395996094, 233.3429718017578, 168.74859619140625, 501.5228271484375, 497.03814697265625, 241.16448974609375, 144.6256103515625, 115.0462646484375, 217.8783416748047, -21.7022705078125, 253.03932189941406, -9.968265533447266, 133.21533203125, 162.45851135253906, 327.96136474609375, 357.69110107421875, 222.7803955078125, -11.135894775390625, 109.69984436035156, -164.8815155029297, 223.02947998046875, 651.4669189453125, 41.97666931152344, 20.634483337402344, 258.89984130859375, -248.83470153808594, 524.5952758789062, 140.98382568359375, -30.28911590576172, -14.666269302368164, 484.3284912109375, 428.4851989746094, -192.6728973388672, -51.23290252685547, 528.8668212890625, 304.49591064453125, -113.27703094482422, 202.29843139648438, -76.50812530517578, 125.81475067138672, 163.66409301757812, 158.39247131347656, -375.91094970703125, 387.05322265625, 116.35906982421875, 263.010009765625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000513.npy"} +{"epoch": 0.7533039647577092, "step": 514, "batch_size": 64, "mean": 165.6499786376953, "std": 217.8463134765625, "min": -268.7690734863281, "p10": -113.83453216552734, "median": 135.15642547607422, "p90": 470.1742340087891, "max": 821.59521484375, "pos_frac": 0.8125, "sample": [-268.7690734863281, -117.5558853149414, 154.82740783691406, -206.37921142578125, 404.36236572265625, 175.23162841796875, 394.25653076171875, 264.4066162109375, 821.59521484375, 552.9913330078125, 41.15904235839844, 220.6139678955078, -130.96881103515625, 94.1712646484375, 330.56121826171875, -35.22807312011719, 1.8576240539550781, 135.7161407470703, 507.7917175292969, 586.2637939453125, -160.68434143066406, 113.05718231201172, 318.4045715332031, -61.70873260498047, 23.43975067138672, 76.79483032226562, 93.16938781738281, 54.59112548828125, 146.8697509765625, 17.659873962402344, 291.3101501464844, 260.1424560546875, 23.25383758544922, 163.92007446289062, 76.80237579345703, 94.35462951660156, 383.0255126953125, -148.22845458984375, 393.9627380371094, -6.879764556884766, 105.24879455566406, -188.45030212402344, -105.15137481689453, 248.93072509765625, 229.78756713867188, 54.7072868347168, -52.394134521484375, 22.330734252929688, 147.66024780273438, 446.8680114746094, 134.59671020507812, 328.5723876953125, 21.027206420898438, 460.1675109863281, 487.11834716796875, 159.3018035888672, 213.8281707763672, 129.7857666015625, 474.46282958984375, 145.14073181152344, 393.91754150390625, 110.83662414550781, 52.0810546875, 501.06231689453125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000514.npy"} +{"epoch": 0.7547723935389133, "step": 515, "batch_size": 64, "mean": 150.60723876953125, "std": 188.081787109375, "min": -399.48956298828125, "p10": -21.165063285827635, "median": 116.44390869140625, "p90": 406.93385314941406, "max": 530.67236328125, "pos_frac": 0.859375, "sample": [31.674335479736328, 107.18632507324219, 363.6842041015625, 471.2610168457031, 103.29766082763672, -399.48956298828125, 321.30682373046875, 198.93968200683594, 19.17462158203125, -201.6609649658203, -21.866500854492188, 119.19868469238281, 0.642608642578125, 408.430419921875, 315.75286865234375, 207.761474609375, 65.45231628417969, 331.8365173339844, 103.49488830566406, -29.3974609375, 183.63601684570312, -127.86288452148438, 173.0978240966797, 354.8746032714844, 128.76329040527344, 51.09465026855469, 461.458251953125, 9.35226821899414, 187.4853515625, 53.550880432128906, 463.4454345703125, 17.384191513061523, -19.52837562561035, 161.9189453125, 340.2764587402344, 378.5174560546875, 241.656494140625, 10.169235229492188, 482.2117919921875, -5.959716796875, 113.68913269042969, 403.4418640136719, -257.77435302734375, 4.978889465332031, -99.33511352539062, 164.23411560058594, 44.23126983642578, 530.67236328125, 51.56166458129883, 85.11465454101562, 131.68453979492188, 415.07421875, 13.029457092285156, 24.94879913330078, 331.8644714355469, 356.7550964355469, 44.195159912109375, 39.932891845703125, 310.39471435546875, 279.1754150390625, 104.1105728149414, 141.01260375976562, 106.86365509033203, 236.78488159179688], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000515.npy"} +{"epoch": 0.7562408223201175, "step": 516, "batch_size": 64, "mean": 213.1667022705078, "std": 188.92970275878906, "min": -113.22004699707031, "p10": -36.00684375762938, "median": 209.97553253173828, "p90": 468.75957336425785, "max": 701.504638671875, "pos_frac": 0.84375, "sample": [170.78976440429688, 174.09786987304688, 234.66177368164062, -44.359779357910156, 411.8899841308594, 296.6397705078125, 289.5389099121094, -58.36225891113281, 258.69818115234375, 398.55206298828125, 471.501953125, 217.58897399902344, 221.644287109375, 322.7784729003906, 88.82298278808594, -21.7891845703125, 342.0362243652344, 104.91215515136719, 133.1707763671875, 529.3195190429688, -69.81373596191406, 83.8443603515625, -57.1353759765625, 69.18574523925781, 277.5112609863281, 173.2631072998047, 54.620933532714844, 517.1201782226562, 701.504638671875, 128.84657287597656, 14.292633056640625, 503.09375, 671.2566528320312, 225.3583984375, 462.3606872558594, 165.38760375976562, 230.23153686523438, 50.488155364990234, 121.78968811035156, 143.82138061523438, 227.35240173339844, -26.203157424926758, 285.7564697265625, 202.36209106445312, 7.671298980712891, 433.9630126953125, 173.06092834472656, 145.4244384765625, 26.260398864746094, 189.42234802246094, 270.0798645019531, 329.05328369140625, 41.44190979003906, 217.66676330566406, 310.91510009765625, -79.2877426147461, -113.22004699707031, -10.26434326171875, 387.773193359375, 324.7392883300781, 389.70379638671875, 528.4795532226562, 411.56646728515625, -40.20842361450195], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000516.npy"} +{"epoch": 0.7577092511013216, "step": 517, "batch_size": 64, "mean": 215.73797607421875, "std": 212.75189208984375, "min": -149.8743896484375, "p10": -54.280464553833, "median": 213.58683013916016, "p90": 472.0979736328125, "max": 877.5679321289062, "pos_frac": 0.859375, "sample": [164.33322143554688, 422.1336669921875, 311.295166015625, 25.295684814453125, 426.42877197265625, 121.34786987304688, 187.3769073486328, 262.744140625, 365.95556640625, -119.6388931274414, 264.144775390625, 609.8967895507812, 288.6907958984375, 239.71331787109375, 251.08155822753906, 53.73389434814453, 244.24522399902344, 151.96044921875, 281.1994323730469, 471.87469482421875, 524.11279296875, 393.79364013671875, 192.02362060546875, 74.89154052734375, 56.050071716308594, -44.76229476928711, 75.70028686523438, 215.27105712890625, 22.228424072265625, 877.5679321289062, 65.97526550292969, 68.96568298339844, -137.08143615722656, 472.19366455078125, 8.20208740234375, 348.86614990234375, -149.8743896484375, 318.51373291015625, -58.35968017578125, 137.3319091796875, 306.99853515625, 64.25577545166016, 182.02357482910156, 281.3670349121094, 260.76971435546875, 309.0311279296875, 273.15545654296875, 252.65963745117188, -8.858085632324219, 699.19287109375, 102.31153106689453, 324.7733154296875, 303.51702880859375, 639.4580078125, -81.56396484375, 619.444580078125, 114.15000915527344, 206.98733520507812, -89.32154083251953, 211.90260314941406, -128.24310302734375, 359.99072265625, 85.15814208984375, 32.646751403808594], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000517.npy"} +{"epoch": 0.7591776798825257, "step": 518, "batch_size": 64, "mean": 182.2362823486328, "std": 212.7080841064453, "min": -389.33612060546875, "p10": -65.74963226318357, "median": 173.2620849609375, "p90": 503.3758605957032, "max": 648.193359375, "pos_frac": 0.8125, "sample": [28.69387435913086, 536.656005859375, 285.13470458984375, 201.11363220214844, 169.6531219482422, 176.8710479736328, 160.14498901367188, 235.41378784179688, 40.99687576293945, 108.79673767089844, 250.57994079589844, 371.18218994140625, 98.39071655273438, 88.16584777832031, 257.62384033203125, 155.61163330078125, 220.70001220703125, 125.58271026611328, 44.1802978515625, 366.91619873046875, 451.173583984375, 75.2912826538086, -142.03097534179688, 168.27899169921875, -72.34320068359375, 271.55645751953125, 514.4713745117188, 150.683349609375, 199.7558135986328, 553.5274047851562, 641.42919921875, -50.36463928222656, 553.9107055664062, 477.486328125, 36.415626525878906, 232.32839965820312, 138.95977783203125, -11.44732666015625, 323.2956237792969, 362.4075012207031, 87.35945129394531, -5.5926513671875, 227.88900756835938, 96.20418548583984, 206.7608642578125, 43.80486297607422, -180.26853942871094, -389.33612060546875, 326.6617736816406, 225.58010864257812, -2.0563812255859375, -3.9121360778808594, -198.72462463378906, 190.22561645507812, -264.86328125, 648.193359375, 287.6657409667969, 82.09860229492188, -86.72805786132812, 541.074951171875, 276.690185546875, 96.56440734863281, 356.6745300292969, 303.96246337890625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000518.npy"} +{"epoch": 0.7606461086637298, "step": 519, "batch_size": 64, "mean": 162.36383056640625, "std": 198.49996948242188, "min": -270.1556091308594, "p10": -42.978946685791, "median": 144.08911895751953, "p90": 432.1249176025391, "max": 782.9617919921875, "pos_frac": 0.859375, "sample": [53.768768310546875, 228.5183563232422, 99.79356384277344, -187.71487426757812, 483.73931884765625, -121.95446014404297, 219.3363494873047, 451.36151123046875, 434.29425048828125, 247.84678649902344, 146.10365295410156, 427.0631408691406, -47.85378646850586, 498.98443603515625, 252.49307250976562, -195.16683959960938, 495.36077880859375, 327.4040832519531, 159.1009521484375, 197.2683563232422, 177.939697265625, 14.367355346679688, 50.04234313964844, 98.94094848632812, 109.44366455078125, 64.57296752929688, 182.895751953125, 86.72108459472656, 188.84556579589844, 34.02092742919922, 5.40643310546875, 420.83770751953125, 119.56381225585938, -65.0882568359375, 229.177001953125, 177.300048828125, 143.0273895263672, 49.88347625732422, -28.42058563232422, 124.46641540527344, 145.15084838867188, 13.141082763671875, 337.86419677734375, 137.03671264648438, -270.1556091308594, 391.4967041015625, 782.9617919921875, 33.320045471191406, 87.78472900390625, 93.28141784667969, 161.2115020751953, 647.4541015625, 31.241161346435547, 336.1473693847656, 40.01324462890625, 177.75474548339844, 395.1045837402344, 9.098934173583984, 183.73226928710938, 236.55508422851562, 100.92933654785156, -31.604320526123047, 181.80764770507812, -183.73379516601562], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000519.npy"} +{"epoch": 0.762114537444934, "step": 520, "batch_size": 64, "mean": 177.554931640625, "std": 180.08615112304688, "min": -169.42083740234375, "p10": -13.278492355346671, "median": 122.2581558227539, "p90": 409.6128387451172, "max": 677.9761352539062, "pos_frac": 0.859375, "sample": [92.3885726928711, 20.5166015625, 650.470458984375, 111.92604064941406, 230.06544494628906, 101.2528076171875, 364.120849609375, -4.83843994140625, -17.230628967285156, 112.5328369140625, 122.20783996582031, -16.6920166015625, 382.1460876464844, 23.751449584960938, 47.189720153808594, 131.79165649414062, 396.33880615234375, 677.9761352539062, -17.889129638671875, 91.46159362792969, 341.92340087890625, 565.0025024414062, -169.42083740234375, 214.166015625, -57.43548583984375, 402.75927734375, 89.1174545288086, 486.23773193359375, 147.92227172851562, 282.10748291015625, -5.313602447509766, 104.43278503417969, 426.0498046875, 122.41523742675781, 156.88861083984375, 252.0967254638672, 388.3218994140625, 349.7788391113281, 342.40478515625, 70.87081909179688, 293.65521240234375, 194.5643310546875, 72.26707458496094, 122.3084716796875, 37.5806884765625, 386.7286682128906, -88.51679229736328, 93.40604400634766, 459.12957763671875, 7.4053955078125, 130.8441162109375, 142.68609619140625, 34.036346435546875, 207.18408203125, 82.44876098632812, 412.5500793457031, 12.93780517578125, 196.4013671875, -17.929733276367188, 276.6475830078125, 52.236358642578125, 95.95213317871094, 82.22879028320312, 66.95135498046875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000520.npy"} +{"epoch": 0.7635829662261381, "step": 521, "batch_size": 64, "mean": 171.84954833984375, "std": 192.42271423339844, "min": -232.79440307617188, "p10": -27.025022506713864, "median": 167.61767578125, "p90": 430.7688293457031, "max": 722.5846557617188, "pos_frac": 0.828125, "sample": [252.9186248779297, 449.1880798339844, 236.1822052001953, 18.136749267578125, -113.6614761352539, -154.56790161132812, 13.796112060546875, 3.3380889892578125, 14.06060791015625, 263.2393493652344, 46.32597351074219, -3.9598922729492188, -32.103824615478516, 35.13123321533203, 423.5941467285156, 512.381103515625, 46.2857666015625, -21.69704818725586, -104.89533996582031, 463.42620849609375, 25.526317596435547, 50.69477844238281, 94.46659088134766, 192.50640869140625, 377.75201416015625, 447.7825927734375, 344.5852355957031, 167.26669311523438, 21.560707092285156, 108.66552734375, 254.79693603515625, 298.0780029296875, 1.1271896362304688, 293.73052978515625, -22.850440979003906, 198.32180786132812, 133.82870483398438, 167.96865844726562, 425.2308654785156, -162.1689453125, 192.79278564453125, 404.8913269042969, 122.98815155029297, -4.139812469482422, 112.3636474609375, 264.416748046875, -28.814128875732422, 243.6192169189453, 51.144195556640625, -232.79440307617188, 275.4158020019531, 246.6677703857422, 485.92547607421875, 85.48954772949219, 56.90447998046875, 369.2842102050781, 204.03260803222656, 403.8179626464844, 14.117591857910156, 722.5846557617188, 296.7137145996094, 217.7669219970703, 294.0522155761719, 433.1422424316406], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000521.npy"} +{"epoch": 0.7650513950073421, "step": 522, "batch_size": 64, "mean": 158.5929412841797, "std": 162.22384643554688, "min": -246.65260314941406, "p10": -20.68228607177734, "median": 165.9637451171875, "p90": 373.797332763672, "max": 509.1732177734375, "pos_frac": 0.828125, "sample": [249.44602966308594, 199.3843231201172, 313.37384033203125, -18.97881317138672, 416.18084716796875, 26.617183685302734, 333.2748107910156, 101.18893432617188, 24.254199981689453, -18.396472930908203, 13.811893463134766, 128.7459716796875, 422.327880859375, 17.90508270263672, 343.219482421875, 195.20266723632812, 268.9268798828125, -29.256946563720703, 229.74623107910156, 227.748046875, 247.5255126953125, 383.5325927734375, 170.4840087890625, 171.6272735595703, -144.74957275390625, 509.1732177734375, 112.47840118408203, 151.7954864501953, 393.56158447265625, 188.09194946289062, 118.17291259765625, 121.04423522949219, 127.35462951660156, 78.34915161132812, 286.39422607421875, -246.65260314941406, 206.527099609375, 23.804773330688477, 203.2672119140625, -73.719482421875, -10.396560668945312, 233.18359375, 311.43438720703125, -21.41234588623047, 11.661884307861328, 244.82208251953125, 219.2763671875, 161.4434814453125, -107.04080200195312, 468.5710754394531, 319.56103515625, 61.494720458984375, 348.9209289550781, 122.83441925048828, 58.013397216796875, 30.43634796142578, -135.08056640625, -12.077262878417969, 351.08172607421875, 190.60403442382812, 424.36505126953125, 30.885074615478516, 281.7857666015625, 92.79566955566406], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000522.npy"} +{"epoch": 0.7665198237885462, "step": 523, "batch_size": 64, "mean": 164.87081909179688, "std": 184.5341339111328, "min": -369.8634033203125, "p10": -39.44512939453124, "median": 177.901123046875, "p90": 378.8673156738281, "max": 575.4981079101562, "pos_frac": 0.8125, "sample": [-5.9966278076171875, 358.53826904296875, 76.04365539550781, 236.60032653808594, 280.61053466796875, -32.78877258300781, 306.506103515625, 353.9908752441406, 269.84613037109375, 181.98052978515625, -345.7998352050781, 242.80145263671875, 5.440488815307617, 255.0087890625, 8.073650360107422, 41.00297546386719, 168.61212158203125, 424.4281005859375, 259.13104248046875, 196.2232666015625, 72.25575256347656, 404.9750061035156, 152.804931640625, -47.62849426269531, 174.91539001464844, 109.98707580566406, 138.19447326660156, 107.62814331054688, -112.2536849975586, 520.2066040039062, 180.88685607910156, 86.94725036621094, 132.36245727539062, -369.8634033203125, 222.7696533203125, 396.01531982421875, -84.6265640258789, -0.7295055389404297, 319.91546630859375, -41.288726806640625, 261.9972839355469, 46.438232421875, 378.5644836425781, 210.44046020507812, 245.9899444580078, 11.362518310546875, 328.9085693359375, 143.9537353515625, 425.5157470703125, 147.4729766845703, 376.6304016113281, 150.63967895507812, 232.5836181640625, 59.45006561279297, 343.4183349609375, 238.71339416503906, 238.7871856689453, -35.143402099609375, 279.7896728515625, 575.4981079101562, 378.9971008300781, 79.39137268066406, -185.0068359375, -26.386775970458984], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000523.npy"} +{"epoch": 0.7679882525697503, "step": 524, "batch_size": 64, "mean": 171.35018920898438, "std": 167.56967163085938, "min": -269.4731140136719, "p10": -55.866487503051744, "median": 176.47528839111328, "p90": 379.30376281738285, "max": 496.6446228027344, "pos_frac": 0.84375, "sample": [140.79847717285156, 496.6446228027344, 106.224853515625, 235.1478271484375, 155.9041748046875, 193.5, 156.71258544921875, 29.569862365722656, 258.6155700683594, 174.8466796875, 305.73968505859375, 33.642120361328125, 155.62136840820312, 417.6602783203125, 316.0618591308594, 169.09408569335938, 14.165115356445312, 244.52467346191406, 129.9425811767578, 238.5078125, -269.4731140136719, 107.82513427734375, 173.0579833984375, 461.873291015625, 84.95845031738281, -77.45330810546875, 237.22787475585938, 225.36691284179688, 421.849853515625, -38.732730865478516, 50.35384750366211, 83.37751770019531, 125.12447357177734, -14.338422775268555, 328.4082946777344, -73.19473266601562, 303.773681640625, 467.03485107421875, 260.6175842285156, 232.58204650878906, 179.20834350585938, -163.93276977539062, 34.82537841796875, -63.20952606201172, 380.2359619140625, -8.561798095703125, 278.12445068359375, 12.757850646972656, 261.57061767578125, 343.8222961425781, 431.88385009765625, 354.4568786621094, 245.7846221923828, -172.49295043945312, 19.494667053222656, 191.95677185058594, 377.1286315917969, 162.30467224121094, 28.27823829650879, 298.7315368652344, 258.6138610839844, 178.10389709472656, 357.75848388671875, -83.5955810546875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000524.npy"} +{"epoch": 0.7694566813509545, "step": 525, "batch_size": 64, "mean": 214.59054565429688, "std": 175.216552734375, "min": -176.97422790527344, "p10": -4.007358741760254, "median": 215.00707244873047, "p90": 452.0044952392579, "max": 609.6181640625, "pos_frac": 0.875, "sample": [216.9359893798828, 381.224853515625, 172.1352081298828, 202.83401489257812, -20.30060386657715, 392.8261413574219, 329.114990234375, 144.19149780273438, 412.0030822753906, 234.66232299804688, 184.50387573242188, -91.49078369140625, 234.5777587890625, 80.06306457519531, 124.40765380859375, -176.97422790527344, 165.22055053710938, 196.97225952148438, 509.46710205078125, 247.3624267578125, 340.3017578125, 114.5185546875, -3.879608154296875, 268.4340515136719, 70.58270263671875, 247.70547485351562, 178.90086364746094, 323.3642883300781, 245.43490600585938, 349.5015869140625, 343.6229248046875, 18.15679168701172, 529.26904296875, 86.5222396850586, 460.9636535644531, 192.0196533203125, 21.322898864746094, 233.99325561523438, 303.84918212890625, 5.678955078125, 609.6181640625, -31.59564208984375, 5.38848876953125, -4.062108993530273, 49.194488525390625, 203.66873168945312, 236.82168579101562, 181.73751831054688, 376.15142822265625, 213.07815551757812, 399.4996337890625, 514.6342163085938, -70.54409790039062, 219.19752502441406, 431.09979248046875, 226.43247985839844, 136.21481323242188, 48.809410095214844, -105.61742401123047, 342.4445495605469, 243.57516479492188, 504.0127868652344, 174.44459533691406, 559.590576171875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000525.npy"} +{"epoch": 0.7709251101321586, "step": 526, "batch_size": 64, "mean": 211.10134887695312, "std": 185.56678771972656, "min": -126.8649673461914, "p10": -18.7349494934082, "median": 213.61046600341797, "p90": 422.26844177246096, "max": 788.072021484375, "pos_frac": 0.875, "sample": [557.8634643554688, 409.84356689453125, 103.03413391113281, 108.47393798828125, 230.928955078125, 128.52212524414062, 16.819997787475586, -51.18592071533203, 243.1490936279297, -126.8649673461914, 347.50909423828125, 311.53472900390625, 216.58926391601562, 92.60025024414062, 412.2973937988281, 600.561767578125, 137.676513671875, 309.14794921875, 185.2852783203125, 361.60711669921875, 161.53936767578125, 262.7479553222656, -36.57536315917969, 426.541748046875, 280.74493408203125, 304.3328552246094, 788.072021484375, 210.6316680908203, 274.8616027832031, 79.6424331665039, -89.75633239746094, 392.92230224609375, 14.203712463378906, 76.23094940185547, -58.84576416015625, 24.09759521484375, 381.7085876464844, 230.2279510498047, 48.816497802734375, -107.23092651367188, 207.2973175048828, 256.099609375, 281.24481201171875, 196.44647216796875, -19.08905029296875, 183.76309204101562, 407.3922424316406, 439.013916015625, -17.908714294433594, 218.82041931152344, 28.526947021484375, 167.3134002685547, 62.920806884765625, 545.40771484375, 317.83782958984375, 315.21246337890625, 396.6102294921875, 247.48838806152344, 244.5543975830078, 52.49128723144531, 40.19733428955078, 190.4529571533203, 15.409835815429688, 472.6747741699219], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000526.npy"} +{"epoch": 0.7723935389133627, "step": 527, "batch_size": 64, "mean": 170.04966735839844, "std": 159.8002471923828, "min": -140.78436279296875, "p10": -30.861424446105957, "median": 138.49231719970703, "p90": 420.4017120361329, "max": 535.508056640625, "pos_frac": 0.859375, "sample": [89.14354705810547, 426.3156433105469, 344.9944152832031, 288.8158874511719, 275.9327087402344, 110.19772338867188, 535.508056640625, -140.78436279296875, -84.29525756835938, 135.49359130859375, 88.87983703613281, -15.660186767578125, 434.0904846191406, 121.46018981933594, 64.40361785888672, -30.55608367919922, 352.8843078613281, 63.50426483154297, 155.5926513671875, 102.6593017578125, -82.45524597167969, 449.4716796875, 118.06770324707031, 98.63262939453125, 113.66567993164062, 304.28106689453125, -48.52888488769531, 57.671302795410156, 212.07785034179688, 244.87171936035156, 237.52822875976562, 276.8670959472656, 177.02975463867188, 256.1673278808594, 491.6998291015625, 463.90380859375, 483.1023254394531, -30.992284774780273, 20.156661987304688, 212.13516235351562, 398.38995361328125, 101.31995391845703, 406.6025390625, 110.57929992675781, -81.61920166015625, 161.16482543945312, 200.71051025390625, 113.4797592163086, 196.36627197265625, 81.08708190917969, 177.87362670898438, 78.564208984375, 263.8476257324219, 186.71258544921875, 52.266448974609375, 68.64549255371094, 377.6037292480469, 206.08961486816406, 157.65771484375, 141.4910430908203, 57.055179595947266, -77.41913604736328, 30.115758895874023, 100.66011047363281], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000527.npy"} +{"epoch": 0.7738619676945668, "step": 528, "batch_size": 64, "mean": 176.34500122070312, "std": 194.7735595703125, "min": -122.09181213378906, "p10": -71.17388610839842, "median": 150.73025512695312, "p90": 460.2392578125001, "max": 701.232177734375, "pos_frac": 0.84375, "sample": [20.468711853027344, 300.87701416015625, 354.7621765136719, 197.01165771484375, 75.650634765625, -86.93829345703125, -116.84033966064453, 469.75726318359375, 283.71148681640625, 148.41845703125, 288.70062255859375, 63.24916076660156, 161.52301025390625, -85.27996826171875, 218.0624542236328, 135.11749267578125, 293.7286376953125, -19.260391235351562, 504.16650390625, 701.232177734375, -93.67768096923828, 379.15631103515625, 153.04205322265625, 160.692626953125, 90.83157348632812, 506.81097412109375, 45.61699676513672, 605.6351318359375, 25.114784240722656, 262.8285827636719, -121.60381317138672, 93.658447265625, 142.0870361328125, 27.50363540649414, 303.1477966308594, 178.8827667236328, 30.391395568847656, 193.5668182373047, 344.1282043457031, 155.63272094726562, 75.74126434326172, 386.4349060058594, 124.5592041015625, 41.745513916015625, 529.0091552734375, 7.476524353027344, 325.26904296875, 528.1249389648438, 206.5459442138672, 322.9004821777344, -78.54811096191406, -53.96736145019531, 27.542343139648438, 367.30194091796875, -12.421873092651367, 438.03057861328125, 170.6783905029297, 10.390377044677734, -122.09181213378906, 6.7139892578125, 52.889404296875, 87.66075134277344, 400.55133056640625, 51.978492736816406], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000528.npy"} +{"epoch": 0.775330396475771, "step": 529, "batch_size": 64, "mean": 139.0117950439453, "std": 181.8086700439453, "min": -232.9820098876953, "p10": -115.77489471435545, "median": 118.76569366455078, "p90": 368.59409790039064, "max": 642.2626953125, "pos_frac": 0.765625, "sample": [22.868728637695312, -9.988876342773438, 70.94194793701172, -121.57318115234375, 187.164794921875, 143.78221130371094, 189.5635986328125, 251.40719604492188, 46.087860107421875, 642.2626953125, 177.51602172851562, 316.39178466796875, 107.2943344116211, 40.23387908935547, 274.2471923828125, -12.322214126586914, 106.97093963623047, 104.7390365600586, 221.6043701171875, 3.3624191284179688, 22.694679260253906, 155.45248413085938, 365.79132080078125, 386.05133056640625, 369.7952880859375, -226.21316528320312, 26.637413024902344, 346.2989196777344, -102.24555969238281, 388.0374755859375, 234.21383666992188, 107.49014282226562, 70.28890228271484, 352.3421630859375, 42.86260986328125, -58.59037780761719, 73.25311279296875, 295.77734375, 358.0423583984375, 272.76202392578125, -142.2682342529297, 262.1827392578125, 285.2601013183594, -232.9820098876953, 371.0069274902344, -168.54432678222656, 158.7740478515625, -11.243408203125, 394.11932373046875, 47.775184631347656, 130.04124450683594, 286.0657653808594, 27.5351505279541, 238.67030334472656, -127.78852844238281, 55.147430419921875, -4.958229064941406, 319.74725341796875, 390.7203674316406, 304.78692626953125, -66.01274871826172, -15.711532592773438, -129.44381713867188, 280.5782775878906], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000529.npy"} +{"epoch": 0.7767988252569751, "step": 530, "batch_size": 64, "mean": 214.80227661132812, "std": 198.728515625, "min": -376.05035400390625, "p10": 3.713556480407725, "median": 187.15264892578125, "p90": 508.13880310058596, "max": 625.6104125976562, "pos_frac": 0.890625, "sample": [452.1865539550781, 242.98406982421875, 13.806159973144531, 212.40948486328125, 129.38681030273438, -79.1671371459961, 518.7642211914062, 473.8296813964844, 105.22323608398438, -83.82986450195312, 174.6234588623047, -102.47222137451172, 163.83029174804688, 49.452362060546875, 409.43896484375, 282.1783447265625, 78.37281799316406, 213.69638061523438, 87.09195709228516, 271.7808837890625, 336.3443603515625, 144.84133911132812, 546.8612060546875, 197.54132080078125, 27.319198608398438, 314.0155029296875, -22.855712890625, 366.3190612792969, 100.63185119628906, 289.59844970703125, -0.6118450164794922, 446.1908264160156, 524.2667236328125, 254.9285125732422, 312.5251770019531, 15.673717498779297, 502.3022155761719, 170.97830200195312, 66.49691772460938, -83.62162780761719, 88.23475646972656, 319.0356750488281, 139.50164794921875, 588.0261840820312, 26.388063430786133, 243.77737426757812, 344.02685546875, 43.08031463623047, 625.6104125976562, 69.95569610595703, 295.78289794921875, 111.82697296142578, 510.64019775390625, -376.05035400390625, 609.335693359375, 150.59376525878906, 123.75039672851562, 257.087158203125, 309.3270568847656, 333.8250427246094, 65.7275390625, 391.76580810546875, 176.00091552734375, 176.76397705078125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000530.npy"} +{"epoch": 0.7782672540381792, "step": 531, "batch_size": 64, "mean": 193.65338134765625, "std": 190.3219451904297, "min": -232.92572021484375, "p10": -3.673493766784665, "median": 168.1113739013672, "p90": 447.6664642333984, "max": 629.5892333984375, "pos_frac": 0.875, "sample": [203.82321166992188, -4.793985366821289, 347.9383850097656, 629.5892333984375, 349.6424560546875, 33.78031921386719, 37.14484405517578, 35.13831329345703, 610.500244140625, 42.753326416015625, 27.16270637512207, 365.10809326171875, 9.953664779663086, 311.7342224121094, 162.87637329101562, 195.7042694091797, 269.8836364746094, 23.887954711914062, -95.30911254882812, 117.201171875, 628.2041625976562, 138.11007690429688, 333.81085205078125, 63.95539855957031, 300.7348937988281, 207.185302734375, 95.37873840332031, 160.3023223876953, 243.61000061035156, -9.819236755371094, 406.77349853515625, 335.5501708984375, 301.04913330078125, 161.87045288085938, 29.31544303894043, 183.88632202148438, -1.0590133666992188, -16.408782958984375, -7.850484848022461, 448.2720947265625, 450.6123352050781, 419.57159423828125, 177.7974853515625, 71.11805725097656, 447.7225036621094, -232.92572021484375, 174.92425537109375, 185.7864990234375, 443.782470703125, 447.53570556640625, 289.194580078125, 99.64586639404297, 608.050048828125, 43.028907775878906, 149.55892944335938, 18.177597045898438, 295.08978271484375, 31.218338012695312, 88.08756256103516, 44.61293029785156, 173.34637451171875, 341.7969665527344, 52.59369659423828, -103.10150146484375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000531.npy"} +{"epoch": 0.7797356828193832, "step": 532, "batch_size": 64, "mean": 205.66433715820312, "std": 217.08494567871094, "min": -335.2635192871094, "p10": -31.017150878906236, "median": 191.54363250732422, "p90": 497.1039947509766, "max": 844.9273681640625, "pos_frac": 0.84375, "sample": [547.7509765625, 316.05450439453125, 501.2924499511719, 102.1438217163086, 551.863525390625, 150.43209838867188, 419.89019775390625, 315.3503723144531, -335.2635192871094, 187.98733520507812, -123.85429382324219, 291.2415771484375, 240.24501037597656, 388.3922119140625, 32.07853698730469, 113.97769165039062, 315.8179626464844, -228.20697021484375, 323.45892333984375, 197.73428344726562, 379.8222351074219, 610.0997314453125, 612.6029052734375, 514.2672119140625, 328.6512145996094, 95.07138061523438, 39.13267517089844, 403.012939453125, 425.4140625, -16.54461669921875, 51.09978485107422, 83.90816497802734, -37.21966552734375, 472.1662902832031, 206.72410583496094, 69.65133666992188, 150.3179931640625, 487.3309326171875, 215.705078125, 217.73614501953125, -39.38477325439453, 363.3064880371094, 107.92131805419922, 241.77310180664062, 254.44667053222656, 189.39151000976562, 119.19757843017578, 387.3729553222656, 94.3305892944336, -13.306617736816406, 31.97595977783203, -68.91693115234375, 48.62239074707031, 844.9273681640625, 103.38691711425781, 111.29550170898438, 193.6957550048828, -179.3260040283203, 205.14459228515625, 228.437255859375, -8.218734741210938, 180.94354248046875, 10.924468994140625, 137.2398681640625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000532.npy"} +{"epoch": 0.7812041116005873, "step": 533, "batch_size": 64, "mean": 165.48873901367188, "std": 191.6353759765625, "min": -312.511962890625, "p10": -61.446154022216774, "median": 141.98855590820312, "p90": 421.2763885498047, "max": 620.9957275390625, "pos_frac": 0.8125, "sample": [-109.42349243164062, -116.92843627929688, 290.00225830078125, 131.79080200195312, 18.907882690429688, 141.36981201171875, 509.71856689453125, 391.0701904296875, 141.20184326171875, 117.96044921875, 485.9082946777344, 324.2655944824219, 377.0428161621094, 353.4859619140625, -146.49993896484375, 281.5077209472656, 360.97125244140625, 187.95999145507812, 78.44009399414062, -41.488372802734375, 256.26220703125, 77.1025390625, 142.6072998046875, 424.4972229003906, 166.0426483154297, 296.34918212890625, 120.36132049560547, -312.511962890625, 79.19384765625, 620.9957275390625, 80.14535522460938, 365.3537292480469, 68.58059692382812, 89.32880401611328, -28.17013931274414, -128.573974609375, 48.232276916503906, 253.7704315185547, 45.8330078125, 210.5804443359375, 28.744007110595703, 210.31121826171875, 541.2225341796875, 413.7611083984375, -0.628082275390625, 193.6163787841797, 2.289581298828125, -5.728363037109375, 180.82675170898438, 71.35543823242188, 174.53433227539062, 59.847991943359375, 338.9169006347656, 82.44966888427734, -69.9994888305664, 34.506858825683594, -93.44737243652344, 147.106201171875, 187.52386474609375, 489.1216735839844, -27.70476531982422, 158.45986938476562, 548.0068969726562, 272.9424133300781], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000533.npy"} +{"epoch": 0.7826725403817915, "step": 534, "batch_size": 64, "mean": 178.01699829101562, "std": 175.44932556152344, "min": -408.9154968261719, "p10": -12.480702972412107, "median": 142.03105926513672, "p90": 388.8825225830078, "max": 601.7130126953125, "pos_frac": 0.859375, "sample": [186.5072021484375, 374.7098083496094, -10.87286376953125, 543.1220703125, 67.88397216796875, 86.80917358398438, -96.04913330078125, 303.5899353027344, 276.83978271484375, 52.60651397705078, 72.81822967529297, 441.7304992675781, -13.169776916503906, 108.35205078125, 388.7687683105469, 477.3525390625, 264.42724609375, 125.06912994384766, 165.02728271484375, 276.8895263671875, 128.79542541503906, -23.20785140991211, 118.31741333007812, 266.59649658203125, 278.2331237792969, 266.40582275390625, 601.7130126953125, 39.58023452758789, 68.71473693847656, 180.88397216796875, 104.53112030029297, 413.9376525878906, 141.9589385986328, -22.278236389160156, 343.1749572753906, -408.9154968261719, 27.990203857421875, 366.7755432128906, 388.9312744140625, 329.27325439453125, 93.90667724609375, 62.44371032714844, 74.24920654296875, 113.67179107666016, 382.47918701171875, 177.34019470214844, 235.7039337158203, 120.55598449707031, 233.69366455078125, 424.50408935546875, 386.3515625, 68.37830352783203, 173.91525268554688, 132.36509704589844, 227.41238403320312, 11.67449951171875, 142.10317993164062, 95.92415618896484, 356.1671142578125, -44.59466552734375, 240.79214477539062, -139.8286895751953, -8.174507141113281, 128.2293701171875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000534.npy"} +{"epoch": 0.7841409691629956, "step": 535, "batch_size": 64, "mean": 210.83477783203125, "std": 179.2676239013672, "min": -206.97418212890625, "p10": 3.7161422729492264, "median": 174.5077667236328, "p90": 436.2331207275391, "max": 588.9719848632812, "pos_frac": 0.90625, "sample": [-53.84523010253906, 564.4599609375, 85.32617950439453, 176.20977783203125, 72.03184509277344, 98.06305694580078, 108.05902099609375, 482.88067626953125, 77.47686767578125, 405.4640808105469, 151.70103454589844, 199.23904418945312, 432.3514404296875, 588.9719848632812, 272.6358337402344, 164.03651428222656, -62.736351013183594, 400.39776611328125, 195.97695922851562, 147.76438903808594, 437.8966979980469, 263.689697265625, 223.3385009765625, 392.2919006347656, 233.01564025878906, 404.457763671875, 169.0163116455078, 249.03854370117188, 68.45956420898438, 202.9688720703125, 452.381591796875, 342.5340576171875, 353.5399169921875, 227.9445037841797, 127.02387237548828, -143.34268188476562, 362.25848388671875, 351.19415283203125, 144.03509521484375, 158.2806396484375, 261.39532470703125, 548.58740234375, 72.39739227294922, 159.35812377929688, 157.86053466796875, 159.36790466308594, 144.9310302734375, 410.4386901855469, 0.4474945068359375, -206.97418212890625, 403.2733154296875, 92.62850189208984, 201.49412536621094, 36.968467712402344, -60.085174560546875, -120.75735473632812, 390.9367370605469, 11.342987060546875, 554.0623779296875, 154.66049194335938, 15.285730361938477, 311.89202880859375, 94.6199951171875, 172.80575561523438], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000535.npy"} +{"epoch": 0.7856093979441997, "step": 536, "batch_size": 64, "mean": 242.001708984375, "std": 208.06781005859375, "min": -117.17591857910156, "p10": 0.015157318115235574, "median": 196.9644012451172, "p90": 519.8614929199218, "max": 950.3262939453125, "pos_frac": 0.890625, "sample": [127.37212371826172, 167.96075439453125, 282.0853271484375, 373.6037292480469, 756.5942993164062, 97.67189025878906, 209.63006591796875, -52.37078857421875, -8.43637466430664, 15.340408325195312, 170.45663452148438, 246.15408325195312, 404.0706481933594, 520.9557495117188, 140.3365936279297, 397.10052490234375, 649.8255615234375, 82.71775817871094, 519.158203125, 358.0950927734375, 478.76611328125, 267.7926940917969, 1.1721878051757812, 191.30628967285156, 350.4734191894531, 580.6785888671875, 20.776657104492188, 405.85308837890625, 325.02923583984375, -0.480712890625, 950.3262939453125, 390.1835632324219, 210.01734924316406, 180.78359985351562, 276.0673828125, 207.17874145507812, 130.0753631591797, 615.7412109375, 3.8539390563964844, -43.68222427368164, 169.9342498779297, -8.799514770507812, 520.1629028320312, 186.76171875, 202.6225128173828, -19.46422004699707, 85.3887939453125, 153.33087158203125, 230.88096618652344, 114.5340576171875, 252.5345458984375, 90.77303314208984, 134.64239501953125, 210.01882934570312, 160.97048950195312, 474.987548828125, 144.85777282714844, 40.3270378112793, -117.17591857910156, 169.77005004882812, 393.9105224609375, 329.1695556640625, 158.505859375, 409.23150634765625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000536.npy"} +{"epoch": 0.7870778267254038, "step": 537, "batch_size": 64, "mean": 205.03135681152344, "std": 208.29222106933594, "min": -235.19488525390625, "p10": -19.251358985900872, "median": 201.9073944091797, "p90": 514.1027008056642, "max": 707.215576171875, "pos_frac": 0.875, "sample": [-235.19488525390625, -63.962066650390625, 5.801811218261719, 44.491004943847656, 8.166061401367188, 208.6682586669922, 34.75581359863281, 550.5905151367188, 28.987777709960938, 67.69076538085938, 265.8339538574219, -21.429367065429688, 309.8273010253906, 112.69956970214844, -142.4069366455078, 359.5294189453125, 541.7386474609375, -49.428253173828125, 314.91156005859375, 251.50933837890625, -14.169340133666992, 475.8887634277344, 67.67810821533203, -155.2501220703125, 136.10411071777344, 11.085769653320312, 620.2044067382812, 313.41351318359375, 223.09255981445312, 271.1813659667969, 617.8900146484375, 168.13995361328125, 530.4801025390625, 249.9248046875, 228.0770263671875, 272.3570861816406, 117.49334716796875, 382.3999938964844, 254.4989013671875, 101.04774475097656, 8.297454833984375, 421.3248596191406, 118.00611877441406, -156.42324829101562, 197.04917907714844, 377.101806640625, 45.37472152709961, 206.76560974121094, 176.37808227539062, 423.60382080078125, 97.14877319335938, 349.0707702636719, 13.561515808105469, 264.92279052734375, 460.2724609375, 113.42040252685547, 707.215576171875, 154.17529296875, 445.873291015625, 260.89825439453125, 127.24661254882812, 551.2952270507812, 13.303831100463867, 281.80548095703125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000537.npy"} +{"epoch": 0.788546255506608, "step": 538, "batch_size": 64, "mean": 199.83676147460938, "std": 178.4113311767578, "min": -154.2666015625, "p10": 5.0141880035400455, "median": 207.87051391601562, "p90": 405.4839233398438, "max": 683.4248046875, "pos_frac": 0.90625, "sample": [25.639480590820312, 314.64599609375, 228.0228271484375, 535.3056030273438, 436.52880859375, -154.2666015625, 387.66741943359375, 136.74612426757812, 231.4209442138672, 20.900672912597656, 19.11090850830078, 251.41259765625, 408.1011962890625, 181.3478240966797, 92.83824157714844, 247.9285125732422, 248.4099578857422, 41.15630340576172, 97.73637390136719, 456.8582458496094, 257.38720703125, 40.896636962890625, 59.239288330078125, 382.38592529296875, 217.64279174804688, 138.86476135253906, 133.6260986328125, 683.4248046875, 2.4659881591796875, -9.200103759765625, -152.7543182373047, 399.376953125, 165.5595245361328, 14.007949829101562, 588.142333984375, 280.35345458984375, 10.95998764038086, 257.92657470703125, 225.5874786376953, 322.681884765625, 168.26678466796875, -102.43335723876953, 18.410823822021484, 398.53289794921875, 362.9136047363281, 358.08758544921875, 20.33017349243164, 198.09823608398438, -5.076444625854492, 103.92274475097656, 331.2232666015625, 295.3465881347656, 270.25311279296875, 254.12533569335938, 47.68126678466797, 91.91519927978516, 237.1868896484375, 194.50669860839844, 518.335693359375, 393.97515869140625, -27.390579223632812, 44.45545196533203, 309.4463806152344, 81.35265350341797], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000538.npy"} +{"epoch": 0.7900146842878121, "step": 539, "batch_size": 64, "mean": 174.60418701171875, "std": 199.82589721679688, "min": -273.3730773925781, "p10": -58.10613632202148, "median": 172.46062469482422, "p90": 433.77124938964846, "max": 565.4817504882812, "pos_frac": 0.75, "sample": [206.95321655273438, 283.91119384765625, 46.23754119873047, 44.63044738769531, 238.7677001953125, 317.0406188964844, -127.15180969238281, -8.467704772949219, -8.248504638671875, 369.66070556640625, 26.300010681152344, -9.963769912719727, 333.85498046875, 235.29200744628906, 417.9728088378906, 314.7920227050781, 546.5304565429688, 174.5831756591797, 149.79217529296875, 443.8666687011719, 550.3408813476562, 169.39398193359375, 436.03826904296875, 404.5898742675781, 79.44506072998047, -8.99648666381836, 34.46788024902344, 445.45562744140625, 172.42471313476562, -33.908668518066406, -155.32843017578125, 151.97250366210938, -110.08148193359375, 385.1771240234375, 153.11669921875, -12.0462646484375, 31.393592834472656, 363.8553466796875, 52.20407485961914, -38.738494873046875, -146.50352478027344, 565.4817504882812, -31.482513427734375, 210.11146545410156, 212.97410583496094, -62.493682861328125, 423.80377197265625, 18.169815063476562, 255.46884155273438, -273.3730773925781, 252.80325317382812, -53.07512664794922, 537.4462890625, 303.5210876464844, 257.04620361328125, 52.82829284667969, 428.4815368652344, 172.4965362548828, 105.59597778320312, 308.4041748046875, -60.26228332519531, 394.88372802734375, 182.60862731933594, 52.60264587402344], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000539.npy"} +{"epoch": 0.7914831130690162, "step": 540, "batch_size": 64, "mean": 234.35610961914062, "std": 210.23281860351562, "min": -241.16552734375, "p10": -42.60294265747065, "median": 214.86673736572266, "p90": 494.48736572265625, "max": 680.6773071289062, "pos_frac": 0.890625, "sample": [265.9650573730469, 322.803955078125, 495.2112731933594, 74.56660461425781, 570.8328247070312, 8.11859130859375, 174.46530151367188, 80.0374755859375, -241.16552734375, -214.1680908203125, -209.66903686523438, 281.8553771972656, 143.1809539794922, 373.93255615234375, 585.1568603515625, 172.08233642578125, 150.72601318359375, 153.15414428710938, 384.44232177734375, 448.3740539550781, -92.28890991210938, 82.58324432373047, 191.54330444335938, 310.474365234375, 334.89508056640625, 427.0599365234375, 310.6537170410156, 646.0408325195312, 400.597900390625, 304.4535827636719, 222.79849243164062, 282.4217834472656, 205.88961791992188, 47.602943420410156, 173.98263549804688, 369.97119140625, 50.303619384765625, 7.1627197265625, 473.4294128417969, 153.51602172851562, 385.9563903808594, 680.6773071289062, 566.7345581054688, 492.7982482910156, -86.75186157226562, 464.3604431152344, 93.72479248046875, 239.23524475097656, 278.7618408203125, 206.9349822998047, 17.573562622070312, 95.20101928710938, 442.22259521484375, 323.2165222167969, 136.50106811523438, 190.0287322998047, 454.4222412109375, 110.67292022705078, 523.3578491210938, 124.93981170654297, -63.93108367919922, 329.8627624511719, 146.92921447753906, -77.6322021484375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000540.npy"} +{"epoch": 0.7929515418502202, "step": 541, "batch_size": 64, "mean": 160.1500244140625, "std": 205.92359924316406, "min": -251.4555206298828, "p10": -101.00747451782225, "median": 154.9323501586914, "p90": 370.93999328613285, "max": 946.278564453125, "pos_frac": 0.796875, "sample": [-29.578582763671875, 137.97802734375, 177.85806274414062, 167.28506469726562, 135.00228881835938, -10.897186279296875, 150.18606567382812, 110.39198303222656, 42.769927978515625, -33.0823974609375, -47.120147705078125, 601.7410888671875, -130.34400939941406, 215.6171417236328, 273.9233093261719, 66.09980010986328, 211.07614135742188, 681.9024047851562, 79.7680892944336, 181.61782836914062, 946.278564453125, 224.30075073242188, 110.03451538085938, -157.3451385498047, 194.86627197265625, -110.26734161376953, 107.00726318359375, 135.45516967773438, -251.4555206298828, 372.0321350097656, 204.29910278320312, 47.227821350097656, 298.3650817871094, 86.13821411132812, -62.828399658203125, 128.0567626953125, 225.6260986328125, 159.6786346435547, 334.7880859375, 164.0265350341797, 63.38679504394531, 268.29022216796875, 257.9847412109375, 368.39166259765625, 276.6413879394531, 307.80352783203125, 39.25960159301758, -82.9127426147461, 396.9014892578125, 277.3840637207031, -112.30570220947266, 252.3900146484375, 240.1995849609375, 557.5325927734375, 451.56427001953125, 28.472171783447266, -167.3162078857422, 134.89999389648438, -108.76235961914062, 52.132545471191406, 14.343269348144531, 239.10226440429688, 179.90945434570312, 175.82969665527344], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000541.npy"} +{"epoch": 0.7944199706314243, "step": 542, "batch_size": 64, "mean": 200.40982055664062, "std": 195.6437530517578, "min": -247.27731323242188, "p10": -29.431737327575682, "median": 209.3413543701172, "p90": 463.1502960205079, "max": 621.4874267578125, "pos_frac": 0.796875, "sample": [270.33526611328125, 175.3935546875, 232.56265258789062, -6.097801208496094, -85.6411361694336, 496.1383056640625, -29.909975051879883, 621.4874267578125, 492.23236083984375, 237.64385986328125, 130.0609588623047, 432.75140380859375, 340.8506774902344, -55.587554931640625, 164.91336059570312, 343.10858154296875, 544.3140258789062, 245.9254608154297, 44.67707061767578, 404.86517333984375, 86.9322509765625, 224.4584503173828, 244.31182861328125, 125.41009521484375, -84.27864837646484, 183.58189392089844, 26.91558837890625, 6.175849914550781, -2.3446502685546875, 471.5530700683594, -28.31584930419922, 381.2861633300781, 263.5230712890625, -15.24334716796875, 528.4678344726562, -11.21530532836914, 390.458251953125, 133.36508178710938, -24.15188980102539, 308.2278137207031, 241.3824462890625, 303.5035705566406, 147.4203643798828, -133.22073364257812, 424.663330078125, 180.77536010742188, 24.428586959838867, -154.69183349609375, 368.4948425292969, 65.94734954833984, 149.81100463867188, 332.0072937011719, 46.58470153808594, 352.54132080078125, 368.67303466796875, 287.63275146484375, 29.44809913635254, -247.27731323242188, 493.3968505859375, 211.65936279296875, 122.18997955322266, 207.02334594726562, 351.14892578125, 443.5438232421875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000542.npy"} +{"epoch": 0.7958883994126285, "step": 543, "batch_size": 64, "mean": 141.28262329101562, "std": 221.51736450195312, "min": -287.81781005859375, "p10": -111.8427909851074, "median": 100.25955581665039, "p90": 423.97726135253913, "max": 699.9432373046875, "pos_frac": 0.78125, "sample": [15.857368469238281, -34.87908935546875, 114.00603485107422, -17.737335205078125, 142.6389923095703, 660.931884765625, 5.396768569946289, -22.950393676757812, -287.81781005859375, 284.5003967285156, -95.7128677368164, -118.755615234375, 23.58413314819336, 290.19451904296875, 627.3961791992188, 248.6609649658203, 45.60504150390625, 699.9432373046875, -149.50013732910156, 11.936653137207031, 181.4257354736328, -18.56871795654297, 6.050380706787109, 10.56153678894043, -266.8697814941406, 80.8043441772461, -132.94049072265625, 350.1614990234375, 109.52410125732422, 105.52723693847656, 356.76690673828125, 93.62251281738281, -230.56787109375, 409.4270324707031, 11.331329345703125, -4.752941131591797, 684.3704833984375, 328.8748779296875, 555.6521606445312, 17.69457244873047, 430.21307373046875, 21.074417114257812, -61.45930099487305, 154.7357635498047, 208.88412475585938, 94.99187469482422, 293.155029296875, 485.8835754394531, -141.6090545654297, 197.996337890625, 240.335205078125, 19.54676055908203, 262.6103515625, 59.03839874267578, 253.07159423828125, 183.85516357421875, 158.79376220703125, 12.578086853027344, 37.53248596191406, 214.66790771484375, 43.03247833251953, 118.95214080810547, 296.8399353027344, 365.9737548828125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000543.npy"} +{"epoch": 0.7973568281938326, "step": 544, "batch_size": 64, "mean": 219.43704223632812, "std": 191.59884643554688, "min": -236.78021240234375, "p10": 16.93291587829591, "median": 204.64896392822266, "p90": 440.75460510253913, "max": 957.3087158203125, "pos_frac": 0.921875, "sample": [428.489501953125, 171.41033935546875, 187.9970703125, 493.8725280761719, 957.3087158203125, -137.05886840820312, 345.1434020996094, 77.83136749267578, 270.903564453125, -48.6925048828125, 712.8486328125, 59.26533508300781, 137.32008361816406, 208.6167755126953, 13.180213928222656, 310.7701721191406, 216.38841247558594, 203.185546875, 185.70001220703125, 303.6990661621094, 271.64984130859375, 245.0660400390625, 74.5757064819336, 247.8973388671875, 153.74826049804688, 154.36795043945312, 571.1945190429688, 178.96160888671875, 61.05891799926758, 82.60399627685547, 334.0869140625, 237.61538696289062, 83.7551040649414, 384.0580749511719, -21.584274291992188, 206.1123809814453, 334.975830078125, 216.35426330566406, 237.18515014648438, -31.66688346862793, 222.93511962890625, 196.537109375, 156.30389404296875, 341.32012939453125, 344.66650390625, 170.92710876464844, 266.6988525390625, 525.7059936523438, 209.8150177001953, 166.126220703125, 1.9330711364746094, 99.18965911865234, 172.319091796875, 488.96728515625, 385.1680603027344, 25.689220428466797, 296.9891052246094, 56.93873596191406, 354.2985534667969, 27.855880737304688, 119.32533264160156, 446.0110778808594, -236.78021240234375, 84.8346176147461], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000544.npy"} +{"epoch": 0.7988252569750367, "step": 545, "batch_size": 64, "mean": 204.42727661132812, "std": 191.4514923095703, "min": -160.59518432617188, "p10": -39.5004940032959, "median": 216.1151885986328, "p90": 448.97792358398436, "max": 698.9456176757812, "pos_frac": 0.8125, "sample": [-36.382965087890625, -67.60345458984375, -158.46548461914062, -30.5716552734375, 205.2354278564453, 348.2803039550781, 271.10205078125, -11.65821647644043, 319.3405456542969, 174.75546264648438, 431.3743591308594, 446.2836608886719, -140.767822265625, 304.89117431640625, 474.3959045410156, 50.66040802001953, 316.959716796875, 458.21600341796875, 520.386962890625, 260.7327575683594, 263.045654296875, 226.9949493408203, 416.40594482421875, 129.06214904785156, 106.68596649169922, 56.186676025390625, -39.66876220703125, 230.62420654296875, 336.4949035644531, -9.995996475219727, 446.798583984375, 74.71839904785156, 117.6540756225586, 33.68379211425781, 449.91192626953125, 282.2367858886719, 167.95729064941406, 244.93185424804688, -39.10786819458008, 328.344970703125, 18.571510314941406, 398.5120544433594, 151.25399780273438, 270.156494140625, 70.67019653320312, 377.24285888671875, 188.519287109375, 698.9456176757812, 192.05551147460938, 272.96051025390625, -50.579322814941406, 42.205657958984375, 50.497169494628906, 283.02783203125, 181.69076538085938, 353.8582763671875, 551.7437133789062, -75.94204711914062, 264.3984375, 105.63005828857422, 338.3976135253906, -160.59518432617188, 132.2206268310547, 467.77349853515625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000545.npy"} +{"epoch": 0.8002936857562408, "step": 546, "batch_size": 64, "mean": 191.35585021972656, "std": 169.6428680419922, "min": -272.64447021484375, "p10": 1.4584869384765797, "median": 210.2536849975586, "p90": 363.21375732421876, "max": 565.556884765625, "pos_frac": 0.890625, "sample": [18.851654052734375, 59.854042053222656, 359.06884765625, 121.52804565429688, 272.8172912597656, 329.76129150390625, 201.07431030273438, 287.36627197265625, 246.5314483642578, 100.20281982421875, -49.189491271972656, -87.347412109375, 70.20195007324219, 50.38140869140625, 210.7098388671875, 122.47367858886719, 50.12525177001953, 334.5061950683594, 356.88189697265625, -5.9957275390625, 38.522071838378906, 522.85693359375, -145.32272338867188, 352.72186279296875, 218.381103515625, 225.1121063232422, 234.0209197998047, 209.7975311279297, 466.9374084472656, 39.683998107910156, 169.00537109375, 28.175338745117188, 105.74488067626953, 247.69683837890625, 363.24444580078125, -201.64915466308594, 363.14215087890625, 193.6132354736328, 345.3188781738281, 529.6465454101562, -272.64447021484375, 264.90814208984375, 321.51812744140625, 151.33639526367188, 239.57131958007812, 565.556884765625, -46.64532470703125, 387.5284423828125, 203.46292114257812, 259.64080810546875, 72.22830963134766, 255.88536071777344, 148.68023681640625, 301.9132385253906, 94.7487564086914, 390.3955383300781, 87.86990356445312, 240.2689971923828, 20.368125915527344, 200.76986694335938, 62.27699279785156, 289.0511474609375, 345.0365905761719, 306.59515380859375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000546.npy"} +{"epoch": 0.801762114537445, "step": 547, "batch_size": 64, "mean": 177.60296630859375, "std": 206.3690185546875, "min": -410.78167724609375, "p10": -43.18483352661132, "median": 151.30179595947266, "p90": 462.54852294921875, "max": 731.87744140625, "pos_frac": 0.75, "sample": [-52.73162841796875, 200.38751220703125, -410.78167724609375, 529.5948486328125, -5.714305877685547, -47.972412109375, 142.64540100097656, 171.51821899414062, -99.122802734375, 407.3891296386719, 98.88726806640625, 555.9658813476562, 195.111572265625, 731.87744140625, -45.30049133300781, 9.380851745605469, 483.5194091796875, 56.624061584472656, -93.83171081542969, 386.2064208984375, 74.21598052978516, 126.74011993408203, 317.5200500488281, 104.19894409179688, 306.4491882324219, 37.463382720947266, 365.9104919433594, 288.32733154296875, 152.9331512451172, 182.73788452148438, 497.995361328125, 152.6359405517578, -28.458580017089844, 324.7756652832031, 300.3485107421875, 195.37860107421875, -3.302112579345703, 149.9676513671875, 227.57717895507812, 24.275146484375, 461.4408874511719, 540.7879028320312, 308.9368591308594, -12.950515747070312, 4.90643310546875, -38.24829864501953, -78.26142883300781, 463.0232238769531, 60.62038040161133, 178.1033935546875, 253.819091796875, -20.035537719726562, 148.55096435546875, 286.9765319824219, 97.47386169433594, 103.14152526855469, 396.3358154296875, -10.109130859375, 364.170654296875, -25.797714233398438, 423.1719970703125, 342.705078125, 127.4173583984375, -20.9325008392334], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000547.npy"} +{"epoch": 0.8032305433186491, "step": 548, "batch_size": 64, "mean": 190.63504028320312, "std": 216.26589965820312, "min": -401.9170837402344, "p10": -49.03263015747069, "median": 184.80425262451172, "p90": 438.10917053222664, "max": 992.0072021484375, "pos_frac": 0.8125, "sample": [126.34623718261719, 352.3087158203125, 134.0555419921875, 209.2287139892578, 42.42046356201172, -52.65991973876953, -401.9170837402344, 654.4378051757812, 305.4031066894531, 251.24270629882812, 276.5625, 309.0495910644531, 376.635009765625, 359.71014404296875, 258.3948974609375, -191.56295776367188, 10.098188400268555, 282.9513854980469, 490.5393371582031, 13.332433700561523, 465.727783203125, 290.2822570800781, 147.11062622070312, 222.935546875, 412.4776306152344, -65.01264190673828, 171.28956604003906, 244.37899780273438, 308.317626953125, 207.2677764892578, -67.58200073242188, 992.0072021484375, 352.316650390625, 97.59832763671875, 174.86361694335938, -95.29004669189453, -29.038421630859375, 349.08258056640625, 18.555877685546875, 518.6746826171875, 155.637939453125, -86.49667358398438, -40.56895446777344, 250.10687255859375, 21.286344528198242, -12.845062255859375, 199.0289764404297, 81.10357666015625, 271.4272766113281, -18.698516845703125, 165.33673095703125, 192.00103759765625, -16.052385330200195, 648.9522094726562, 173.84542846679688, 82.36297607421875, 86.74232482910156, 230.33224487304688, 177.6074676513672, 257.2541198730469, 142.39956665039062, 221.72427368164062, 46.52037811279297, 449.0941162109375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000548.npy"} +{"epoch": 0.8046989720998532, "step": 549, "batch_size": 64, "mean": 182.7948760986328, "std": 183.0711212158203, "min": -237.69508361816406, "p10": -25.781418609619124, "median": 142.8939666748047, "p90": 395.6239624023438, "max": 709.0800170898438, "pos_frac": 0.84375, "sample": [98.38626098632812, 491.06103515625, 87.13493347167969, 154.08670043945312, 216.43605041503906, 359.28680419921875, 23.729660034179688, 298.6722412109375, 129.2724609375, 265.9540710449219, -2.408336639404297, 270.12542724609375, 66.3993911743164, 165.45596313476562, 99.20193481445312, -7.3607940673828125, 435.90576171875, -117.5111312866211, -66.63777923583984, 146.76632690429688, 238.48931884765625, 257.2204284667969, -33.67597198486328, 361.97650146484375, 240.5916748046875, 127.89736938476562, 116.48974609375, 136.18841552734375, 257.15985107421875, 362.99676513671875, 137.394775390625, 77.85914611816406, 210.14852905273438, 111.36707305908203, 270.7792663574219, -3.6871109008789062, 88.07682037353516, 435.4564514160156, 146.49111938476562, -103.91313171386719, -37.212188720703125, 59.687660217285156, 329.6578674316406, 256.4512023925781, 709.0800170898438, 118.14915466308594, 87.76075744628906, 327.4604797363281, 139.29681396484375, 592.4710083007812, 377.33099365234375, -237.69508361816406, 17.43910026550293, 403.46380615234375, 295.60498046875, 26.2059383392334, 93.28992462158203, -51.537776947021484, 239.11109924316406, 683.6848754882812, 23.926437377929688, 325.8062744140625, 92.79307556152344, 277.3817443847656], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000549.npy"} +{"epoch": 0.8061674008810573, "step": 550, "batch_size": 64, "mean": 160.8795623779297, "std": 167.1549835205078, "min": -197.494384765625, "p10": -50.03217964172363, "median": 161.15219116210938, "p90": 398.9084075927734, "max": 504.6216735839844, "pos_frac": 0.828125, "sample": [181.94290161132812, 84.53060913085938, 429.3679504394531, 127.49056243896484, 161.82852172851562, 397.88641357421875, 228.7786407470703, 54.31898498535156, 35.9201774597168, 14.881942749023438, 246.080810546875, 149.982421875, 399.3464050292969, 104.53905487060547, 133.5237274169922, 472.4071960449219, 148.48895263671875, -46.59364700317383, 54.89189147949219, 177.6876220703125, -197.494384765625, -19.93878173828125, 349.54327392578125, 254.0095977783203, -176.9570770263672, 333.89300537109375, 159.21661376953125, 167.06463623046875, 202.4217071533203, 29.086898803710938, 16.683258056640625, -51.505836486816406, 160.47586059570312, -3.8073959350585938, 156.05343627929688, 504.6216735839844, 341.0078125, 220.35498046875, 483.62628173828125, 233.45765686035156, 297.4662780761719, 179.6092071533203, -14.084453582763672, -64.58920288085938, 201.0389862060547, 250.271728515625, 190.10328674316406, 339.83123779296875, 134.85079956054688, 216.7572021484375, 78.1062240600586, 446.9581298828125, 225.3997344970703, 414.49786376953125, -195.7957763671875, 106.1162109375, 165.27488708496094, 348.403076171875, 115.9728775024414, -88.45867919921875, 92.29823303222656, -162.4495849609375, 225.71766662597656, 73.88206481933594], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000550.npy"} +{"epoch": 0.8076358296622613, "step": 551, "batch_size": 64, "mean": 237.9665069580078, "std": 190.44442749023438, "min": -197.34292602539062, "p10": 7.065000152587891, "median": 255.38268280029297, "p90": 467.9608489990235, "max": 620.681640625, "pos_frac": 0.90625, "sample": [19.551212310791016, 6.959808349609375, 290.0582275390625, 191.70497131347656, 300.6253662109375, 430.1789245605469, -71.7689208984375, 111.65055847167969, 523.8304443359375, 293.9955749511719, 26.641868591308594, 109.58006286621094, -42.06074523925781, 482.035888671875, 361.2762451171875, 450.7356872558594, 407.9281311035156, 153.296630859375, 329.529052734375, 590.473876953125, 434.8470153808594, 212.1029052734375, 471.8212890625, 458.9531555175781, 335.252197265625, 115.15098571777344, 491.9920654296875, 233.89512634277344, -85.92101287841797, 328.26190185546875, 159.94036865234375, 62.79598617553711, 346.9674377441406, 100.98159790039062, 226.5537872314453, 389.6087341308594, 339.136962890625, 356.08111572265625, 320.365234375, -31.782182693481445, -152.05209350585938, 429.2960205078125, 249.6811981201172, 218.76016235351562, 20.638744354248047, 620.681640625, 36.450408935546875, -197.34292602539062, 534.681884765625, 21.802770614624023, 7.310447692871094, 97.09188842773438, 261.08416748046875, 369.279052734375, 205.91873168945312, 12.243637084960938, 436.366943359375, 345.8227844238281, 276.56005859375, 236.70465087890625, 157.70457458496094, 440.51141357421875, 266.0034484863281, 101.42896270751953], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000551.npy"} +{"epoch": 0.8091042584434655, "step": 552, "batch_size": 64, "mean": 268.2510986328125, "std": 233.95657348632812, "min": -131.51611328125, "p10": 15.12550678253174, "median": 240.6553955078125, "p90": 605.7608337402345, "max": 871.7969360351562, "pos_frac": 0.90625, "sample": [156.94906616210938, 211.38296508789062, 136.59869384765625, 612.4056396484375, 239.02304077148438, 653.4361572265625, 456.7603454589844, 285.2084045410156, 149.80926513671875, 152.595703125, 351.1197204589844, 388.4986572265625, -44.44316101074219, 322.605224609375, 144.7043914794922, 473.8599853515625, 267.2587585449219, 590.2562866210938, 838.4674682617188, 82.88430786132812, 140.48147583007812, 321.366943359375, 466.5950622558594, 275.3067626953125, -131.51611328125, 24.187965393066406, 30.28105926513672, 428.4709167480469, 14.032270431518555, -120.87641143798828, 499.2243957519531, -54.74577331542969, 198.56700134277344, 101.52095794677734, 310.61395263671875, 202.7633819580078, 373.47857666015625, 649.0921020507812, 871.7969360351562, 464.77447509765625, 342.1239013671875, 21.13593292236328, 165.08595275878906, 426.4856872558594, 421.4065856933594, 808.4089965820312, 286.5975646972656, 31.171646118164062, 699.0415649414062, 458.72021484375, 145.8038787841797, 43.32439041137695, 110.46013641357422, -91.80221557617188, 17.6763916015625, 214.05889892578125, 28.178543090820312, 150.7283477783203, 164.867431640625, 242.28775024414062, 330.20166015625, 300.69451904296875, 347.0009765625, -30.38385772705078], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000552.npy"} +{"epoch": 0.8105726872246696, "step": 553, "batch_size": 64, "mean": 178.16009521484375, "std": 198.6342010498047, "min": -170.5230712890625, "p10": -104.24310073852538, "median": 164.0085906982422, "p90": 446.77106323242197, "max": 864.3414916992188, "pos_frac": 0.8125, "sample": [864.3414916992188, 545.0469970703125, -44.0439453125, 159.11495971679688, 213.55767822265625, -2.4736709594726562, 36.17620849609375, -110.55609893798828, 194.45960998535156, 165.6279296875, 76.60574340820312, 96.81266784667969, 159.8608856201172, 241.75704956054688, 28.94164276123047, 235.66651916503906, -104.81558227539062, 17.852996826171875, 297.2744140625, -118.73704528808594, 202.68814086914062, 218.0777130126953, 325.6929016113281, -112.31693267822266, 303.0924072265625, 47.82334899902344, 475.97479248046875, 25.9696044921875, 58.02552032470703, 288.76483154296875, -128.75379943847656, 241.23065185546875, 290.36566162109375, -146.4615936279297, 150.43374633789062, 122.02374267578125, 165.0794677734375, 125.4588623046875, 382.89300537109375, -102.90731048583984, 164.15835571289062, 35.87385559082031, 313.9533996582031, 228.89852905273438, 54.021324157714844, 163.85882568359375, -22.455856323242188, 140.79351806640625, 510.8006591796875, 162.25064086914062, 501.285888671875, 257.18536376953125, 416.5042419433594, 154.06639099121094, 491.72796630859375, 426.5489501953125, 85.00397491455078, 405.4443359375, -12.66448974609375, 455.43768310546875, -170.5230712890625, 276.83746337890625, 195.31048583984375, 282.30218505859375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000553.npy"} +{"epoch": 0.8120411160058737, "step": 554, "batch_size": 64, "mean": 236.0059356689453, "std": 225.51168823242188, "min": -401.1003112792969, "p10": -49.05091857910154, "median": 259.9373474121094, "p90": 535.4107727050782, "max": 732.1500244140625, "pos_frac": 0.875, "sample": [117.80604553222656, -26.865737915039062, 267.4223327636719, 111.11569213867188, 354.7522277832031, 286.6401672363281, 373.3771667480469, 355.8357238769531, 207.10830688476562, 271.1990966796875, 550.2936401367188, 246.27410888671875, 376.40704345703125, 249.86566162109375, -58.55885314941406, 329.5907287597656, 428.4570617675781, 99.95126342773438, 253.73959350585938, 266.1351013183594, -78.24120330810547, -233.31832885742188, 86.8795166015625, 37.558021545410156, 681.1900634765625, 9.288749694824219, 198.38449096679688, 272.8525085449219, 217.9354248046875, 302.634765625, 330.19268798828125, 499.4552001953125, 615.6937866210938, 100.69784545898438, 543.4443969726562, 235.5736083984375, 599.0328369140625, 84.13137817382812, 456.7386779785156, -96.26460266113281, 99.77560424804688, 331.60601806640625, 314.1939392089844, 372.4498291015625, 353.7559509277344, 76.21635437011719, 294.4638977050781, 22.9305419921875, 311.87353515625, 202.13970947265625, -157.8077392578125, 162.16329956054688, 284.5869445800781, -401.1003112792969, 516.6656494140625, 71.8583984375, -134.94375610351562, 117.23474884033203, 477.43829345703125, 323.1368408203125, 732.1500244140625, 679.7835693359375, 107.02819061279297, 22.373836517333984], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000554.npy"} +{"epoch": 0.8135095447870778, "step": 555, "batch_size": 64, "mean": 225.95533752441406, "std": 239.44918823242188, "min": -272.9366760253906, "p10": -15.382216835021971, "median": 169.50648498535156, "p90": 609.2062133789063, "max": 777.6156005859375, "pos_frac": 0.875, "sample": [162.9748077392578, 138.9681854248047, 341.6907043457031, 372.8086853027344, 285.71588134765625, 593.16064453125, 181.0276641845703, 211.02084350585938, -16.00665283203125, 75.0339126586914, 19.076614379882812, 478.445556640625, 477.4117736816406, 336.56182861328125, 84.264404296875, -13.925199508666992, 120.07910919189453, 43.4716796875, 727.50390625, 575.329833984375, 176.43331909179688, 616.0828857421875, 168.57577514648438, 488.8057861328125, 339.462158203125, 55.2288818359375, 426.4959716796875, 45.78419494628906, 183.9578857421875, 7.6168060302734375, 155.15379333496094, 316.95526123046875, 169.847412109375, 275.1302185058594, 21.446029663085938, 308.3558044433594, 222.93907165527344, 1.4194564819335938, 69.77114868164062, 69.86760711669922, -272.9366760253906, 304.13177490234375, 676.6915283203125, 670.2115478515625, 1.2270832061767578, -124.52403259277344, 661.7677001953125, 22.566328048706055, 169.16555786132812, 166.31640625, 726.27783203125, -120.62289428710938, 457.757568359375, -69.01170349121094, -19.115394592285156, 214.73806762695312, 777.6156005859375, 50.677093505859375, 103.30691528320312, 141.699951171875, 434.74853515625, 205.59481811523438, -83.65061950683594, 52.56597900390625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000555.npy"} +{"epoch": 0.8149779735682819, "step": 556, "batch_size": 64, "mean": 175.44180297851562, "std": 221.1058349609375, "min": -226.6982421875, "p10": -49.04304733276367, "median": 147.698974609375, "p90": 489.05208740234394, "max": 821.03857421875, "pos_frac": 0.8125, "sample": [415.8344421386719, 80.30164337158203, 66.85934448242188, 23.9027099609375, 177.99978637695312, 70.31890106201172, 409.6485290527344, 205.83297729492188, -105.15831756591797, 328.3754577636719, 0.7109298706054688, 372.02874755859375, 201.62359619140625, 120.84750366210938, 277.591552734375, 148.3648681640625, 739.450927734375, 106.02774047851562, 334.0389709472656, 9.694755554199219, -215.48301696777344, 76.28824615478516, -226.6982421875, 55.55243682861328, 51.31011962890625, 315.43328857421875, 183.58621215820312, -19.850051879882812, 440.7056884765625, -99.8773193359375, 197.8138427734375, 58.18650817871094, 168.6310577392578, -112.33361053466797, 509.77197265625, -48.32164764404297, 147.0330810546875, 100.04524230957031, 177.61233520507812, 80.82282257080078, -18.49609375, 300.2160949707031, 219.14537048339844, 208.96533203125, -164.99270629882812, 137.0425567626953, 681.6634521484375, 88.74754333496094, 559.4771118164062, 20.394996643066406, 193.9111328125, 3.8590316772460938, 379.4041748046875, 114.60076904296875, 821.03857421875, -8.344291687011719, 560.845947265625, 613.1337890625, 174.08892822265625, 174.0091094970703, 173.4612579345703, -17.092884063720703, -49.35221862792969, 238.0247802734375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000556.npy"} +{"epoch": 0.8164464023494861, "step": 557, "batch_size": 64, "mean": 197.9497833251953, "std": 212.25421142578125, "min": -233.74900817871094, "p10": -36.252143096923824, "median": 172.07102966308594, "p90": 517.804119873047, "max": 753.585205078125, "pos_frac": 0.84375, "sample": [-114.978271484375, 167.4939422607422, -233.74900817871094, 97.93243408203125, 44.81123733520508, 251.911376953125, -20.461410522460938, 389.4822082519531, 321.47015380859375, 97.61981201171875, 257.8198547363281, 314.06048583984375, 379.6212463378906, -37.172607421875, 41.9034538269043, 271.18731689453125, 271.73992919921875, 205.0571746826172, 22.391666412353516, 75.86609649658203, 49.50567626953125, 753.585205078125, 464.03131103515625, 562.54541015625, 122.38134765625, 380.00091552734375, 138.70458984375, 68.59617614746094, -59.952232360839844, 492.7930908203125, 176.3284912109375, 200.37118530273438, 536.8567504882812, 204.36376953125, 90.34117126464844, 65.83229064941406, 254.9758758544922, 52.42656707763672, 336.15203857421875, 174.47299194335938, 111.01154327392578, -46.920387268066406, 171.06265258789062, 87.33346557617188, 79.02839660644531, -165.37677001953125, 420.5220642089844, 607.8558349609375, -19.860504150390625, 528.5231323242188, 123.13093566894531, 554.0166625976562, 702.712158203125, 173.81369018554688, 220.46798706054688, 21.165939331054688, 10.941036224365234, 418.4145202636719, -117.06340026855469, 451.89056396484375, -34.104393005371094, 218.17630004882812, 173.07940673828125, 110.64482116699219], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000557.npy"} +{"epoch": 0.8179148311306902, "step": 558, "batch_size": 64, "mean": 279.3931579589844, "std": 228.474609375, "min": -286.64239501953125, "p10": 9.398254394531257, "median": 260.9794616699219, "p90": 590.3374267578125, "max": 759.4757690429688, "pos_frac": 0.90625, "sample": [299.62451171875, 163.59877014160156, 265.5112609863281, 380.0782165527344, 102.9677734375, 249.2166290283203, 404.1200256347656, 117.23829650878906, 96.7000961303711, -41.488494873046875, -85.35885620117188, 483.8787536621094, 759.4757690429688, 615.6082763671875, 128.4854278564453, 289.8674621582031, 268.698974609375, 284.3193664550781, 590.9202880859375, 438.2096252441406, 182.82582092285156, 256.0933837890625, 175.29302978515625, 363.8800354003906, 16.55596160888672, 233.86068725585938, 568.193115234375, 98.47590637207031, 418.5343017578125, 167.4319610595703, 138.36782836914062, 256.4476623535156, 308.722900390625, 701.0674438476562, 588.9774169921875, 337.90875244140625, -167.0374755859375, 80.01278686523438, 192.61758422851562, 327.48419189453125, 162.8302001953125, 41.856414794921875, 203.15013122558594, 246.7802734375, 167.32408142089844, -25.01085090637207, -286.64239501953125, 708.8184204101562, 455.6827087402344, 6.330665588378906, 450.1526184082031, 571.810302734375, 472.08868408203125, 366.20770263671875, 708.09130859375, -76.55076599121094, 292.1759338378906, 167.21096801757812, 142.41448974609375, 39.813995361328125, 285.666015625, 729.2020263671875, 498.6388854980469, 495.7354431152344], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000558.npy"} +{"epoch": 0.8193832599118943, "step": 559, "batch_size": 64, "mean": 210.38426208496094, "std": 204.4213104248047, "min": -128.2694549560547, "p10": -55.72831726074218, "median": 196.2493667602539, "p90": 504.81188049316415, "max": 680.2167358398438, "pos_frac": 0.859375, "sample": [370.6992492675781, 360.1683654785156, 339.8375244140625, 356.54547119140625, 3.0376663208007812, 33.36344909667969, 100.05476379394531, 75.30940246582031, 117.3567123413086, 680.2167358398438, 338.76708984375, 293.66729736328125, 215.57142639160156, 223.91744995117188, 114.78512573242188, 470.7445983886719, 593.48046875, 477.8816833496094, -64.969970703125, -82.4200439453125, 516.3533935546875, 99.79716491699219, 37.048919677734375, 214.20188903808594, 190.88185119628906, 201.61688232421875, -74.3496322631836, 60.216552734375, 261.703125, 238.65472412109375, 54.03589630126953, -46.727657318115234, 17.84815216064453, 126.11975860595703, 129.53688049316406, 295.68115234375, -104.61862182617188, 43.577003479003906, 287.167724609375, 185.64364624023438, 439.76361083984375, 105.03575134277344, 27.4471435546875, 42.259559631347656, -128.2694549560547, 190.00042724609375, 101.71359252929688, 172.236328125, 565.4959106445312, -36.04198455810547, 270.7010803222656, 368.7669372558594, 289.9339599609375, 295.231689453125, 580.9425659179688, 670.2532348632812, -81.13206481933594, 571.1653442382812, 389.32232666015625, 218.84014892578125, 212.0670623779297, 37.78465270996094, -59.58574295043945, 468.2571105957031], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000559.npy"} +{"epoch": 0.8208516886930984, "step": 560, "batch_size": 64, "mean": 154.923828125, "std": 160.76820373535156, "min": -193.82382202148438, "p10": -37.61505775451659, "median": 141.4957504272461, "p90": 381.4240692138672, "max": 512.8580322265625, "pos_frac": 0.828125, "sample": [332.64697265625, 103.89665985107422, 163.61703491210938, 263.7518615722656, -19.8048038482666, 379.01531982421875, 4.523157119750977, -122.34351348876953, 179.27212524414062, 258.6912536621094, 82.78021240234375, 78.07510375976562, 277.1448974609375, 215.40550231933594, 512.8580322265625, 67.76648712158203, 468.32891845703125, 414.05609130859375, 123.13873291015625, 214.09521484375, 276.12774658203125, 65.96624755859375, 143.38717651367188, 455.9507141113281, 169.60548400878906, -23.74224090576172, 71.9522705078125, 335.831787109375, 153.82496643066406, -48.75080871582031, 382.4563903808594, -42.77943420410156, 23.70393180847168, 213.96615600585938, 175.53504943847656, -119.8799057006836, 37.64563751220703, 505.9234313964844, 117.72470092773438, 392.6749572753906, 139.6043243408203, -58.38507080078125, 188.128173828125, 24.969635009765625, 298.54486083984375, 187.63294982910156, 76.72096252441406, 88.28543853759766, -9.78288459777832, 300.6554260253906, 189.1270751953125, 46.5018310546875, 189.78494262695312, 85.16165924072266, 66.78639221191406, 69.8804702758789, -25.56484603881836, 352.81524658203125, 267.4781494140625, 292.5447998046875, -193.82382202148438, 35.43572235107422, -68.74427032470703, 87.3282470703125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000560.npy"} +{"epoch": 0.8223201174743024, "step": 561, "batch_size": 64, "mean": 166.27667236328125, "std": 178.9585418701172, "min": -373.56689453125, "p10": -43.50627822875976, "median": 167.90450286865234, "p90": 368.3968505859376, "max": 691.6019287109375, "pos_frac": 0.828125, "sample": [303.36639404296875, 155.55194091796875, -46.203529357910156, -70.85648345947266, -78.4411392211914, 317.5741882324219, 297.1369934082031, 45.29966735839844, 228.08883666992188, 463.6809387207031, -12.035770416259766, 300.30670166015625, 469.29986572265625, 216.6763153076172, 132.63864135742188, 165.85598754882812, 66.76248168945312, 77.67691802978516, 261.4525146484375, 170.0640411376953, 527.3169555664062, 43.09927749633789, 301.9748840332031, 259.8816223144531, 58.473297119140625, 209.31394958496094, 445.4002990722656, 194.92239379882812, 169.95301818847656, 381.7810363769531, 337.1670837402344, 691.6019287109375, 6.2953948974609375, 215.82363891601562, 310.23175048828125, 123.04302215576172, -104.50740051269531, 69.52717590332031, 19.360197067260742, 221.329833984375, 73.40717315673828, -37.21269226074219, -13.326698303222656, 289.58953857421875, 103.70742797851562, -85.69091033935547, 186.537353515625, 502.4197998046875, 273.4765625, 251.51705932617188, -373.56689453125, 121.20855712890625, 236.47474670410156, 51.36902618408203, 256.23590087890625, 18.6328125, 285.17315673828125, 145.6338653564453, 131.75096130371094, 22.985107421875, 202.69537353515625, -84.26593017578125, 163.38327026367188, -26.313209533691406], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000561.npy"} +{"epoch": 0.8237885462555066, "step": 562, "batch_size": 64, "mean": 231.1669158935547, "std": 222.22869873046875, "min": -188.68917846679688, "p10": 6.328999519348148, "median": 186.84603881835938, "p90": 479.8970977783203, "max": 951.66796875, "pos_frac": 0.90625, "sample": [381.7158508300781, 951.66796875, 314.2706604003906, 98.84008026123047, 94.38327026367188, 147.45217895507812, 146.26788330078125, 133.3414764404297, 262.9179992675781, 299.62530517578125, 214.09661865234375, 581.0208129882812, 132.89328002929688, 368.4930725097656, 452.5050048828125, -55.90617370605469, 118.8289794921875, 257.96240234375, 181.3939666748047, 308.1055603027344, 161.8881378173828, 297.85186767578125, 411.6837158203125, 4.970067977905273, 9.499839782714844, 447.85406494140625, 531.1405639648438, 12.869400024414062, 634.8427734375, 56.28759765625, -187.92831420898438, -32.03742218017578, 133.5207977294922, 379.6306457519531, -51.94928741455078, 894.5244750976562, 76.370361328125, 250.16033935546875, 238.83181762695312, 402.96087646484375, 87.64891815185547, 128.56541442871094, 378.8882751464844, 299.87255859375, -188.68917846679688, 236.72866821289062, 36.10716247558594, 478.0532531738281, 319.4627685546875, 690.3909301757812, 73.03472900390625, 192.29811096191406, 333.8349914550781, 154.4653778076172, 170.30972290039062, 203.1035919189453, 95.26078796386719, 94.30384063720703, -85.11811065673828, 51.10238265991211, 84.94667053222656, 114.33897399902344, 480.68731689453125, 302.2375183105469], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000562.npy"} +{"epoch": 0.8252569750367107, "step": 563, "batch_size": 64, "mean": 171.15786743164062, "std": 202.13336181640625, "min": -311.2334899902344, "p10": -112.15496368408199, "median": 176.87477111816406, "p90": 431.49711608886724, "max": 608.6920166015625, "pos_frac": 0.78125, "sample": [96.12352752685547, -129.03488159179688, -128.315185546875, 203.12457275390625, 319.8251037597656, -143.2132568359375, 469.9493103027344, 140.66226196289062, 294.5863037109375, 278.9668273925781, 196.88931274414062, 173.07229614257812, -18.469566345214844, 306.1220397949219, 232.61236572265625, 232.3485870361328, 57.86857604980469, 608.6920166015625, 376.8004150390625, 65.00874328613281, 414.1623229980469, 2.1162185668945312, -46.920501708984375, -127.46015930175781, 145.9253692626953, -2.605571746826172, 282.54901123046875, 353.25811767578125, 317.21771240234375, 172.2495574951172, 25.64617919921875, 159.67982482910156, -268.5535888671875, 325.0238952636719, -0.4327545166015625, 289.8957214355469, -45.25008773803711, -33.263465881347656, 438.32122802734375, 274.020751953125, 392.25836181640625, 453.8130798339844, 446.0992126464844, 180.67724609375, 331.9723205566406, 210.74676513671875, -311.2334899902344, 595.044677734375, 8.316520690917969, 415.5741882324219, -76.44284057617188, 481.5067138671875, 105.90430450439453, 39.31767272949219, 200.72500610351562, -143.57217407226562, 359.4183349609375, 48.91617202758789, 124.79450988769531, 71.41039276123047, 223.37339782714844, 79.67977905273438, 275.3030090332031, 131.30174255371094], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000563.npy"} +{"epoch": 0.8267254038179148, "step": 564, "batch_size": 64, "mean": 213.35166931152344, "std": 158.0843505859375, "min": -164.20814514160156, "p10": 18.682535934448246, "median": 205.54314422607422, "p90": 456.51244812011726, "max": 581.0137939453125, "pos_frac": 0.921875, "sample": [259.20635986328125, 254.67282104492188, 481.0886535644531, -15.099075317382812, 472.1432800292969, 462.501708984375, 371.1931457519531, -48.075172424316406, 190.76065063476562, 49.153350830078125, 150.01254272460938, 303.22979736328125, 133.57354736328125, 223.53018188476562, 344.379638671875, 23.265655517578125, 86.22206115722656, 217.02212524414062, 374.20733642578125, 182.8987579345703, 238.3350067138672, 128.83229064941406, 236.3340301513672, -164.20814514160156, 85.62284851074219, 167.09512329101562, -45.449981689453125, -50.427345275878906, 12.190216064453125, 29.831787109375, 194.05606079101562, 393.76373291015625, 282.3770751953125, 73.17041778564453, 138.79959106445312, 220.7984619140625, 194.0641632080078, 138.02578735351562, 148.47393798828125, 16.718341827392578, 442.5375061035156, 441.7903747558594, 63.8160400390625, 160.19415283203125, 398.02801513671875, 257.0265197753906, 471.3704528808594, 328.55767822265625, 467.9290466308594, 275.7664794921875, 180.3485107421875, 286.5277404785156, 125.1080093383789, 50.02024841308594, 270.9827880859375, 106.33958435058594, 326.31561279296875, 462.7164611816406, 308.5884094238281, 91.61381530761719, 581.0137939453125, 117.71734619140625, 233.44300842285156, 252.46414184570312], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000564.npy"} +{"epoch": 0.8281938325991189, "step": 565, "batch_size": 64, "mean": 217.47377014160156, "std": 247.77931213378906, "min": -139.36817932128906, "p10": -64.69062957763671, "median": 190.6804656982422, "p90": 592.3961120605469, "max": 875.9595336914062, "pos_frac": 0.765625, "sample": [-108.31529235839844, 875.9595336914062, 570.81787109375, -106.38276672363281, 111.3212890625, 347.112548828125, 615.6429443359375, 58.21739196777344, 28.518814086914062, -54.93977355957031, 222.09164428710938, 371.75274658203125, -11.769439697265625, 148.57470703125, -5.149662017822266, -139.36817932128906, -73.04354095458984, 95.2257080078125, -31.162353515625, 293.8182067871094, 322.6280517578125, 195.03610229492188, 307.1310119628906, 730.6754150390625, -2.0441131591796875, 57.85541534423828, 174.96673583984375, 189.17974853515625, -138.74063110351562, 345.56842041015625, 245.93392944335938, 232.0522003173828, 269.87548828125, 348.6090087890625, -12.165237426757812, 0.5769500732421875, 533.7991333007812, 343.0699462890625, -68.86956787109375, 194.30352783203125, -3.7767486572265625, 253.1917724609375, 239.58328247070312, 777.58154296875, 265.31158447265625, 814.6182861328125, 128.41192626953125, 581.908203125, 401.91278076171875, -117.81704711914062, 156.857421875, 19.966537475585938, 428.4618835449219, 359.05517578125, 601.0166015625, 192.18118286132812, 355.4490661621094, 37.931339263916016, 141.62033081054688, 113.2646255493164, 102.16604614257812, 38.67202377319336, -44.50154113769531, 596.8909301757812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000565.npy"} +{"epoch": 0.8296622613803231, "step": 566, "batch_size": 64, "mean": 205.38894653320312, "std": 242.2856903076172, "min": -241.48538208007812, "p10": -115.67440490722656, "median": 225.59204864501953, "p90": 534.8379089355469, "max": 767.9952392578125, "pos_frac": 0.75, "sample": [-23.27978515625, -65.45197296142578, 303.918701171875, 432.32342529296875, 27.081451416015625, -70.97561645507812, 540.960205078125, 169.75369262695312, -230.26904296875, 400.1883544921875, 278.8823547363281, -18.427413940429688, -3.86712646484375, 169.62979125976562, 294.5208740234375, 263.8710021972656, 328.8702392578125, 29.146961212158203, 22.96310806274414, -241.48538208007812, 409.9062194824219, 428.9527587890625, 302.20391845703125, -117.59661865234375, 209.23245239257812, -65.8371810913086, 184.11680603027344, 373.1910095214844, 99.19950866699219, 572.6727294921875, 273.4634704589844, 767.9952392578125, 419.6341857910156, 38.891632080078125, 628.3896484375, 29.980140686035156, -157.2658233642578, 582.6168212890625, 269.6200866699219, -41.13618469238281, -142.33804321289062, 350.3728332519531, 86.0438461303711, 461.7764892578125, 335.41241455078125, 495.163818359375, 552.7889404296875, 356.345703125, -111.18923950195312, -38.98310089111328, 4.674289703369141, 551.8890991210938, 176.82537841796875, 501.0154113769531, 6.536905288696289, -122.46263885498047, 390.49700927734375, 338.95574951171875, 520.5525512695312, -153.8419189453125, 148.96722412109375, 77.58529663085938, 241.95164489746094, 299.76837158203125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000566.npy"} +{"epoch": 0.8311306901615272, "step": 567, "batch_size": 64, "mean": 200.66995239257812, "std": 207.567626953125, "min": -164.55628967285156, "p10": -60.123818969726535, "median": 188.61919403076172, "p90": 482.166879272461, "max": 662.517822265625, "pos_frac": 0.8125, "sample": [24.81395721435547, 214.42359924316406, 630.6150512695312, 349.2638854980469, 356.11895751953125, 326.89251708984375, 100.51447296142578, 181.623046875, 521.3881225585938, 274.6286315917969, 444.463623046875, 18.25403594970703, 602.6163330078125, 292.2881774902344, 131.93875122070312, 152.72564697265625, 405.22369384765625, 434.99554443359375, 381.7782897949219, 64.87168884277344, 330.99102783203125, 238.3553009033203, -139.32228088378906, 251.45726013183594, 178.75770568847656, 310.19769287109375, 140.8562774658203, 519.1831665039062, -32.89619445800781, 375.7236328125, -22.409677505493164, 458.33660888671875, 171.0504608154297, -133.7476806640625, 88.629150390625, 1.2929496765136719, 129.32623291015625, 339.7725830078125, 662.517822265625, 493.54876708984375, 25.378387451171875, -164.55628967285156, -32.49462890625, -16.90526580810547, 71.09352111816406, 492.3798522949219, 208.0301513671875, 104.78516387939453, 18.40563201904297, -88.33830261230469, -148.68934631347656, 387.04840087890625, 29.87493133544922, 402.414306640625, -71.79280090332031, 194.06179809570312, 100.4404067993164, 317.9058837890625, -16.198402404785156, 185.50965881347656, 236.0995635986328, 191.72872924804688, -149.8403778076172, 295.4775390625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000567.npy"} +{"epoch": 0.8325991189427313, "step": 568, "batch_size": 64, "mean": 237.3043975830078, "std": 207.4449920654297, "min": -352.886962890625, "p10": -9.463046264648417, "median": 222.49490356445312, "p90": 522.6878112792969, "max": 795.0197143554688, "pos_frac": 0.890625, "sample": [273.59637451171875, 21.415203094482422, 406.0661926269531, 216.4410400390625, 299.61383056640625, 291.60809326171875, 296.301025390625, 286.3671875, 409.4459228515625, -352.886962890625, 76.17301940917969, -69.53961944580078, 343.83050537109375, 588.41943359375, 10.920448303222656, 276.2139587402344, 135.86297607421875, 41.090885162353516, 194.30796813964844, 180.5999298095703, 187.55197143554688, 232.67593383789062, 555.4605712890625, 351.26177978515625, 212.37060546875, 239.42044067382812, 175.01321411132812, 795.0197143554688, 45.960121154785156, -43.36722946166992, 310.8284912109375, 295.09100341796875, 336.06524658203125, 529.0869750976562, 411.28912353515625, 367.9830322265625, 202.24180603027344, 272.9722900390625, 203.5835723876953, 401.0428161621094, 650.9923706054688, 57.540958404541016, 152.13877868652344, 212.3202362060547, 188.94651794433594, 148.1558837890625, -50.12039566040039, -212.6193389892578, 160.59744262695312, 525.9356079101562, 118.7141342163086, 599.4174194335938, 254.83135986328125, -130.65853881835938, 142.6744384765625, 192.4422607421875, 108.953369140625, 495.8736572265625, 259.6750793457031, 515.109619140625, 228.54876708984375, 485.1943054199219, -18.198829650878906, 93.61717224121094], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000568.npy"} +{"epoch": 0.8340675477239354, "step": 569, "batch_size": 64, "mean": 193.74807739257812, "std": 227.28184509277344, "min": -487.4178466796875, "p10": -10.05993423461914, "median": 162.97211456298828, "p90": 448.8573608398438, "max": 957.9197387695312, "pos_frac": 0.859375, "sample": [-119.4218978881836, 216.69589233398438, 957.9197387695312, 129.0575408935547, 434.4576416015625, -4.4915618896484375, 160.71829223632812, 162.06008911132812, -121.94518280029297, 9.462017059326172, 499.42535400390625, 523.4183349609375, 175.07376098632812, 300.4884033203125, 328.8807373046875, 50.799774169921875, 68.42158508300781, 418.893310546875, 118.05316162109375, 93.01959991455078, -487.4178466796875, 408.51702880859375, 144.58901977539062, 13.990718841552734, -9.101255416870117, 383.7492370605469, 546.8785400390625, 42.86296081542969, 35.801544189453125, 6.184047698974609, 235.1317138671875, 541.998291015625, 453.7783203125, 72.26986694335938, 421.3492126464844, 324.0366516113281, 32.05474853515625, 270.9817199707031, 350.44012451171875, -141.220458984375, 401.1901550292969, 18.311473846435547, -10.470796585083008, 467.15869140625, 208.6075897216797, 100.70575714111328, 382.716796875, 265.9951477050781, -79.37358856201172, 57.86949157714844, 37.25547790527344, 437.3751220703125, 74.08296966552734, 111.97457122802734, 163.88414001464844, -259.1024475097656, 143.19903564453125, 183.58534240722656, 79.28852844238281, 311.59991455078125, 190.93472290039062, 220.57443237304688, 422.19512939453125, 422.4590759277344], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000569.npy"} +{"epoch": 0.8355359765051396, "step": 570, "batch_size": 64, "mean": 215.5782470703125, "std": 186.26356506347656, "min": -114.8064193725586, "p10": -29.31507015228271, "median": 187.0723114013672, "p90": 499.2525848388673, "max": 596.6212158203125, "pos_frac": 0.859375, "sample": [144.1869354248047, 507.4245910644531, 191.27207946777344, 512.7949829101562, 553.4423828125, 168.8174285888672, 366.8981628417969, 508.75091552734375, 466.9952697753906, 212.34217834472656, 179.1697998046875, 431.9551086425781, 217.23330688476562, 283.4495849609375, 15.091268539428711, 263.56207275390625, 596.6212158203125, 243.93548583984375, 423.6235046386719, -49.58204650878906, 176.60015869140625, -114.8064193725586, -26.432954788208008, 80.27371978759766, 60.65741729736328, 560.816650390625, 101.63706970214844, 181.04818725585938, 129.79058837890625, 213.29969787597656, 134.12612915039062, 242.4585418701172, 28.21851348876953, 29.537050247192383, 127.70907592773438, 475.576904296875, 91.035400390625, 364.62066650390625, 61.684410095214844, 475.7701721191406, 206.5359649658203, -63.44220733642578, -57.560150146484375, 142.13226318359375, 281.81646728515625, 280.168212890625, 182.87254333496094, 391.20391845703125, 323.9735412597656, 437.4876708984375, -8.804218292236328, 71.60885620117188, 480.1845703125, 218.00709533691406, 201.664794921875, -30.550262451171875, 46.35820770263672, 169.2552032470703, 194.6728973388672, -42.58243179321289, -50.62700653076172, 554.422607421875, 103.24322509765625, 133.36082458496094], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000570.npy"} +{"epoch": 0.8370044052863436, "step": 571, "batch_size": 64, "mean": 225.44558715820312, "std": 227.0445098876953, "min": -316.82525634765625, "p10": -36.941321945190424, "median": 201.2151870727539, "p90": 489.8732604980469, "max": 810.5743408203125, "pos_frac": 0.875, "sample": [513.391357421875, 335.65631103515625, 98.70033264160156, 352.5158386230469, 162.45144653320312, 197.42251586914062, 26.383159637451172, 356.5569152832031, 225.2019805908203, 459.03802490234375, 810.5743408203125, 197.2290802001953, 102.8357925415039, 383.35211181640625, 8.549022674560547, 227.32313537597656, 396.07537841796875, 463.225341796875, 140.90805053710938, 430.1593017578125, 491.58282470703125, 670.6793212890625, 499.092041015625, 200.86260986328125, -124.5040054321289, 140.1937255859375, 292.38946533203125, 194.6985321044922, 130.10739135742188, 394.6748046875, 103.29850006103516, 25.933868408203125, 338.51275634765625, 673.8269653320312, 552.9776611328125, 172.95103454589844, -73.0181884765625, 83.08233642578125, 485.88427734375, 406.1737060546875, 240.62091064453125, 51.78792190551758, 471.3552551269531, 160.7300567626953, -32.316471099853516, 221.49252319335938, 448.0247802734375, 201.56776428222656, 130.6565399169922, 189.5510711669922, -108.06602478027344, -315.61468505859375, 234.2686767578125, 188.30462646484375, -316.82525634765625, 264.0229797363281, 176.92832946777344, 320.5289001464844, -294.95611572265625, 218.7570037841797, 8.800445556640625, 407.8967590332031, 52.976104736328125, -38.92340087890625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000571.npy"} +{"epoch": 0.8384728340675477, "step": 572, "batch_size": 64, "mean": 207.0398406982422, "std": 263.8523864746094, "min": -354.2936706542969, "p10": -36.30773544311522, "median": 154.01842498779297, "p90": 611.196923828125, "max": 904.3502197265625, "pos_frac": 0.8125, "sample": [-179.78646850585938, -214.6869659423828, 21.14886474609375, 343.719970703125, 806.7547607421875, 142.2099609375, -14.069122314453125, 52.82099151611328, 54.96677017211914, -21.669532775878906, 113.07967376708984, 153.81854248046875, 408.5132141113281, 103.85316467285156, 338.863525390625, 329.8343200683594, 393.9075012207031, 672.38232421875, 176.13070678710938, 295.4156188964844, 5.7530975341796875, 92.47938537597656, 179.32618713378906, 366.0829772949219, 61.35845184326172, 116.75785064697266, 30.635372161865234, -49.59669494628906, -156.55621337890625, 443.29534912109375, 295.1125183105469, -41.337013244628906, 194.61318969726562, 100.35574340820312, 74.19378662109375, 894.79443359375, 119.68502044677734, 208.90493774414062, -12.733053207397461, 166.7313995361328, 445.03240966796875, 31.723766326904297, -149.74755859375, 350.3186950683594, 12.396305084228516, -354.2936706542969, 904.3502197265625, 154.2183074951172, 174.7628631591797, -5.369514465332031, 510.19873046875, 158.30609130859375, 233.94552612304688, 657.6846313476562, 102.23741149902344, 218.70840454101562, 439.00439453125, 729.0052490234375, 29.76171875, -24.57275390625, 256.20196533203125, 605.501220703125, 613.637939453125, 90.47340393066406], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000572.npy"} +{"epoch": 0.8399412628487518, "step": 573, "batch_size": 64, "mean": 210.18882751464844, "std": 204.98233032226562, "min": -201.79513549804688, "p10": -22.246588897705077, "median": 217.30941009521484, "p90": 483.3437957763675, "max": 685.7117919921875, "pos_frac": 0.828125, "sample": [316.0776062011719, -12.963560104370117, 79.8342514038086, 65.1685791015625, 4.9063873291015625, 595.6414794921875, 232.35647583007812, 51.04163360595703, 361.8669128417969, 359.1055908203125, 277.8775939941406, 214.87501525878906, -7.89671516418457, 656.554931640625, 219.74380493164062, 228.11341857910156, 685.7117919921875, -179.32754516601562, 131.32186889648438, -21.159690856933594, -62.89366912841797, 512.10009765625, 53.418846130371094, 242.441162109375, -0.8493576049804688, 177.6937255859375, 414.484619140625, 632.7733154296875, 221.49984741210938, 303.8836669921875, -126.98558044433594, 180.12095642089844, 353.0603332519531, 101.10322570800781, 597.5518798828125, -201.79513549804688, 297.6785888671875, -58.059478759765625, 224.439697265625, -48.234886169433594, 609.0032348632812, 12.675086975097656, 280.4117431640625, 71.69925689697266, 179.5318603515625, 58.41770935058594, 236.69158935546875, 352.0171813964844, 356.9156494140625, -22.71240234375, 157.68133544921875, 136.77499389648438, 24.40313720703125, 242.02821350097656, 136.2031707763672, 152.22921752929688, 235.10433959960938, 388.1385803222656, 397.1304016113281, 188.6450958251953, 294.90765380859375, 388.5480041503906, 416.2457580566406, 87.11268615722656], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000573.npy"} +{"epoch": 0.8414096916299559, "step": 574, "batch_size": 64, "mean": 228.71356201171875, "std": 215.7228546142578, "min": -181.90855407714844, "p10": -15.22149314880371, "median": 201.91565704345703, "p90": 542.1769653320314, "max": 746.4078369140625, "pos_frac": 0.875, "sample": [58.132545471191406, 238.51463317871094, -77.65298461914062, 152.26129150390625, 314.3641662597656, 57.31505584716797, 40.436279296875, 746.4078369140625, 511.68402099609375, 389.420654296875, 117.68926239013672, 570.6657104492188, 564.287109375, 378.631591796875, 163.42623901367188, 181.11854553222656, 168.28704833984375, 190.2618865966797, -164.70132446289062, 611.5042114257812, 226.5760498046875, 208.9689178466797, 90.14350891113281, 507.04443359375, 226.241943359375, 485.6040954589844, 217.7884521484375, -92.62065887451172, -14.383869171142578, 11.830501556396484, 101.47747802734375, -29.338397979736328, 260.1702575683594, 388.1202392578125, 31.03667449951172, 401.5076599121094, -102.9420394897461, 606.5372314453125, 126.04934692382812, 387.5876159667969, 492.26104736328125, 47.85362243652344, 483.39801025390625, 228.791748046875, -15.580474853515625, 109.83541870117188, 194.86239624023438, 109.12046813964844, -181.90855407714844, 75.5743408203125, 269.3813171386719, 30.990074157714844, 22.908098220825195, 514.8203125, 326.8030090332031, 238.71994018554688, 185.54922485351562, 553.9012451171875, 158.22520446777344, 327.8925476074219, 213.3890838623047, 694.2860107421875, 234.5117950439453, 72.62845611572266], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000574.npy"} +{"epoch": 0.8428781204111601, "step": 575, "batch_size": 64, "mean": 199.388427734375, "std": 248.0028839111328, "min": -413.5680236816406, "p10": -84.68728561401367, "median": 247.22763061523438, "p90": 522.951318359375, "max": 691.1770629882812, "pos_frac": 0.796875, "sample": [327.66668701171875, 283.97650146484375, 332.9739074707031, 324.19921875, 329.5633850097656, -228.92013549804688, 267.70623779296875, 268.94256591796875, 235.16763305664062, 145.6078338623047, 111.7637710571289, 128.38211059570312, -85.08981323242188, 273.23779296875, 271.67486572265625, 153.35374450683594, -16.17913818359375, 596.8245239257812, 78.64598083496094, 279.04327392578125, 275.7889099121094, 268.2657470703125, 526.3543701171875, 29.351699829101562, 139.00149536132812, 2.3272323608398438, -413.5680236816406, 214.96810913085938, -83.74805450439453, 354.0701599121094, -51.633079528808594, 602.8638305664062, 336.1439514160156, 508.4385986328125, 44.31378936767578, 12.558792114257812, 179.27279663085938, -67.59737396240234, 558.8862915039062, 475.71697998046875, 120.95367431640625, -8.811943054199219, -266.1317443847656, 414.920166015625, 691.1770629882812, -346.26318359375, 395.0184631347656, 380.10272216796875, 299.46380615234375, 550.82666015625, 74.6538314819336, 515.0108642578125, 342.5936279296875, 82.6494140625, 372.9972839355469, 259.2876281738281, 224.58016967773438, -328.10736083984375, 594.2323608398438, -7.136280059814453, -284.0282897949219, 124.9620590209961, 189.99810791015625, 377.5931091308594], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000575.npy"} +{"epoch": 0.8443465491923642, "step": 576, "batch_size": 64, "mean": 201.89810180664062, "std": 234.3874053955078, "min": -318.228515625, "p10": -88.78559265136718, "median": 178.4701156616211, "p90": 520.0535644531251, "max": 694.3473510742188, "pos_frac": 0.796875, "sample": [411.46258544921875, 657.3067626953125, -318.228515625, 120.42239379882812, -70.08988952636719, 124.34580993652344, -53.13658142089844, 570.5038452148438, 233.57669067382812, 109.71295928955078, 237.8289337158203, 492.5496826171875, 173.86996459960938, 70.05403137207031, 357.3548278808594, 267.2772216796875, 471.9590759277344, -236.2653350830078, 307.4091491699219, 263.07904052734375, 321.4066162109375, 363.7835693359375, 531.8409423828125, 609.2027587890625, 77.75535583496094, 644.74560546875, -110.54893493652344, -96.70606994628906, 254.52590942382812, 436.34979248046875, -92.69053649902344, 183.0702667236328, 156.40740966796875, 20.546524047851562, 334.10430908203125, 450.3121337890625, 694.3473510742188, 171.25042724609375, -14.127019882202148, 367.50787353515625, 44.63097381591797, 336.28851318359375, 238.93777465820312, 313.56512451171875, 0.8400497436523438, -267.5798034667969, -10.602203369140625, 260.1786804199219, -131.8363037109375, 168.65086364746094, 208.57986450195312, 650.58203125, 459.1601867675781, 343.36016845703125, -79.67405700683594, 93.77299499511719, 97.11389923095703, 128.92660522460938, 126.32743072509766, 77.716064453125, -49.450157165527344, 76.21539306640625, 72.26614379882812, 269.4308776855469], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000576.npy"} +{"epoch": 0.8458149779735683, "step": 577, "batch_size": 64, "mean": 220.69618225097656, "std": 198.37405395507812, "min": -337.20928955078125, "p10": 19.667361450195337, "median": 229.21475219726562, "p90": 448.24671630859376, "max": 731.4049072265625, "pos_frac": 0.921875, "sample": [-337.20928955078125, 699.1193237304688, 371.608642578125, 243.575439453125, 85.6099624633789, 91.53071594238281, 249.7360382080078, 232.57261657714844, 153.08197021484375, 8.527984619140625, 472.1893615722656, 162.67282104492188, 225.8568878173828, 443.37713623046875, 84.55858612060547, 383.3862609863281, 450.33367919921875, 376.213623046875, 53.44172668457031, 387.25994873046875, 186.13307189941406, 206.33041381835938, -166.51675415039062, 104.23443603515625, 332.4267578125, 260.5342102050781, 136.0951385498047, 370.47125244140625, 305.4090576171875, 323.37744140625, 290.8561096191406, 185.78733825683594, 180.15554809570312, 303.9698791503906, 395.615478515625, 52.8082275390625, 323.477783203125, 4.102207183837891, 203.3269805908203, -253.57115173339844, 61.778717041015625, 139.90859985351562, 313.35626220703125, 145.3286590576172, -58.27753448486328, 49.71638107299805, 109.48655700683594, 466.6201477050781, 307.25018310546875, 478.8713684082031, 179.66246032714844, 116.82384490966797, 514.203369140625, 70.99436950683594, 731.4049072265625, 367.3658447265625, 295.454833984375, 45.65924072265625, -233.87400817871094, 331.81756591796875, 235.2960662841797, 394.7135314941406, 280.3793029785156, 198.1481170654297], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000577.npy"} +{"epoch": 0.8472834067547724, "step": 578, "batch_size": 64, "mean": 209.7040252685547, "std": 188.189697265625, "min": -190.18270874023438, "p10": 7.301729583740236, "median": 213.330810546875, "p90": 480.01564941406264, "max": 653.2373046875, "pos_frac": 0.921875, "sample": [653.2373046875, 12.417240142822266, 253.64425659179688, 242.66717529296875, 308.1128234863281, 208.08642578125, 602.0894165039062, 211.16671752929688, 49.21890640258789, 53.7889518737793, 35.025299072265625, 584.2732543945312, 232.80101013183594, 243.88101196289062, 157.90316772460938, -190.18270874023438, 272.57159423828125, 47.475730895996094, 494.8379821777344, 399.34039306640625, 414.9622802734375, 245.36595153808594, 321.44720458984375, 5.9340362548828125, -145.48062133789062, 233.04310607910156, 393.9814453125, 70.0350570678711, 28.579181671142578, 140.14004516601562, 8.241622924804688, 91.9754638671875, 228.4662628173828, 184.48355102539062, 10.631118774414062, 358.0972595214844, 130.7236328125, 57.726524353027344, -145.28842163085938, 86.67874908447266, 137.61839294433594, 241.91690063476562, 239.765380859375, 152.79063415527344, 409.2757568359375, 215.49490356445312, 58.38140106201172, 16.601091384887695, 357.40521240234375, 132.3123321533203, 92.5005111694336, 495.00140380859375, 225.77993774414062, 315.6222839355469, 328.477783203125, 445.4302062988281, -27.557830810546875, 368.4637451171875, 523.5986328125, 516.7899169921875, 188.3620147705078, 6.898918151855469, -27.162734985351562, 415.19171142578125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000578.npy"} +{"epoch": 0.8487518355359766, "step": 579, "batch_size": 64, "mean": 170.311767578125, "std": 197.3756561279297, "min": -319.40966796875, "p10": -77.75754966735838, "median": 178.65999603271484, "p90": 402.38423156738287, "max": 668.9669189453125, "pos_frac": 0.796875, "sample": [7.230949401855469, -26.60483741760254, -110.7431640625, 302.2812805175781, -7.18109130859375, 140.74319458007812, 328.0102844238281, 187.62615966796875, 285.9949645996094, 424.11968994140625, -12.064430236816406, 51.06648635864258, -319.40966796875, 200.8219757080078, 216.80044555664062, 353.2398376464844, 269.44793701171875, 309.8540344238281, 662.1376953125, 40.18150329589844, -26.262107849121094, 84.61419677734375, 140.65292358398438, 210.24072265625, 219.0596160888672, 254.37759399414062, 276.882568359375, -88.32997131347656, 265.0181579589844, 248.0746612548828, 158.71688842773438, 94.46519470214844, 316.964111328125, 444.6628112792969, 668.9669189453125, 378.29681396484375, 237.3236541748047, -136.4746551513672, -53.088565826416016, 299.3644714355469, 376.3173522949219, 115.69137573242188, 411.0126647949219, -158.5931396484375, 537.28759765625, 46.18052673339844, 103.37571716308594, 100.50212097167969, -187.19020080566406, 206.2152557373047, 382.251220703125, 157.22171020507812, 169.69383239746094, 260.90545654296875, 244.26507568359375, -243.57003784179688, 101.77323150634766, 127.04861450195312, 41.872737884521484, -7.9766998291015625, 20.572158813476562, 418.2279052734375, 232.23660278320312, 147.55340576171875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000579.npy"} +{"epoch": 0.8502202643171806, "step": 580, "batch_size": 64, "mean": 240.44398498535156, "std": 236.74366760253906, "min": -362.1604919433594, "p10": -3.846459197998041, "median": 225.63176727294922, "p90": 571.1842712402344, "max": 983.210693359375, "pos_frac": 0.890625, "sample": [124.26460266113281, 610.9647827148438, 89.88163757324219, 277.05804443359375, -58.774715423583984, 252.05609130859375, 167.99754333496094, 570.8261108398438, 245.76162719726562, 273.6534729003906, 334.5607604980469, 95.93804931640625, 807.7736206054688, 65.5615234375, 40.061161041259766, 405.5904541015625, 212.67947387695312, 147.9399871826172, 400.09710693359375, 313.5465087890625, 225.61524963378906, 270.3841247558594, 340.6028137207031, 317.35821533203125, 266.2523498535156, 158.29776000976562, 246.93667602539062, 319.50189208984375, 420.72735595703125, 63.997039794921875, 225.64828491210938, 150.8760528564453, 164.56460571289062, 89.85444641113281, 594.167724609375, 432.7642517089844, 89.49716186523438, -362.1604919433594, 507.0333557128906, 312.8234558105469, 571.3377685546875, 151.87179565429688, -259.1800842285156, 340.1004638671875, 94.28462982177734, 201.71961975097656, -266.6208801269531, -157.41639709472656, 313.6285400390625, 983.210693359375, 251.5125274658203, 2.1524658203125, 660.0036010742188, 211.87643432617188, 230.765380859375, 142.72225952148438, -6.417427062988281, 171.83355712890625, -13.784215927124023, 65.50022888183594, 216.69622802734375, 626.8521118164062, 427.2744445800781, 216.310546875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000580.npy"} +{"epoch": 0.8516886930983847, "step": 581, "batch_size": 64, "mean": 175.04466247558594, "std": 193.9200897216797, "min": -242.10585021972656, "p10": -28.243380355834958, "median": 157.74681091308594, "p90": 392.2498962402344, "max": 708.2516479492188, "pos_frac": 0.84375, "sample": [258.9923400878906, 27.905410766601562, 68.07827758789062, 6.6541290283203125, 152.60650634765625, 128.86830139160156, -242.10585021972656, 198.3238525390625, -21.671058654785156, 44.369747161865234, 126.59259033203125, 266.9617614746094, 332.7699890136719, 152.7261962890625, 383.760986328125, -234.79379272460938, 602.34716796875, 202.8948516845703, 364.40057373046875, 598.408447265625, -70.33311462402344, 397.9586486816406, 75.23329162597656, 254.0503387451172, 158.2139434814453, 347.5561828613281, 198.14161682128906, 96.62610626220703, 191.664794921875, 332.1979064941406, -147.0103302001953, 708.2516479492188, 9.14284896850586, 369.79510498046875, 146.14190673828125, 225.6156005859375, 395.88800048828125, 157.0157012939453, 128.53341674804688, 270.8268737792969, 162.35076904296875, 36.713226318359375, 50.8013916015625, 221.02976989746094, 266.72625732421875, 104.62736511230469, 380.7540283203125, 340.30670166015625, 536.0449829101562, 157.27967834472656, -192.41162109375, -23.580360412597656, 26.560848236083984, -106.53555297851562, 328.527587890625, 96.43760681152344, -25.221065521240234, 50.78373718261719, 180.051513671875, 25.92003631591797, 241.34896850585938, 422.3615417480469, -29.538658142089844, 287.9189758300781], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000581.npy"} +{"epoch": 0.8531571218795888, "step": 582, "batch_size": 64, "mean": 226.93896484375, "std": 230.09548950195312, "min": -424.28131103515625, "p10": -20.954928588867187, "median": 211.28174591064453, "p90": 539.9937072753906, "max": 673.1987915039062, "pos_frac": 0.828125, "sample": [-424.28131103515625, 152.33038330078125, 673.1987915039062, 428.14385986328125, 429.1812438964844, -23.84832763671875, 258.0645446777344, 325.080078125, 280.4598388671875, 108.95974731445312, 214.1455535888672, 208.41793823242188, 24.454910278320312, 478.5120849609375, 180.0419921875, 303.9603576660156, 80.21455383300781, -20.242111206054688, 231.73976135253906, 567.168701171875, 284.91644287109375, 117.39601135253906, 398.9735107421875, -84.79330444335938, 174.6807861328125, 640.5838623046875, 86.72974395751953, 268.0506896972656, 523.749267578125, 464.3987731933594, 201.12448120117188, 279.6459655761719, 509.4539794921875, 50.007225036621094, -111.33038330078125, 7.882286071777344, -2.5117416381835938, -47.70671081542969, 199.97286987304688, 357.9739990234375, 40.59112548828125, -12.590866088867188, 595.3108520507812, 613.0579833984375, 174.54200744628906, 543.498291015625, 7.305015563964844, 130.2725372314453, 39.35374450683594, 145.24549865722656, -1.0539360046386719, 531.5328369140625, 550.72314453125, 294.776611328125, -21.260421752929688, -291.8085021972656, 295.9139709472656, 55.837074279785156, 398.05474853515625, 408.4471740722656, 531.8163452148438, 144.0994110107422, 334.50811767578125, 221.02088928222656], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000582.npy"} +{"epoch": 0.8546255506607929, "step": 583, "batch_size": 64, "mean": 165.63748168945312, "std": 212.22793579101562, "min": -248.15357971191406, "p10": -123.70702743530272, "median": 144.8111343383789, "p90": 443.19486083984384, "max": 746.5250854492188, "pos_frac": 0.828125, "sample": [171.09695434570312, 116.70635986328125, 170.82650756835938, 154.55259704589844, 9.407045364379883, 141.4368438720703, 426.7957458496094, 233.0437774658203, 746.5250854492188, 105.14397430419922, 71.14178466796875, 45.74690246582031, -140.9965362548828, 176.46556091308594, 133.744384765625, -127.3484115600586, 362.9073486328125, 272.9460754394531, 178.03689575195312, 257.00384521484375, 268.1392822265625, 54.19501495361328, 137.807373046875, -248.15357971191406, 57.72887420654297, -169.18838500976562, 276.16717529296875, 78.8759765625, -236.52362060546875, 7.341251373291016, 148.88563537597656, 64.28445434570312, 236.34988403320312, 374.8746337890625, 323.0087585449219, 647.4482421875, 234.19866943359375, 7.4236907958984375, 450.2230529785156, 80.22882843017578, -8.850677490234375, 137.35340881347656, 189.61346435546875, -49.723960876464844, -24.05279541015625, -115.21046447753906, 678.5795288085938, 245.2376251220703, -142.63162231445312, 249.6335906982422, 477.88360595703125, 490.58612060546875, 26.79169464111328, 311.7953186035156, 310.30047607421875, -162.6912841796875, 101.58992004394531, 297.8688049316406, 148.1854248046875, 167.21353149414062, 662.98486328125, 119.06948852539062, 86.37970733642578, 104.39529418945312], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000583.npy"} +{"epoch": 0.856093979441997, "step": 584, "batch_size": 64, "mean": 265.22979736328125, "std": 206.04635620117188, "min": -104.50790405273438, "p10": 31.708298873901384, "median": 244.1134490966797, "p90": 585.3760314941406, "max": 726.5732421875, "pos_frac": 0.9375, "sample": [237.89862060546875, 171.15296936035156, 10.797191619873047, 279.2886657714844, 85.560791015625, -83.80946350097656, 433.6907958984375, 363.16693115234375, 320.38177490234375, 79.56233215332031, -10.536102294921875, 75.24443817138672, 54.261077880859375, 268.3987731933594, 313.1209716796875, 530.9207153320312, 246.62213134765625, 631.2939453125, 487.5432434082031, 97.62996673583984, 252.42300415039062, 160.2972412109375, 122.53218078613281, 199.1206512451172, 169.03701782226562, -104.50790405273438, 576.2990112304688, 164.23110961914062, 94.43362426757812, 242.52056884765625, 404.6940002441406, 256.7590637207031, 163.17327880859375, 8.202972412109375, 342.88232421875, 314.75213623046875, 78.08592224121094, 292.4724426269531, 25.185943603515625, 245.70632934570312, 342.6041259765625, 46.927127838134766, 130.19476318359375, 616.6372680664062, 199.67823791503906, 196.352294921875, 475.6379089355469, 715.151611328125, -99.91519165039062, 657.7220458984375, 585.89892578125, 73.21150207519531, 726.5732421875, 584.1559448242188, 405.34576416015625, 439.7679138183594, 218.2854766845703, 382.40521240234375, 93.53067779541016, 289.1512756347656, 362.9259033203125, 237.6346435546875, 613.8806762695312, 80.45811462402344], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000584.npy"} +{"epoch": 0.8575624082232012, "step": 585, "batch_size": 64, "mean": 207.1536865234375, "std": 227.9725341796875, "min": -301.8384094238281, "p10": -59.46738128662109, "median": 203.4548568725586, "p90": 507.9173797607422, "max": 906.69775390625, "pos_frac": 0.8125, "sample": [222.8062744140625, -13.423080444335938, 295.90093994140625, 230.5608367919922, -62.2813835144043, 906.69775390625, 343.7340393066406, -73.62091064453125, 174.17642211914062, 99.02694702148438, 599.4469604492188, 69.93885803222656, 114.12680053710938, 272.79736328125, 117.20638275146484, 344.4244384765625, 359.08697509765625, 419.96075439453125, 785.05810546875, 283.50042724609375, 174.0448760986328, 539.9151611328125, 23.46398162841797, 261.6197204589844, 339.4535827636719, 3.6984710693359375, 223.6246795654297, 293.7051086425781, 411.4783935546875, 87.74795532226562, 60.38697814941406, 215.8907470703125, 89.45659637451172, -101.14227294921875, 500.9735412597656, 125.37743377685547, 512.6806640625, -136.07052612304688, 68.89949035644531, 354.5568542480469, 135.83453369140625, 44.154720306396484, 444.2934265136719, -30.013893127441406, 324.18115234375, 191.0189666748047, 253.52066040039062, -29.44854736328125, 346.56640625, 70.3799057006836, -57.169342041015625, -301.8384094238281, 18.678539276123047, -236.49819946289062, -60.45225524902344, 574.78857421875, 389.641357421875, 181.6737518310547, 370.1095275878906, -16.096847534179688, 220.9795379638672, 226.0294189453125, 147.7229766845703, 510.893310546875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000585.npy"} +{"epoch": 0.8590308370044053, "step": 586, "batch_size": 64, "mean": 185.64146423339844, "std": 225.00393676757812, "min": -331.0683898925781, "p10": -62.11248741149901, "median": 162.7815399169922, "p90": 452.72645568847656, "max": 728.3416748046875, "pos_frac": 0.796875, "sample": [629.0169677734375, 169.07138061523438, -29.078914642333984, 11.662910461425781, 345.70733642578125, 206.13568115234375, -43.102821350097656, 218.90162658691406, 35.398277282714844, 446.8611145019531, 52.63240432739258, 195.4718017578125, 229.93153381347656, 499.68707275390625, -81.52416229248047, 172.9957733154297, 407.0718078613281, -331.0683898925781, 88.93115997314453, 190.4832000732422, 141.4693145751953, 7.623132705688477, 111.69771575927734, 455.24017333984375, 417.00457763671875, -143.40892028808594, 427.912353515625, -147.87554931640625, 163.51495361328125, 24.962921142578125, 361.120361328125, -14.711845397949219, 365.37847900390625, -47.452796936035156, 66.95216369628906, 86.70230102539062, 356.28887939453125, 12.258384704589844, 142.50262451171875, 72.22345733642578, 575.1614990234375, -51.4819450378418, 491.3817138671875, 370.4949951171875, -268.4900817871094, -208.80149841308594, 224.10568237304688, 409.56805419921875, 334.2060241699219, 106.00939178466797, -47.26191711425781, 30.511253356933594, 333.45849609375, 728.3416748046875, 441.9161071777344, 161.72019958496094, 411.29437255859375, 544.38671875, 73.2373046875, -66.6684341430664, 274.38714599609375, 415.408447265625, 162.04812622070312, 161.53201293945312], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000586.npy"} +{"epoch": 0.8604992657856094, "step": 587, "batch_size": 64, "mean": 204.01170349121094, "std": 213.73391723632812, "min": -349.0215148925781, "p10": -7.588995933532713, "median": 194.9998016357422, "p90": 473.6946166992187, "max": 729.942626953125, "pos_frac": 0.859375, "sample": [286.4754333496094, 305.7410888671875, 259.03485107421875, 52.354888916015625, 393.16986083984375, -5.810216903686523, 255.49066162109375, -8.351329803466797, -0.606353759765625, 299.48370361328125, 16.548810958862305, 264.2508239746094, 666.1258544921875, 121.2784194946289, 251.4161376953125, 301.75390625, 6.0740814208984375, 126.06501770019531, 170.54330444335938, 296.0978698730469, -99.41340637207031, 208.69210815429688, 466.65899658203125, 15.017827987670898, 181.4741973876953, 630.8428955078125, -79.58302307128906, 462.2259521484375, 183.74301147460938, 36.14164733886719, 229.0619354248047, 161.14169311523438, 42.76908874511719, 20.574005126953125, 352.24664306640625, 209.7763214111328, 134.74929809570312, -349.0215148925781, 118.82466125488281, 59.72536849975586, -84.3070297241211, 532.4547119140625, 13.42481803894043, 474.2311706542969, 84.58897399902344, 656.3402709960938, 136.33412170410156, -157.1921844482422, 206.256591796875, 239.49746704101562, 389.8311462402344, 729.942626953125, 315.98590087890625, 375.34130859375, 302.3248291015625, 329.2957458496094, 511.3467102050781, -115.77013397216797, 171.77386474609375, 16.586620330810547, 65.41827392578125, 35.622039794921875, 312.1940002441406, 472.4426574707031], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000587.npy"} +{"epoch": 0.8619676945668135, "step": 588, "batch_size": 64, "mean": 193.74520874023438, "std": 207.644775390625, "min": -209.73870849609375, "p10": -24.070416641235347, "median": 140.3182601928711, "p90": 470.9127838134766, "max": 751.7249755859375, "pos_frac": 0.84375, "sample": [233.3413543701172, 130.16212463378906, -72.1985855102539, 46.24172592163086, 305.4654541015625, 342.9049377441406, 335.4679260253906, 38.061981201171875, 329.78973388671875, 474.7217712402344, -25.723388671875, 46.14313507080078, 16.714935302734375, 122.56409454345703, -67.67074584960938, -19.46181869506836, 112.75253295898438, 42.683746337890625, 104.76557159423828, 91.11996459960938, 462.025146484375, 42.824676513671875, -151.74803161621094, 365.4225158691406, 138.36170959472656, 134.36936950683594, 655.2444458007812, 146.0498046875, -151.2807159423828, 7.825462341308594, 170.02688598632812, 87.23674011230469, 115.64265441894531, 546.2166748046875, 367.5950927734375, 137.33718872070312, 197.87684631347656, -16.567710876464844, 328.529052734375, 250.45303344726562, 549.444091796875, 319.3030090332031, 113.72039794921875, 351.76153564453125, 194.3732452392578, 44.197914123535156, 90.88676452636719, 576.0679931640625, -209.73870849609375, 368.41192626953125, -20.213481903076172, 751.7249755859375, 688.7512817382812, 311.98895263671875, 404.06927490234375, 205.38296508789062, 207.2388916015625, 84.9481201171875, -92.15111541748047, 179.6767578125, 240.51278686523438, 142.27481079101562, 391.60577392578125, 84.16706848144531], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000588.npy"} +{"epoch": 0.8634361233480177, "step": 589, "batch_size": 64, "mean": 259.88519287109375, "std": 216.98187255859375, "min": -206.84754943847656, "p10": 16.163293266296392, "median": 226.66895294189453, "p90": 546.5589233398438, "max": 747.85302734375, "pos_frac": 0.921875, "sample": [500.7704162597656, 336.5050048828125, 269.38287353515625, 103.04444885253906, 304.549072265625, 40.2545280456543, 21.859024047851562, 500.4849548339844, 219.61083984375, -2.3144454956054688, 102.64710998535156, 155.57943725585938, 274.7359924316406, 225.57142639160156, 193.8006591796875, 691.975830078125, 710.223388671875, -101.71907806396484, 181.8289031982422, 114.82797241210938, 525.4463500976562, 11.61590576171875, 73.07382202148438, 660.9581298828125, 94.53399658203125, 225.362548828125, 284.7107849121094, 227.7664794921875, 430.55657958984375, 80.4237060546875, 574.686767578125, 454.43634033203125, 198.48886108398438, 441.3990173339844, 747.85302734375, 75.79409790039062, 14.08917236328125, 57.746376037597656, 500.4474792480469, 34.26240921020508, 386.6233825683594, 54.47998046875, 279.3848571777344, 300.9206237792969, 405.3825378417969, 175.19818115234375, 552.4307861328125, 84.27897644042969, 299.8562927246094, 623.903564453125, 532.85791015625, 207.82565307617188, 315.23699951171875, -38.95709228515625, 188.7644500732422, -98.84104919433594, 440.4276123046875, 495.2042541503906, 21.00290870666504, 161.81307983398438, 264.3313903808594, 268.08056640625, -206.84754943847656, 362.02313232421875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000589.npy"} +{"epoch": 0.8649045521292217, "step": 590, "batch_size": 64, "mean": 225.7188720703125, "std": 199.24403381347656, "min": -231.9090576171875, "p10": -33.8900032043457, "median": 210.0173568725586, "p90": 478.25585327148445, "max": 696.8587646484375, "pos_frac": 0.84375, "sample": [172.91810607910156, 664.9293823242188, 487.14019775390625, 99.66924285888672, 198.50833129882812, -32.695770263671875, 306.9424743652344, 88.034912109375, 408.31719970703125, -122.80386352539062, 148.1613006591797, -57.550838470458984, 261.4365234375, 484.85455322265625, 128.7908935546875, 139.77784729003906, -77.37796783447266, -5.439140319824219, 377.30633544921875, 103.59408569335938, 189.9735107421875, -34.401817321777344, 50.600555419921875, 255.24668884277344, -66.7420654296875, 108.90387725830078, 462.85888671875, 120.13530731201172, 696.8587646484375, -15.110300064086914, 88.5477294921875, 303.39349365234375, 172.49362182617188, 623.895751953125, 259.7715148925781, 349.3472595214844, 202.35316467285156, 297.71844482421875, 285.2787170410156, 288.1724548339844, 211.772705078125, 164.07119750976562, -132.37582397460938, -231.9090576171875, 240.32237243652344, 43.35078430175781, 368.21868896484375, 306.11566162109375, 208.2620086669922, 269.845458984375, 341.7642517089844, 390.69775390625, 133.65966796875, 161.97109985351562, 177.07797241210938, 523.6695556640625, 440.7247314453125, 669.176025390625, 338.12725830078125, 306.9355773925781, 406.9700012207031, 292.0985412597656, 48.77595138549805, 352.8768310546875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000590.npy"} +{"epoch": 0.8663729809104258, "step": 591, "batch_size": 64, "mean": 224.37022399902344, "std": 190.1165313720703, "min": -287.195556640625, "p10": 4.000942420959476, "median": 243.23467254638672, "p90": 419.38966064453126, "max": 732.4713134765625, "pos_frac": 0.90625, "sample": [-0.29320526123046875, 732.4713134765625, 411.4574279785156, 342.9906005859375, 476.4389343261719, 9.416824340820312, 242.6864471435547, -190.79824829101562, 365.98699951171875, 588.9775390625, 228.8746337890625, 204.73898315429688, 410.43365478515625, 251.6466064453125, 183.03675842285156, 616.096923828125, 419.0166320800781, 282.2677917480469, 106.70843505859375, 243.78289794921875, 118.57872009277344, 442.603759765625, 2.5368785858154297, 84.08226776123047, 199.84869384765625, 115.9273452758789, -86.84349060058594, 335.9449462890625, 177.7333984375, 234.55816650390625, -169.22837829589844, 145.18731689453125, 258.7408447265625, -54.82600402832031, 338.1978759765625, 78.05107116699219, 48.43876647949219, 120.26657104492188, 348.8887939453125, 115.21985626220703, -287.195556640625, 387.08770751953125, 306.34521484375, 314.2486572265625, 45.460060119628906, 419.5495300292969, 329.3803405761719, 35.948482513427734, 302.8040466308594, 272.71112060546875, 301.932373046875, 260.0828857421875, 334.9383239746094, 117.01141357421875, 335.98004150390625, 259.8021545410156, 589.327880859375, 252.59283447265625, 7.417091369628906, 321.71380615234375, 133.6289520263672, 162.39007568359375, 195.11622619628906, 181.57522583007812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000591.npy"} +{"epoch": 0.8678414096916299, "step": 592, "batch_size": 64, "mean": 205.769287109375, "std": 255.82884216308594, "min": -484.8076171875, "p10": -107.32505111694334, "median": 236.88382720947266, "p90": 478.63973693847663, "max": 735.658935546875, "pos_frac": 0.828125, "sample": [128.66175842285156, 445.9571533203125, -56.46527099609375, 299.12689208984375, -229.39662170410156, 59.07624816894531, 298.57196044921875, 367.246826171875, 670.2863159179688, 318.7159729003906, -88.70314025878906, 657.4447631835938, 74.9380874633789, 462.1199951171875, -406.9331359863281, 94.87289428710938, 161.90435791015625, 221.71749877929688, 81.55548095703125, 735.658935546875, 111.31269073486328, 332.3960876464844, -74.54402923583984, 467.7706604003906, -13.151105880737305, 72.0281982421875, 172.11273193359375, 451.39190673828125, 538.0309448242188, 275.5980529785156, 71.15707397460938, 12.306808471679688, 100.50006103515625, 351.8642883300781, 514.0125732421875, 50.24675750732422, 27.95556640625, 256.7803649902344, -115.30587005615234, 396.32208251953125, 252.05015563964844, 212.82791137695312, 410.95281982421875, 378.55230712890625, 153.45556640625, -137.66448974609375, 297.1826477050781, -484.8076171875, 394.5281677246094, 254.98121643066406, 431.849853515625, 112.51869201660156, 535.6765747070312, 291.3429260253906, -194.533203125, 422.7757263183594, 412.52838134765625, 201.02603149414062, 483.29791259765625, 134.23294067382812, -475.97625732421875, 288.32208251953125, 340.41607666015625, 158.554443359375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000592.npy"} +{"epoch": 0.869309838472834, "step": 593, "batch_size": 64, "mean": 157.88589477539062, "std": 213.43161010742188, "min": -336.8236083984375, "p10": -78.00479888916016, "median": 164.5595245361328, "p90": 389.59934387207034, "max": 759.2660522460938, "pos_frac": 0.765625, "sample": [-8.776641845703125, 132.40695190429688, -63.035499572753906, -7.01922607421875, 16.861000061035156, -159.88848876953125, 257.78216552734375, -24.472667694091797, 370.6529541015625, 474.6742248535156, 294.767333984375, 421.36981201171875, 416.1650390625, 341.0605773925781, 46.260337829589844, 3.9899463653564453, 387.0408020019531, 390.69586181640625, 316.8177185058594, 136.95863342285156, 23.575878143310547, 68.12429809570312, 73.86161041259766, 57.15497589111328, 315.4431457519531, 759.2660522460938, 219.04132080078125, 174.27703857421875, 279.02374267578125, -193.0798797607422, 0.6730098724365234, 381.2150573730469, -75.82263946533203, 25.4637451171875, 290.9302673339844, 372.5643310546875, -27.031524658203125, 530.1651611328125, 32.02483367919922, -29.398921966552734, -336.8236083984375, 367.903076171875, -288.0967712402344, -95.83222961425781, 157.805908203125, -6.5233154296875, 32.773582458496094, 204.12408447265625, 410.1141052246094, 377.0029602050781, 215.17356872558594, 209.46340942382812, 374.4366455078125, 364.91748046875, 218.3609619140625, 327.35107421875, 171.31314086914062, 14.320741653442383, -225.32028198242188, 298.53314208984375, 37.18798828125, -78.94001007080078, 19.347652435302734, 314.3214111328125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000593.npy"} +{"epoch": 0.8707782672540382, "step": 594, "batch_size": 64, "mean": 216.91842651367188, "std": 242.40472412109375, "min": -319.7406005859375, "p10": -77.16597633361813, "median": 206.81334686279297, "p90": 574.2287963867188, "max": 817.2074584960938, "pos_frac": 0.796875, "sample": [337.3593444824219, 90.67526245117188, -6.431047439575195, 5.469087600708008, -28.637229919433594, -34.50524139404297, 503.66510009765625, 261.576416015625, 250.60528564453125, 153.72784423828125, 137.7333526611328, 283.0268859863281, 698.1422119140625, 484.00201416015625, 115.8675537109375, 509.9976501464844, 457.5846862792969, 227.2734832763672, 248.1997528076172, -41.56350326538086, -92.42417907714844, 214.66957092285156, -114.26512908935547, 583.8650512695312, -37.2724723815918, 324.1055908203125, -207.69943237304688, 326.1312561035156, 414.5370788574219, 125.41299438476562, 112.88282775878906, 292.0652160644531, 173.41934204101562, 340.492919921875, 290.33441162109375, 355.9345397949219, 221.5178985595703, 6.537166595458984, 198.95712280273438, -319.7406005859375, 67.77157592773438, 614.4656982421875, 226.1172637939453, 169.39642333984375, 179.47015380859375, 691.595703125, 551.7442016601562, 248.71026611328125, 154.31390380859375, 316.7613830566406, -93.0404281616211, 17.25543212890625, 49.557865142822266, -165.43280029296875, 408.31524658203125, -19.79513931274414, 419.6402587890625, 590.7423095703125, 647.851318359375, 135.00047302246094, 817.2074584960938, -126.3609390258789, 51.21856689453125, 67.04303741455078], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000594.npy"} +{"epoch": 0.8722466960352423, "step": 595, "batch_size": 64, "mean": 193.73770141601562, "std": 206.18321228027344, "min": -155.0855712890625, "p10": -41.75386505126952, "median": 170.7867889404297, "p90": 444.00838317871103, "max": 906.998291015625, "pos_frac": 0.84375, "sample": [-139.3776397705078, 349.1163330078125, 171.25595092773438, 123.45319366455078, -3.3846511840820312, 621.48095703125, 107.1680908203125, 278.9719543457031, -59.446224212646484, 184.130126953125, 332.8001708984375, 174.1707000732422, 109.57156372070312, 425.8049011230469, -98.92237854003906, 309.95306396484375, 170.317626953125, 772.3074951171875, 220.0437774658203, 67.9788818359375, -67.63510131835938, -155.0855712890625, 178.91412353515625, 289.8413391113281, -47.07227325439453, -0.9195213317871094, 307.10467529296875, 64.20557403564453, 259.1490478515625, 166.394287109375, 69.78731536865234, 238.55581665039062, 88.47225952148438, 233.0135498046875, 415.8063049316406, 152.1891632080078, 121.06454467773438, 168.71739196777344, 463.82470703125, 312.48077392578125, 4.8210296630859375, 5.918247222900391, 269.8976745605469, -138.89022827148438, 211.58749389648438, 168.12168884277344, 208.69883728027344, 7.9646148681640625, 274.00592041015625, 384.5820007324219, -29.34424591064453, 451.80987548828125, 10.52215576171875, 113.41004943847656, 152.76507568359375, 203.09609985351562, 264.7972412109375, 45.367828369140625, 19.135986328125, 80.01612854003906, 553.1837158203125, 394.2660827636719, 460.27960205078125, 906.998291015625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000595.npy"} +{"epoch": 0.8737151248164464, "step": 596, "batch_size": 64, "mean": 174.1080322265625, "std": 218.0355224609375, "min": -492.026611328125, "p10": -65.81622734069823, "median": 179.30972290039062, "p90": 462.8183868408203, "max": 741.7048950195312, "pos_frac": 0.796875, "sample": [-123.2099609375, 2.99310302734375, 192.29879760742188, 22.349105834960938, 249.3113555908203, 421.8265380859375, 7.547325134277344, 222.02252197265625, 304.3143005371094, 99.09548950195312, 536.6925048828125, 204.30123901367188, -70.05596923828125, 346.3841552734375, 506.8474426269531, 572.134033203125, 80.69776916503906, 94.77864074707031, 460.1896057128906, 302.4393310546875, -61.47272872924805, 463.94500732421875, 192.39830017089844, 187.0637664794922, -67.67772674560547, 515.990234375, -47.102439880371094, 222.51669311523438, 238.04722595214844, 44.442779541015625, -27.162399291992188, 138.71026611328125, -3.1845932006835938, 452.8289489746094, 63.77783203125, 399.1469421386719, 520.845458984375, 298.13336181640625, 248.51727294921875, -43.83965301513672, 179.6060791015625, 45.276329040527344, 113.3472900390625, 121.9308853149414, 145.7550048828125, 349.59381103515625, 359.4246826171875, 111.85942077636719, 64.3307113647461, 203.93370056152344, 261.3103332519531, 54.50672149658203, -492.026611328125, 173.37796020507812, 179.01336669921875, -240.10977172851562, 26.358413696289062, -38.7169303894043, 281.30462646484375, 340.58538818359375, 741.7048950195312, -70.71279907226562, 271.5040283203125, -209.12623596191406], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000596.npy"} +{"epoch": 0.8751835535976505, "step": 597, "batch_size": 64, "mean": 220.94094848632812, "std": 233.67666625976562, "min": -338.7794189453125, "p10": -21.619601440429676, "median": 184.88504791259766, "p90": 501.09338684082036, "max": 982.6571044921875, "pos_frac": 0.859375, "sample": [183.91738891601562, 174.48968505859375, 252.449951171875, 206.8774871826172, 134.70501708984375, 665.8358154296875, 13.118274688720703, 480.1827697753906, 248.21646118164062, 166.69224548339844, -0.05028533935546875, 91.90911865234375, -9.066993713378906, 71.79873657226562, 293.8399658203125, 142.30172729492188, 286.49847412109375, 141.50303649902344, 172.93301391601562, 615.4296875, 117.69775390625, 296.05401611328125, 71.38922119140625, 484.0558166503906, 120.88169860839844, 193.04696655273438, 11.160701751708984, 22.453384399414062, 159.0236053466797, -302.86328125, -338.7794189453125, -34.1890869140625, 298.40130615234375, 95.28901672363281, 308.6814270019531, 774.1411743164062, 342.427734375, -26.999290466308594, -67.15824890136719, 213.2689208984375, 366.1057434082031, 376.46087646484375, 114.58170318603516, 287.83074951171875, 93.54546356201172, 285.1376037597656, 277.7728576660156, 414.60107421875, 121.37922668457031, 547.7562866210938, 178.56439208984375, -140.58990478515625, -179.85476684570312, 982.6571044921875, 659.6328125, 185.8527069091797, 393.428466796875, 160.3397216796875, 392.6593933105469, 508.39520263671875, 226.03294372558594, 338.02703857421875, 115.5553970336914, 362.782958984375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000597.npy"} +{"epoch": 0.8766519823788547, "step": 598, "batch_size": 64, "mean": 191.70065307617188, "std": 229.20547485351562, "min": -500.3950500488281, "p10": -79.3244129180908, "median": 176.971923828125, "p90": 475.3672332763672, "max": 597.372314453125, "pos_frac": 0.84375, "sample": [182.66078186035156, 447.4585876464844, 63.4024543762207, 500.6312255859375, -194.73373413085938, 239.24209594726562, 371.225830078125, 396.96856689453125, 111.66595458984375, 440.817626953125, 25.996124267578125, -244.12069702148438, -500.3950500488281, 92.94300842285156, 391.05023193359375, 0.6466579437255859, 553.0655517578125, 253.2823486328125, 597.372314453125, 312.6406555175781, 103.17080688476562, 289.88421630859375, 477.65673828125, 117.6416244506836, 124.73451232910156, 289.8408508300781, 343.45428466796875, 89.979736328125, -382.65509033203125, 33.95368194580078, 148.69412231445312, 205.37680053710938, 161.7886199951172, -40.983802795410156, -97.20399475097656, 343.1829528808594, -19.938030242919922, 481.4952697753906, 88.16954040527344, 226.852783203125, 165.22360229492188, 136.89859008789062, -308.6273498535156, 363.9593505859375, -88.95335388183594, 534.654541015625, 171.28306579589844, 217.5603485107422, 168.60633850097656, 115.8360366821289, 321.2182312011719, 376.24334716796875, 531.9622802734375, 245.8300323486328, 350.09600830078125, 104.9653091430664, 353.71661376953125, 426.5120849609375, 377.1339111328125, 154.07650756835938, 55.09996032714844, 55.46221160888672, 470.0250549316406, -56.85688400268555], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000598.npy"} +{"epoch": 0.8781204111600588, "step": 599, "batch_size": 64, "mean": 191.21209716796875, "std": 168.37648010253906, "min": -222.0301513671875, "p10": -15.61826782226561, "median": 178.03377532958984, "p90": 421.7710510253907, "max": 558.7544555664062, "pos_frac": 0.875, "sample": [206.64199829101562, 151.37550354003906, 167.3382110595703, 98.62470245361328, 125.04584503173828, 395.760009765625, 366.109130859375, 112.14093017578125, 246.40386962890625, 292.2486877441406, 463.17523193359375, 80.17884063720703, 9.462282180786133, 157.8428955078125, 9.840307235717773, 177.33592224121094, 348.29840087890625, 210.94894409179688, 187.04434204101562, 439.767822265625, 77.80242919921875, 356.8439025878906, 291.8473815917969, 143.51840209960938, 155.13392639160156, -85.65274810791016, 170.4857177734375, 237.54310607910156, 146.3667755126953, -222.0301513671875, 443.24371337890625, -22.173187255859375, 379.43170166015625, 364.9517822265625, 447.34649658203125, 432.91864013671875, 373.35687255859375, -0.323455810546875, 558.7544555664062, 24.231300354003906, 228.26019287109375, 44.301116943359375, 395.50146484375, 160.49107360839844, 238.9995574951172, 127.13507080078125, 354.3257751464844, -88.1925277709961, 188.95399475097656, 194.86196899414062, 312.0023193359375, 44.766815185546875, -50.90862274169922, 8.650627136230469, -121.38046264648438, 180.34222412109375, 316.1549987792969, 178.73162841796875, 197.8526153564453, 517.0157470703125, 82.02352905273438, -109.89984893798828, 166.3647003173828, 152.03970336914062], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000599.npy"} +{"epoch": 0.8795888399412628, "step": 600, "batch_size": 64, "mean": 200.9642333984375, "std": 209.95469665527344, "min": -264.81121826171875, "p10": -25.680631065368633, "median": 187.8054428100586, "p90": 488.32214355468756, "max": 679.581787109375, "pos_frac": 0.875, "sample": [51.86615753173828, 134.0846405029297, 337.964599609375, -182.05752563476562, 422.7728271484375, 266.30828857421875, 86.13287353515625, 226.61167907714844, 7.253183364868164, 419.76251220703125, 141.78143310546875, 215.41534423828125, 436.50482177734375, 64.32386016845703, 206.00253295898438, 443.7244567871094, -33.99054718017578, 545.20263671875, 59.134666442871094, 189.81243896484375, 396.6393127441406, 496.4413146972656, 0.7779998779296875, 211.76693725585938, 86.25746154785156, 51.11769104003906, 310.5928955078125, 679.581787109375, 619.7191162109375, 226.019775390625, 123.54129028320312, 177.45632934570312, 421.6044921875, 230.8607635498047, 424.24530029296875, 264.4963073730469, -121.22053527832031, 163.3179168701172, -6.290826797485352, 380.1529541015625, 503.0849914550781, 141.01234436035156, 40.01362609863281, 126.70818328857422, 626.5941772460938, 94.84532928466797, 469.3774108886719, 165.7022247314453, -219.00144958496094, 238.34048461914062, 192.5136260986328, 282.7049865722656, 324.1881103515625, 103.51668548583984, 206.0673065185547, 509.4216003417969, -170.82864379882812, 86.42373657226562, 45.70067596435547, -125.87079620361328, -264.81121826171875, 97.28587341308594, 27.23418426513672, 185.79844665527344], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000600.npy"} +{"epoch": 0.8810572687224669, "step": 601, "batch_size": 64, "mean": 206.70999145507812, "std": 221.90296936035156, "min": -283.50244140625, "p10": -25.57880744934081, "median": 206.50877380371094, "p90": 470.96009826660173, "max": 850.0047607421875, "pos_frac": 0.8125, "sample": [130.28643798828125, 307.13201904296875, 850.0047607421875, 427.5804138183594, -29.194873809814453, 493.0544738769531, 221.68533325195312, 226.80694580078125, -11.857261657714844, 372.08251953125, -38.444664001464844, -124.00108337402344, 489.5513916015625, 295.92657470703125, 186.4058837890625, 272.6669006347656, 155.9471435546875, 275.69537353515625, 61.407814025878906, 727.6036987304688, 252.1281280517578, 374.22259521484375, -13.9337158203125, 80.58497619628906, 173.34356689453125, 226.47718811035156, 10.600385665893555, 169.92401123046875, 540.795166015625, 191.33221435546875, -1.0568618774414062, 174.41392517089844, -172.82040405273438, 130.29315185546875, -15.734710693359375, 414.4703063964844, 69.98468780517578, 290.78265380859375, 54.055633544921875, 303.72265625, 82.54590606689453, 323.1934509277344, 92.70549774169922, -270.3137512207031, 140.54344177246094, -283.50244140625, 19.740264892578125, 639.2828369140625, 380.4698486328125, -17.141319274902344, 285.7795104980469, 277.7181091308594, 75.80290222167969, 304.97784423828125, -143.1930694580078, 624.3316040039062, 12.723714828491211, 349.3819274902344, 385.5821838378906, 282.45477294921875, 390.3625793457031, 132.30215454101562, 283.27532958984375, 316.49114990234375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000601.npy"} +{"epoch": 0.882525697503671, "step": 602, "batch_size": 64, "mean": 248.43075561523438, "std": 220.58334350585938, "min": -252.203125, "p10": -6.942386627197249, "median": 205.16057586669922, "p90": 550.2981750488283, "max": 828.4927978515625, "pos_frac": 0.890625, "sample": [143.19625854492188, 81.70696258544922, -25.310626983642578, 134.79965209960938, 197.21762084960938, 146.57659912109375, 504.5538330078125, 345.19049072265625, 105.22868347167969, 102.08470153808594, 265.94903564453125, 277.94561767578125, 155.3697967529297, 283.60693359375, 645.2698974609375, 158.30841064453125, 569.9028930664062, 228.8719482421875, 305.1427917480469, 375.8503723144531, 28.09967041015625, 171.81106567382812, -19.01165771484375, -124.07028198242188, 83.23152160644531, 247.0748291015625, 446.12744140625, 212.3009490966797, -252.203125, 20.184186935424805, 9.663414001464844, 361.3409118652344, 360.490966796875, 94.88270568847656, 432.2911376953125, 392.464599609375, 189.62185668945312, 289.710205078125, 168.47140502929688, -105.40592193603516, 622.50537109375, 346.4651184082031, 399.0804443359375, 152.229248046875, 449.79376220703125, 404.460693359375, 828.4927978515625, -14.059158325195312, 456.1627197265625, -113.44618225097656, 584.73974609375, 455.3703308105469, 180.068603515625, 198.02020263671875, 700.7349243164062, 107.38690185546875, 52.20331573486328, 298.2598876953125, 117.51571655273438, 134.115234375, 477.8583984375, 755.3399047851562, 223.64651489257812, 74.08610534667969], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000602.npy"} +{"epoch": 0.8839941262848752, "step": 603, "batch_size": 64, "mean": 210.02691650390625, "std": 223.47769165039062, "min": -204.7823944091797, "p10": -39.01070098876953, "median": 176.07644653320312, "p90": 529.253973388672, "max": 876.066650390625, "pos_frac": 0.796875, "sample": [-43.7440185546875, -48.66346740722656, 285.58038330078125, 23.49238395690918, 601.3864135742188, 9.84368896484375, 376.396484375, 154.56053161621094, 191.0591278076172, 387.945068359375, 127.6915283203125, 324.49468994140625, -204.7823944091797, 355.54058837890625, 42.36188888549805, 165.10617065429688, 30.408599853515625, 500.3912353515625, 465.53814697265625, 203.92605590820312, 541.6237182617188, 343.3856201171875, 259.5971984863281, 2.216938018798828, -28.767623901367188, 198.076416015625, 423.1174621582031, 175.76718139648438, 176.38571166992188, 128.47027587890625, 482.12908935546875, -160.17127990722656, 409.76593017578125, 38.485687255859375, 26.62798309326172, 360.4049072265625, 455.3253173828125, -95.56453704833984, 92.99545288085938, 207.91934204101562, 356.1316833496094, 128.4925537109375, -33.63755798339844, 56.88412094116211, -6.735679626464844, 221.384033203125, 144.6773223876953, 141.76380920410156, -30.6312255859375, 158.34915161132812, -4.624624252319336, 404.8217468261719, 426.083740234375, -41.3134765625, 254.74520874023438, -98.57162475585938, 876.066650390625, 194.36209106445312, 582.9351196289062, -2.88092041015625, 588.6790771484375, 545.67529296875, 557.855712890625, 34.886966705322266], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000603.npy"} +{"epoch": 0.8854625550660793, "step": 604, "batch_size": 64, "mean": 238.44009399414062, "std": 223.31759643554688, "min": -462.18450927734375, "p10": -14.440049743652343, "median": 241.2116241455078, "p90": 528.7655975341797, "max": 752.7164916992188, "pos_frac": 0.84375, "sample": [-18.952077865600586, 83.80230712890625, 34.58997344970703, -13.671615600585938, 350.51788330078125, 264.7066955566406, 452.09796142578125, 423.8373718261719, 208.2447052001953, 231.35675048828125, -200.07119750976562, 504.0403747558594, 285.21844482421875, -54.725257873535156, 454.54156494140625, 323.89215087890625, 352.62957763671875, 18.950428009033203, 204.13021850585938, 440.73284912109375, 181.06707763671875, -198.36712646484375, 141.4077911376953, 246.70361328125, 72.3196029663086, -462.18450927734375, 580.5166625976562, -13.125564575195312, 259.1054382324219, 35.45367431640625, -86.9413070678711, 191.4763946533203, 197.29954528808594, 371.0829162597656, 214.07839965820312, 574.6005859375, 299.52288818359375, 16.29834747314453, 234.06227111816406, 169.45565795898438, 476.4557800292969, 357.90960693359375, 271.9794616699219, 45.84209442138672, 428.8996276855469, 547.7559204101562, 227.10130310058594, 752.7164916992188, 299.2238464355469, 206.71424865722656, 417.935546875, 558.0712280273438, 375.6705322265625, 189.7387237548828, 539.3621215820312, 280.1117858886719, 54.60936737060547, -2.849884033203125, 235.71963500976562, 333.64013671875, -14.769378662109375, 327.9144287109375, 705.6693115234375, 275.0420837402344], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000604.npy"} +{"epoch": 0.8869309838472834, "step": 605, "batch_size": 64, "mean": 207.88526916503906, "std": 224.07066345214844, "min": -343.9155578613281, "p10": -12.215621948242188, "median": 177.49118041992188, "p90": 458.1726348876954, "max": 1140.482666015625, "pos_frac": 0.875, "sample": [54.10554504394531, 79.98041534423828, 11.53668212890625, 69.55782318115234, 278.329345703125, 141.9385223388672, 136.8401641845703, 361.71771240234375, 443.2171630859375, 105.98635864257812, 339.52239990234375, -37.500640869140625, 13.775148391723633, 39.31494903564453, 274.4526672363281, 268.14044189453125, 152.6310272216797, 569.7776489257812, 632.7374877929688, 30.858993530273438, 1140.482666015625, 163.79318237304688, 128.934326171875, 234.845947265625, 209.11973571777344, 61.4398193359375, 102.98822021484375, 255.88937377929688, -12.137622833251953, 188.94308471679688, 153.18609619140625, -90.02510070800781, 368.2388916015625, -57.691192626953125, -31.164859771728516, 257.3203430175781, 643.6632690429688, 75.65953063964844, 183.56515502929688, 464.5821228027344, 103.63575744628906, 770.8514404296875, -127.42256164550781, -343.9155578613281, 399.34197998046875, 242.28199768066406, -12.24905014038086, 312.009033203125, 108.84778594970703, 190.4939422607422, 110.75614166259766, 249.92108154296875, 174.19134521484375, 244.41989135742188, 106.87867736816406, 261.158447265625, 180.791015625, 246.44915771484375, 277.4964599609375, 95.19038391113281, 352.850341796875, 283.6719055175781, 142.50863647460938, 525.9457397460938], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000605.npy"} +{"epoch": 0.8883994126284875, "step": 606, "batch_size": 64, "mean": 230.42236328125, "std": 206.36439514160156, "min": -201.84262084960938, "p10": -11.028369522094726, "median": 205.40987396240234, "p90": 499.7105407714845, "max": 890.30126953125, "pos_frac": 0.859375, "sample": [-0.012340545654296875, -44.117279052734375, 340.44354248046875, -44.960174560546875, 284.5291442871094, 408.39996337890625, -11.20663070678711, 85.62188720703125, 458.43792724609375, 529.3759155273438, 120.53396606445312, 890.30126953125, 359.79010009765625, -10.6124267578125, -20.662994384765625, 78.50640869140625, 188.51397705078125, 178.0638427734375, 297.7148132324219, 473.3016052246094, 207.08580017089844, -35.59284973144531, 530.6983032226562, 6.203956604003906, 590.361083984375, 408.0616760253906, 203.73394775390625, 154.39906311035156, 86.72791290283203, -32.15974426269531, 117.97114562988281, 348.0954284667969, 67.819580078125, 82.21194458007812, -201.84262084960938, 100.29563903808594, 216.61956787109375, 250.24090576171875, 3.9692630767822266, 3.908203125, 182.53988647460938, 244.78094482421875, 632.5348510742188, 382.6184387207031, 247.30252075195312, 296.58111572265625, 175.12139892578125, 403.040283203125, 358.0831298828125, 171.64309692382812, 126.14414978027344, 511.0286560058594, 0.1733245849609375, 230.8859100341797, 284.9866638183594, 701.2816162109375, 182.4760284423828, 310.11822509765625, 111.41515350341797, 319.827392578125, 464.2574462890625, 269.93402099609375, 351.94140625, 117.54580688476562], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000606.npy"} +{"epoch": 0.8898678414096917, "step": 607, "batch_size": 64, "mean": 184.58773803710938, "std": 183.45977783203125, "min": -193.13888549804688, "p10": -22.028811836242674, "median": 173.91915893554688, "p90": 444.8914062500001, "max": 626.232177734375, "pos_frac": 0.859375, "sample": [192.5768585205078, 274.2718811035156, 219.26853942871094, 107.62181091308594, 97.69417572021484, 95.14413452148438, 495.1866455078125, 467.2987060546875, 9.772689819335938, 498.40521240234375, -100.47344970703125, 179.02108764648438, 234.0203857421875, 2.8842697143554688, 128.47763061523438, 626.232177734375, 62.3925666809082, 87.6280746459961, 87.7646255493164, 64.67292022705078, 74.52069091796875, -21.693769454956055, 128.55035400390625, 202.13856506347656, -160.61407470703125, 400.69281005859375, -117.26563262939453, 343.39312744140625, 323.1130065917969, 188.4459991455078, 287.1936340332031, 20.81147003173828, 510.38812255859375, 139.43777465820312, 421.04620361328125, 308.517822265625, 251.68597412109375, 8.333251953125, 455.11077880859375, 168.81723022460938, -193.13888549804688, 350.8962707519531, -167.686767578125, 396.4417419433594, 224.1097412109375, 68.81304931640625, -5.342063903808594, 259.2011413574219, -53.166954040527344, 244.021240234375, 61.949127197265625, 248.99136352539062, 95.77828979492188, 202.02276611328125, 70.3888168334961, 46.79280090332031, 365.59686279296875, 469.0025634765625, 340.4056396484375, 389.48162841796875, 159.85433959960938, -22.172401428222656, 355.3100891113281, 143.58103942871094], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000607.npy"} +{"epoch": 0.8913362701908958, "step": 608, "batch_size": 64, "mean": 199.4820556640625, "std": 176.6732177734375, "min": -171.54856872558594, "p10": -25.424427413940414, "median": 179.53932189941406, "p90": 414.7523345947266, "max": 725.8709716796875, "pos_frac": 0.859375, "sample": [417.9019775390625, 320.9483642578125, 93.06681060791016, 178.409423828125, 17.906394958496094, 423.8438415527344, 142.42617797851562, 62.21794509887695, 205.09315490722656, 323.9159851074219, 335.76531982421875, -82.60890197753906, 152.21133422851562, 323.30755615234375, -31.544082641601562, 83.76727294921875, 20.489288330078125, 283.50103759765625, 12.58574104309082, 263.8724670410156, 407.4031677246094, 162.3840789794922, 54.277374267578125, 180.66921997070312, -147.41659545898438, 229.47557067871094, 150.860595703125, 252.08804321289062, 401.54071044921875, -47.48548889160156, 108.37936401367188, -6.812980651855469, 140.5633087158203, 218.2710723876953, 152.22157287597656, 302.5477294921875, 146.5124969482422, -47.29035949707031, 387.0177001953125, 1.5760574340820312, 725.8709716796875, 281.6695861816406, 484.1011962890625, 60.3284912109375, 291.2890625, 219.5955810546875, 6.567779541015625, 332.4366760253906, 382.8648376464844, 169.3974151611328, 449.7251281738281, -31.08989715576172, 276.9657287597656, 440.5937194824219, -171.54856872558594, 406.6728515625, -12.204998016357422, 96.20614624023438, 399.63604736328125, 346.0977783203125, 321.91400146484375, 438.23944091796875, 133.513916015625, 124.14849853515625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000608.npy"} +{"epoch": 0.8928046989720999, "step": 609, "batch_size": 64, "mean": 229.5865020751953, "std": 203.2161865234375, "min": -364.376220703125, "p10": 39.04909362792969, "median": 222.65142822265625, "p90": 452.84071044921876, "max": 887.6013793945312, "pos_frac": 0.921875, "sample": [150.84356689453125, 147.4427490234375, 61.951759338378906, -35.44316101074219, 259.289306640625, 348.793701171875, 427.775146484375, 292.7773132324219, 272.74652099609375, 71.77493286132812, 455.1901550292969, 421.1182861328125, 44.723358154296875, 296.8899841308594, 337.9992370605469, -364.376220703125, -36.675323486328125, 207.51199340820312, 236.64080810546875, 99.18367767333984, 173.23587036132812, 277.9339599609375, 188.76193237304688, 447.3586730957031, 168.3311767578125, 327.49627685546875, 624.0448608398438, 93.38552856445312, 42.77853012084961, 368.99200439453125, 159.0535888671875, 230.1363067626953, 515.5391845703125, 346.55303955078125, 887.6013793945312, 612.4312744140625, -69.92224884033203, 305.0618896484375, 125.34171295166016, 341.46002197265625, 183.73794555664062, 233.3997344970703, 239.99832153320312, 715.51123046875, 261.42852783203125, 249.2763671875, 219.75933837890625, 44.83259201049805, -131.08958435058594, 102.77790832519531, 332.478271484375, 117.80624389648438, 174.85118103027344, 110.65419006347656, 25.520408630371094, 77.03240203857422, 153.53375244140625, 85.412109375, 361.7148742675781, 604.8792724609375, 94.06981658935547, 311.2242431640625, 225.54351806640625, 37.45076370239258], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000609.npy"} +{"epoch": 0.8942731277533039, "step": 610, "batch_size": 64, "mean": 194.17398071289062, "std": 188.55816650390625, "min": -245.99185180664062, "p10": 0.0648876190185641, "median": 200.4319305419922, "p90": 438.63706359863284, "max": 665.4229125976562, "pos_frac": 0.890625, "sample": [252.70184326171875, 212.29312133789062, 223.3088836669922, 232.31935119628906, -26.0640869140625, 218.75831604003906, 39.0877685546875, 251.4971160888672, 171.8084716796875, 68.6734619140625, 20.45233917236328, 188.57073974609375, 43.64221954345703, 416.67889404296875, 525.0101318359375, 318.3875732421875, 144.22994995117188, 121.99886322021484, -122.59151458740234, -3.8328704833984375, 182.2195587158203, 413.24945068359375, 173.23220825195312, 512.053955078125, -53.399993896484375, 230.68927001953125, 41.962520599365234, 107.14604187011719, 665.4229125976562, 496.0547180175781, 611.346923828125, 33.60096740722656, 41.34930419921875, 435.6598205566406, 304.58294677734375, 248.96868896484375, 103.8904800415039, 340.87506103515625, 178.4512939453125, 234.5056610107422, 323.37542724609375, 134.5392608642578, 9.159656524658203, 63.28892517089844, 230.49862670898438, 89.3830337524414, 393.66900634765625, 71.03131103515625, -166.44781494140625, 286.4298095703125, 25.19586181640625, 337.2108154296875, 108.34712219238281, 44.50505447387695, 293.2908935546875, 246.70440673828125, 296.555419921875, 221.68942260742188, 113.96904754638672, 227.3754119873047, -245.99185180664062, -225.57052612304688, 439.91302490234375, 510.220458984375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000610.npy"} +{"epoch": 0.895741556534508, "step": 611, "batch_size": 64, "mean": 217.61814880371094, "std": 216.18695068359375, "min": -196.0240478515625, "p10": -31.206892776489237, "median": 195.97540283203125, "p90": 500.5283691406251, "max": 655.9007568359375, "pos_frac": 0.875, "sample": [303.6317443847656, 193.1886749267578, 481.1869812011719, 443.0170593261719, 405.4892578125, 302.1930847167969, 6.396171569824219, 354.47723388671875, 141.20437622070312, 37.958168029785156, 218.43106079101562, 630.95166015625, -90.33442687988281, -11.229717254638672, -118.18646240234375, 4.609670639038086, -39.76853942871094, -121.71937561035156, 222.73477172851562, 508.8175354003906, -196.0240478515625, 440.5495300292969, 456.2921142578125, -162.34170532226562, 367.1251525878906, 352.4244384765625, 224.31295776367188, 640.1685180664062, 206.40139770507812, 248.3478546142578, 135.74853515625, 478.2092590332031, 223.79676818847656, 7.708124160766602, 4.207763671875, 187.8914794921875, 47.3466796875, 127.22161865234375, 8.97027587890625, 194.22552490234375, 388.0750732421875, 197.72528076171875, 24.940963745117188, 575.6876220703125, 191.4155731201172, 221.66830444335938, 76.52692413330078, 178.55795288085938, 398.7425537109375, 87.22469329833984, 345.575439453125, -141.3912811279297, 167.9302978515625, 214.2197265625, 105.84640502929688, 131.9979248046875, 398.7089538574219, 457.53778076171875, 153.31507873535156, 560.6845092773438, 630.9602661132812, 6.615211486816406, 655.9007568359375, 33.46413803100586], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000611.npy"} +{"epoch": 0.8972099853157122, "step": 612, "batch_size": 64, "mean": 234.01136779785156, "std": 176.507568359375, "min": -106.71668243408203, "p10": 2.7530059814453143, "median": 253.62757110595703, "p90": 452.07932739257814, "max": 632.80322265625, "pos_frac": 0.90625, "sample": [460.06463623046875, 249.39622497558594, 352.5169677734375, -99.56884765625, 9.629878997802734, 259.3722839355469, 436.21728515625, 319.9719543457031, 134.0349578857422, -85.91618347167969, 1.896575927734375, 455.32049560546875, 95.955078125, 619.612548828125, 318.95562744140625, 408.9361572265625, 398.18646240234375, 215.79739379882812, 244.68820190429688, 444.5166015625, 276.8966369628906, 84.0735855102539, -106.71668243408203, 359.5430908203125, 542.7406616210938, 126.85205078125, 202.84457397460938, 4.7513427734375, 281.85638427734375, 382.9432373046875, 355.4317321777344, 160.88333129882812, 75.36781311035156, 56.469642639160156, 393.0572204589844, 129.39161682128906, 295.3707275390625, 279.41290283203125, 142.46356201171875, 61.92938232421875, 207.81651306152344, 257.8589172363281, 35.296875, 240.652587890625, 392.37030029296875, -11.684383392333984, 467.4263610839844, 373.00677490234375, 632.80322265625, 203.7305908203125, 55.40522766113281, 370.5611572265625, 456.75250244140625, 123.24565887451172, -54.19511413574219, 129.8925323486328, 274.3606262207031, 81.49491882324219, 274.90997314453125, 100.42311096191406, 432.35546875, 279.0595703125, 331.233154296875, -23.19622802734375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000612.npy"} +{"epoch": 0.8986784140969163, "step": 613, "batch_size": 64, "mean": 216.34506225585938, "std": 171.2978973388672, "min": -112.21524810791016, "p10": 21.181314086914067, "median": 194.83810424804688, "p90": 437.1473419189453, "max": 692.4522094726562, "pos_frac": 0.921875, "sample": [331.1004638671875, 202.80303955078125, 271.357421875, -15.800243377685547, 297.5, 346.56829833984375, 60.90071487426758, 317.2805480957031, 280.2414855957031, 135.47377014160156, 380.4214782714844, 66.61299896240234, 323.16375732421875, 46.28893280029297, 622.1858520507812, 240.7770233154297, 34.40797424316406, 401.06427001953125, 235.01502990722656, 82.38890075683594, 321.21807861328125, 110.98229217529297, 225.8555450439453, 469.7174072265625, 435.8042297363281, 158.0686492919922, 414.4327392578125, 449.981689453125, 25.10858154296875, 249.758544921875, 182.1450958251953, 158.27694702148438, 68.06024932861328, 326.15521240234375, 110.48771667480469, 437.72296142578125, 125.5808334350586, 277.94775390625, 647.7911376953125, -0.41819000244140625, 118.7320327758789, 478.2595520019531, 378.56732177734375, 27.08867645263672, -16.61260986328125, 112.48513793945312, 111.38966369628906, 692.4522094726562, -10.892963409423828, 76.86627197265625, 95.49986267089844, 197.90716552734375, -112.21524810791016, 196.891357421875, 329.2010498046875, 293.0054016113281, 124.166748046875, 150.77944946289062, 12.810218811035156, 192.78485107421875, 70.57713317871094, 141.68289184570312, 310.7304382324219, 19.498199462890625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000613.npy"} +{"epoch": 0.9001468428781204, "step": 614, "batch_size": 64, "mean": 229.2654571533203, "std": 202.93690490722656, "min": -277.4686279296875, "p10": -35.14746208190917, "median": 221.26158905029297, "p90": 499.7027160644532, "max": 661.502197265625, "pos_frac": 0.859375, "sample": [182.774169921875, 452.13226318359375, 45.036773681640625, 316.8032531738281, 133.4257354736328, 372.8975830078125, 72.17605590820312, 295.391357421875, 75.26777648925781, -38.93551254272461, 354.4496765136719, 29.316112518310547, 257.9624938964844, 195.16448974609375, 333.41009521484375, -96.8906478881836, 364.04766845703125, 478.94757080078125, 206.36244201660156, -55.5218391418457, 588.7359619140625, 10.296905517578125, 221.5305633544922, 350.14263916015625, 27.93109130859375, -26.308677673339844, 279.1474609375, 164.74618530273438, 235.24095153808594, -62.10615539550781, 220.99261474609375, 440.58270263671875, -21.776371002197266, 375.23858642578125, -57.502777099609375, 97.54436492919922, 390.702392578125, 51.86725997924805, 215.5644073486328, 330.4791259765625, 453.4047546386719, 123.34620666503906, 414.7685546875, 256.2843017578125, 608.2171020507812, 517.7822875976562, 197.77963256835938, 180.3096923828125, 250.8170166015625, 352.2992858886719, 73.29442596435547, 24.53645896911621, 466.4958190917969, 91.38214111328125, 553.0511474609375, 661.502197265625, 537.3701782226562, 110.137939453125, -73.8761215209961, -277.4686279296875, 508.5977783203125, 116.52587890625, 301.3283386230469, 417.8362731933594], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000614.npy"} +{"epoch": 0.9016152716593245, "step": 615, "batch_size": 64, "mean": 176.6024932861328, "std": 185.09478759765625, "min": -262.17449951171875, "p10": 5.0747100830078145, "median": 153.07540893554688, "p90": 387.67908020019536, "max": 746.5266723632812, "pos_frac": 0.90625, "sample": [40.900909423828125, 498.72998046875, 147.56149291992188, 284.99468994140625, 88.8552017211914, 339.6221618652344, 6.7407989501953125, 93.44208526611328, 114.806884765625, 56.184112548828125, 327.3707580566406, 4.3606719970703125, 200.6096954345703, -11.05517578125, 381.8424072265625, -70.17538452148438, 72.015869140625, 44.66040802001953, 180.34376525878906, 214.56402587890625, 146.01324462890625, 38.983917236328125, 684.6895141601562, 72.84447479248047, 50.640716552734375, 274.0600280761719, 84.0866470336914, 420.46697998046875, 227.99696350097656, 61.54119873046875, 185.64602661132812, 198.6831512451172, 180.74951171875, 166.45396423339844, 135.78622436523438, 8.360816955566406, 283.78265380859375, 158.58932495117188, 62.71138000488281, 95.83544158935547, -262.17449951171875, 347.40960693359375, 173.487548828125, 20.460861206054688, 390.1805114746094, 275.558837890625, 222.18167114257812, 618.4907836914062, 78.86080932617188, 334.1113586425781, 303.965576171875, -114.09001159667969, 169.99403381347656, -6.492340087890625, 111.7856674194336, 35.04462432861328, 104.61561584472656, 202.17913818359375, 192.28915405273438, 625.0773315429688, -46.41200256347656, 746.5266723632812, 170.8002166748047, 54.41090393066406], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000615.npy"} +{"epoch": 0.9030837004405287, "step": 616, "batch_size": 64, "mean": 158.7562255859375, "std": 202.3321075439453, "min": -221.38937377929688, "p10": -119.70584564208984, "median": 135.65179443359375, "p90": 422.4989410400391, "max": 887.6134033203125, "pos_frac": 0.8125, "sample": [433.05181884765625, 23.163665771484375, 121.26428985595703, 382.3591003417969, 250.72927856445312, 115.13858795166016, 126.33233642578125, -34.665008544921875, 162.778076171875, 531.4822387695312, 453.5950927734375, 343.4500427246094, 159.96603393554688, 136.4618377685547, 131.92967224121094, 134.8417510986328, 162.82553100585938, 153.19186401367188, 134.57354736328125, -99.4645004272461, 467.8367004394531, 391.1755676269531, 10.930938720703125, 63.060462951660156, -112.05313110351562, -221.38937377929688, 270.5347900390625, 164.564697265625, 156.24359130859375, 365.2743225097656, 887.6134033203125, 70.99793243408203, -177.43234252929688, 166.18154907226562, 33.13047409057617, -122.98558044433594, -22.142724990844727, 94.82376098632812, 167.91342163085938, 254.72410583496094, 119.81266784667969, 338.9721374511719, 83.40052795410156, 161.798828125, -160.42623901367188, 341.0674133300781, 118.80238342285156, -23.134963989257812, 79.20160675048828, 211.66709899902344, 11.077346801757812, -138.09481811523438, 361.51019287109375, 13.738998413085938, 224.63116455078125, 119.96749877929688, 415.62457275390625, 223.22084045410156, 32.385986328125, 294.91888427734375, 425.4450988769531, -145.94288635253906, 463.6048583984375, -144.85910034179688], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000616.npy"} +{"epoch": 0.9045521292217328, "step": 617, "batch_size": 64, "mean": 161.53744506835938, "std": 180.4005889892578, "min": -233.53880310058594, "p10": -14.32687854766845, "median": 128.89208984375, "p90": 376.4504272460938, "max": 671.559814453125, "pos_frac": 0.875, "sample": [76.04585266113281, 131.37298583984375, 25.4833984375, 337.6031188964844, -17.043256759643555, -33.85179901123047, 115.42491912841797, -180.3028106689453, 316.95355224609375, 90.21049499511719, 83.93241119384766, 173.54830932617188, 236.70895385742188, 370.5715637207031, 124.02333068847656, 158.109130859375, 39.389068603515625, 346.7322998046875, 74.19825744628906, 584.6294555664062, 285.3663330078125, 126.41119384765625, 496.02410888671875, 94.51470184326172, 180.7418212890625, 258.6524353027344, 118.2593002319336, 32.02293014526367, 297.4986572265625, 180.8580780029297, 28.471529006958008, 302.34527587890625, -7.9886627197265625, 33.068519592285156, 323.9809875488281, 288.52886962890625, 225.14840698242188, 230.1918487548828, 265.7973327636719, 193.986083984375, 132.45591735839844, -71.50707244873047, 378.9699401855469, 264.4739074707031, 626.446044921875, 394.7838134765625, 56.37160110473633, -233.53880310058594, 108.501953125, 167.4743194580078, -150.6476593017578, 23.329147338867188, 16.03685760498047, 135.37982177734375, 67.2370834350586, 1.1736412048339844, 425.6300354003906, 68.43997192382812, 10.142282485961914, 48.07080078125, 671.559814453125, 232.7518310546875, 64.95552062988281, -107.71231079101562], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000617.npy"} +{"epoch": 0.9060205580029369, "step": 618, "batch_size": 64, "mean": 218.5650634765625, "std": 208.6680908203125, "min": -205.2748260498047, "p10": -49.45308532714842, "median": 208.97681427001953, "p90": 490.518310546875, "max": 731.0361328125, "pos_frac": 0.828125, "sample": [-176.44369506835938, 338.5655212402344, 151.65667724609375, 90.6574478149414, 208.80935668945312, 174.62962341308594, 217.70208740234375, 51.222957611083984, 564.9814453125, 242.67996215820312, 311.4578857421875, -56.947669982910156, 226.08590698242188, 198.25013732910156, 329.3247985839844, 155.76376342773438, -173.329833984375, 526.0623779296875, 90.83501434326172, -132.83596801757812, -31.965721130371094, 79.71133422851562, 3.5994186401367188, 27.586349487304688, -10.92633056640625, -123.66685485839844, 210.84060668945312, 421.6152038574219, 456.75390625, 221.28579711914062, 78.45741271972656, 82.49559783935547, 386.50439453125, 484.15789794921875, 404.0015869140625, -205.2748260498047, -119.71871948242188, 183.1783447265625, 200.95277404785156, 210.51513671875, 199.76475524902344, 330.31280517578125, 499.4875793457031, 380.6046142578125, 453.43017578125, -24.21997833251953, 299.56512451171875, 231.65211486816406, 206.7731475830078, 553.3488159179688, 354.7769775390625, -29.6997013092041, 731.0361328125, 437.13909912109375, 493.24420166015625, 182.2277069091797, 209.14427185058594, 355.3895568847656, 503.1593017578125, 70.51284790039062, 141.59393310546875, 483.3891296386719, 418.13861083984375, 208.16139221191406], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000618.npy"} +{"epoch": 0.9074889867841409, "step": 619, "batch_size": 64, "mean": 174.41763305664062, "std": 232.4015655517578, "min": -327.4848937988281, "p10": -96.29234924316407, "median": 157.80319213867188, "p90": 502.39216613769537, "max": 814.137451171875, "pos_frac": 0.75, "sample": [-253.9483642578125, -13.293899536132812, 199.4505615234375, 238.73809814453125, 338.29974365234375, -96.36026763916016, 121.36253356933594, -22.006690979003906, 347.09222412109375, 27.863744735717773, 260.50616455078125, 401.1950988769531, 89.60490417480469, 814.137451171875, 7.7553253173828125, 218.97113037109375, 489.4106140136719, 146.7645263671875, 568.3807983398438, 391.2976379394531, -169.80633544921875, 9.67097282409668, 478.03857421875, 507.9556884765625, -96.69132995605469, -118.04852294921875, -96.13387298583984, 460.16351318359375, 237.833740234375, -47.47711181640625, -83.61868286132812, 184.67849731445312, 94.66390228271484, -23.34698486328125, 386.7145080566406, -8.673065185546875, 113.9237060546875, -22.182968139648438, 234.79505920410156, 187.1540985107422, 180.10791015625, 563.7008056640625, 293.79119873046875, 476.09405517578125, 222.67022705078125, 61.91797637939453, -123.49581909179688, 551.2868041992188, 329.7170715332031, 168.84185791015625, 68.64981842041016, 16.39895248413086, 216.17929077148438, 587.3143310546875, -327.4848937988281, 45.83343505859375, 321.8696594238281, -84.7769775390625, 534.8938598632812, 93.090576171875, 103.9217529296875, 75.56741333007812, 10.675985336303711, 271.1288757324219], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000619.npy"} +{"epoch": 0.908957415565345, "step": 620, "batch_size": 64, "mean": 192.46888732910156, "std": 214.85707092285156, "min": -266.64776611328125, "p10": -31.466917037963867, "median": 171.10147857666016, "p90": 456.3808959960937, "max": 819.9296875, "pos_frac": 0.828125, "sample": [260.0913391113281, -238.94732666015625, 304.24078369140625, 217.34011840820312, 108.17314910888672, 468.3790283203125, 286.263671875, 276.8333740234375, -13.077690124511719, 531.2754516601562, 16.412635803222656, -2.8162155151367188, -30.767234802246094, 455.60595703125, 338.045166015625, 118.36091613769531, 481.8648376464844, 130.42869567871094, 277.5621032714844, 438.89044189453125, 277.15582275390625, 130.26199340820312, 98.171142578125, 281.6708984375, -153.359130859375, 238.1348876953125, 283.1405029296875, 117.3612060546875, 255.56898498535156, 22.999755859375, -266.64776611328125, 450.5993347167969, 294.38262939453125, 798.7205200195312, 88.88616943359375, 208.87010192871094, 45.252952575683594, 157.2034149169922, 262.8634033203125, 227.0277099609375, 25.711538314819336, 240.1851348876953, -76.7861328125, -1.092071533203125, 739.826904296875, 122.72100830078125, 456.7130126953125, -96.20695495605469, 161.3934326171875, 256.8836669921875, -31.766780853271484, 56.58732604980469, 819.9296875, 37.510337829589844, 124.86155700683594, 43.144744873046875, 226.50057983398438, 87.6064224243164, 45.81953430175781, 320.8966064453125, 129.48672485351562, 253.88363647460938, -49.035308837890625, 180.8095245361328], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000620.npy"} +{"epoch": 0.9104258443465492, "step": 621, "batch_size": 64, "mean": 235.76600646972656, "std": 208.9789581298828, "min": -268.7210388183594, "p10": 22.50883560180665, "median": 217.9726333618164, "p90": 508.6497283935547, "max": 755.1241455078125, "pos_frac": 0.921875, "sample": [390.941650390625, -214.61892700195312, -268.7210388183594, -51.76728820800781, 283.6722412109375, 70.60953521728516, 229.61907958984375, 216.9062042236328, 159.54226684570312, 542.6184692382812, 378.58758544921875, 119.82234191894531, 170.28541564941406, 359.40966796875, 190.50607299804688, 62.96949005126953, 11.392114639282227, 219.23110961914062, 63.706336975097656, 226.94818115234375, 755.1241455078125, 309.31689453125, 476.9869079589844, 232.7121124267578, 503.0301818847656, 670.5567626953125, 144.99356079101562, 444.1219177246094, 451.8643798828125, 59.820186614990234, 19.464942932128906, 386.68560791015625, 370.48797607421875, 132.71917724609375, 203.05029296875, 535.1773681640625, 96.97637176513672, 318.99212646484375, 164.45199584960938, 64.24197387695312, 29.611251831054688, 412.92535400390625, -147.29473876953125, 291.13232421875, 287.3983459472656, 193.23773193359375, 362.1988525390625, 221.90516662597656, 117.38117218017578, 511.05810546875, -137.84886169433594, 78.21434783935547, 212.19247436523438, 290.9451904296875, 92.97441101074219, 167.20408630371094, 326.2295227050781, 643.8428344726562, 82.77291870117188, 205.48516845703125, 714.6908569335938, 254.40231323242188, 219.0390625, 156.89149475097656], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000621.npy"} +{"epoch": 0.9118942731277533, "step": 622, "batch_size": 64, "mean": 204.817626953125, "std": 203.19898986816406, "min": -121.26095581054688, "p10": -22.317111968994137, "median": 204.00540161132812, "p90": 441.3841735839845, "max": 880.2883911132812, "pos_frac": 0.828125, "sample": [281.1438903808594, 264.2371520996094, 95.10282135009766, 287.2088623046875, 204.21360778808594, 301.4615783691406, -66.08058166503906, 220.2080078125, 688.545166015625, 230.2344207763672, 133.81228637695312, 86.08222198486328, 107.55061340332031, -51.49613952636719, 134.04006958007812, 393.26629638671875, 135.3400421142578, 207.935791015625, 118.45913696289062, 203.7971954345703, 327.3067932128906, 34.166900634765625, 105.83955383300781, 880.2883911132812, 208.09295654296875, 22.056365966796875, 234.11685180664062, 240.86868286132812, 299.9621887207031, -103.97116088867188, 197.70046997070312, -10.9815673828125, 568.7113647460938, 167.07058715820312, 128.6187744140625, 558.44677734375, -23.462814331054688, 4.994819641113281, 452.655029296875, 247.18405151367188, 406.1807556152344, 286.795166015625, -14.493034362792969, -19.64380645751953, 179.42141723632812, 24.69744873046875, 207.36294555664062, -121.26095581054688, 14.322463989257812, 166.29681396484375, 514.6886596679688, 407.00726318359375, 415.08551025390625, 358.8062438964844, 233.85256958007812, 266.7911682128906, -116.89471435546875, -57.80408477783203, 91.88661193847656, 705.6851196289062, 296.33203125, 246.2186737060547, 110.47346496582031, -8.206352233886719], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000622.npy"} +{"epoch": 0.9133627019089574, "step": 623, "batch_size": 64, "mean": 167.99696350097656, "std": 221.33395385742188, "min": -443.862548828125, "p10": -20.835553741455072, "median": 131.35900115966797, "p90": 479.8459259033206, "max": 853.627685546875, "pos_frac": 0.875, "sample": [166.7950439453125, 154.247802734375, -229.64938354492188, 531.52001953125, 166.16290283203125, 68.53880310058594, 333.1898193359375, 324.1171569824219, 329.5730285644531, 46.225250244140625, 578.1415405273438, 85.71121215820312, 136.06607055664062, 15.469467163085938, 212.65060424804688, -16.208179473876953, 511.0638427734375, 18.53216552734375, 54.1060791015625, 72.646728515625, 126.65193176269531, 179.15423583984375, 165.18508911132812, 68.4530029296875, -443.862548828125, 150.7613525390625, 296.4179382324219, 19.786537170410156, 218.3019561767578, 240.021728515625, 81.06491088867188, 95.71688842773438, 26.476234436035156, 18.0307559967041, -105.84085083007812, 23.916912078857422, 621.073974609375, 198.75787353515625, 173.48609924316406, 121.79146575927734, 59.45655822753906, 144.45565795898438, 234.41148376464844, 349.6709289550781, -133.97665405273438, -22.818714141845703, 853.627685546875, 390.29217529296875, 35.16667556762695, -141.01300048828125, 689.0071411132812, 407.0041198730469, 123.35505676269531, 49.02814483642578, 68.31863403320312, 591.0875854492188, -153.79864501953125, 170.8369598388672, 242.43856811523438, 122.24757385253906, 392.9094543457031, 62.84686279296875, 324.54827880859375, 58.4578857421875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000623.npy"} +{"epoch": 0.9148311306901615, "step": 624, "batch_size": 64, "mean": 228.5789337158203, "std": 211.294921875, "min": -311.6298522949219, "p10": -46.161084747314405, "median": 252.8228302001953, "p90": 508.9970550537111, "max": 613.0419311523438, "pos_frac": 0.890625, "sample": [291.3034973144531, -94.30964660644531, 374.1925048828125, 38.48816680908203, 272.44317626953125, 394.3426513671875, 338.7167663574219, 613.0419311523438, 527.752685546875, 29.95185089111328, 99.04714965820312, 18.1971435546875, 247.33889770507812, 431.6060485839844, 318.4247741699219, 64.45819091796875, 431.3183898925781, 451.89263916015625, 535.7327880859375, -311.6298522949219, 193.7393035888672, 420.83404541015625, 2.0855789184570312, 132.15614318847656, 224.74966430664062, 177.8787841796875, 429.5107727050781, 584.6476440429688, 167.85427856445312, 33.7076416015625, 172.61668395996094, 411.5329284667969, -104.14801025390625, 285.7229309082031, 91.2779541015625, 304.5013427734375, 416.2672119140625, 41.17564392089844, 297.6347351074219, 260.29376220703125, 73.35599517822266, 70.00330352783203, 302.06787109375, -73.31318664550781, 465.2339172363281, 335.1851501464844, 204.0912628173828, 374.963623046875, 531.9283447265625, 347.4401550292969, 115.61714172363281, -165.1271209716797, 74.53074645996094, 612.1218872070312, 48.358665466308594, 36.30418395996094, -66.83822631835938, 83.00850677490234, -172.38023376464844, 258.3067626953125, 543.6407470703125, 177.8292236328125, 462.900390625, 373.47393798828125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000624.npy"} +{"epoch": 0.9162995594713657, "step": 625, "batch_size": 64, "mean": 202.92684936523438, "std": 209.3148956298828, "min": -134.86526489257812, "p10": -5.085464477539062, "median": 175.16551971435547, "p90": 476.6017456054688, "max": 939.4347534179688, "pos_frac": 0.859375, "sample": [-117.7895278930664, 41.611236572265625, 366.8362121582031, 363.77252197265625, 80.83938598632812, 178.65130615234375, 391.8375549316406, 800.5572509765625, 194.1472930908203, 382.7674255371094, -4.328704833984375, -127.07821655273438, -5.4097900390625, 132.9619598388672, 457.69183349609375, 273.546630859375, -1.5133228302001953, -13.305522918701172, 939.4347534179688, 294.96221923828125, 617.573974609375, 153.5429229736328, 62.350013732910156, 171.6797332763672, 1.3046493530273438, 218.05337524414062, 484.70599365234375, 35.52415466308594, 39.99922180175781, 136.18856811523438, 157.6168212890625, 221.57516479492188, 395.9007568359375, -14.640384674072266, 374.18524169921875, 141.42453002929688, 40.147254943847656, 321.44085693359375, 179.6024169921875, 89.78961181640625, 18.970138549804688, 125.80663299560547, 504.3719482421875, 209.5049285888672, 237.6912384033203, 255.9561767578125, 31.552701950073242, 297.1690673828125, 191.4336395263672, 136.4359588623047, 33.55536651611328, -9.97186279296875, 334.7122802734375, 196.9169158935547, 111.20732116699219, 70.82238006591797, 89.02275085449219, 185.57510375976562, 615.3999633789062, -134.86526489257812, 270.8877868652344, 16.546539306640625, 504.532958984375, 235.9263916015625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000625.npy"} +{"epoch": 0.9177679882525698, "step": 626, "batch_size": 64, "mean": 154.83029174804688, "std": 186.25460815429688, "min": -182.77786254882812, "p10": -63.41778450012206, "median": 129.2095947265625, "p90": 400.96693420410156, "max": 636.8036499023438, "pos_frac": 0.78125, "sample": [287.1801452636719, 96.79092407226562, -52.563232421875, 144.04542541503906, 96.90885925292969, 203.65919494628906, 435.02783203125, 93.89413452148438, 231.42156982421875, 24.85070037841797, 359.2861022949219, 207.48988342285156, 539.2716064453125, 468.684326171875, -37.51287841796875, -16.651817321777344, 296.19464111328125, -95.9579086303711, 22.371124267578125, 75.669677734375, 18.1021728515625, -25.046737670898438, 137.9235382080078, 19.23166847229004, -67.64802551269531, 120.49565124511719, 92.93292236328125, -122.56538391113281, 6.692966461181641, 369.88067626953125, -4.13245964050293, 324.814208984375, -53.54722213745117, -132.47003173828125, 284.1272888183594, 324.98663330078125, 403.00677490234375, 179.95654296875, 160.33839416503906, 299.2237854003906, 382.62060546875, 1.5013427734375, 636.8036499023438, 366.6805419921875, 490.2425537109375, 151.99383544921875, -84.24713897705078, 165.95254516601562, 24.830718994140625, 233.52020263671875, 65.538330078125, 436.86932373046875, 338.7362060546875, 396.2073059082031, 1.26043701171875, 49.53456115722656, 109.27559661865234, 247.94349670410156, -182.77786254882812, -107.56047058105469, 164.27220153808594, 52.105323791503906, 262.83709716796875, -11.36562728881836], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000626.npy"} +{"epoch": 0.9192364170337739, "step": 627, "batch_size": 64, "mean": 223.52023315429688, "std": 272.48663330078125, "min": -296.0367736816406, "p10": -63.359281158447246, "median": 172.86180114746094, "p90": 580.9934509277343, "max": 1085.9390869140625, "pos_frac": 0.8125, "sample": [216.94512939453125, 206.14129638671875, 753.8206787109375, 363.1896057128906, 97.62726593017578, -33.69093322753906, 160.36117553710938, 268.4672546386719, 581.36865234375, 168.25250244140625, 484.072998046875, -159.29061889648438, 133.49310302734375, -20.400484085083008, -80.93062591552734, 482.6817626953125, 31.51506805419922, 130.66534423828125, 487.00762939453125, 413.5819396972656, -188.65882873535156, 423.4998474121094, 431.21759033203125, -15.926176071166992, 687.6978759765625, 90.22686767578125, 299.38214111328125, 494.001953125, 94.27485656738281, 10.537603378295898, 69.82230377197266, 36.33534622192383, 152.87649536132812, 763.1826171875, 457.1688232421875, 193.28509521484375, 388.1671142578125, 186.19515991210938, 291.426025390625, -159.48675537109375, 62.08197784423828, 9.084043502807617, 1085.9390869140625, 114.53079223632812, 209.302734375, -43.105140686035156, 722.5047607421875, -15.762138366699219, 49.588775634765625, 73.2745361328125, 580.1179809570312, 41.572044372558594, 356.8228759765625, -296.0367736816406, 45.57822799682617, 319.4747619628906, 236.63063049316406, 70.83934020996094, -152.52261352539062, 815.7924194335938, 329.2164001464844, -72.03962707519531, 177.47109985351562, 194.83328247070312], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000627.npy"} +{"epoch": 0.920704845814978, "step": 628, "batch_size": 64, "mean": 232.15545654296875, "std": 208.61180114746094, "min": -315.15240478515625, "p10": -18.73267879486084, "median": 210.9044189453125, "p90": 489.64844055175786, "max": 659.69775390625, "pos_frac": 0.84375, "sample": [273.61029052734375, -0.887359619140625, 353.928466796875, 622.0986328125, 391.75640869140625, 118.38143920898438, 304.6409606933594, 220.19622802734375, 1.3314361572265625, -92.132080078125, -18.87641143798828, 111.10974884033203, -315.15240478515625, 542.5599975585938, 295.3073425292969, 433.2351379394531, 49.35835266113281, -67.09630584716797, 659.69775390625, 429.646240234375, 466.45782470703125, 436.60235595703125, 208.72286987304688, 222.96728515625, 94.64234161376953, -49.745094299316406, 178.68727111816406, 218.74618530273438, 40.00220489501953, 67.27339172363281, 497.2034912109375, 315.69500732421875, 389.15362548828125, 395.2939147949219, 463.776123046875, 469.9833068847656, 213.08596801757812, 353.29107666015625, 56.867820739746094, 165.23318481445312, -18.397302627563477, 185.82366943359375, 90.91655731201172, 260.98944091796875, 327.5606689453125, 282.71502685546875, 472.0199890136719, 194.95751953125, 29.17603302001953, -4.49359130859375, 162.62362670898438, 124.92749786376953, 457.7311096191406, 579.7015380859375, -52.94732666015625, 45.45970916748047, 531.880126953125, 164.00717163085938, 452.0396423339844, 163.160400390625, 631.0389404296875, 152.52780151367188, 129.12425231933594, -21.246837615966797], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000628.npy"} +{"epoch": 0.922173274596182, "step": 629, "batch_size": 64, "mean": 198.65170288085938, "std": 167.18450927734375, "min": -62.813201904296875, "p10": -6.47862491607666, "median": 171.3987274169922, "p90": 442.0304718017578, "max": 761.2892456054688, "pos_frac": 0.875, "sample": [243.01922607421875, 252.76364135742188, -21.87650489807129, -29.07415199279785, 443.239013671875, 105.19697570800781, 79.88362121582031, 321.5460510253906, 308.3548889160156, 296.04443359375, 199.94085693359375, 137.926513671875, 277.85791015625, 68.40838623046875, 278.396728515625, -20.13526153564453, 169.62753295898438, 106.59295654296875, -6.022848129272461, 288.4337158203125, 24.866127014160156, 491.84521484375, 267.9434509277344, -56.52601623535156, 40.93745803833008, 209.95614624023438, 19.017257690429688, -6.673957824707031, 377.2380065917969, 209.57855224609375, 123.87747955322266, 143.6171112060547, 761.2892456054688, 467.7276611328125, 302.060546875, 110.3122329711914, -60.70597839355469, 447.52105712890625, 292.1484375, 38.24244689941406, 48.64616394042969, 69.50238037109375, 117.82435607910156, 61.09727478027344, 340.6885986328125, 458.9184265136719, 215.4552459716797, 127.91166687011719, 165.89158630371094, 15.898702621459961, 245.23683166503906, 138.6670684814453, 173.169921875, 418.9521789550781, 93.77183532714844, 439.2105407714844, 130.5365753173828, 506.7532653808594, 296.6685485839844, 337.8368835449219, 305.27227783203125, 154.56478881835938, 209.65133666992188, -62.813201904296875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000629.npy"} +{"epoch": 0.9236417033773862, "step": 630, "batch_size": 64, "mean": 160.61888122558594, "std": 196.10609436035156, "min": -217.9891357421875, "p10": -82.9333515167236, "median": 135.39012145996094, "p90": 395.0394104003907, "max": 771.3785400390625, "pos_frac": 0.859375, "sample": [470.3343811035156, -217.9891357421875, 94.73662567138672, 154.463623046875, -41.86688232421875, 348.4046630859375, 197.99209594726562, 212.4967041015625, 771.3785400390625, 139.88140869140625, 256.7682800292969, 157.09422302246094, 217.01235961914062, 333.6595153808594, 404.0460205078125, -111.01220703125, 24.618070602416992, 88.32135009765625, 120.96265411376953, 171.6875457763672, 77.01346588134766, 99.86203002929688, 68.45135498046875, 608.4202880859375, -142.01051330566406, 627.5818481445312, 156.86155700683594, 329.9886474609375, 45.865989685058594, 79.22758483886719, 22.480043411254883, 76.07207489013672, 222.83526611328125, -216.03195190429688, 343.3219909667969, 16.189069747924805, -96.81758117675781, 339.0408020019531, 67.55984497070312, 188.0412139892578, 37.6629638671875, 150.07821655273438, 57.044883728027344, -99.8931884765625, 113.7215576171875, 127.62896728515625, 213.10333251953125, 165.95523071289062, 239.41741943359375, 13.691551208496094, 201.03150939941406, 438.3962707519531, 278.4153747558594, 118.57261657714844, -50.53681564331055, 150.0846710205078, 130.89883422851562, 14.399673461914062, 320.6181945800781, -97.51337432861328, 601.0951538085938, 46.721168518066406, 374.02398681640625, 28.047300338745117], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000630.npy"} +{"epoch": 0.9251101321585903, "step": 631, "batch_size": 64, "mean": 176.74578857421875, "std": 227.75619506835938, "min": -417.2519226074219, "p10": -90.08995285034177, "median": 174.0776824951172, "p90": 456.8575561523438, "max": 703.4926147460938, "pos_frac": 0.78125, "sample": [621.667724609375, 235.2025604248047, -31.505950927734375, 287.092041015625, 120.01673126220703, 366.167236328125, 133.93930053710938, 122.86852264404297, -126.29296875, -39.59480285644531, 222.83358764648438, 703.4926147460938, -219.22799682617188, 394.2892761230469, -40.91019821166992, -67.80912780761719, 115.36299896240234, 587.654541015625, 39.531410217285156, 123.3080062866211, 44.22618865966797, 21.304048538208008, 328.66900634765625, -290.8728942871094, 266.3725891113281, 215.95947265625, 325.408203125, 21.545730590820312, 344.154541015625, 124.77297973632812, -417.2519226074219, 393.8056640625, 442.9940185546875, 233.03565979003906, 324.98126220703125, 44.4517822265625, 178.4798583984375, -52.20219421386719, 170.72518920898438, 326.5177307128906, 177.43017578125, -99.22803497314453, 586.4510498046875, 424.0065002441406, 316.9504699707031, -68.76776123046875, 532.5474243164062, 204.0513916015625, 462.799072265625, 315.42816162109375, 144.80661010742188, 103.68717956542969, 10.679744720458984, 40.476768493652344, -167.88214111328125, -17.6572265625, 350.546630859375, 252.11154174804688, 559.7561645507812, -159.33163452148438, 117.674560546875, 261.0144958496094, 120.61038970947266, 248.40614318847656], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000631.npy"} +{"epoch": 0.9265785609397944, "step": 632, "batch_size": 64, "mean": 187.42959594726562, "std": 222.80389404296875, "min": -426.7119445800781, "p10": -43.08137969970703, "median": 202.99563598632812, "p90": 474.65592041015634, "max": 674.0104370117188, "pos_frac": 0.84375, "sample": [230.47067260742188, -358.47637939453125, 130.7645263671875, 246.11630249023438, 427.9430236816406, -426.7119445800781, 457.043212890625, 481.27276611328125, 459.21661376953125, 374.4814453125, 346.14849853515625, 190.8587188720703, 255.52618408203125, 515.3038330078125, 228.74957275390625, 99.18669891357422, 435.2679748535156, 307.3848571777344, 20.363765716552734, -9.607994079589844, 415.13018798828125, 254.30145263671875, 309.45465087890625, 279.02978515625, -169.06272888183594, 502.2877502441406, 306.98480224609375, -177.62014770507812, 270.8939514160156, 70.4924087524414, -282.71343994140625, -42.536170959472656, 532.8630981445312, 175.60816955566406, 136.54574584960938, 167.84715270996094, 490.4132080078125, 319.4331359863281, 127.47013092041016, 15.070213317871094, 297.9874267578125, 21.21459197998047, -43.315040588378906, 84.40469360351562, 229.2156524658203, 206.29345703125, 337.1781005859375, 73.63890838623047, 525.5072631835938, 674.0104370117188, 325.18560791015625, 1.366241455078125, -202.29039001464844, 65.65412139892578, 98.90275573730469, 116.57508087158203, -7.021631240844727, 342.5163879394531, 52.060569763183594, 199.69781494140625, 135.4899444580078, 41.14726257324219, 24.603708267211914, 282.2755432128906], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000632.npy"} +{"epoch": 0.9280469897209985, "step": 633, "batch_size": 64, "mean": 154.74440002441406, "std": 210.4459686279297, "min": -310.533203125, "p10": -103.12013549804686, "median": 168.32492065429688, "p90": 335.94842834472655, "max": 695.8942260742188, "pos_frac": 0.765625, "sample": [-96.99923706054688, 322.56488037109375, 247.5592498779297, 38.29901885986328, 245.31884765625, 332.70050048828125, -183.6879425048828, -105.74337768554688, 572.4171142578125, 177.04067993164062, 300.3331298828125, 202.16015625, 250.79249572753906, -153.82778930664062, -72.33740234375, 46.82135772705078, 207.6427459716797, 515.6943359375, 167.88693237304688, 89.69625854492188, 86.51720428466797, 491.5900573730469, 215.0577850341797, -14.154808044433594, 307.05810546875, 83.22430419921875, 54.43592834472656, 157.48599243164062, -310.533203125, 47.900291442871094, 315.5384521484375, 158.55303955078125, 303.4766845703125, 161.91036987304688, 117.07444763183594, -19.628650665283203, -55.519287109375, -296.88763427734375, -197.29344177246094, 174.5755615234375, 695.8942260742188, 335.6390380859375, -66.42816162109375, -224.20547485351562, 334.46771240234375, 236.83004760742188, 329.86798095703125, 572.8037109375, 336.0810241699219, 148.90176391601562, -63.967498779296875, 170.33273315429688, -41.00273132324219, 280.26788330078125, 37.25819396972656, 70.02204895019531, 209.15147399902344, 547.1162109375, 103.3549575805664, 68.23684692382812, 309.70318603515625, 168.76290893554688, 181.01910400390625, 278.82147216796875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000633.npy"} +{"epoch": 0.9295154185022027, "step": 634, "batch_size": 64, "mean": 157.60775756835938, "std": 201.89881896972656, "min": -191.72691345214844, "p10": -55.940705871582026, "median": 122.10478210449219, "p90": 448.4654083251954, "max": 653.3046875, "pos_frac": 0.734375, "sample": [-80.58781433105469, 341.24969482421875, 107.70633697509766, 558.14208984375, 73.30644226074219, 16.208005905151367, 11.671724319458008, 195.06695556640625, 247.5668182373047, 367.52386474609375, 37.600128173828125, 169.28948974609375, 49.068687438964844, 167.92486572265625, 418.851806640625, 299.7915344238281, -25.422183990478516, -57.88836669921875, 105.87228393554688, 42.974884033203125, 461.1569519042969, -191.72691345214844, -100.94807434082031, 63.90895080566406, -139.48934936523438, 151.71786499023438, 52.033966064453125, 463.59552001953125, 315.396240234375, 84.14836120605469, 503.0829162597656, 131.71066284179688, 154.78607177734375, -128.99063110351562, 653.3046875, 390.9659729003906, 628.760009765625, -161.12074279785156, 119.66815185546875, 43.494571685791016, 196.07872009277344, -24.504409790039062, -32.96088409423828, -48.17325210571289, 81.20967102050781, -36.93280792236328, 268.3081970214844, -31.076492309570312, 497.543701171875, 362.8617858886719, -4.474485397338867, 268.0185546875, 346.64501953125, 365.0145263671875, 381.2369079589844, 126.31089782714844, 215.51742553710938, 318.7663269042969, -36.37181854248047, 124.54141235351562, -26.258087158203125, -51.39616394042969, 63.45313262939453, 222.16702270507812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000634.npy"} +{"epoch": 0.9309838472834068, "step": 635, "batch_size": 64, "mean": 168.44796752929688, "std": 181.3225555419922, "min": -195.991455078125, "p10": -38.71995925903318, "median": 133.6643524169922, "p90": 456.56730651855474, "max": 626.7081909179688, "pos_frac": 0.859375, "sample": [118.51290893554688, 222.66737365722656, 580.6842041015625, -70.97064208984375, 185.00283813476562, -60.706504821777344, 95.93161010742188, 55.56354904174805, 6.122398376464844, 92.93633270263672, 148.8157958984375, 626.7081909179688, 116.29756164550781, 151.8903045654297, -65.20892333984375, 208.23703002929688, 53.11688232421875, 39.474945068359375, 573.5759887695312, 42.36985778808594, 264.26312255859375, 16.536911010742188, 425.62274169921875, 510.940673828125, 209.94149780273438, 84.8569107055664, 6.1066741943359375, 279.3269958496094, 153.79441833496094, 521.6593627929688, 3.864065170288086, 277.1654052734375, 34.048370361328125, 439.8821716308594, 319.84588623046875, 115.85761260986328, 74.66230010986328, 71.69654846191406, -66.11666107177734, 240.6383819580078, 95.70088195800781, 272.13555908203125, 568.3905639648438, 219.84512329101562, 54.21874237060547, 163.6270751953125, 301.87298583984375, -9.980491638183594, -195.991455078125, 214.4967041015625, 114.22998046875, -17.447830200195312, 254.12893676757812, 247.02664184570312, 261.7586364746094, 153.68655395507812, 91.62728881835938, 98.93203735351562, 70.453857421875, 463.71807861328125, -102.9245376586914, -47.836585998535156, 202.84628295898438, 200.53981018066406], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000635.npy"} +{"epoch": 0.9324522760646109, "step": 636, "batch_size": 64, "mean": 192.24244689941406, "std": 188.2211456298828, "min": -183.7495574951172, "p10": -65.73992614746092, "median": 185.7210235595703, "p90": 389.3109039306641, "max": 853.431640625, "pos_frac": 0.828125, "sample": [44.59840393066406, -4.0381927490234375, 60.275177001953125, 83.041015625, 155.54910278320312, -85.08090209960938, 301.68115234375, 97.05860137939453, 345.66961669921875, 24.878562927246094, 143.1800994873047, 124.41568756103516, 204.66961669921875, 220.12380981445312, 179.01101684570312, 184.24847412109375, -103.64105224609375, 522.5953369140625, 368.2938232421875, 157.0594482421875, 283.1182861328125, 374.1434631347656, 121.94149780273438, -89.26246643066406, 291.3251037597656, 175.15682983398438, 365.2006530761719, 254.59080505371094, 191.22178649902344, 853.431640625, -130.4774627685547, -28.215225219726562, 360.1881408691406, 244.0739288330078, -91.22421264648438, 562.0452270507812, 399.2304382324219, 282.4279479980469, 290.885009765625, -75.479736328125, 368.082763671875, 135.22886657714844, 129.9981689453125, 321.9433288574219, 250.2469482421875, 324.28753662109375, 409.79248046875, 271.8365173339844, 41.21416473388672, -183.7495574951172, 133.54933166503906, -43.013702392578125, -24.8966064453125, 95.93244934082031, 274.59600830078125, 204.2164764404297, 391.28240966796875, 503.0627746582031, 127.4524917602539, 272.38848876953125, 12.16510009765625, 58.08549880981445, 187.19357299804688, 384.7107238769531], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000636.npy"} +{"epoch": 0.933920704845815, "step": 637, "batch_size": 64, "mean": 211.48562622070312, "std": 200.90463256835938, "min": -171.2229766845703, "p10": -8.570950317382806, "median": 180.7552032470703, "p90": 492.54697875976564, "max": 714.3002319335938, "pos_frac": 0.875, "sample": [177.1295166015625, 155.02627563476562, 61.705169677734375, 201.5836181640625, 40.7326545715332, 381.179443359375, 353.6954345703125, -171.03494262695312, 483.32269287109375, 396.92724609375, 472.71063232421875, -25.180173873901367, 190.82322692871094, 183.39251708984375, -37.34889221191406, 496.500244140625, 214.36314392089844, 199.61093139648438, 463.80877685546875, 257.10528564453125, 549.43310546875, 35.39073944091797, 99.5096435546875, 714.3002319335938, 131.29452514648438, 222.5825958251953, 102.54104614257812, 101.34603881835938, 372.4716491699219, 142.34078979492188, 259.9051818847656, -139.53756713867188, -11.253807067871094, 289.24365234375, 286.7339782714844, -23.793577194213867, 69.78962707519531, 200.5648193359375, 435.216552734375, 93.36083984375, 251.993896484375, 540.9772338867188, 127.7118148803711, 178.11788940429688, 346.8121643066406, 313.30181884765625, 456.95904541015625, 25.91230010986328, 535.9947509765625, -2.3109512329101562, 524.6905517578125, -171.2229766845703, 103.08795166015625, 324.17156982421875, 109.26971435546875, 62.98982238769531, 122.31840515136719, 615.522216796875, 40.38377380371094, 468.75592041015625, 10.013700485229492, 51.25578308105469, 63.658599853515625, 7.222188949584961], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000637.npy"} +{"epoch": 0.9353891336270191, "step": 638, "batch_size": 64, "mean": 212.10604858398438, "std": 179.75425720214844, "min": -99.80333709716797, "p10": -2.6114883422851456, "median": 187.73301696777344, "p90": 444.01737976074224, "max": 666.4597778320312, "pos_frac": 0.890625, "sample": [624.567138671875, -93.170166015625, 23.4241943359375, 207.79794311523438, 239.2899627685547, 38.497711181640625, 128.2093048095703, 115.96722412109375, 294.5614013671875, 100.33511352539062, 276.9775390625, 666.4597778320312, 118.35478210449219, 507.7843933105469, 234.53050231933594, 183.6592559814453, 236.04310607910156, 146.71408081054688, 308.65399169921875, 157.58262634277344, 103.1744613647461, 37.68256378173828, 132.30108642578125, 217.09048461914062, 144.07115173339844, 427.4736633300781, 94.44741821289062, 191.80677795410156, 537.2674560546875, 365.2970886230469, 421.44049072265625, 418.6014404296875, 273.37841796875, 213.43954467773438, 311.72955322265625, 97.4294662475586, 8.320297241210938, -11.930095672607422, -7.296539306640625, -21.809642791748047, -97.07508850097656, 165.83224487304688, 304.6072692871094, 486.09136962890625, 12.120674133300781, -17.020448684692383, 433.4871520996094, 228.25340270996094, 223.19635009765625, 164.37179565429688, 65.93508911132812, 21.263914108276367, 343.3990478515625, 168.1254119873047, 314.29913330078125, 107.44677734375, 116.12027740478516, 114.70984649658203, 410.9671630859375, 634.75341796875, -99.80333709716797, 234.34561157226562, 448.53033447265625, 320.6749267578125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000638.npy"} +{"epoch": 0.9368575624082232, "step": 639, "batch_size": 64, "mean": 160.06594848632812, "std": 195.8649444580078, "min": -382.5997619628906, "p10": -79.38620300292968, "median": 173.70648956298828, "p90": 437.26673583984376, "max": 608.1993408203125, "pos_frac": 0.8125, "sample": [503.7920837402344, 18.52119255065918, 50.14503479003906, 311.7951354980469, 389.6470642089844, 7.6360931396484375, 326.8287048339844, 211.48223876953125, 2.0906753540039062, 224.96844482421875, 256.5979919433594, -114.80181884765625, 297.0916442871094, -23.505035400390625, 8.373458862304688, 108.51217651367188, -83.16346740722656, -151.7591552734375, 179.75991821289062, 28.939231872558594, 457.1045837402344, -0.8151760101318359, 392.4843444824219, 268.8179016113281, 243.2722930908203, -57.80782699584961, -187.0832061767578, 192.0811004638672, 94.0448989868164, 196.7289581298828, 437.6488037109375, 80.6082534790039, 54.03398895263672, -70.57258605957031, 460.28985595703125, 223.04859924316406, 140.6103057861328, 445.1388244628906, 169.3972930908203, 183.24166870117188, 178.01568603515625, 295.1606750488281, 354.6300048828125, 269.62762451171875, 608.1993408203125, -245.96986389160156, 154.36129760742188, 186.70108032226562, -110.06399536132812, 117.91033935546875, 110.8123779296875, 31.543136596679688, 312.7952880859375, 86.00244903564453, -382.5997619628906, -69.2173843383789, 91.51852416992188, 225.88539123535156, 221.81666564941406, 462.8720397949219, 125.69049072265625, 436.375244140625, 352.7012939453125, 154.228271484375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000639.npy"} +{"epoch": 0.9383259911894273, "step": 640, "batch_size": 64, "mean": 240.2373809814453, "std": 230.48500061035156, "min": -158.01272583007812, "p10": -40.59599914550781, "median": 222.8864974975586, "p90": 562.7909973144532, "max": 947.5530395507812, "pos_frac": 0.859375, "sample": [589.1386108398438, 274.9664611816406, -41.765869140625, 162.4958953857422, 147.5434112548828, 16.039642333984375, 470.4107666015625, 555.40087890625, 271.84228515625, 36.845359802246094, 510.74981689453125, 628.0315551757812, 368.723388671875, 286.3754577636719, 29.292194366455078, 148.86260986328125, 242.82591247558594, -112.94805908203125, -22.005752563476562, 259.24365234375, 22.51458740234375, 335.662353515625, 263.85479736328125, 96.08207702636719, 88.28436279296875, -74.60032653808594, 153.39993286132812, 345.31878662109375, 947.5530395507812, 394.53594970703125, 368.56201171875, 226.581787109375, 7.15191650390625, 340.2569580078125, 933.276611328125, 54.283302307128906, -158.01272583007812, 165.1238555908203, -45.0792236328125, 205.43798828125, 149.5564727783203, 277.3740234375, 135.30784606933594, 565.9581909179688, 243.4217071533203, 128.64883422851562, 219.1912078857422, 300.55426025390625, -45.28839111328125, 422.8502502441406, 652.2518310546875, 230.42153930664062, -37.866302490234375, 350.4986267089844, -61.94651794433594, 112.83204650878906, 159.32765197753906, 187.14108276367188, 131.2151641845703, 332.93011474609375, 343.0828857421875, 679.9683837890625, 233.05845642089844, 172.44741821289062], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000640.npy"} +{"epoch": 0.9397944199706314, "step": 641, "batch_size": 64, "mean": 170.12400817871094, "std": 210.6383056640625, "min": -266.2994079589844, "p10": -93.45084381103514, "median": 160.91754913330078, "p90": 406.84335632324235, "max": 986.9483642578125, "pos_frac": 0.8125, "sample": [95.80083465576172, 178.7285919189453, 281.26617431640625, 638.8883666992188, -128.87063598632812, 178.59347534179688, 212.6827392578125, 362.42047119140625, 168.085693359375, 6.0314483642578125, -24.327159881591797, 141.97824096679688, -100.48271179199219, -120.34061431884766, 986.9483642578125, 47.053775787353516, -20.022926330566406, -110.29106903076172, 17.064254760742188, 193.0494384765625, 264.7832336425781, 96.11994171142578, 89.100830078125, 195.6671600341797, 217.75576782226562, 129.62210083007812, -266.2994079589844, 63.53468322753906, 171.85772705078125, 112.63017272949219, 101.59196472167969, 222.82069396972656, 164.70672607421875, -77.04315185546875, 99.74140167236328, 184.46340942382812, 539.3836669921875, 158.6442413330078, 57.71376037597656, 163.19085693359375, 167.36163330078125, -71.59506225585938, 36.49871063232422, -151.8480224609375, 99.09558868408203, 73.77787780761719, -186.09266662597656, 140.6579132080078, 328.373291015625, 497.58697509765625, 477.1280517578125, 421.56890869140625, -9.8719482421875, 301.7022705078125, 249.52505493164062, 318.6780090332031, 343.3250732421875, 365.83966064453125, 150.1139373779297, 136.02197265625, 237.60708618164062, 357.74285888671875, 372.4837341308594, 538.0135498046875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000641.npy"} +{"epoch": 0.9412628487518355, "step": 642, "batch_size": 64, "mean": 202.28163146972656, "std": 242.1876983642578, "min": -213.87701416015625, "p10": -64.95450668334958, "median": 175.60137939453125, "p90": 498.7808410644533, "max": 1096.93017578125, "pos_frac": 0.796875, "sample": [164.82472229003906, -77.83607482910156, 225.68695068359375, 344.43292236328125, 182.08743286132812, -144.6363525390625, 456.86376953125, -3.391998291015625, -34.89751434326172, 185.5986328125, 83.90117645263672, 25.9627685546875, 302.9444580078125, -83.799560546875, 243.66006469726562, 650.6611938476562, 173.43963623046875, 393.41497802734375, -157.08953857421875, 60.796199798583984, -157.83212280273438, 13.13116455078125, 17.555259704589844, 69.45933532714844, 177.76312255859375, 253.57627868652344, 129.26382446289062, 214.6392822265625, 68.33497619628906, 245.25228881835938, 414.21832275390625, 181.9056396484375, 446.3160705566406, 627.2544555664062, -22.680877685546875, 261.7708435058594, 170.1641845703125, 407.0386962890625, -213.87701416015625, 302.1199645996094, 333.33978271484375, 168.97984313964844, 192.17221069335938, 516.7453002929688, 12.078433990478516, 676.1356811523438, 398.82867431640625, 549.8711547851562, -34.52925109863281, 204.3162841796875, 1096.93017578125, 226.67465209960938, -30.088333129882812, -18.182327270507812, 36.84565734863281, -148.3971405029297, 151.94334411621094, 708.8631591796875, 51.133026123046875, 99.04964447021484, 171.1531982421875, 384.658447265625, 162.47467041015625, 437.03070068359375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000642.npy"} +{"epoch": 0.9427312775330396, "step": 643, "batch_size": 64, "mean": 187.63221740722656, "std": 185.28456115722656, "min": -246.0331268310547, "p10": -64.29046707153319, "median": 172.54158782958984, "p90": 420.4883361816407, "max": 610.30322265625, "pos_frac": 0.875, "sample": [171.07843017578125, 178.0089874267578, 148.34625244140625, 428.15087890625, 253.33502197265625, 479.6251220703125, 369.5740051269531, -70.62541961669922, -123.38371276855469, 107.38575744628906, 7.8034515380859375, 85.94749450683594, 230.43783569335938, 111.920654296875, -89.84632873535156, 340.5643615722656, 402.60906982421875, 388.75323486328125, 335.3093566894531, 333.61077880859375, 187.5418243408203, 34.307098388671875, 64.42363739013672, 115.65030670166016, 37.69319534301758, 115.69091796875, -149.59083557128906, 260.8354797363281, 241.92852783203125, 307.829833984375, -246.0331268310547, 3.9499950408935547, -172.15362548828125, 349.5018310546875, 491.6728820800781, 513.2865600585938, 170.56808471679688, 70.3818359375, 168.9185791015625, 320.7330627441406, 127.51896667480469, 570.443603515625, 487.1767578125, 292.71392822265625, 208.29095458984375, 334.47967529296875, 4.260402679443359, 174.00474548339844, 610.30322265625, 119.62471008300781, 312.4907531738281, 308.25421142578125, 114.38856506347656, -49.5089111328125, 327.9207763671875, 33.91973114013672, 92.00846862792969, 20.038002014160156, 157.8038330078125, 212.56350708007812, 137.44439697265625, 296.7081298828125, -75.77156829833984, 215.644287109375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000643.npy"} +{"epoch": 0.9441997063142438, "step": 644, "batch_size": 64, "mean": 182.40570068359375, "std": 228.03553771972656, "min": -406.736572265625, "p10": -89.46753005981445, "median": 140.9361801147461, "p90": 514.1600158691407, "max": 732.505859375, "pos_frac": 0.828125, "sample": [650.2742919921875, -103.05843353271484, 416.16094970703125, -406.736572265625, 41.06227111816406, 279.542724609375, -158.05844116210938, 11.442092895507812, 200.23348999023438, 82.99459838867188, 29.284217834472656, 222.0054473876953, -146.7449951171875, 453.6593322753906, 114.33299255371094, 614.1153564453125, 148.247314453125, 138.35113525390625, 141.10818481445312, 57.201011657714844, 118.03125, 458.360107421875, 140.76417541503906, -64.74776458740234, 379.22003173828125, 459.9611511230469, -5.016838073730469, 43.54210662841797, 173.70220947265625, 557.7742309570312, 34.47101593017578, -89.96194458007812, 383.9934997558594, 92.71369934082031, 170.4317626953125, 51.07530975341797, 732.505859375, 104.69192504882812, 146.82826232910156, 198.703125, 35.333518981933594, -20.080299377441406, 376.4293212890625, 14.053489685058594, 333.9388732910156, 284.5560607910156, 82.07762145996094, -88.31389617919922, 85.17005920410156, 518.503662109375, 534.36669921875, 325.807861328125, -134.84739685058594, -118.09568786621094, 280.0333251953125, 504.02484130859375, 253.6387481689453, 165.19532775878906, 32.96662902832031, 95.11905670166016, 216.67959594726562, 659.896728515625, 94.59078979492188, 270.4596252441406], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000644.npy"} +{"epoch": 0.9456681350954479, "step": 645, "batch_size": 64, "mean": 151.87600708007812, "std": 225.63226318359375, "min": -319.826171875, "p10": -99.14875488281248, "median": 121.24322509765625, "p90": 427.2677185058594, "max": 815.3126831054688, "pos_frac": 0.734375, "sample": [36.405029296875, 345.1513977050781, -136.28060913085938, 411.92620849609375, -198.18370056152344, 61.55888366699219, -163.6839599609375, 132.65469360351562, 312.53631591796875, 60.23601150512695, -11.803722381591797, 28.93096923828125, 56.883827209472656, -6.007463455200195, -3.3992042541503906, 129.01553344726562, 209.32972717285156, -43.918487548828125, 536.9099731445312, 154.93536376953125, 303.5914611816406, -12.375434875488281, 119.13046264648438, 137.7696075439453, 95.53248596191406, 420.8627014160156, 403.63525390625, 158.60836791992188, -209.3719940185547, -319.826171875, 74.6194839477539, 130.65570068359375, -87.49920654296875, 364.6120300292969, 548.7095336914062, -9.838104248046875, 414.4939270019531, 430.0127258300781, -104.14141845703125, -27.07268524169922, 815.3126831054688, 123.35598754882812, 24.8424072265625, 13.777332305908203, 84.08497619628906, 280.92462158203125, 222.3773956298828, 232.3546905517578, 205.59347534179688, 200.93617248535156, 286.64093017578125, 13.395339965820312, 497.701904296875, 21.47655487060547, 528.9827880859375, 763.0114135742188, 112.92957305908203, 302.18017578125, -176.1848602294922, 199.8670654296875, 235.59735107421875, -25.7056941986084, 42.93568801879883, -31.629268646240234], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000645.npy"} +{"epoch": 0.947136563876652, "step": 646, "batch_size": 64, "mean": 189.32763671875, "std": 177.27403259277344, "min": -184.42095947265625, "p10": -3.514230346679674, "median": 175.60989379882812, "p90": 495.5795715332032, "max": 597.8863525390625, "pos_frac": 0.890625, "sample": [583.763427734375, 188.901611328125, 221.27584838867188, 276.76605224609375, 379.6044616699219, 175.00599670410156, 540.4391479492188, 330.46588134765625, 519.750732421875, 9.780044555664062, 191.40377807617188, 216.47669982910156, -184.42095947265625, 191.85287475585938, 181.41128540039062, 186.2559356689453, 183.51844787597656, 62.158058166503906, 479.05706787109375, 129.38525390625, 247.20770263671875, 130.82583618164062, 15.012260437011719, 189.96481323242188, 32.39663314819336, 164.52682495117188, 283.67694091796875, 225.3447723388672, -65.83840942382812, 285.184326171875, 165.65243530273438, 246.55233764648438, -38.922027587890625, 255.69557189941406, -9.211776733398438, 282.52020263671875, 40.2454833984375, 444.0407409667969, 147.7357940673828, -80.49485778808594, 184.57249450683594, 597.8863525390625, 110.00575256347656, 13.728809356689453, -86.17265319824219, -31.179702758789062, 100.58116149902344, 40.520286560058594, 64.29426574707031, 555.54931640625, 502.66064453125, 99.04035186767578, 218.84092712402344, 119.70585632324219, 362.1390075683594, 45.00129699707031, 117.57200622558594, 59.021339416503906, 176.2137908935547, 121.92765045166016, 81.16071319580078, 571.2699584960938, 94.58605194091797, 173.07669067382812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000646.npy"} +{"epoch": 0.9486049926578561, "step": 647, "batch_size": 64, "mean": 184.9781494140625, "std": 210.51329040527344, "min": -313.9056396484375, "p10": -67.70788192749022, "median": 179.26246643066406, "p90": 474.27304077148443, "max": 696.4304809570312, "pos_frac": 0.84375, "sample": [184.11280822753906, 341.4125061035156, 228.0560302734375, 140.61415100097656, 96.8248519897461, 403.580322265625, 413.9488830566406, 244.73284912109375, 287.244384765625, 30.19153594970703, 599.2477416992188, 167.7034912109375, 21.237091064453125, 529.9818725585938, 426.3487548828125, 218.25750732421875, 58.16227722167969, 481.7391662597656, 314.0860595703125, 53.829986572265625, 84.83377075195312, -57.19012451171875, 4.114213943481445, 8.650848388671875, 336.52069091796875, 342.2645568847656, -32.43351745605469, 696.4304809570312, -41.17628479003906, 522.8759765625, 95.65947723388672, 14.334709167480469, 128.01309204101562, 355.32867431640625, 504.57452392578125, 12.421096801757812, 198.11924743652344, 73.68083190917969, -179.93980407714844, -211.4499053955078, 321.292236328125, -137.6405029296875, 538.4166259765625, -94.79412841796875, -170.12753295898438, 113.04035186767578, 295.3373718261719, 176.16189575195312, 154.83560180664062, 347.49688720703125, 113.41239929199219, 182.363037109375, 456.8520812988281, 236.8404541015625, 68.96855163574219, 241.06065368652344, -313.9056396484375, 195.18692016601562, -72.21549224853516, 392.22772216796875, 220.66116333007812, 286.77301025390625, 34.6217155456543, 154.7911376953125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000647.npy"} +{"epoch": 0.9500734214390602, "step": 648, "batch_size": 64, "mean": 244.16232299804688, "std": 270.6612854003906, "min": -346.6820373535156, "p10": -62.45386734008789, "median": 227.82299041748047, "p90": 598.3100219726563, "max": 1067.294921875, "pos_frac": 0.828125, "sample": [85.6718521118164, 24.119522094726562, 63.822044372558594, 290.7289123535156, 130.50408935546875, 357.7978210449219, 544.7921142578125, 863.7838134765625, 234.21080017089844, 747.9654541015625, 67.38653564453125, -59.14757537841797, 145.37319946289062, 480.89630126953125, -63.870849609375, -139.7578887939453, 421.2977294921875, 177.9580841064453, 76.36576843261719, -211.75604248046875, -48.37904357910156, 311.118896484375, 137.04315185546875, 336.62945556640625, 39.66304016113281, 605.77783203125, 293.2330322265625, 212.1864471435547, 125.51734924316406, 59.282752990722656, 407.7447509765625, 48.08123016357422, -10.272117614746094, 1067.294921875, 448.193359375, 114.62195587158203, -83.53211975097656, 226.18443298339844, 100.11083984375, -159.72299194335938, 41.9202880859375, 241.9263458251953, 781.2511596679688, 273.9044189453125, 82.17730712890625, 429.357421875, 727.512451171875, -84.12506103515625, 245.6271209716797, 262.56243896484375, 466.13226318359375, 229.4615478515625, -11.885406494140625, 287.6460266113281, 436.3551330566406, 113.82164764404297, 338.0263977050781, 407.5718078613281, 195.19320678710938, 580.8851318359375, 456.0163879394531, 342.93450927734375, -346.6820373535156, 659.8800048828125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000648.npy"} +{"epoch": 0.9515418502202643, "step": 649, "batch_size": 64, "mean": 157.4774627685547, "std": 216.9331512451172, "min": -193.53952026367188, "p10": -59.20358009338379, "median": 88.38896560668945, "p90": 431.39488830566427, "max": 956.5933837890625, "pos_frac": 0.796875, "sample": [-59.1582145690918, 15.257667541503906, 103.1583480834961, -71.0357666015625, 386.9860534667969, 37.81733322143555, 36.62790298461914, 210.10821533203125, 310.1700439453125, 78.75030517578125, 380.8262939453125, -59.2230224609375, 88.51993560791016, 63.81321716308594, -27.285682678222656, -41.732879638671875, 6.7733917236328125, 450.42724609375, -191.32528686523438, 265.41278076171875, 164.596435546875, 57.901039123535156, -173.12945556640625, -17.97144889831543, 251.6746826171875, -90.9964828491211, 298.7698974609375, 30.18280029296875, -17.360668182373047, 294.93414306640625, 335.6865234375, -42.87822723388672, 68.18353271484375, 85.1829605102539, 473.96051025390625, 192.33856201171875, 353.7748107910156, 238.2694091796875, 386.8070068359375, 88.25799560546875, 737.5322265625, 8.135000228881836, 0.8077201843261719, 488.5226135253906, 227.4010009765625, 65.73859405517578, 365.707275390625, 956.5933837890625, 140.4156494140625, 185.16607666015625, 48.53923797607422, 91.51853942871094, 64.77900695800781, 474.941162109375, 70.86102294921875, 354.8818054199219, 462.1495056152344, -193.53952026367188, 10.427104949951172, 167.82330322265625, -174.0902862548828, 123.38339233398438, 354.3416442871094, 83.45037078857422], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000649.npy"} +{"epoch": 0.9530102790014684, "step": 650, "batch_size": 64, "mean": 223.4412841796875, "std": 231.90771484375, "min": -265.58062744140625, "p10": -74.22417945861815, "median": 211.3965072631836, "p90": 533.6227600097657, "max": 711.9141845703125, "pos_frac": 0.84375, "sample": [-4.398124694824219, -159.111083984375, 386.8988037109375, 416.18133544921875, 248.97903442382812, 370.4916076660156, 286.02984619140625, -60.30207443237305, 621.348876953125, 184.9087677001953, 533.3010864257812, -140.8423309326172, 711.9141845703125, 319.861083984375, 491.83331298828125, 418.9464111328125, 244.09571838378906, 614.6974487304688, 215.65074157714844, 704.6171264648438, 380.69073486328125, 104.24188995361328, 481.20196533203125, 254.04122924804688, 152.8882598876953, 112.05973815917969, 84.9012451171875, 77.7558364868164, 60.564483642578125, 279.0166320800781, 429.30670166015625, 156.7917938232422, -80.1907958984375, -136.9647979736328, 74.70999908447266, 539.518310546875, 20.139026641845703, 639.2340087890625, -49.68427658081055, 168.67263793945312, 207.14227294921875, 533.7606201171875, -242.63641357421875, 347.877197265625, 86.13436126708984, 259.2713928222656, 198.1632843017578, 369.1857604980469, 262.5638732910156, 493.82513427734375, 301.88665771484375, 71.4432144165039, 465.02313232421875, 178.2056884765625, 286.00738525390625, 100.85051727294922, 26.809524536132812, -265.58062744140625, -132.61569213867188, 129.21109008789062, 313.365966796875, 29.78748321533203, 61.11526107788086, 65.44806671142578], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000650.npy"} +{"epoch": 0.9544787077826725, "step": 651, "batch_size": 64, "mean": 230.05149841308594, "std": 197.73291015625, "min": -147.1245574951172, "p10": -13.961513710021968, "median": 232.8486785888672, "p90": 487.9522186279297, "max": 612.10693359375, "pos_frac": 0.859375, "sample": [396.27642822265625, 282.0247802734375, 433.492919921875, 612.10693359375, 21.012680053710938, 85.16497802734375, 236.55157470703125, 476.9375305175781, 154.09849548339844, 72.91865539550781, 433.04595947265625, 372.14349365234375, 190.51377868652344, 134.58966064453125, 180.759765625, 249.53765869140625, 58.66971206665039, 29.057266235351562, 9.400545120239258, 211.78245544433594, 488.6324462890625, -147.1245574951172, 601.1448974609375, 232.64962768554688, 369.1724853515625, 40.063941955566406, -0.8042583465576172, -89.8446044921875, 173.75537109375, 332.1452331542969, 150.89952087402344, -25.185020446777344, 322.18902587890625, -9.344755172729492, 606.710205078125, 344.7373046875, 382.123779296875, 184.30465698242188, 313.9063415527344, 5.312507629394531, 263.465576171875, 442.82830810546875, -84.15577697753906, 165.439697265625, -15.94012451171875, 345.4627380371094, 2.62298583984375, 591.716064453125, -30.761432647705078, 137.3173370361328, 93.2389144897461, 282.9241943359375, 586.0462646484375, 435.44342041015625, 233.0477294921875, -75.175537109375, 275.4951477050781, 97.19328308105469, 486.3650207519531, 277.0000305175781, 294.869873046875, 238.25106811523438, 587.495849609375, 177.57566833496094], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000651.npy"} +{"epoch": 0.9559471365638766, "step": 652, "batch_size": 64, "mean": 201.07382202148438, "std": 214.36643981933594, "min": -184.3486328125, "p10": -19.527482795715322, "median": 172.1037368774414, "p90": 505.04883422851566, "max": 750.6529541015625, "pos_frac": 0.84375, "sample": [119.39434814453125, 133.59271240234375, 494.20135498046875, -184.3486328125, 52.03859329223633, 228.98504638671875, 244.27354431152344, -27.096107482910156, 379.1990966796875, 131.73001098632812, 413.73193359375, 517.4052124023438, 548.2177734375, -37.38191223144531, 28.77178192138672, 15.708099365234375, -6.375030517578125, 457.1795654296875, 638.0401611328125, 42.74907684326172, 251.08836364746094, 96.1069107055664, 48.126792907714844, 215.8665771484375, 177.9528045654297, -7.77911376953125, 204.46502685546875, 218.3454132080078, 129.1629638671875, 146.82818603515625, 150.09385681152344, 714.85546875, -111.3759994506836, 687.1548461914062, 239.71273803710938, 313.0727844238281, 206.6119842529297, -32.5447998046875, -24.562498092651367, 20.019256591796875, 207.17164611816406, 25.38445281982422, 5.608188629150391, 5.461631774902344, 367.9622802734375, 263.5119934082031, 502.4102478027344, 13.361228942871094, 310.333251953125, 166.25466918945312, 750.6529541015625, 9.472728729248047, 486.73321533203125, 96.6902084350586, 18.517887115478516, 74.23731994628906, 506.1796569824219, -112.58601379394531, 227.44647216796875, 341.2975769042969, 198.87437438964844, 281.67083740234375, 293.1866760253906, -4.326488494873047], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000652.npy"} +{"epoch": 0.9574155653450808, "step": 653, "batch_size": 64, "mean": 215.99537658691406, "std": 220.5646209716797, "min": -217.56024169921875, "p10": -34.73289794921874, "median": 189.85691833496094, "p90": 516.0339996337891, "max": 817.0328979492188, "pos_frac": 0.859375, "sample": [277.80084228515625, 6.672660827636719, 59.28367614746094, 817.0328979492188, 333.7984619140625, -86.84122467041016, 56.76982116699219, 52.8780517578125, 15.73748779296875, 328.98175048828125, 666.4010009765625, 196.0928192138672, -217.56024169921875, 500.11639404296875, 267.3002624511719, 374.1707763671875, 372.5052185058594, 328.439208984375, 537.41845703125, 113.9781494140625, 47.75851821899414, 223.86228942871094, 509.0602111816406, 194.20831298828125, 205.0811004638672, 253.14024353027344, 475.44244384765625, 131.49786376953125, 187.7022705078125, -114.7273941040039, 469.28338623046875, 210.55921936035156, 503.89306640625, -77.04013061523438, 247.27645874023438, 97.68876647949219, 110.6932373046875, 331.0965881347656, -57.705955505371094, 571.487060546875, 729.6492309570312, 415.24371337890625, -38.96870422363281, 7.004508972167969, 555.6480712890625, 318.7415771484375, 16.923385620117188, 117.36585998535156, 174.603515625, 114.34033203125, -4.801300048828125, 102.51155090332031, 519.0227661132812, 148.89907836914062, 221.8614044189453, 112.95396423339844, 55.265586853027344, 10.2557373046875, 383.26025390625, 176.3228302001953, 192.01156616210938, -115.92744445800781, 115.13177490234375, -24.849349975585938], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000653.npy"} +{"epoch": 0.9588839941262849, "step": 654, "batch_size": 64, "mean": 180.4493408203125, "std": 248.49661254882812, "min": -285.9581298828125, "p10": -121.43778610229492, "median": 154.3584747314453, "p90": 460.8765991210938, "max": 1071.8570556640625, "pos_frac": 0.75, "sample": [-9.236946105957031, 486.83184814453125, -144.84112548828125, 322.74786376953125, 120.15849304199219, 43.68574523925781, -181.3197021484375, 154.316650390625, -104.88352966308594, 417.49993896484375, 154.40029907226562, 159.01663208007812, 135.70376586914062, 47.51007080078125, 106.05267333984375, -122.20569610595703, 111.65845489501953, 580.5069580078125, 315.34808349609375, 389.8509216308594, 241.84156799316406, 1071.8570556640625, 217.48448181152344, 116.957275390625, 223.75543212890625, 300.94091796875, -95.69252014160156, 232.00921630859375, 100.20394897460938, -43.17472839355469, -8.215530395507812, 342.2994689941406, 298.725830078125, 237.40048217773438, 32.692138671875, 20.98938751220703, 604.5279541015625, -149.17568969726562, 44.469581604003906, 330.7279052734375, 467.474365234375, 347.91259765625, 441.1328125, -33.280662536621094, -285.9581298828125, 390.1595458984375, 152.46368408203125, 387.6236267089844, 445.4818115234375, 99.24412536621094, 189.08856201171875, 468.0475158691406, 405.7512512207031, 24.390968322753906, 181.25714111328125, -119.64599609375, 281.569580078125, -123.31332397460938, 788.45263671875, 236.64505004882812, 71.04684448242188, -98.82421875, -256.5241394042969, -14.863533020019531], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000654.npy"} +{"epoch": 0.960352422907489, "step": 655, "batch_size": 64, "mean": 187.52426147460938, "std": 234.6633758544922, "min": -259.3392028808594, "p10": -86.63518753051757, "median": 141.1035385131836, "p90": 552.2408386230469, "max": 736.0499267578125, "pos_frac": 0.828125, "sample": [-99.86981201171875, -61.46065139770508, 706.5086669921875, 123.71652221679688, 32.03900909423828, 690.7723999023438, 12.71331787109375, 436.04278564453125, 21.700407028198242, 2.6163578033447266, 286.36993408203125, 15.369295120239258, 62.53424072265625, -259.3392028808594, 114.8009262084961, 212.870849609375, 16.394386291503906, 736.0499267578125, 44.40385437011719, 465.3730163574219, 160.8709716796875, 219.74566650390625, 537.0869750976562, 183.98397827148438, 22.69801902770996, 119.09626007080078, 239.14305114746094, 231.28201293945312, -198.41712951660156, 196.782958984375, -74.9361801147461, 622.8472900390625, 165.1442108154297, 66.11155700683594, 379.3702392578125, 265.1930236816406, -139.70492553710938, 350.527099609375, 313.9530029296875, 304.2586975097656, 48.106292724609375, 24.212135314941406, 336.1145935058594, 656.2172241210938, 558.7353515625, 140.37669372558594, 139.05184936523438, -110.17463684082031, 343.74786376953125, 292.49871826171875, -24.252365112304688, -91.6490478515625, 141.83038330078125, 5.621257781982422, 521.5594482421875, 33.0279426574707, 624.793212890625, -104.54499816894531, -6.3860015869140625, 321.50640869140625, 93.65021514892578, 204.2560577392578, 104.05728149414062, 224.55276489257812], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000655.npy"} +{"epoch": 0.9618208516886931, "step": 656, "batch_size": 64, "mean": 277.0999450683594, "std": 201.01937866210938, "min": -55.363372802734375, "p10": 34.408545684814456, "median": 248.06629180908203, "p90": 574.5714843750002, "max": 755.751220703125, "pos_frac": 0.9375, "sample": [755.751220703125, 340.34527587890625, -15.812355041503906, 19.796306610107422, 103.26514434814453, 279.63946533203125, 119.91542053222656, 419.99273681640625, 198.69627380371094, 380.1402587890625, 269.1612548828125, 282.58782958984375, 232.45619201660156, 227.02371215820312, 154.10830688476562, 442.8541564941406, 105.11065673828125, 173.92996215820312, 492.33990478515625, 211.02963256835938, 398.6091003417969, 21.0758056640625, 534.7813720703125, -50.98431396484375, -55.363372802734375, 31.69818878173828, 392.7061767578125, 590.5460815429688, 370.98919677734375, 636.204833984375, 162.20909118652344, 669.108642578125, 344.6318054199219, -19.867877960205078, 637.193603515625, 336.06744384765625, 373.91802978515625, 266.2573547363281, 507.3448791503906, 112.872314453125, 687.599853515625, 171.7727813720703, 537.2974243164062, 189.78021240234375, 263.6763916015625, 466.75506591796875, 169.28851318359375, 629.7605590820312, 83.63250732421875, 357.6956481933594, 383.57904052734375, 178.8526153564453, 63.64781951904297, 116.65946960449219, 493.1126708984375, 211.89535522460938, 40.73271179199219, 44.5804443359375, 99.70447540283203, 79.57795715332031, 328.16705322265625, 232.13082885742188, 156.6778564453125, 295.4896545410156], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000656.npy"} +{"epoch": 0.9632892804698973, "step": 657, "batch_size": 64, "mean": 255.2210235595703, "std": 213.33184814453125, "min": -197.03720092773438, "p10": 31.710523986816412, "median": 239.14884185791016, "p90": 512.2172363281251, "max": 867.8160400390625, "pos_frac": 0.9375, "sample": [115.76577758789062, 9.103866577148438, 73.67535400390625, 275.50079345703125, 533.312744140625, 867.8160400390625, 283.4982604980469, 89.40180969238281, 402.43914794921875, -37.6756591796875, -115.13104248046875, 178.55580139160156, 326.609375, 38.06050109863281, 290.08575439453125, 336.655517578125, 28.989105224609375, 317.4761657714844, 161.52191162109375, 519.402587890625, 492.3364562988281, 400.73175048828125, 129.00814819335938, 798.4364013671875, 388.89715576171875, 656.3406372070312, 106.29472351074219, 157.40899658203125, 235.4693603515625, 203.2266082763672, 194.01214599609375, 300.31085205078125, 19.85301971435547, -45.71046447753906, 373.6634521484375, 454.26220703125, 283.96710205078125, 85.02845001220703, 143.5672607421875, 134.37994384765625, 244.07418823242188, 94.14480590820312, 52.513633728027344, 242.86962890625, 664.00830078125, 116.9625473022461, 197.6405487060547, 315.07867431640625, 359.054443359375, 150.7528076171875, 495.451416015625, 366.01275634765625, 305.1728515625, 364.97265625, 238.00994873046875, 147.28359985351562, 360.7832946777344, 826.5579833984375, 240.28773498535156, -197.03720092773438, 61.089988708496094, 45.61195373535156, 170.2801513671875, 266.0218505859375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000657.npy"} +{"epoch": 0.9647577092511013, "step": 658, "batch_size": 64, "mean": 200.5369873046875, "std": 180.16241455078125, "min": -187.98873901367188, "p10": -46.97070999145505, "median": 191.05763244628906, "p90": 403.8044647216797, "max": 731.3876953125, "pos_frac": 0.859375, "sample": [243.59597778320312, 247.39443969726562, 152.77352905273438, -90.57917785644531, 5.862140655517578, 143.3468017578125, 383.73687744140625, 176.7501678466797, -19.487533569335938, 342.2554626464844, 273.58270263671875, 597.5596923828125, 78.62059020996094, 296.7287292480469, 144.47396850585938, -13.045761108398438, 34.08878707885742, 128.81427001953125, 99.31961822509766, -138.72962951660156, 364.5753479003906, 35.80156707763672, 231.06382751464844, 119.51681518554688, -97.97447204589844, 404.5793762207031, 134.75326538085938, 321.4144592285156, 453.91436767578125, -81.6580581665039, 43.68861389160156, 77.70526123046875, 97.6959228515625, 401.996337890625, 189.5314178466797, 731.3876953125, 55.50048065185547, 302.0085144042969, 179.68692016601562, 134.62918090820312, 192.58384704589844, 246.18972778320312, 431.8006286621094, 270.0964660644531, 304.18438720703125, 325.5565185546875, 362.1897888183594, 255.51644897460938, 284.9960632324219, 401.7200927734375, 215.81578063964844, 312.9141845703125, -76.71795654296875, 467.9768981933594, 189.01907348632812, 241.24343872070312, 320.43182373046875, 527.082275390625, -187.98873901367188, 243.15953063964844, 137.4220733642578, 115.5086441040039, 125.53716278076172, -58.74921417236328], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000658.npy"} +{"epoch": 0.9662261380323054, "step": 659, "batch_size": 64, "mean": 215.85671997070312, "std": 209.70188903808594, "min": -190.11407470703125, "p10": -78.12952651977538, "median": 212.80684661865234, "p90": 512.1656280517578, "max": 680.02587890625, "pos_frac": 0.828125, "sample": [343.8583679199219, -88.51754760742188, 147.57269287109375, 329.52734375, 218.5227508544922, 119.29759216308594, 613.67626953125, 82.40380859375, -165.82984924316406, 45.10206604003906, 680.02587890625, 127.99272918701172, 349.1449279785156, -98.34684753417969, 253.1240997314453, 62.26378631591797, 260.28173828125, 265.3046569824219, 188.34619140625, 176.16818237304688, 130.82952880859375, 208.83865356445312, -111.78562927246094, 667.9763793945312, 73.9850082397461, 654.9569091796875, 316.0902404785156, 42.459495544433594, 563.728271484375, 293.4295654296875, 231.64981079101562, 234.3502197265625, -0.8464946746826172, 188.7173309326172, 320.6605224609375, 237.99732971191406, 362.03863525390625, 107.14291381835938, 31.220596313476562, 499.87921142578125, 100.38502502441406, -32.994895935058594, 358.1173095703125, -55.4322395324707, 513.2608642578125, -79.068603515625, -75.93834686279297, 141.70135498046875, 365.65814208984375, 552.9315185546875, 476.00689697265625, 240.9506378173828, 159.41552734375, 276.8338623046875, 270.3324890136719, 157.4106903076172, 390.53167724609375, 336.7508239746094, 509.6100769042969, -87.57246398925781, 216.77503967285156, 140.9616241455078, -190.11407470703125, 165.07958984375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000659.npy"} +{"epoch": 0.9676945668135095, "step": 660, "batch_size": 64, "mean": 240.26258850097656, "std": 193.74392700195312, "min": -248.89501953125, "p10": 19.15821189880371, "median": 238.84903717041016, "p90": 451.9656646728516, "max": 837.3460693359375, "pos_frac": 0.90625, "sample": [93.05894470214844, 209.1713104248047, 284.1724853515625, 324.3511047363281, 18.978012084960938, 253.4054718017578, 352.3712158203125, 234.2387237548828, 205.74346923828125, 393.0523376464844, 130.88369750976562, 19.578678131103516, 301.2532653808594, 302.26171875, 262.18890380859375, 631.0958862304688, 146.51344299316406, -12.166946411132812, 254.1867218017578, 204.27194213867188, 254.34512329101562, 309.12713623046875, 213.89205932617188, 329.758056640625, 428.252197265625, 651.8312377929688, 54.902732849121094, -49.44731903076172, 346.1275634765625, 341.6935119628906, 76.5625228881836, 519.1892700195312, -45.13006591796875, 755.51318359375, 99.16561889648438, 301.1444091796875, 386.3872375488281, 91.95964050292969, 239.33160400390625, 230.1561737060547, 158.29124450683594, 238.49510192871094, 335.0459289550781, 251.85406494140625, 88.79238891601562, 133.81198120117188, 142.31890869140625, 837.3460693359375, 481.27813720703125, 310.14337158203125, 47.57081604003906, 222.45962524414062, -136.0136260986328, 453.2152404785156, 449.04998779296875, 77.54545593261719, 212.53744506835938, 194.78546142578125, 239.20297241210938, 364.11370849609375, 245.73065185546875, -248.89501953125, -35.87609100341797, 170.63043212890625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000660.npy"} +{"epoch": 0.9691629955947136, "step": 661, "batch_size": 64, "mean": 215.5369873046875, "std": 224.0557861328125, "min": -373.63427734375, "p10": -10.4930362701416, "median": 212.36197662353516, "p90": 473.9339447021485, "max": 843.7229614257812, "pos_frac": 0.875, "sample": [146.41786193847656, 42.88475799560547, 334.59075927734375, 17.40711784362793, 239.69862365722656, -9.043163299560547, 272.21978759765625, -137.03787231445312, 303.5181579589844, 351.192138671875, 477.7587890625, 344.86859130859375, 37.83125305175781, 607.615478515625, 843.7229614257812, 240.2265625, 230.77227783203125, 36.94573974609375, -11.114410400390625, 90.90573120117188, 234.0057830810547, 358.58795166015625, 147.98410034179688, 118.79061889648438, 27.37151527404785, 350.1729736328125, 127.55052185058594, 352.7639465332031, 690.2147216796875, 28.599075317382812, -88.44221496582031, 379.34912109375, 37.91070556640625, 513.8236083984375, 202.2251434326172, 128.25979614257812, 268.1425476074219, 207.38858032226562, 205.45751953125, 160.63958740234375, -132.56187438964844, 185.0229034423828, 22.183303833007812, 397.09844970703125, 248.17437744140625, -189.41357421875, 301.2527160644531, 277.23077392578125, 19.61122703552246, 379.80828857421875, 633.9257202148438, -373.63427734375, 145.7464141845703, -18.026689529418945, 373.3932189941406, 307.7130432128906, 72.7519760131836, 465.0093078613281, 29.962465286254883, 739.9600830078125, 430.92742919921875, 307.7751159667969, 217.3353729248047, 40.945106506347656], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000661.npy"} +{"epoch": 0.9706314243759178, "step": 662, "batch_size": 64, "mean": 189.50390625, "std": 210.74281311035156, "min": -249.718505859375, "p10": -104.35340347290035, "median": 182.49099731445312, "p90": 486.97780456542966, "max": 666.0914306640625, "pos_frac": 0.796875, "sample": [99.17557525634766, 245.89138793945312, 666.0914306640625, -249.718505859375, 464.5307922363281, 133.5683135986328, 263.73004150390625, 158.58917236328125, 606.2396240234375, 136.21945190429688, 434.3280334472656, -49.05267333984375, -182.9694366455078, 155.27513122558594, 276.1632385253906, 87.45171356201172, 156.10472106933594, 123.64219665527344, 620.3301391601562, 213.62164306640625, 554.8056640625, 33.41607666015625, 266.16339111328125, 102.21987915039062, 130.7418212890625, 271.40283203125, 142.780029296875, 443.45977783203125, 96.50045776367188, -5.61517333984375, 201.8182373046875, -167.9674072265625, 200.28416442871094, -134.1566162109375, 222.5934600830078, 369.5362243652344, 149.7958984375, 323.8348693847656, 377.9135437011719, 303.10638427734375, 31.0123291015625, 176.27415466308594, 486.5639953613281, 505.08538818359375, 266.1583557128906, 405.445068359375, -201.54823303222656, 207.91000366210938, 188.7078399658203, 158.22312927246094, 113.43033599853516, -16.018985748291016, -11.519363403320312, 198.6378173828125, -118.00899505615234, -35.111427307128906, 491.704833984375, 487.1551513671875, 230.513916015625, 260.4500427246094, 263.4366455078125, -136.3475341796875, -72.4903564453125, 6.7412567138671875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000662.npy"} +{"epoch": 0.9720998531571219, "step": 663, "batch_size": 64, "mean": 216.28622436523438, "std": 222.6233673095703, "min": -309.46478271484375, "p10": -19.3818977355957, "median": 185.84868621826172, "p90": 518.5176513671876, "max": 1008.6831665039062, "pos_frac": 0.875, "sample": [470.24798583984375, 228.4047393798828, 377.66607666015625, 52.667694091796875, 349.30712890625, 505.42852783203125, 98.28520202636719, 37.649078369140625, -17.64031982421875, 203.08566284179688, 107.50041961669922, 73.28408813476562, 524.3504028320312, 388.59442138671875, 47.67021179199219, 304.96539306640625, 278.1474304199219, 250.3929901123047, -162.95742797851562, 34.32464599609375, 123.12421417236328, 272.36767578125, 16.719730377197266, 180.1549072265625, 275.531982421875, 225.42724609375, 273.3645935058594, 100.89450073242188, 316.6328125, 1008.6831665039062, -23.90266227722168, 571.562255859375, 712.0941162109375, -26.694686889648438, 123.5751953125, 563.8997192382812, 456.1930847167969, 130.8081817626953, 47.36224365234375, 489.81048583984375, 513.146484375, 203.37118530273438, 225.835205078125, 105.04607391357422, 227.677001953125, 191.54246520996094, 139.6951141357422, 115.75465393066406, 136.25726318359375, 607.1748046875, 301.6785583496094, 203.3538818359375, 520.819580078125, -309.46478271484375, 298.7106628417969, 20.967967987060547, -37.376983642578125, -25.19114112854004, 169.80068969726562, -20.12828826904297, 19.008193969726562, 122.2840805053711, 10.072967529296875, 113.29871368408203], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000663.npy"} +{"epoch": 0.973568281938326, "step": 664, "batch_size": 64, "mean": 238.5847625732422, "std": 244.26019287109375, "min": -639.4927368164062, "p10": 5.569326019287111, "median": 220.1680908203125, "p90": 551.6020263671875, "max": 842.2698974609375, "pos_frac": 0.90625, "sample": [560.4552612304688, 171.26974487304688, 176.63522338867188, 586.7753295898438, 396.6155090332031, 394.4172058105469, 176.35263061523438, 533.1785278320312, 5.108497619628906, 548.1185302734375, -167.41900634765625, 543.124755859375, 256.53106689453125, 61.148048400878906, 697.4086303710938, 125.49076843261719, -58.144287109375, 312.61328125, 118.59037017822266, 278.89813232421875, 204.08468627929688, 404.0287170410156, 6.64459228515625, 576.9885864257812, -72.34292602539062, 66.01034545898438, 636.2540893554688, 18.010013580322266, 72.27024841308594, -134.41183471679688, 526.0150146484375, 236.25149536132812, 83.1326904296875, 241.46597290039062, 182.90023803710938, 327.3023681640625, -639.4927368164062, 151.99948120117188, 503.587646484375, 271.1911315917969, 67.86293029785156, 283.4335632324219, 200.131103515625, 161.5206298828125, 480.8816223144531, 842.2698974609375, 257.18603515625, 91.2805404663086, 552.6583251953125, 113.96270751953125, 65.70437622070312, 549.1373291015625, 318.3724365234375, 60.11991882324219, 247.62478637695312, 257.6827087402344, 239.79025268554688, 59.28769302368164, 124.79646301269531, 34.921722412109375, 282.1672668457031, 159.79629516601562, 493.6960144042969, -53.918296813964844], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000664.npy"} +{"epoch": 0.9750367107195301, "step": 665, "batch_size": 64, "mean": 212.38864135742188, "std": 197.04576110839844, "min": -151.5409393310547, "p10": -4.725917053222654, "median": 164.99365234375, "p90": 441.0594055175781, "max": 883.9987182617188, "pos_frac": 0.859375, "sample": [22.52051544189453, 117.59878540039062, 223.6453857421875, 110.31869506835938, 135.32180786132812, 355.6841125488281, 357.33856201171875, 97.614501953125, 490.1099853515625, 165.8376922607422, 335.3275146484375, 354.34197998046875, -2.1646728515625, 297.1141357421875, 146.70309448242188, -53.568206787109375, -65.452392578125, 162.01577758789062, 164.1496124267578, 303.77447509765625, 268.6643371582031, 84.89521789550781, 332.708740234375, 276.2417297363281, -62.12522888183594, 98.18358612060547, -151.5409393310547, 442.51580810546875, 445.9633483886719, -2.0151100158691406, 41.38072967529297, 139.2833251953125, 420.015625, 279.5614013671875, 423.7835388183594, 107.56214141845703, 437.6611328125, 433.39349365234375, 50.910911560058594, 166.33758544921875, 125.98727416992188, 504.8291931152344, 430.2130126953125, -125.34820556640625, 179.64862060546875, 611.4617919921875, -16.273117065429688, 159.55682373046875, 256.30413818359375, 280.9985046386719, 92.99507904052734, 263.9017028808594, 328.18646240234375, 99.29656982421875, 663.0098876953125, 19.351974487304688, 59.83805847167969, 59.340171813964844, 196.27835083007812, 373.5907897949219, 883.9987182617188, -5.8235931396484375, 117.43511962890625, 82.48367309570312], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000665.npy"} +{"epoch": 0.9765051395007343, "step": 666, "batch_size": 64, "mean": 210.448974609375, "std": 184.1608123779297, "min": -275.409912109375, "p10": -10.846952056884758, "median": 236.49000549316406, "p90": 411.0364044189454, "max": 657.0706787109375, "pos_frac": 0.875, "sample": [308.0421142578125, 63.55003356933594, 394.78839111328125, 256.7340393066406, -91.21871948242188, 245.1796112060547, 395.713623046875, 331.6983947753906, 76.3132095336914, 534.922119140625, -13.821022033691406, 133.53469848632812, 192.59762573242188, 30.35802459716797, 68.73583221435547, 116.56636047363281, 416.81884765625, 72.29784393310547, -18.0289306640625, 241.20016479492188, 523.46826171875, 146.93386840820312, 322.32086181640625, 262.465087890625, 657.0706787109375, -3.9074554443359375, 379.0577392578125, 117.20906066894531, 3.0734481811523438, -275.409912109375, 281.3290100097656, 244.5028533935547, 231.77984619140625, 510.73248291015625, 365.4489440917969, 212.09091186523438, 381.5937805175781, -95.54208374023438, 146.98292541503906, 317.4851989746094, 67.10946655273438, 90.5077896118164, -104.30614471435547, 504.51043701171875, 241.70787048339844, 346.04608154296875, -199.79745483398438, 478.9071960449219, 180.6829071044922, 116.671630859375, 315.1097412109375, 313.177978515625, 168.10751342773438, 86.0501708984375, 79.67230987548828, 91.85061645507812, 163.41522216796875, 303.3704528808594, 397.5440368652344, 262.8309326171875, 97.16130828857422, 335.6383972167969, 272.4544677734375, 375.6438293457031], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000666.npy"} +{"epoch": 0.9779735682819384, "step": 667, "batch_size": 64, "mean": 210.88682556152344, "std": 168.20181274414062, "min": -189.2760009765625, "p10": 8.052841949462891, "median": 202.95575714111328, "p90": 485.7379486083985, "max": 565.02099609375, "pos_frac": 0.921875, "sample": [160.20980834960938, -7.442756652832031, 214.5915069580078, -189.2760009765625, 7.894073486328125, 321.1673278808594, 225.08370971679688, 209.381591796875, 269.88873291015625, 363.96270751953125, 190.4865264892578, 92.61227416992188, 83.60565185546875, 196.9795379638672, 87.06342315673828, 66.9258804321289, 280.85614013671875, 162.4248046875, 538.1019287109375, 155.7869415283203, 305.4777526855469, 130.22500610351562, 470.9209899902344, 209.86041259765625, 275.404052734375, 101.7769775390625, 373.7989501953125, 124.02985382080078, 280.7568664550781, 411.84332275390625, 187.41610717773438, 236.22369384765625, 492.08807373046875, 318.0165710449219, 301.35723876953125, 10.517181396484375, 75.74444580078125, 206.09071350097656, 272.2483825683594, 12.083732604980469, -72.19441223144531, 565.02099609375, 511.37835693359375, 7.745857238769531, 506.93280029296875, 533.2022094726562, 263.1767883300781, 101.45186614990234, 212.35879516601562, 290.2676086425781, 8.423301696777344, 315.1063232421875, -88.76142883300781, 456.5992431640625, 500.1366882324219, 199.82080078125, 70.90167999267578, 186.15444946289062, 48.080894470214844, 116.22604370117188, -15.342239379882812, 119.79559326171875, 168.14190673828125, 265.94891357421875], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000667.npy"} +{"epoch": 0.9794419970631424, "step": 668, "batch_size": 64, "mean": 181.56585693359375, "std": 224.90020751953125, "min": -289.83624267578125, "p10": -65.5842430114746, "median": 174.7934112548828, "p90": 450.6506225585938, "max": 860.1803588867188, "pos_frac": 0.8125, "sample": [347.6592712402344, 163.7660675048828, 773.4384765625, -289.83624267578125, 29.863842010498047, 199.25985717773438, 278.1476135253906, 156.89109802246094, 420.7977294921875, 195.47695922851562, 70.45362854003906, -188.20460510253906, 86.40795135498047, 114.01565551757812, 270.42401123046875, 184.34815979003906, 175.7165069580078, 6.7532806396484375, -170.83990478515625, 12.606243133544922, 47.35961151123047, 345.0230407714844, -68.48627471923828, 334.394775390625, 238.33462524414062, 164.80426025390625, 48.436187744140625, 40.017425537109375, 13.961898803710938, 347.21978759765625, 256.1815490722656, 270.4308166503906, 239.58734130859375, -58.812835693359375, 485.5888977050781, 173.8703155517578, -249.63909912109375, 442.3076171875, 198.95614624023438, -15.768255233764648, 283.4886169433594, 293.59686279296875, 485.33135986328125, 132.8571319580078, -215.6239013671875, 557.1276245117188, 146.93096923828125, -125.89857482910156, 159.84765625, -7.431051254272461, 191.84402465820312, 0.5021286010742188, 141.96673583984375, 280.6355895996094, -37.985809326171875, 416.2891845703125, 320.8626403808594, 317.7890625, 860.1803588867188, 228.6190643310547, -2.29837703704834, 454.2261962890625, 40.79241943359375, 605.651611328125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000668.npy"} +{"epoch": 0.9809104258443465, "step": 669, "batch_size": 64, "mean": 235.43093872070312, "std": 223.2918701171875, "min": -339.2726745605469, "p10": -32.40299758911128, "median": 247.85086822509766, "p90": 512.5798645019531, "max": 890.4578247070312, "pos_frac": 0.890625, "sample": [241.87071228027344, 34.615478515625, 33.363990783691406, 515.505615234375, -255.96847534179688, 552.9152221679688, 268.11566162109375, 340.7945251464844, 112.94263458251953, -74.46610260009766, 890.4578247070312, -53.66611099243164, 17.2109375, 226.80555725097656, -59.71630096435547, 256.5859069824219, 505.75311279296875, -79.11717987060547, 112.31067657470703, 257.71112060546875, 300.67340087890625, 255.4932098388672, 582.691162109375, 221.79615783691406, 251.5776824951172, 490.1866455078125, 160.927978515625, 450.468017578125, 561.8748779296875, 78.81094360351562, 276.0255126953125, 690.8767700195312, 319.3158264160156, 454.1521911621094, 536.7244262695312, -339.2726745605469, 57.627872467041016, 258.83349609375, 482.1073913574219, 244.12405395507812, 128.94158935546875, 218.78652954101562, 57.21539306640625, 289.5600280761719, -284.6728515625, 297.4924621582031, 165.89686584472656, 459.8333740234375, 227.93222045898438, 113.57827758789062, 27.230552673339844, 271.3161315917969, 115.87277221679688, 188.19863891601562, 290.9233703613281, 397.631591796875, 57.571044921875, 261.1757507324219, 344.1536865234375, 385.7154541015625, 139.69509887695312, 355.415283203125, 151.21640014648438, 227.861572265625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000669.npy"} +{"epoch": 0.9823788546255506, "step": 670, "batch_size": 64, "mean": 252.3150634765625, "std": 215.02674865722656, "min": -194.6228790283203, "p10": 12.395456886291504, "median": 236.8810806274414, "p90": 579.9424560546876, "max": 669.4306030273438, "pos_frac": 0.921875, "sample": [175.94528198242188, 360.13360595703125, 187.6103515625, 581.2470092773438, 221.7383270263672, 12.209478378295898, -152.97616577148438, 3.068510055541992, 403.6255187988281, 542.3219604492188, 253.0212860107422, 123.12384033203125, -194.6228790283203, 601.9380493164062, 495.20489501953125, 663.4459838867188, 327.2867431640625, 172.46749877929688, 496.0530090332031, 83.66439819335938, 180.3867950439453, -126.93611907958984, 608.351806640625, 47.87963104248047, 300.10333251953125, 284.5395202636719, 234.97433471679688, 136.20938110351562, 302.13885498046875, 37.240142822265625, 465.6418762207031, 543.2235107421875, 57.12138366699219, 337.94122314453125, 294.60394287109375, 669.4306030273438, 469.9674072265625, 21.432300567626953, 251.04983520507812, 442.74334716796875, -8.074962615966797, 576.8984985351562, 167.11143493652344, 113.375244140625, 618.4825439453125, 331.53363037109375, 248.6724395751953, 238.78782653808594, -128.33912658691406, 38.09351348876953, 142.57968139648438, 621.279052734375, 194.35321044921875, 31.396728515625, 362.03021240234375, 43.71813201904297, 147.73977661132812, 287.1593017578125, 151.11849975585938, 371.09014892578125, 153.1345672607422, 323.9218444824219, 12.82940673828125, 194.7220458984375], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000670.npy"} +{"epoch": 0.9838472834067548, "step": 671, "batch_size": 64, "mean": 200.64739990234375, "std": 207.5231170654297, "min": -291.468017578125, "p10": -82.95695877075195, "median": 207.53913116455078, "p90": 508.5959167480469, "max": 661.8035888671875, "pos_frac": 0.859375, "sample": [21.440109252929688, -81.48747253417969, 178.83908081054688, 230.415283203125, -88.86859130859375, 413.6890869140625, 216.40200805664062, 91.22572326660156, 91.95941162109375, 375.8087158203125, 213.33050537109375, 59.901573181152344, 314.07086181640625, 303.5915832519531, 259.9033203125, 11.161575317382812, 102.69947814941406, 77.09217071533203, 404.9666748046875, 288.2181396484375, 7.198095321655273, 603.7454833984375, -291.468017578125, 587.06494140625, 454.81634521484375, 201.7477569580078, 503.95367431640625, 355.6311950683594, 321.72821044921875, 257.6329040527344, 53.333438873291016, 9.863059997558594, -128.75384521484375, 193.88385009765625, 372.3069763183594, -83.58673858642578, 244.3923797607422, 98.8968505859375, 307.83221435546875, 247.012451171875, 522.481201171875, 125.74325561523438, 111.60206604003906, 166.67050170898438, 327.8052673339844, 127.56336975097656, 250.6981201171875, 568.63916015625, -108.3604507446289, 28.77733612060547, 409.6550598144531, 239.67385864257812, -21.865060806274414, 8.91876220703125, 231.86074829101562, 186.51206970214844, -177.66708374023438, -87.04367065429688, 94.02085876464844, 519.838623046875, 60.65911102294922, 661.8035888671875, 510.58544921875, 281.2713317871094], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000671.npy"} +{"epoch": 0.9853157121879589, "step": 672, "batch_size": 64, "mean": 183.14840698242188, "std": 221.39488220214844, "min": -294.7835388183594, "p10": -72.8859909057617, "median": 172.1819305419922, "p90": 502.12529296875005, "max": 772.6377563476562, "pos_frac": 0.828125, "sample": [147.63038635253906, 13.479583740234375, -41.08946990966797, 115.5376968383789, 19.116313934326172, 39.84059143066406, 341.2218933105469, 259.88433837890625, 507.5906982421875, -13.533348083496094, 91.3928451538086, 58.28734588623047, -294.7835388183594, 86.71666717529297, 190.11297607421875, 98.98673248291016, 77.47213745117188, -81.46113586425781, 256.9522705078125, 344.3035583496094, 63.996158599853516, 522.0155029296875, 0.02843475341796875, -100.27005004882812, 341.478515625, 47.68975067138672, 593.46923828125, -92.7896728515625, 175.19149780273438, 772.6377563476562, -52.8773193359375, 595.6994018554688, 585.0364990234375, 70.96783447265625, 399.7062072753906, 365.7352294921875, 231.5238494873047, 174.8809814453125, 197.53213500976562, 489.3726806640625, -199.537109375, 124.63651275634766, 249.76417541503906, 318.6065368652344, 221.67794799804688, 357.67364501953125, 169.48287963867188, 122.56771850585938, 214.247802734375, 132.11471557617188, 259.8273620605469, 230.14346313476562, -0.6514892578125, 339.1412048339844, 105.8773193359375, 25.568355560302734, 185.03553771972656, 265.8392028808594, 160.74180603027344, -279.8701477050781, 379.6746520996094, 391.49383544921875, 610.511474609375, -261.75274658203125], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000672.npy"} +{"epoch": 0.986784140969163, "step": 673, "batch_size": 64, "mean": 232.09201049804688, "std": 176.2749481201172, "min": -208.53250122070312, "p10": -2.853785896301267, "median": 225.24545288085938, "p90": 437.97309875488287, "max": 684.8131713867188, "pos_frac": 0.875, "sample": [-53.86638259887695, 90.9210205078125, 147.63218688964844, 332.15167236328125, 85.47492218017578, 414.4759521484375, 256.3399353027344, 215.2488555908203, 563.9085693359375, 237.97927856445312, -134.76084899902344, 520.6912841796875, 203.95220947265625, 417.4468078613281, 221.80845642089844, 317.10772705078125, -96.27581787109375, 258.76708984375, 684.8131713867188, 212.13421630859375, 159.7642364501953, 198.63201904296875, 290.9716796875, -35.66877746582031, 324.75885009765625, -0.5731582641601562, 224.79046630859375, 370.30255126953125, -103.45684814453125, 79.50532531738281, 340.7193298339844, 325.34906005859375, 180.34751892089844, 139.41294860839844, 243.30958557128906, 516.6347045898438, 358.00787353515625, 147.5012664794922, 409.4271240234375, 292.14892578125, 371.4346618652344, -3.831197738647461, 443.3904113769531, 349.6898193359375, 225.700439453125, 86.43402862548828, 84.33460998535156, 92.42131042480469, 284.3025817871094, 518.8087158203125, 212.30502319335938, 169.0315704345703, 341.74090576171875, 378.5941467285156, -208.53250122070312, 273.7994384765625, 165.4154510498047, 425.33270263671875, 240.6552734375, 151.55197143554688, 130.19949340820312, 87.20185089111328, 484.0193786621094, 192.05380249023438], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000673.npy"} +{"epoch": 0.9882525697503671, "step": 674, "batch_size": 64, "mean": 243.77340698242188, "std": 218.91998291015625, "min": -124.27944946289062, "p10": -28.868311309814445, "median": 226.7104034423828, "p90": 522.0181915283204, "max": 1111.0888671875, "pos_frac": 0.875, "sample": [184.02157592773438, 50.44559860229492, 320.321044921875, 222.8170623779297, 193.21377563476562, 396.52996826171875, 130.04469299316406, 118.30503845214844, -111.8160171508789, 510.2712097167969, 349.85833740234375, 535.5682373046875, -70.55516052246094, 545.9259033203125, 48.02684020996094, 52.774696350097656, -32.544219970703125, 751.9943237304688, 313.34149169921875, 1111.0888671875, 292.6250915527344, 120.2673110961914, -93.80232238769531, 416.1117858886719, 170.06695556640625, 134.82119750976562, 355.28875732421875, 277.7824401855469, 335.478759765625, 112.29180908203125, 284.90252685546875, 632.494384765625, 53.44039535522461, 230.60374450683594, 220.83319091796875, 538.4407348632812, -47.517120361328125, -20.29119110107422, 148.95994567871094, 400.1002197265625, 219.21051025390625, 188.70770263671875, 347.0191650390625, 315.676025390625, 116.46281433105469, 480.926025390625, 245.06472778320312, 300.95721435546875, 114.69844055175781, 250.7020721435547, -96.36026763916016, 23.471019744873047, 527.0526123046875, -124.27944946289062, 211.34014892578125, 303.28582763671875, 267.5603942871094, 496.404052734375, 337.4120178222656, 273.07098388671875, 251.27764892578125, 15.593706130981445, 199.32980346679688, 154.38267517089844], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000674.npy"} +{"epoch": 0.9897209985315712, "step": 675, "batch_size": 64, "mean": 264.8315124511719, "std": 215.10855102539062, "min": -100.6742935180664, "p10": -7.880447006225584, "median": 279.3965301513672, "p90": 551.1552978515625, "max": 906.2046508789062, "pos_frac": 0.875, "sample": [210.60374450683594, 282.7980041503906, 335.7771911621094, 172.46649169921875, 186.62210083007812, 190.78651428222656, 501.65789794921875, 11.265121459960938, 197.6897735595703, 309.85003662109375, 406.7689208984375, -100.6742935180664, 401.8761901855469, 539.6819458007812, 78.80268096923828, -8.780349731445312, 10.888763427734375, 506.8248596191406, 292.4803466796875, 196.75662231445312, 296.9351501464844, 56.036643981933594, 906.2046508789062, 378.33111572265625, 23.719566345214844, 74.48722076416016, -47.642337799072266, 421.01300048828125, 298.2601013183594, 70.25534057617188, 336.3941650390625, 398.92449951171875, 313.82305908203125, 307.0628356933594, 97.73800659179688, 408.69146728515625, 597.8971557617188, -71.94367980957031, 556.0724487304688, 654.193603515625, 158.73219299316406, 129.96115112304688, 19.50017547607422, 251.553955078125, 295.5823974609375, -28.270355224609375, -49.78202819824219, 253.87841796875, 394.6025390625, 315.1009826660156, 183.34677124023438, 512.8770141601562, 209.80435180664062, -5.780673980712891, -22.67559051513672, 313.7610778808594, 234.1557159423828, 275.99505615234375, 291.4430236816406, 623.0143432617188, 108.62008666992188, 672.2962646484375, 727.0750732421875, 283.8271179199219], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000675.npy"} +{"epoch": 0.9911894273127754, "step": 676, "batch_size": 64, "mean": 221.587890625, "std": 185.0238800048828, "min": -204.75978088378906, "p10": 7.922611999511726, "median": 219.52853393554688, "p90": 443.74252319335943, "max": 665.7999877929688, "pos_frac": 0.921875, "sample": [220.03497314453125, 432.0218505859375, 395.5724792480469, 342.739501953125, 377.89324951171875, 20.86248016357422, 434.3532409667969, 227.53005981445312, 2.824993133544922, -204.75978088378906, 166.0916748046875, 50.0711669921875, 61.52317810058594, 255.02529907226562, 366.47998046875, 556.1682739257812, 174.51690673828125, 194.7499542236328, 106.24098205566406, -37.07644271850586, 26.909154891967773, 34.36112976074219, 447.0794372558594, 21.856658935546875, 110.18911743164062, 252.914794921875, 69.52848052978516, 219.0220947265625, 165.44491577148438, 600.3628540039062, 129.41732788085938, 350.72711181640625, 224.339599609375, 420.4491882324219, -22.47588348388672, 73.77456665039062, 474.86663818359375, 249.8501739501953, 209.95352172851562, 665.7999877929688, 253.80264282226562, 141.46530151367188, 116.43235778808594, 136.81114196777344, 373.0807800292969, -21.457244873046875, 14.90179443359375, 366.78509521484375, 148.588134765625, 103.51689910888672, 573.7109985351562, 304.35919189453125, 418.408935546875, 280.919189453125, 17.378517150878906, 306.114501953125, 526.8067626953125, -59.55261993408203, 435.9563903808594, 27.2882080078125, 227.303466796875, 290.1015625, 4.9315338134765625, 326.7369384765625], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000676.npy"} +{"epoch": 0.9926578560939795, "step": 677, "batch_size": 64, "mean": 224.55026245117188, "std": 183.73178100585938, "min": -138.5392303466797, "p10": -41.10122509002684, "median": 227.70767974853516, "p90": 460.0816864013672, "max": 591.519775390625, "pos_frac": 0.859375, "sample": [-93.63243865966797, 590.9208984375, 248.1072998046875, 173.08889770507812, 173.51629638671875, 289.6485900878906, -78.18030548095703, -89.03665924072266, 401.98175048828125, 26.328125, 227.9947967529297, 269.7815856933594, -50.72358703613281, 422.9791564941406, 129.34596252441406, 397.48529052734375, 484.1257629394531, -138.5392303466797, 90.00602722167969, 190.8552703857422, 453.5527038574219, 392.9798278808594, 357.62030029296875, 25.134567260742188, 180.1815185546875, 312.59088134765625, 379.7015380859375, 270.0909729003906, 308.402587890625, 449.2285461425781, 168.61573791503906, 136.22769165039062, -24.76869773864746, 190.96435546875, 335.62957763671875, -90.75323486328125, 451.0285339355469, 146.92356872558594, 251.45619201660156, 227.42056274414062, 362.6070556640625, 229.640869140625, 310.9725341796875, 480.96075439453125, -16.615692138671875, 493.6527099609375, 123.1197280883789, 36.870452880859375, 591.519775390625, 82.44148254394531, 516.4908447265625, -48.10087966918945, 184.62521362304688, 139.49496459960938, 416.77313232421875, 214.18258666992188, 81.38528442382812, 462.87982177734375, 305.8608703613281, 312.9781494140625, 290.5519104003906, 8.903656005859375, 69.23126220703125, 132.50936889648438], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000677.npy"} +{"epoch": 0.9941262848751835, "step": 678, "batch_size": 64, "mean": 228.6033477783203, "std": 228.0395965576172, "min": -290.8135681152344, "p10": -38.233072662353514, "median": 183.65972137451172, "p90": 527.572021484375, "max": 679.174072265625, "pos_frac": 0.84375, "sample": [50.24901580810547, 657.869384765625, 306.3175354003906, 297.1796569824219, 170.3922882080078, -100.47683715820312, 390.8541564941406, -290.8135681152344, 43.711700439453125, 156.88580322265625, -33.61668395996094, 443.32952880859375, 114.57077026367188, 489.9757385253906, 104.02454376220703, 134.82176208496094, 302.1100769042969, -172.760009765625, 176.27894592285156, 524.7175903320312, 466.2821960449219, 69.95362854003906, 4.332798004150391, 538.534912109375, 417.9076843261719, 679.174072265625, 313.06207275390625, 33.68626403808594, -10.230274200439453, 528.7953491210938, -48.517112731933594, -75.0353775024414, 5.561767578125, 414.8968505859375, 83.39004516601562, 529.94580078125, 105.9303970336914, 637.9427490234375, 119.90509033203125, 355.92071533203125, -120.72062683105469, 191.04049682617188, 391.04290771484375, 432.91827392578125, 359.8258056640625, 60.47444534301758, 484.639892578125, 658.373046875, 124.60592651367188, 302.6697692871094, 431.7529296875, -40.211524963378906, 202.8174591064453, 3.0076446533203125, -17.802038192749023, 100.62733459472656, 127.35902404785156, 363.068603515625, 284.6091003417969, 245.35775756835938, 440.7750549316406, 85.22439575195312, 504.0308532714844, 78.0666275024414], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000678.npy"} +{"epoch": 0.9955947136563876, "step": 679, "batch_size": 64, "mean": 191.6798095703125, "std": 198.15069580078125, "min": -210.4315185546875, "p10": -22.994201850891105, "median": 160.37981414794922, "p90": 413.2328643798828, "max": 691.2268676757812, "pos_frac": 0.8125, "sample": [413.6873779296875, 81.14801788330078, 283.27325439453125, 475.0793151855469, 121.51620483398438, 293.8789978027344, 371.9747619628906, 495.6528625488281, 387.0341796875, 139.72750854492188, 324.2143859863281, 372.6943054199219, 170.40647888183594, -86.84657287597656, 660.9566650390625, 3.6382827758789062, 77.5868148803711, -1.235748291015625, 180.8961181640625, 361.760498046875, 326.2406921386719, 681.6865234375, 68.61247253417969, 76.40823364257812, 53.88978958129883, -147.72247314453125, 143.9468994140625, 282.8680114746094, 50.38315963745117, 96.08097076416016, 215.10540771484375, 38.957916259765625, -38.164886474609375, 290.8733215332031, 170.431884765625, 54.344970703125, 489.81695556640625, 106.3123779296875, -1.8715534210205078, 25.49060821533203, -26.865997314453125, -8.945602416992188, 412.1723327636719, 150.3531494140625, 330.4325866699219, 320.99658203125, 34.807716369628906, -7.698394775390625, -132.99319458007812, 93.18069458007812, 350.274169921875, -210.4315185546875, 112.58779907226562, 57.53705978393555, 319.9848327636719, 326.1250305175781, 281.25537109375, 237.997802734375, 300.3197326660156, 691.2268676757812, 240.64068603515625, 337.82342529296875, -40.04841232299805, -13.960012435913086], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000679.npy"} +{"epoch": 0.9970631424375918, "step": 680, "batch_size": 64, "mean": 219.90866088867188, "std": 234.1836700439453, "min": -461.0483093261719, "p10": -23.55131988525387, "median": 174.34294891357422, "p90": 533.143914794922, "max": 814.233642578125, "pos_frac": 0.890625, "sample": [559.3265991210938, -116.59141540527344, 188.7498016357422, 358.19317626953125, 112.79039001464844, 382.2013854980469, 727.03271484375, 173.97630310058594, 470.1224060058594, 130.99081420898438, 229.3258056640625, 141.0559539794922, 126.72874450683594, 74.09072875976562, 357.3044738769531, 683.24169921875, 125.11589050292969, -272.48223876953125, 10.309310913085938, 120.87984466552734, 711.4902954101562, 174.7095947265625, 11.537464141845703, 39.02025604248047, 122.64579010009766, 306.7646484375, 137.77293395996094, 72.72274017333984, 337.8365478515625, 376.45123291015625, 173.1016845703125, 257.0477600097656, 64.9632568359375, 423.667236328125, 387.475830078125, 110.34504699707031, -74.05223083496094, 339.55889892578125, 455.0523681640625, -82.13372802734375, 226.02774047851562, 545.7362060546875, 291.2197265625, 191.93829345703125, 114.56491088867188, -461.0483093261719, 308.2578430175781, -74.11438751220703, 104.27444458007812, 503.76190185546875, 208.4468231201172, 814.233642578125, 11.440628051757812, 58.60966491699219, 257.1711120605469, 198.10488891601562, 443.6771240234375, 625.6815185546875, 28.458736419677734, 135.33563232421875, -38.063018798828125, 132.12518310546875, 152.94546508789062, 367.0285339355469], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000680.npy"} +{"epoch": 0.9985315712187959, "step": 681, "batch_size": 64, "mean": 210.1411590576172, "std": 217.13018798828125, "min": -191.2996826171875, "p10": -37.33971557617187, "median": 211.02891540527344, "p90": 535.0164428710938, "max": 861.7386474609375, "pos_frac": 0.828125, "sample": [144.2663116455078, -165.50997924804688, 278.50909423828125, 111.206787109375, 429.1376647949219, 91.9530258178711, 324.9046325683594, 259.0286560058594, -27.440391540527344, 109.06219482421875, -8.8726806640625, 211.4141845703125, -45.85395812988281, -28.026626586914062, 283.188232421875, 861.7386474609375, 276.1739196777344, -86.95836639404297, 31.988388061523438, 458.7796630859375, 128.55572509765625, 3.9635696411132812, 29.69641876220703, 80.96951293945312, 285.791748046875, 247.4071044921875, 210.64364624023438, 552.8243408203125, 275.99859619140625, 366.7225646972656, 167.34832763671875, 330.2803039550781, 278.69384765625, 83.06842041015625, 43.92963409423828, 230.57518005371094, 6.339235305786133, -123.39431762695312, 302.73602294921875, -49.07380294799805, -41.33103942871094, 583.3427734375, 120.31920623779297, 402.5660705566406, -191.2996826171875, 50.55120849609375, -14.70720100402832, 94.68011474609375, 144.4928741455078, 15.234443664550781, 466.83355712890625, 213.13790893554688, 449.34344482421875, 315.2378234863281, 213.03363037109375, 253.2039031982422, 537.4722900390625, 529.2861328125, 557.2875366210938, 333.9270935058594, 85.74532318115234, 553.01123046875, 680.7440795898438, 135.1562042236328], "npy": "/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs/step_0000681.npy"} diff --git a/margin_logs/step_0000001.npy b/margin_logs/step_0000001.npy new file mode 100644 index 0000000..248c095 --- /dev/null +++ b/margin_logs/step_0000001.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cb7ed5e9b5d6de6c4e509dd17cf5d9c91337fabd0c174e116c5e60872823ad93 +size 384 diff --git a/margin_logs/step_0000002.npy b/margin_logs/step_0000002.npy new file mode 100644 index 0000000..984e4c2 --- /dev/null +++ b/margin_logs/step_0000002.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fc23171824afa57340cda53f69d83aef67c7c0b95175e9ec4a3a7bc3c221bc4f +size 384 diff --git a/margin_logs/step_0000003.npy b/margin_logs/step_0000003.npy new file mode 100644 index 0000000..171f8af --- /dev/null +++ b/margin_logs/step_0000003.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:891d56878dc8b74bcefcfbf7ffb8fbe79af80925b889150e280b1c3a283c25c7 +size 384 diff --git a/margin_logs/step_0000004.npy b/margin_logs/step_0000004.npy new file mode 100644 index 0000000..b743a57 --- /dev/null +++ b/margin_logs/step_0000004.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:42775267d5983851a10f1207ebe671db3c02449711a2bb4af776c65de1564a99 +size 384 diff --git a/margin_logs/step_0000005.npy b/margin_logs/step_0000005.npy new file mode 100644 index 0000000..c6160bd --- /dev/null +++ b/margin_logs/step_0000005.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:045dd2f1195978ee1d6d6acd572363cfbe0977458b6f69cb5acea10b17c1916e +size 384 diff --git a/margin_logs/step_0000006.npy b/margin_logs/step_0000006.npy new file mode 100644 index 0000000..151266d --- /dev/null +++ b/margin_logs/step_0000006.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bbce2dd45f047ebdc841deac7ceea0988616c786a366ec9b0d8cc822a0326aef +size 384 diff --git a/margin_logs/step_0000007.npy b/margin_logs/step_0000007.npy new file mode 100644 index 0000000..4fa9381 --- /dev/null +++ b/margin_logs/step_0000007.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a15dd5784fd6debf62a34d99d4f6598b8721969849adaccd8f9d1abcc5815d65 +size 384 diff --git a/margin_logs/step_0000008.npy b/margin_logs/step_0000008.npy new file mode 100644 index 0000000..c36855f --- /dev/null +++ b/margin_logs/step_0000008.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c29bc92fcae8a1184ba0749467ed7d512a58e2862fd947a38061daf6d23aa7db +size 384 diff --git a/margin_logs/step_0000009.npy b/margin_logs/step_0000009.npy new file mode 100644 index 0000000..a0c9b27 --- /dev/null +++ b/margin_logs/step_0000009.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7a3cc86de8109788deef5ede1ea7c2e891b7c4d3bf52d28bb652307cef3a6bde +size 384 diff --git a/margin_logs/step_0000010.npy b/margin_logs/step_0000010.npy new file mode 100644 index 0000000..c5086f0 --- /dev/null +++ b/margin_logs/step_0000010.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0cb4300645a6e1602236a6801ee56bb635d75a9ffac0a6fba33c75eb4a00efb0 +size 384 diff --git a/margin_logs/step_0000011.npy b/margin_logs/step_0000011.npy new file mode 100644 index 0000000..6f7f902 --- /dev/null +++ b/margin_logs/step_0000011.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9ef6949495a73f4f500ec1a6c8334cd43bdbe1ae4b9f34ce3c797f2cbf2209ee +size 384 diff --git a/margin_logs/step_0000012.npy b/margin_logs/step_0000012.npy new file mode 100644 index 0000000..94b5127 --- /dev/null +++ b/margin_logs/step_0000012.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2dd99b1f16b659428ac29b7761265766dcb32a1c10ab1116e5a7f22aaa7a2b90 +size 384 diff --git a/margin_logs/step_0000013.npy b/margin_logs/step_0000013.npy new file mode 100644 index 0000000..f7de1d3 --- /dev/null +++ b/margin_logs/step_0000013.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4b214ee56cbfa942e759fcf72e51ab9ee0b2cf587e6761196ed725b347abe43c +size 384 diff --git a/margin_logs/step_0000014.npy b/margin_logs/step_0000014.npy new file mode 100644 index 0000000..a13a443 --- /dev/null +++ b/margin_logs/step_0000014.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:620f8736f3fc84206ca657d20da3eb4a21bb3d900a3dbe986f72a5a49e6a503c +size 384 diff --git a/margin_logs/step_0000015.npy b/margin_logs/step_0000015.npy new file mode 100644 index 0000000..45b59b7 --- /dev/null +++ b/margin_logs/step_0000015.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:31ada52616c04c7cbfe2efe48031e075ebaf88b0e6934564f16fcd6f9a8b010b +size 384 diff --git a/margin_logs/step_0000016.npy b/margin_logs/step_0000016.npy new file mode 100644 index 0000000..5b098e0 --- /dev/null +++ b/margin_logs/step_0000016.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:285c219cabb5f84dd6fed1d21c8fe9317dfdc4ad7c5794ae29d186f48dd5d23e +size 384 diff --git a/margin_logs/step_0000017.npy b/margin_logs/step_0000017.npy new file mode 100644 index 0000000..4ace517 --- /dev/null +++ b/margin_logs/step_0000017.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:53385f7e940bcb6b02dfdc549d6ffc0e4319e58e7c460d998d8ffe636ac3942f +size 384 diff --git a/margin_logs/step_0000018.npy b/margin_logs/step_0000018.npy new file mode 100644 index 0000000..b663447 --- /dev/null +++ b/margin_logs/step_0000018.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8970525201b214baf7ee4e048303d8a18d1bb26361b3084951061e18f760060c +size 384 diff --git a/margin_logs/step_0000019.npy b/margin_logs/step_0000019.npy new file mode 100644 index 0000000..68621b2 --- /dev/null +++ b/margin_logs/step_0000019.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:42b1df07b40de86c57f49cf98863a8795c6f12c9d268d71fb83ff0802692cc04 +size 384 diff --git a/margin_logs/step_0000020.npy b/margin_logs/step_0000020.npy new file mode 100644 index 0000000..18a73b4 --- /dev/null +++ b/margin_logs/step_0000020.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1065d7e03863ee813ddd9aef342f3d4ee8a11f6ce36230aacd31532832b1bc93 +size 384 diff --git a/margin_logs/step_0000021.npy b/margin_logs/step_0000021.npy new file mode 100644 index 0000000..c0e7986 --- /dev/null +++ b/margin_logs/step_0000021.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b93f6d7e8e15d9e12f99a51c4467c3133ff7b0e836bbbb7b2491ea3691f6215a +size 384 diff --git a/margin_logs/step_0000022.npy b/margin_logs/step_0000022.npy new file mode 100644 index 0000000..267c952 --- /dev/null +++ b/margin_logs/step_0000022.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6a734adb9b2c7d417d3c37e594874b2e902e3fd6547a1eb22fd8adc741d7ad6 +size 384 diff --git a/margin_logs/step_0000023.npy b/margin_logs/step_0000023.npy new file mode 100644 index 0000000..37cf7c2 --- /dev/null +++ b/margin_logs/step_0000023.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b057bcd5738423e1acec0584244417d6773264f1cfb39a14de507ec954ac6a85 +size 384 diff --git a/margin_logs/step_0000024.npy b/margin_logs/step_0000024.npy new file mode 100644 index 0000000..0ff1725 --- /dev/null +++ b/margin_logs/step_0000024.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e0bad5e97e325a35c1fad7f506bfd19170de8778a21a16588bb8a1c237daa981 +size 384 diff --git a/margin_logs/step_0000025.npy b/margin_logs/step_0000025.npy new file mode 100644 index 0000000..7f20a9c --- /dev/null +++ b/margin_logs/step_0000025.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:afd6e1cce282cb7681a5789597d7b2c9f897805f482484c30ec8f17db7ea97fb +size 384 diff --git a/margin_logs/step_0000026.npy b/margin_logs/step_0000026.npy new file mode 100644 index 0000000..35a8f2b --- /dev/null +++ b/margin_logs/step_0000026.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dac0c26702072dec8076a288ffb7cb8ca2766b831bac0f976865c13118197572 +size 384 diff --git a/margin_logs/step_0000027.npy b/margin_logs/step_0000027.npy new file mode 100644 index 0000000..ef64de6 --- /dev/null +++ b/margin_logs/step_0000027.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f4961520d6896290afa39fb2d4acc26fb9e9847cf067ac67cbf955b162c7ea7e +size 384 diff --git a/margin_logs/step_0000028.npy b/margin_logs/step_0000028.npy new file mode 100644 index 0000000..40e4231 --- /dev/null +++ b/margin_logs/step_0000028.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bd6673ff769870c700880b7b05b7ac67a8dc97ccb14e0fbed5c14dc7668c6499 +size 384 diff --git a/margin_logs/step_0000029.npy b/margin_logs/step_0000029.npy new file mode 100644 index 0000000..25e9115 --- /dev/null +++ b/margin_logs/step_0000029.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4a3b1b88131c12fbe4fff72172bcc2479fe13f58ee849ae838ee237883ba237e +size 384 diff --git a/margin_logs/step_0000030.npy b/margin_logs/step_0000030.npy new file mode 100644 index 0000000..9ab3c22 --- /dev/null +++ b/margin_logs/step_0000030.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d1229c30414accf4cd34eb64bec3da44a49a0c4491b829550b4eeb38dd02becc +size 384 diff --git a/margin_logs/step_0000031.npy b/margin_logs/step_0000031.npy new file mode 100644 index 0000000..a7d7058 --- /dev/null +++ b/margin_logs/step_0000031.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fbf119ee85ae30c708a844145efcd9a0cf19e1c9859d9b4ba5803a9958146fd1 +size 384 diff --git a/margin_logs/step_0000032.npy b/margin_logs/step_0000032.npy new file mode 100644 index 0000000..fbff442 --- /dev/null +++ b/margin_logs/step_0000032.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:389bd084f917345146deea2f94c5f7959d5d7d2d59ff49c950584cca8018f20c +size 384 diff --git a/margin_logs/step_0000033.npy b/margin_logs/step_0000033.npy new file mode 100644 index 0000000..374291c --- /dev/null +++ b/margin_logs/step_0000033.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6b510c56e6b9b01851cdcbf19dc57df968f251cd1ce37b5165a261c26528ae4d +size 384 diff --git a/margin_logs/step_0000034.npy b/margin_logs/step_0000034.npy new file mode 100644 index 0000000..b4b287b --- /dev/null +++ b/margin_logs/step_0000034.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1154adc5ce86da60e94d2bbf5c64ecfe165171d3fe0b1ec8c268456d537ac5e7 +size 384 diff --git a/margin_logs/step_0000035.npy b/margin_logs/step_0000035.npy new file mode 100644 index 0000000..258d920 --- /dev/null +++ b/margin_logs/step_0000035.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0aa17e0eff2641a03915bf5e130cf628fa562f571090debc5465d7b8b4bb8082 +size 384 diff --git a/margin_logs/step_0000036.npy b/margin_logs/step_0000036.npy new file mode 100644 index 0000000..22b7dd1 --- /dev/null +++ b/margin_logs/step_0000036.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9810f311d8c61f89dc4a98a6f31ce9bb1e298b49630b252beec3d45213b376f0 +size 384 diff --git a/margin_logs/step_0000037.npy b/margin_logs/step_0000037.npy new file mode 100644 index 0000000..ee824e8 --- /dev/null +++ b/margin_logs/step_0000037.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c00e1410302cdad58a12e02381eb34575d8a5db7a432e08200bed24fb588a28b +size 384 diff --git a/margin_logs/step_0000038.npy b/margin_logs/step_0000038.npy new file mode 100644 index 0000000..dc849e3 --- /dev/null +++ b/margin_logs/step_0000038.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4ad9edbda7973db188ee42aab02f9a75c9d032bfe24d21a6e0fdc6a1cc2cf71b +size 384 diff --git a/margin_logs/step_0000039.npy b/margin_logs/step_0000039.npy new file mode 100644 index 0000000..02a9bdf --- /dev/null +++ b/margin_logs/step_0000039.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c636c8d1a1f70bd4a68c2006e8350d6a6eba70f7959ceb8e4a5982ed01e7c70 +size 384 diff --git a/margin_logs/step_0000040.npy b/margin_logs/step_0000040.npy new file mode 100644 index 0000000..db7dbea --- /dev/null +++ b/margin_logs/step_0000040.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6bf6f33f927bab7d5d68518fdf1adfbcd107cbd0f37ed280d227dee8a85f827 +size 384 diff --git a/margin_logs/step_0000041.npy b/margin_logs/step_0000041.npy new file mode 100644 index 0000000..01412df --- /dev/null +++ b/margin_logs/step_0000041.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5f630ed0b84d2dfa1739ef05d7cbf1f06b7f8c6d23052ad6f9407570426c7b94 +size 384 diff --git a/margin_logs/step_0000042.npy b/margin_logs/step_0000042.npy new file mode 100644 index 0000000..9789fc4 --- /dev/null +++ b/margin_logs/step_0000042.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c8cff7cacf39349c78572a90623d69e6711866fbd715c4284ed9d4c318b98d2d +size 384 diff --git a/margin_logs/step_0000043.npy b/margin_logs/step_0000043.npy new file mode 100644 index 0000000..db0f2d2 --- /dev/null +++ b/margin_logs/step_0000043.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0369b53056c83408c0ebaff3970ffd8f85841a4005f04faa6de5fec1e4f40b96 +size 384 diff --git a/margin_logs/step_0000044.npy b/margin_logs/step_0000044.npy new file mode 100644 index 0000000..fb9203e --- /dev/null +++ b/margin_logs/step_0000044.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4dc675c4e72fb543364388f948b884be2858e1c07fce0d017fe1fe20a9d0839c +size 384 diff --git a/margin_logs/step_0000045.npy b/margin_logs/step_0000045.npy new file mode 100644 index 0000000..60003a6 --- /dev/null +++ b/margin_logs/step_0000045.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8c6699f330cc764e8f86e0f56476804c6506ee024ed715d5fd7c799c19467df2 +size 384 diff --git a/margin_logs/step_0000046.npy b/margin_logs/step_0000046.npy new file mode 100644 index 0000000..f96fe5e --- /dev/null +++ b/margin_logs/step_0000046.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cf2612888a956a6de81b5ddfed7adea5e45454bf05659b00e71867cb56a8d144 +size 384 diff --git a/margin_logs/step_0000047.npy b/margin_logs/step_0000047.npy new file mode 100644 index 0000000..346c834 --- /dev/null +++ b/margin_logs/step_0000047.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cc0d71585f3074ae89a26b50d0d110ce376e9b6bc9785b6b82ab8e823f953318 +size 384 diff --git a/margin_logs/step_0000048.npy b/margin_logs/step_0000048.npy new file mode 100644 index 0000000..42791ce --- /dev/null +++ b/margin_logs/step_0000048.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c47fe05ab3a4b23035568e66e5a553446442d8e94ce9b662f685435d4d9d5321 +size 384 diff --git a/margin_logs/step_0000049.npy b/margin_logs/step_0000049.npy new file mode 100644 index 0000000..da880f5 --- /dev/null +++ b/margin_logs/step_0000049.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f581abee272926766fc21f1b9b8fd64364facc9b9e304585103f428a1ff40f52 +size 384 diff --git a/margin_logs/step_0000050.npy b/margin_logs/step_0000050.npy new file mode 100644 index 0000000..ab4d73b --- /dev/null +++ b/margin_logs/step_0000050.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2719f0edc2a463913dff0d1255f2f71ec88d3ff2edc7fe6c58d68b5d81a8d76c +size 384 diff --git a/margin_logs/step_0000051.npy b/margin_logs/step_0000051.npy new file mode 100644 index 0000000..f0efcd9 --- /dev/null +++ b/margin_logs/step_0000051.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:34af6f17955e90916c47cb6175f93ce72c601b4c01003421db125b1bee627bab +size 384 diff --git a/margin_logs/step_0000052.npy b/margin_logs/step_0000052.npy new file mode 100644 index 0000000..3b55274 --- /dev/null +++ b/margin_logs/step_0000052.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fd6ebdb1b87d0de6b4f0c793a7fe4909fa4df6c45f16eccf1be1e78247cb90bf +size 384 diff --git a/margin_logs/step_0000053.npy b/margin_logs/step_0000053.npy new file mode 100644 index 0000000..f837867 --- /dev/null +++ b/margin_logs/step_0000053.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:13a669c43dcb515335faad06222ce375d36f4a2933a6764f392403c83790aa5d +size 384 diff --git a/margin_logs/step_0000054.npy b/margin_logs/step_0000054.npy new file mode 100644 index 0000000..c4f949f --- /dev/null +++ b/margin_logs/step_0000054.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f0d9e4c3a1e06f42e36b96d1c89b2571bf899170c76b3f3ac90dcd5a609bbb15 +size 384 diff --git a/margin_logs/step_0000055.npy b/margin_logs/step_0000055.npy new file mode 100644 index 0000000..faa2a74 --- /dev/null +++ b/margin_logs/step_0000055.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:786764ff2d7f58f8c06d5eec94285b78f7948c164e8849b5f7f2e46b83f371e6 +size 384 diff --git a/margin_logs/step_0000056.npy b/margin_logs/step_0000056.npy new file mode 100644 index 0000000..701d38e --- /dev/null +++ b/margin_logs/step_0000056.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:73913eaa71e1c477e6f50d213f08ae21411fcb444873c6e6bd978f22dbab7af2 +size 384 diff --git a/margin_logs/step_0000057.npy b/margin_logs/step_0000057.npy new file mode 100644 index 0000000..2b3dccb --- /dev/null +++ b/margin_logs/step_0000057.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:23806b2ac887647a166868cb611b75ecd6f026e16292844774380b373928f86e +size 384 diff --git a/margin_logs/step_0000058.npy b/margin_logs/step_0000058.npy new file mode 100644 index 0000000..7e2f7bb --- /dev/null +++ b/margin_logs/step_0000058.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b54b2b369eb799d8e3e07201ad0e3cbc39c5d0a1a8f9f617ee56c876af896827 +size 384 diff --git a/margin_logs/step_0000059.npy b/margin_logs/step_0000059.npy new file mode 100644 index 0000000..b80136c --- /dev/null +++ b/margin_logs/step_0000059.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:af618a382bfe49b96359b38224360b13f10ecf640e4cb60cf1b34ad7e75bd877 +size 384 diff --git a/margin_logs/step_0000060.npy b/margin_logs/step_0000060.npy new file mode 100644 index 0000000..afb6eb1 --- /dev/null +++ b/margin_logs/step_0000060.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b07e3d64776ff714fab836ba93c8e3b4eea7c9fb2946208ecbf77565f0458cce +size 384 diff --git a/margin_logs/step_0000061.npy b/margin_logs/step_0000061.npy new file mode 100644 index 0000000..d12207e --- /dev/null +++ b/margin_logs/step_0000061.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:66e7372b859e3b5cbce6d9f7d4123072315c54518d6b0621bade9ca377241e27 +size 384 diff --git a/margin_logs/step_0000062.npy b/margin_logs/step_0000062.npy new file mode 100644 index 0000000..086018d --- /dev/null +++ b/margin_logs/step_0000062.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:21d241b8e1e5c17b1ea8d3b1784d033629b849b7cfe1271d78843ebb4d5eb4e0 +size 384 diff --git a/margin_logs/step_0000063.npy b/margin_logs/step_0000063.npy new file mode 100644 index 0000000..95392f4 --- /dev/null +++ b/margin_logs/step_0000063.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f9774c264b7b4901513486a3c8447f7f77783a43119689dd6d2ed41a109c30da +size 384 diff --git a/margin_logs/step_0000064.npy b/margin_logs/step_0000064.npy new file mode 100644 index 0000000..d067c04 --- /dev/null +++ b/margin_logs/step_0000064.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:49ec82a2fd12894d50dc44c4bd4d97ab22410a3bf85761d16c93727c17beb226 +size 384 diff --git a/margin_logs/step_0000065.npy b/margin_logs/step_0000065.npy new file mode 100644 index 0000000..127a574 --- /dev/null +++ b/margin_logs/step_0000065.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:778533881998b4c42b9477617b024b52a4610bbf7b6e3fcbf9e9600b72abf6eb +size 384 diff --git a/margin_logs/step_0000066.npy b/margin_logs/step_0000066.npy new file mode 100644 index 0000000..f0bbc8a --- /dev/null +++ b/margin_logs/step_0000066.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2ffdbb14c4da829c5a0d2f39e9a495ce9f291b206e92a2d686507b07b8fe0403 +size 384 diff --git a/margin_logs/step_0000067.npy b/margin_logs/step_0000067.npy new file mode 100644 index 0000000..77921d0 --- /dev/null +++ b/margin_logs/step_0000067.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4d3d49b0bbd2a587ae52f342f6b46862a53152db58a43f5b1c127a2227efc69b +size 384 diff --git a/margin_logs/step_0000068.npy b/margin_logs/step_0000068.npy new file mode 100644 index 0000000..8978888 --- /dev/null +++ b/margin_logs/step_0000068.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:00b1dceb3b0f91fb5edbe9eeb4ac5d4f82d7494ae8939f09879e2af88297358a +size 384 diff --git a/margin_logs/step_0000069.npy b/margin_logs/step_0000069.npy new file mode 100644 index 0000000..e71b77a --- /dev/null +++ b/margin_logs/step_0000069.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:03351cd239c6d1c5e7b20c2a3842988725358e774f80722cf6a9896bf7a23b0f +size 384 diff --git a/margin_logs/step_0000070.npy b/margin_logs/step_0000070.npy new file mode 100644 index 0000000..151740b --- /dev/null +++ b/margin_logs/step_0000070.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:43891d8c90c73df332ff58c1484a61f8d50f4f101cda3e49c47c641fd0bcfbd4 +size 384 diff --git a/margin_logs/step_0000071.npy b/margin_logs/step_0000071.npy new file mode 100644 index 0000000..80b3389 --- /dev/null +++ b/margin_logs/step_0000071.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:60beb47dca741e621acbc667c741b3e26a6a9bc06bac09b3566ffe0bb1d7f9dd +size 384 diff --git a/margin_logs/step_0000072.npy b/margin_logs/step_0000072.npy new file mode 100644 index 0000000..f5cfbbc --- /dev/null +++ b/margin_logs/step_0000072.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f9d022da5582e208a9df94dab3c8e86f39189df5f8381f577730a95658dab2c5 +size 384 diff --git a/margin_logs/step_0000073.npy b/margin_logs/step_0000073.npy new file mode 100644 index 0000000..40d95e3 --- /dev/null +++ b/margin_logs/step_0000073.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ba56f82157040e1da7bc40c48c24078649f716b8e60c3af8e930f39fe1b02af3 +size 384 diff --git a/margin_logs/step_0000074.npy b/margin_logs/step_0000074.npy new file mode 100644 index 0000000..946b1ce --- /dev/null +++ b/margin_logs/step_0000074.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5db980fca145b6cf130026712ebfdb5fd99d350ae6b7bfc2e76ccbfb6843c38a +size 384 diff --git a/margin_logs/step_0000075.npy b/margin_logs/step_0000075.npy new file mode 100644 index 0000000..7a7d0c7 --- /dev/null +++ b/margin_logs/step_0000075.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ed13fda34b8c9c2f7e49cedd23c58b1b604e71700c243ed5e91b5b9e54d95be9 +size 384 diff --git a/margin_logs/step_0000076.npy b/margin_logs/step_0000076.npy new file mode 100644 index 0000000..8c0c3da --- /dev/null +++ b/margin_logs/step_0000076.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0963b476d03ed51ca477604d9e3ce34ffc1a3311be2634f48f651620df38784d +size 384 diff --git a/margin_logs/step_0000077.npy b/margin_logs/step_0000077.npy new file mode 100644 index 0000000..4412831 --- /dev/null +++ b/margin_logs/step_0000077.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e433f4091481c3a9134b5ec5a973ad9a7a9e59c27a5ec413f483b58c18d2bfac +size 384 diff --git a/margin_logs/step_0000078.npy b/margin_logs/step_0000078.npy new file mode 100644 index 0000000..5027e7a --- /dev/null +++ b/margin_logs/step_0000078.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f745e9113926f107365573c2c22b49069162a1509a336ed8aa3fb9043adace6 +size 384 diff --git a/margin_logs/step_0000079.npy b/margin_logs/step_0000079.npy new file mode 100644 index 0000000..8d2df40 --- /dev/null +++ b/margin_logs/step_0000079.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bb756c420636c794ddb67e0da49473e92a6f0c841893c2474cd530667a6f3577 +size 384 diff --git a/margin_logs/step_0000080.npy b/margin_logs/step_0000080.npy new file mode 100644 index 0000000..6f19d62 --- /dev/null +++ b/margin_logs/step_0000080.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16fca18bf7f7ffeeedf6cc780c6e47e1163898304aaa7c33a6537fd908c12734 +size 384 diff --git a/margin_logs/step_0000081.npy b/margin_logs/step_0000081.npy new file mode 100644 index 0000000..9d777d8 --- /dev/null +++ b/margin_logs/step_0000081.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5cd499eb849e4fbf2b950e54158d3c78e61924950c042952c50fc6858ebcafc3 +size 384 diff --git a/margin_logs/step_0000082.npy b/margin_logs/step_0000082.npy new file mode 100644 index 0000000..a05763c --- /dev/null +++ b/margin_logs/step_0000082.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:32e84594e90d93247ec62302119179454111c65798efb33012cde89fad800dcf +size 384 diff --git a/margin_logs/step_0000083.npy b/margin_logs/step_0000083.npy new file mode 100644 index 0000000..3869c4b --- /dev/null +++ b/margin_logs/step_0000083.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c4ea72331cf7575e9d3db2f80fbe2e3c0d629e8b3d473932a89d585af57c5338 +size 384 diff --git a/margin_logs/step_0000084.npy b/margin_logs/step_0000084.npy new file mode 100644 index 0000000..0edc2f0 --- /dev/null +++ b/margin_logs/step_0000084.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:83b7cc9df5e03fce1f39c38b3ccb446718713adc4faa0dfcf766b808593248d8 +size 384 diff --git a/margin_logs/step_0000085.npy b/margin_logs/step_0000085.npy new file mode 100644 index 0000000..8c55a37 --- /dev/null +++ b/margin_logs/step_0000085.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d51b5ff2607ce0898572fda2493fe91a06546600b769267b09b0ab83598c0a25 +size 384 diff --git a/margin_logs/step_0000086.npy b/margin_logs/step_0000086.npy new file mode 100644 index 0000000..b982a11 --- /dev/null +++ b/margin_logs/step_0000086.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bd6cc361f2c5ec6487fcf7fd04ea85c59fe16a992f5d2926e099d01f21e74766 +size 384 diff --git a/margin_logs/step_0000087.npy b/margin_logs/step_0000087.npy new file mode 100644 index 0000000..7c151a0 --- /dev/null +++ b/margin_logs/step_0000087.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a068ceaa00a21465d9fb3480f5d42eeae613abc7934a102bda703544cb4a3eae +size 384 diff --git a/margin_logs/step_0000088.npy b/margin_logs/step_0000088.npy new file mode 100644 index 0000000..940f396 --- /dev/null +++ b/margin_logs/step_0000088.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5078694c4a6d18bbcee518e1c20f3406f6425a6c2b6acafee14399b8fcef7868 +size 384 diff --git a/margin_logs/step_0000089.npy b/margin_logs/step_0000089.npy new file mode 100644 index 0000000..50bd68a --- /dev/null +++ b/margin_logs/step_0000089.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:39e2f4dfe4ca3057ad4d1f1954ee12686dd83f48ca04baeecd345264214500e9 +size 384 diff --git a/margin_logs/step_0000090.npy b/margin_logs/step_0000090.npy new file mode 100644 index 0000000..1b3d86b --- /dev/null +++ b/margin_logs/step_0000090.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2ec2578505700273d81b4f7736ceda95bc92ffb50897a11bfaa2fbc72a983b02 +size 384 diff --git a/margin_logs/step_0000091.npy b/margin_logs/step_0000091.npy new file mode 100644 index 0000000..4c123b0 --- /dev/null +++ b/margin_logs/step_0000091.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d17853226b3b02a2bd7b063f5e9df574a3b3603d034fdbfc6eec5be80befe423 +size 384 diff --git a/margin_logs/step_0000092.npy b/margin_logs/step_0000092.npy new file mode 100644 index 0000000..c26d8f4 --- /dev/null +++ b/margin_logs/step_0000092.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e40d328b0d915cce96c54696e0812998a92e934cb52a98338414cba5d186a122 +size 384 diff --git a/margin_logs/step_0000093.npy b/margin_logs/step_0000093.npy new file mode 100644 index 0000000..8542955 --- /dev/null +++ b/margin_logs/step_0000093.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ab07cfc95084d90a2080568af4bd69165bad96b71ee1de904ba174974b9f6e76 +size 384 diff --git a/margin_logs/step_0000094.npy b/margin_logs/step_0000094.npy new file mode 100644 index 0000000..9d66764 --- /dev/null +++ b/margin_logs/step_0000094.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e957fd1803a7178a6ca13b8ec651a54162f1389659befe15efef289489dca6f7 +size 384 diff --git a/margin_logs/step_0000095.npy b/margin_logs/step_0000095.npy new file mode 100644 index 0000000..9b10d02 --- /dev/null +++ b/margin_logs/step_0000095.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c23ba65efb0ed6feac66b2a9f9dca0a30bd20b152d956d6a9f852e0f292c93a2 +size 384 diff --git a/margin_logs/step_0000096.npy b/margin_logs/step_0000096.npy new file mode 100644 index 0000000..c8d914b --- /dev/null +++ b/margin_logs/step_0000096.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:23475aa2cfe8a682f020f4359a353b5fd5381e5fe022fd0b680aee9d38b3b877 +size 384 diff --git a/margin_logs/step_0000097.npy b/margin_logs/step_0000097.npy new file mode 100644 index 0000000..47ebac0 --- /dev/null +++ b/margin_logs/step_0000097.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:52ed5285baf9029b0be22f2add756b1bae1918584207d74308ede3c08530775b +size 384 diff --git a/margin_logs/step_0000098.npy b/margin_logs/step_0000098.npy new file mode 100644 index 0000000..1cb6bbd --- /dev/null +++ b/margin_logs/step_0000098.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:38882f3f4ab98ee5e08edfa30a90c75dbcb1e267bae81ba6f24e2687ab29eec4 +size 384 diff --git a/margin_logs/step_0000099.npy b/margin_logs/step_0000099.npy new file mode 100644 index 0000000..8abc977 --- /dev/null +++ b/margin_logs/step_0000099.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7a77aa58fb6d0aa28f78958a481b4573609cf4a29d6596365e0670144e2d060 +size 384 diff --git a/margin_logs/step_0000100.npy b/margin_logs/step_0000100.npy new file mode 100644 index 0000000..344b43d --- /dev/null +++ b/margin_logs/step_0000100.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3011d57fc7a949a35b175f56228cc25b9c23d5e4ef48b41a732f344e8a693fcf +size 384 diff --git a/margin_logs/step_0000101.npy b/margin_logs/step_0000101.npy new file mode 100644 index 0000000..271017e --- /dev/null +++ b/margin_logs/step_0000101.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41b2d01702b7e2b9633446451dffa0b0f5f7ee95932f87a8553ff70303e28c30 +size 384 diff --git a/margin_logs/step_0000102.npy b/margin_logs/step_0000102.npy new file mode 100644 index 0000000..5d50899 --- /dev/null +++ b/margin_logs/step_0000102.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e419aa3ca04b3df851b194a327a765c445f503d66a3c2001d1b55d4b2d5e86a +size 384 diff --git a/margin_logs/step_0000103.npy b/margin_logs/step_0000103.npy new file mode 100644 index 0000000..69ef34a --- /dev/null +++ b/margin_logs/step_0000103.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1a67c46bd820c2716e620b06b80b048e7b0c52da4f9bcde6a64b79753ddca744 +size 384 diff --git a/margin_logs/step_0000104.npy b/margin_logs/step_0000104.npy new file mode 100644 index 0000000..3e5f851 --- /dev/null +++ b/margin_logs/step_0000104.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4896875be1d8d0ef053425e18a0f4ac7f8cb2122789028b19bfdff3a4e9fe9e5 +size 384 diff --git a/margin_logs/step_0000105.npy b/margin_logs/step_0000105.npy new file mode 100644 index 0000000..c59f9a1 --- /dev/null +++ b/margin_logs/step_0000105.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:621585fbe62c695fce7dc98887c58601ae42d2925af859cd14df8a2b2eaeac02 +size 384 diff --git a/margin_logs/step_0000106.npy b/margin_logs/step_0000106.npy new file mode 100644 index 0000000..c7f1568 --- /dev/null +++ b/margin_logs/step_0000106.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:89ee8c61586a567d36154161d642534cdfe8e9d12fe13f36db67b682e4537886 +size 384 diff --git a/margin_logs/step_0000107.npy b/margin_logs/step_0000107.npy new file mode 100644 index 0000000..5ee5c27 --- /dev/null +++ b/margin_logs/step_0000107.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bdc9c691ef78c93b9a01c1d803a765c67d0d37d0c45e253fc7e4207c36c28e9c +size 384 diff --git a/margin_logs/step_0000108.npy b/margin_logs/step_0000108.npy new file mode 100644 index 0000000..71d53f7 --- /dev/null +++ b/margin_logs/step_0000108.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dfc794d4043861feae50d3d79d5eb6985a72c626b5054029773f6f214545ce44 +size 384 diff --git a/margin_logs/step_0000109.npy b/margin_logs/step_0000109.npy new file mode 100644 index 0000000..9c15b15 --- /dev/null +++ b/margin_logs/step_0000109.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c2c4227af47575ca5007fbfa0b606d179974edeae9f1fb5fe7e77762f8733561 +size 384 diff --git a/margin_logs/step_0000110.npy b/margin_logs/step_0000110.npy new file mode 100644 index 0000000..33cecff --- /dev/null +++ b/margin_logs/step_0000110.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:870d9679fdb316e93b0416cf124d25c03404aedb9aa540b5c30fbe82df6ab35b +size 384 diff --git a/margin_logs/step_0000111.npy b/margin_logs/step_0000111.npy new file mode 100644 index 0000000..949eb73 --- /dev/null +++ b/margin_logs/step_0000111.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2b6ca76694b5cd38a6f4610ef874a0990d93512007cee6eebde1a14eccdcc566 +size 384 diff --git a/margin_logs/step_0000112.npy b/margin_logs/step_0000112.npy new file mode 100644 index 0000000..a1cc84f --- /dev/null +++ b/margin_logs/step_0000112.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5a7705884dbb851acf472d0d2319b5e9f7246ff045a81046bda35096bd841c5a +size 384 diff --git a/margin_logs/step_0000113.npy b/margin_logs/step_0000113.npy new file mode 100644 index 0000000..3030b52 --- /dev/null +++ b/margin_logs/step_0000113.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b3be92aa5c50e63c8b5a8af69b37f5bc5a7fe5ce270f9a49dc4188110876dfeb +size 384 diff --git a/margin_logs/step_0000114.npy b/margin_logs/step_0000114.npy new file mode 100644 index 0000000..cbd9420 --- /dev/null +++ b/margin_logs/step_0000114.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7df7a0bf8508cde56a924492f8bc6e0ea50b5960b27e8182820675c495996a91 +size 384 diff --git a/margin_logs/step_0000115.npy b/margin_logs/step_0000115.npy new file mode 100644 index 0000000..d0d1658 --- /dev/null +++ b/margin_logs/step_0000115.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f71b15c3e23b8359cfd0f62408e009cf88cebd689773ffdf61c33897f0c61c10 +size 384 diff --git a/margin_logs/step_0000116.npy b/margin_logs/step_0000116.npy new file mode 100644 index 0000000..43094b3 --- /dev/null +++ b/margin_logs/step_0000116.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ad092c963ce102e0a9f9a75bcd0cf32f410cfce70688dc5f3b2e2e3b5acf0d51 +size 384 diff --git a/margin_logs/step_0000117.npy b/margin_logs/step_0000117.npy new file mode 100644 index 0000000..8037a1f --- /dev/null +++ b/margin_logs/step_0000117.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e8d3d92f66b658c49c8be1d66884a0628a705bc847490806122610d06d6d0d47 +size 384 diff --git a/margin_logs/step_0000118.npy b/margin_logs/step_0000118.npy new file mode 100644 index 0000000..3bc7b0f --- /dev/null +++ b/margin_logs/step_0000118.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:06f1f83010bd71c3b49507ca99cfcda681ab945616be014ee2468c5e968a936d +size 384 diff --git a/margin_logs/step_0000119.npy b/margin_logs/step_0000119.npy new file mode 100644 index 0000000..633347c --- /dev/null +++ b/margin_logs/step_0000119.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0dd5dc02751edd3788a18f3451a1cc21fa08b851d48fefc88749eb7768727e4c +size 384 diff --git a/margin_logs/step_0000120.npy b/margin_logs/step_0000120.npy new file mode 100644 index 0000000..be917f1 --- /dev/null +++ b/margin_logs/step_0000120.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:973db6a27450a6e6ba3e0eb32ce17ab2050f17118f161d66ba3e83f98715b244 +size 384 diff --git a/margin_logs/step_0000121.npy b/margin_logs/step_0000121.npy new file mode 100644 index 0000000..d84a328 --- /dev/null +++ b/margin_logs/step_0000121.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:98a543c7f3dec599dcc909ca44cd9fbfed803c93ecb38b1ae1e613ca24a8b694 +size 384 diff --git a/margin_logs/step_0000122.npy b/margin_logs/step_0000122.npy new file mode 100644 index 0000000..68e5de3 --- /dev/null +++ b/margin_logs/step_0000122.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3f338d039a777afb8ca51f987046f210bd3b1bdd4ecebaf2aeb93294c01b40e4 +size 384 diff --git a/margin_logs/step_0000123.npy b/margin_logs/step_0000123.npy new file mode 100644 index 0000000..9099589 --- /dev/null +++ b/margin_logs/step_0000123.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b58c15ce7256130a25dd9616a3b7f4f60482589a6e60aac9282d324bb49033d7 +size 384 diff --git a/margin_logs/step_0000124.npy b/margin_logs/step_0000124.npy new file mode 100644 index 0000000..7126c18 --- /dev/null +++ b/margin_logs/step_0000124.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cfbecd38ffd516d0f8a6bfa913101b9a6497a081524607ea3f3da227921ff193 +size 384 diff --git a/margin_logs/step_0000125.npy b/margin_logs/step_0000125.npy new file mode 100644 index 0000000..a9810df --- /dev/null +++ b/margin_logs/step_0000125.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0889ea3b6ffc764bf18841897a62fe996a16cc659696be8a3990e31e422d2c42 +size 384 diff --git a/margin_logs/step_0000126.npy b/margin_logs/step_0000126.npy new file mode 100644 index 0000000..1d2915d --- /dev/null +++ b/margin_logs/step_0000126.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bfa9f897dcf596e1a3a88b53f166cecd57eb7138d20356fe5b68e3aa79919f95 +size 384 diff --git a/margin_logs/step_0000127.npy b/margin_logs/step_0000127.npy new file mode 100644 index 0000000..6e18a24 --- /dev/null +++ b/margin_logs/step_0000127.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:94a8102402edeb2743872759fee79e5b2ea8091901c127e4ac5f9ec094410d1f +size 384 diff --git a/margin_logs/step_0000128.npy b/margin_logs/step_0000128.npy new file mode 100644 index 0000000..4b91853 --- /dev/null +++ b/margin_logs/step_0000128.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:68143bfea798ca689a73ba171279f002bc7e3b9160a7d69143bbd07ba4340e33 +size 384 diff --git a/margin_logs/step_0000129.npy b/margin_logs/step_0000129.npy new file mode 100644 index 0000000..c622647 --- /dev/null +++ b/margin_logs/step_0000129.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c06f0dffc4b6f996cbca560323aee8f8c1fa5ca67a330eb84121cae2726d709b +size 384 diff --git a/margin_logs/step_0000130.npy b/margin_logs/step_0000130.npy new file mode 100644 index 0000000..27e1d1f --- /dev/null +++ b/margin_logs/step_0000130.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e361661b3a80569341306e7ee29dc2cb9fb874d94b200e1bea4cf3dd745771cd +size 384 diff --git a/margin_logs/step_0000131.npy b/margin_logs/step_0000131.npy new file mode 100644 index 0000000..ab3317b --- /dev/null +++ b/margin_logs/step_0000131.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:861938743fe873ad108dcb82851b5baf17b1d860db0e8336a111e94839f2faab +size 384 diff --git a/margin_logs/step_0000132.npy b/margin_logs/step_0000132.npy new file mode 100644 index 0000000..ea204f5 --- /dev/null +++ b/margin_logs/step_0000132.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6984531c4e9fbf2df7cb7fc97d62966e2370b56764b1f55babf860a4b6c7a5e8 +size 384 diff --git a/margin_logs/step_0000133.npy b/margin_logs/step_0000133.npy new file mode 100644 index 0000000..54396ca --- /dev/null +++ b/margin_logs/step_0000133.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2a305fca8193ff31e615fe3862388c108c5b0f7bd24004043a741b397a317372 +size 384 diff --git a/margin_logs/step_0000134.npy b/margin_logs/step_0000134.npy new file mode 100644 index 0000000..8ea93ed --- /dev/null +++ b/margin_logs/step_0000134.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c68394017f3bcfd6ce851f38734ce6f5960d7f4e84d34409d5d01c0c11aea56b +size 384 diff --git a/margin_logs/step_0000135.npy b/margin_logs/step_0000135.npy new file mode 100644 index 0000000..7d9e9af --- /dev/null +++ b/margin_logs/step_0000135.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a85fcb24f1cadfaec3c4b4d0baddd83a44bbdf85bd57e9dc03903428c2352f7d +size 384 diff --git a/margin_logs/step_0000136.npy b/margin_logs/step_0000136.npy new file mode 100644 index 0000000..a8dd0b3 --- /dev/null +++ b/margin_logs/step_0000136.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:422985525a0dd2a399a489a0f2893537e33bf524d91b17f7d8a7dbf175287aa8 +size 384 diff --git a/margin_logs/step_0000137.npy b/margin_logs/step_0000137.npy new file mode 100644 index 0000000..a2cefeb --- /dev/null +++ b/margin_logs/step_0000137.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7dbc715f087ac32e461bf70ba7925273b67bb6bb4a988eb22bbb67d0292ec994 +size 384 diff --git a/margin_logs/step_0000138.npy b/margin_logs/step_0000138.npy new file mode 100644 index 0000000..bbc2558 --- /dev/null +++ b/margin_logs/step_0000138.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1bfd193de6968ebe220ccc01fcb058236708537813f655b267474b1945eb1308 +size 384 diff --git a/margin_logs/step_0000139.npy b/margin_logs/step_0000139.npy new file mode 100644 index 0000000..0e22d61 --- /dev/null +++ b/margin_logs/step_0000139.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f50e635f41b8f5e1a6b7adcc48562a87108cdf07255a4ee0a71ffea6c08a6856 +size 384 diff --git a/margin_logs/step_0000140.npy b/margin_logs/step_0000140.npy new file mode 100644 index 0000000..6a6eb3f --- /dev/null +++ b/margin_logs/step_0000140.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eb5809fac4bb6a2457a14d8899772ffccbf0b9fd7864e3c891c55046eb29f0b1 +size 384 diff --git a/margin_logs/step_0000141.npy b/margin_logs/step_0000141.npy new file mode 100644 index 0000000..d130095 --- /dev/null +++ b/margin_logs/step_0000141.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0bb6b9aeed44f410d7e15ddfe5cc8f7a9c0e4da7cca93e7a3dd0431b5724907a +size 384 diff --git a/margin_logs/step_0000142.npy b/margin_logs/step_0000142.npy new file mode 100644 index 0000000..9e0dc21 --- /dev/null +++ b/margin_logs/step_0000142.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fefe0d0ca350fba33619300cc31ce30cfcaf4e6148cd92bd4fa65cab76db998b +size 384 diff --git a/margin_logs/step_0000143.npy b/margin_logs/step_0000143.npy new file mode 100644 index 0000000..ccb6b82 --- /dev/null +++ b/margin_logs/step_0000143.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6062633eec96b4bc1ebf57293e8dde2b7edcfec55f9c842e1a2392d15bdcf3f4 +size 384 diff --git a/margin_logs/step_0000144.npy b/margin_logs/step_0000144.npy new file mode 100644 index 0000000..fc29dd0 --- /dev/null +++ b/margin_logs/step_0000144.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a4f38f79f7f93669f59a0fe50645898f51d052fae01e6da03aea4335af1eddf7 +size 384 diff --git a/margin_logs/step_0000145.npy b/margin_logs/step_0000145.npy new file mode 100644 index 0000000..b4f047b --- /dev/null +++ b/margin_logs/step_0000145.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:74e5b61750c9f29cf3fd01842e1d8445c4838b976538b9713696ef02b838842a +size 384 diff --git a/margin_logs/step_0000146.npy b/margin_logs/step_0000146.npy new file mode 100644 index 0000000..85e6ac7 --- /dev/null +++ b/margin_logs/step_0000146.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9f9ff4ca84a0c79c2c3a30c9554cb12c1be8d55c1b8f19bcf640942ff9f02b4d +size 384 diff --git a/margin_logs/step_0000147.npy b/margin_logs/step_0000147.npy new file mode 100644 index 0000000..0a69aa7 --- /dev/null +++ b/margin_logs/step_0000147.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:65f758e6de95ecef2b00dcb60e474b29a1ab9c64153aec56f3cf8cbf0646a808 +size 384 diff --git a/margin_logs/step_0000148.npy b/margin_logs/step_0000148.npy new file mode 100644 index 0000000..413018d --- /dev/null +++ b/margin_logs/step_0000148.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a76aec09a36efced83d72ed16d8a718418121e31b30c63f3ff1c50b1bedb2db9 +size 384 diff --git a/margin_logs/step_0000149.npy b/margin_logs/step_0000149.npy new file mode 100644 index 0000000..2ae3e2b --- /dev/null +++ b/margin_logs/step_0000149.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:29cd0b353e766dc145c9fe8ddb22c9cfef52ce83670ae56a9ff54c865c44942f +size 384 diff --git a/margin_logs/step_0000150.npy b/margin_logs/step_0000150.npy new file mode 100644 index 0000000..d0809f7 --- /dev/null +++ b/margin_logs/step_0000150.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6986408bd7b592013391c8fd55d3c4c24c8bc64ed68bfbde5b12554d5dc3a458 +size 384 diff --git a/margin_logs/step_0000151.npy b/margin_logs/step_0000151.npy new file mode 100644 index 0000000..6c76072 --- /dev/null +++ b/margin_logs/step_0000151.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5c333635daea23f852b56afcc2b6c832925dc5700019f9d665f095737306a1fc +size 384 diff --git a/margin_logs/step_0000152.npy b/margin_logs/step_0000152.npy new file mode 100644 index 0000000..3fd19ce --- /dev/null +++ b/margin_logs/step_0000152.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e4f48caac568fad23fac5509f4f143b40581ce757d6d088a2f17a8ae0dec938d +size 384 diff --git a/margin_logs/step_0000153.npy b/margin_logs/step_0000153.npy new file mode 100644 index 0000000..7c0eb8b --- /dev/null +++ b/margin_logs/step_0000153.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:af8b0e4c7f97036f74b696ab9e52544285a97ce170704c68a8b63270b4cfba28 +size 384 diff --git a/margin_logs/step_0000154.npy b/margin_logs/step_0000154.npy new file mode 100644 index 0000000..8034c65 --- /dev/null +++ b/margin_logs/step_0000154.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dda8088af41ded1f2af033ea79f0c22810c1ed06f645816d645f1f2aa35ddc87 +size 384 diff --git a/margin_logs/step_0000155.npy b/margin_logs/step_0000155.npy new file mode 100644 index 0000000..1aaa8f7 --- /dev/null +++ b/margin_logs/step_0000155.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0884d9f6d959e25cf3bd3fc99e5e4395b137febedb71cd855f7915fc5be1b331 +size 384 diff --git a/margin_logs/step_0000156.npy b/margin_logs/step_0000156.npy new file mode 100644 index 0000000..f34fed4 --- /dev/null +++ b/margin_logs/step_0000156.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1165b440f4a321c8917e9a40d85efe6650ebbec4e269185253d1bbbd3f604a35 +size 384 diff --git a/margin_logs/step_0000157.npy b/margin_logs/step_0000157.npy new file mode 100644 index 0000000..8670438 --- /dev/null +++ b/margin_logs/step_0000157.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6a8aec639744129c66b7fa3a0ac80989633cb2264b7163b9ad3c3f88f6778faf +size 384 diff --git a/margin_logs/step_0000158.npy b/margin_logs/step_0000158.npy new file mode 100644 index 0000000..6355990 --- /dev/null +++ b/margin_logs/step_0000158.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d25db41700aca07c5817b881c64d4d74bcb444bbf8987b1264b90b99e8796634 +size 384 diff --git a/margin_logs/step_0000159.npy b/margin_logs/step_0000159.npy new file mode 100644 index 0000000..8235c2a --- /dev/null +++ b/margin_logs/step_0000159.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4599245950687a19cd632b846b9c06565a3fcbaf0dc4a9bd3e8e92de8c45fa29 +size 384 diff --git a/margin_logs/step_0000160.npy b/margin_logs/step_0000160.npy new file mode 100644 index 0000000..8b1d306 --- /dev/null +++ b/margin_logs/step_0000160.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8d338a2bd9a18b1f24d1e1e58ba7a1e6e8b1c277ca326a13247913a224be9098 +size 384 diff --git a/margin_logs/step_0000161.npy b/margin_logs/step_0000161.npy new file mode 100644 index 0000000..871cb72 --- /dev/null +++ b/margin_logs/step_0000161.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ccbc12197db93889a9b2743103168a8e5628a9af7cedb704f1892acf38768535 +size 384 diff --git a/margin_logs/step_0000162.npy b/margin_logs/step_0000162.npy new file mode 100644 index 0000000..1d4e68e --- /dev/null +++ b/margin_logs/step_0000162.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:335d7807e968a1576e10723987f1da95decbd4b435e09a3f4e55810a28462210 +size 384 diff --git a/margin_logs/step_0000163.npy b/margin_logs/step_0000163.npy new file mode 100644 index 0000000..f528bec --- /dev/null +++ b/margin_logs/step_0000163.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1ebef63462163f9c3c9c131c26e6c9ad6f1bb5b88e6b430312e7ae71062f85d6 +size 384 diff --git a/margin_logs/step_0000164.npy b/margin_logs/step_0000164.npy new file mode 100644 index 0000000..84dbb84 --- /dev/null +++ b/margin_logs/step_0000164.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:179d695b486d57f71e163d0223af09a42c769469ad6108223f45202307e9e964 +size 384 diff --git a/margin_logs/step_0000165.npy b/margin_logs/step_0000165.npy new file mode 100644 index 0000000..bcddc71 --- /dev/null +++ b/margin_logs/step_0000165.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8f2a3041e651b096ce1dba8aea74f4260ca455ad437ff13ffeb94a6a07dbd7e5 +size 384 diff --git a/margin_logs/step_0000166.npy b/margin_logs/step_0000166.npy new file mode 100644 index 0000000..d84bbb4 --- /dev/null +++ b/margin_logs/step_0000166.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3f1981d49d66386193cccdaed917061fc2b6fd94fc6c201feab19e5c6a467473 +size 384 diff --git a/margin_logs/step_0000167.npy b/margin_logs/step_0000167.npy new file mode 100644 index 0000000..92e72e9 --- /dev/null +++ b/margin_logs/step_0000167.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:df8bdd3aee68cfacb0a49244dd3bf6c7869f317255b011cf1ed5b16f1e039244 +size 384 diff --git a/margin_logs/step_0000168.npy b/margin_logs/step_0000168.npy new file mode 100644 index 0000000..96d9f69 --- /dev/null +++ b/margin_logs/step_0000168.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6b0c12b4de76f345b2dbda2bff2e263e2a761cbcadcdc7ebd3f5e8acd6f3933 +size 384 diff --git a/margin_logs/step_0000169.npy b/margin_logs/step_0000169.npy new file mode 100644 index 0000000..400e5be --- /dev/null +++ b/margin_logs/step_0000169.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86d4a986f885ea1c045ff0c8c42124fd0efd5b230ec626590c577eec642301c6 +size 384 diff --git a/margin_logs/step_0000170.npy b/margin_logs/step_0000170.npy new file mode 100644 index 0000000..5d94c2c --- /dev/null +++ b/margin_logs/step_0000170.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a982de44a153bf0a109132f0823e6fb05d327bde7bfd622968fdd6d04a8d1639 +size 384 diff --git a/margin_logs/step_0000171.npy b/margin_logs/step_0000171.npy new file mode 100644 index 0000000..ba9477a --- /dev/null +++ b/margin_logs/step_0000171.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:38131ab914cf0accc7e2cd117eaf9c78c8fd8850ff12811e0f87d22e5a219019 +size 384 diff --git a/margin_logs/step_0000172.npy b/margin_logs/step_0000172.npy new file mode 100644 index 0000000..674187b --- /dev/null +++ b/margin_logs/step_0000172.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:63099641761fb3de4478be49d70cccfdcc9b328ac822bead0f3855bde508781e +size 384 diff --git a/margin_logs/step_0000173.npy b/margin_logs/step_0000173.npy new file mode 100644 index 0000000..735e454 --- /dev/null +++ b/margin_logs/step_0000173.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:00f14d16c0e37420cc228484d48adc13d99953d03404171831b41fa677c6c0c9 +size 384 diff --git a/margin_logs/step_0000174.npy b/margin_logs/step_0000174.npy new file mode 100644 index 0000000..4a294f9 --- /dev/null +++ b/margin_logs/step_0000174.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8aa59dc1013a6649393e63ac1fa0ce80064c862d745fbc8bf29a6b95d717140a +size 384 diff --git a/margin_logs/step_0000175.npy b/margin_logs/step_0000175.npy new file mode 100644 index 0000000..99d0f07 --- /dev/null +++ b/margin_logs/step_0000175.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:85193f04d652c815892b62ea1538d05d4dc3aa0ae1d0275be1f5fa1db51b4fd9 +size 384 diff --git a/margin_logs/step_0000176.npy b/margin_logs/step_0000176.npy new file mode 100644 index 0000000..6fb653f --- /dev/null +++ b/margin_logs/step_0000176.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aabe75b051df86d49298ace990fe0bb48b5e7d3b55a4adbddbb9ed6775ef1486 +size 384 diff --git a/margin_logs/step_0000177.npy b/margin_logs/step_0000177.npy new file mode 100644 index 0000000..6d69a6f --- /dev/null +++ b/margin_logs/step_0000177.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:979d3015b4d5cfc57064ff88084a3ac82575a4bf5f0760195a4b7ef71275cd65 +size 384 diff --git a/margin_logs/step_0000178.npy b/margin_logs/step_0000178.npy new file mode 100644 index 0000000..e84c272 --- /dev/null +++ b/margin_logs/step_0000178.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:917a08f1d4c6992bd1dddced61f77d955280fdcfdcaef7b4059f372c68ae96ff +size 384 diff --git a/margin_logs/step_0000179.npy b/margin_logs/step_0000179.npy new file mode 100644 index 0000000..6236747 --- /dev/null +++ b/margin_logs/step_0000179.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5edb8ce71d9d59c687db53e5192dac7574413da9bb1148ad9a50f97bf6848549 +size 384 diff --git a/margin_logs/step_0000180.npy b/margin_logs/step_0000180.npy new file mode 100644 index 0000000..52b8c4a --- /dev/null +++ b/margin_logs/step_0000180.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f67d6591d6dda8d9f00ae43c97ace65d5f92659c16fb660460b8e763f955a5e2 +size 384 diff --git a/margin_logs/step_0000181.npy b/margin_logs/step_0000181.npy new file mode 100644 index 0000000..3878292 --- /dev/null +++ b/margin_logs/step_0000181.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7c2610a12f1b3f45d828f50ba317fce367650b503fbe3d2ea6a71e5c73c78d41 +size 384 diff --git a/margin_logs/step_0000182.npy b/margin_logs/step_0000182.npy new file mode 100644 index 0000000..8cdfdf1 --- /dev/null +++ b/margin_logs/step_0000182.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4afc98daec856da31b4a157697e869a31f335a3f89191d2eb4eb221392a83ed3 +size 384 diff --git a/margin_logs/step_0000183.npy b/margin_logs/step_0000183.npy new file mode 100644 index 0000000..6667814 --- /dev/null +++ b/margin_logs/step_0000183.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:18d6eafe35a78d06a7c4167a21d547340bddba6eec92f62963c9245890eed9fb +size 384 diff --git a/margin_logs/step_0000184.npy b/margin_logs/step_0000184.npy new file mode 100644 index 0000000..330ae18 --- /dev/null +++ b/margin_logs/step_0000184.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:06a9202e7949d8dabc7f8e859e90edb956225d5168486bf52970129ededa1685 +size 384 diff --git a/margin_logs/step_0000185.npy b/margin_logs/step_0000185.npy new file mode 100644 index 0000000..7c79f0b --- /dev/null +++ b/margin_logs/step_0000185.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:44df0b06f5ce58ad78df642166fd8da936d5eb64db7d1a78b5c29168cc616cbb +size 384 diff --git a/margin_logs/step_0000186.npy b/margin_logs/step_0000186.npy new file mode 100644 index 0000000..4fad575 --- /dev/null +++ b/margin_logs/step_0000186.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2928274e4bd4483ce07920fd7d69c4f167b6046e40833af703dc48d71d65c291 +size 384 diff --git a/margin_logs/step_0000187.npy b/margin_logs/step_0000187.npy new file mode 100644 index 0000000..3d344bc --- /dev/null +++ b/margin_logs/step_0000187.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:88b77a394520b85a456dd75f6e5e811ea896e0a663de8c36608351c2ee3087cf +size 384 diff --git a/margin_logs/step_0000188.npy b/margin_logs/step_0000188.npy new file mode 100644 index 0000000..7a6ffe4 --- /dev/null +++ b/margin_logs/step_0000188.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4f77fbc2caab582587c8259cd16495a87ff04f1ad787b0896b9af68f88dd3db5 +size 384 diff --git a/margin_logs/step_0000189.npy b/margin_logs/step_0000189.npy new file mode 100644 index 0000000..f83758e --- /dev/null +++ b/margin_logs/step_0000189.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c00e28f8251a008d1cbd36e47158a53c19db5ca72bd888eff4af539f29d28360 +size 384 diff --git a/margin_logs/step_0000190.npy b/margin_logs/step_0000190.npy new file mode 100644 index 0000000..d29b99d --- /dev/null +++ b/margin_logs/step_0000190.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c930176f7a54d073574d1a0ba8ef09f1eb08573c32807c2ac39dea26728b1687 +size 384 diff --git a/margin_logs/step_0000191.npy b/margin_logs/step_0000191.npy new file mode 100644 index 0000000..9cbe30d --- /dev/null +++ b/margin_logs/step_0000191.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2cfca25500b9de027b4067b34d9af5e4c1b0b24420796adc6d15444a369b8c51 +size 384 diff --git a/margin_logs/step_0000192.npy b/margin_logs/step_0000192.npy new file mode 100644 index 0000000..6aa14ad --- /dev/null +++ b/margin_logs/step_0000192.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7e487c9c9a14be2cf7171d195550d9227f8b709e9ad86708171dea11ae1ef19b +size 384 diff --git a/margin_logs/step_0000193.npy b/margin_logs/step_0000193.npy new file mode 100644 index 0000000..585e339 --- /dev/null +++ b/margin_logs/step_0000193.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ff9951fce807fe345afb7760a47b85548dce23c33cf6d0566b099389ce3286ac +size 384 diff --git a/margin_logs/step_0000194.npy b/margin_logs/step_0000194.npy new file mode 100644 index 0000000..fcb9829 --- /dev/null +++ b/margin_logs/step_0000194.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0f39bf13cefb79394995ac107d055ebadbdcb659a545de680651be74a92c2daf +size 384 diff --git a/margin_logs/step_0000195.npy b/margin_logs/step_0000195.npy new file mode 100644 index 0000000..65989c6 --- /dev/null +++ b/margin_logs/step_0000195.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:014e7d8ba466c92299d3ae3525b20ae19a985db3a4c48fe27fb69a7f555e0be9 +size 384 diff --git a/margin_logs/step_0000196.npy b/margin_logs/step_0000196.npy new file mode 100644 index 0000000..e602786 --- /dev/null +++ b/margin_logs/step_0000196.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d1888d69e00755a512d542210abe203d02621844b00240bb4203ea7f1e3daf5d +size 384 diff --git a/margin_logs/step_0000197.npy b/margin_logs/step_0000197.npy new file mode 100644 index 0000000..6e8337c --- /dev/null +++ b/margin_logs/step_0000197.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b196e35331095129d07eaacca403a17bc9bcaa5162369156f6e157f58641668 +size 384 diff --git a/margin_logs/step_0000198.npy b/margin_logs/step_0000198.npy new file mode 100644 index 0000000..04411f9 --- /dev/null +++ b/margin_logs/step_0000198.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c548187f0df56088267ab20135f90bbb13f51c6e61e6f690a3132d95d12a1461 +size 384 diff --git a/margin_logs/step_0000199.npy b/margin_logs/step_0000199.npy new file mode 100644 index 0000000..ccac84f --- /dev/null +++ b/margin_logs/step_0000199.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:09865882dc683a4998164f84c2a2ed8639e1f46cbdb41667d50417333b85f731 +size 384 diff --git a/margin_logs/step_0000200.npy b/margin_logs/step_0000200.npy new file mode 100644 index 0000000..5abb369 --- /dev/null +++ b/margin_logs/step_0000200.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1d393ceef31f582b2cee5f76530dda68d4bc8ef748fb23b2c6d673811bffbe4f +size 384 diff --git a/margin_logs/step_0000201.npy b/margin_logs/step_0000201.npy new file mode 100644 index 0000000..05d47d1 --- /dev/null +++ b/margin_logs/step_0000201.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2e8426e32b0666d9f5fef183f0942340bc6b8d1796afbbb28bec0ff2f54b3c1a +size 384 diff --git a/margin_logs/step_0000202.npy b/margin_logs/step_0000202.npy new file mode 100644 index 0000000..692928e --- /dev/null +++ b/margin_logs/step_0000202.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:91d8ba6b2e0646e1af21a03f55165c987884a3970b4bd3e6031b00845a123d30 +size 384 diff --git a/margin_logs/step_0000203.npy b/margin_logs/step_0000203.npy new file mode 100644 index 0000000..3f8b954 --- /dev/null +++ b/margin_logs/step_0000203.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:59c3df8b84d1a622ee323a7ec9c8e72f40945f1febba94afb28d29a97f68b6bb +size 384 diff --git a/margin_logs/step_0000204.npy b/margin_logs/step_0000204.npy new file mode 100644 index 0000000..7b83d29 --- /dev/null +++ b/margin_logs/step_0000204.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:56f85530345a0e74993a1223f65df8890435a990bfb337469ec3622b0a93a466 +size 384 diff --git a/margin_logs/step_0000205.npy b/margin_logs/step_0000205.npy new file mode 100644 index 0000000..6870fb9 --- /dev/null +++ b/margin_logs/step_0000205.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7d654cbc65f36bb21ed43901089dc4b477b9f95125fb138f17df71b5999b9177 +size 384 diff --git a/margin_logs/step_0000206.npy b/margin_logs/step_0000206.npy new file mode 100644 index 0000000..391c7a0 --- /dev/null +++ b/margin_logs/step_0000206.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c7a85347804b569394ae7ffe0a60222f0a1248982c3adae393f821bb983595b5 +size 384 diff --git a/margin_logs/step_0000207.npy b/margin_logs/step_0000207.npy new file mode 100644 index 0000000..19a2fc1 --- /dev/null +++ b/margin_logs/step_0000207.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:beb6d6a5d80ca5864375429c197fa53ffdbcbb7e44520d28466c19514955e89c +size 384 diff --git a/margin_logs/step_0000208.npy b/margin_logs/step_0000208.npy new file mode 100644 index 0000000..7b0536f --- /dev/null +++ b/margin_logs/step_0000208.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a85ad75d73612a59bb4c074d7e6ecca23527163fb2e1ad1ebd0db1146acba0c5 +size 384 diff --git a/margin_logs/step_0000209.npy b/margin_logs/step_0000209.npy new file mode 100644 index 0000000..7f88140 --- /dev/null +++ b/margin_logs/step_0000209.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:547672707ef6366a650a0d63d2ba4bd10d0cb16e8445dc10a9a812f6f7ae8770 +size 384 diff --git a/margin_logs/step_0000210.npy b/margin_logs/step_0000210.npy new file mode 100644 index 0000000..a4c312e --- /dev/null +++ b/margin_logs/step_0000210.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5698364c0f3bea8148226d9c8404c36a2f89a29c37f32a6e519e41ea730d9fbd +size 384 diff --git a/margin_logs/step_0000211.npy b/margin_logs/step_0000211.npy new file mode 100644 index 0000000..4e0d514 --- /dev/null +++ b/margin_logs/step_0000211.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2e549a51af9837c401d84ef5b212afd6fc2be89ca3f8c242080eb053cc08ddd1 +size 384 diff --git a/margin_logs/step_0000212.npy b/margin_logs/step_0000212.npy new file mode 100644 index 0000000..32e0fb4 --- /dev/null +++ b/margin_logs/step_0000212.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:be2c022b59b4f261924b03330e2b9c843f3d0f371f708df49cddd4ae70ed25c1 +size 384 diff --git a/margin_logs/step_0000213.npy b/margin_logs/step_0000213.npy new file mode 100644 index 0000000..ebf4931 --- /dev/null +++ b/margin_logs/step_0000213.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:80169c4e6efe6ce060fb2e84da9550079beb0be966f9138aa3944e53f841bc8d +size 384 diff --git a/margin_logs/step_0000214.npy b/margin_logs/step_0000214.npy new file mode 100644 index 0000000..7adc33a --- /dev/null +++ b/margin_logs/step_0000214.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cdb3cfeb4f63d712132279d9bc7edabc11b761fada1508a7ef6c3cfb22c0a93e +size 384 diff --git a/margin_logs/step_0000215.npy b/margin_logs/step_0000215.npy new file mode 100644 index 0000000..137e9f2 --- /dev/null +++ b/margin_logs/step_0000215.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:118ac7ea7db09543de1450eea594700f5a104231989f105f18d5cb4c0efd8581 +size 384 diff --git a/margin_logs/step_0000216.npy b/margin_logs/step_0000216.npy new file mode 100644 index 0000000..0513773 --- /dev/null +++ b/margin_logs/step_0000216.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e369b98a8577a4350edb4beeb3a3bae2ee96e2eb3055d64a6ffa9e957db6dc0 +size 384 diff --git a/margin_logs/step_0000217.npy b/margin_logs/step_0000217.npy new file mode 100644 index 0000000..822d054 --- /dev/null +++ b/margin_logs/step_0000217.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:55cdb131aad532d9b7ab446e5699cb2ae81bac303446a02db3c797446e09a931 +size 384 diff --git a/margin_logs/step_0000218.npy b/margin_logs/step_0000218.npy new file mode 100644 index 0000000..8be20f6 --- /dev/null +++ b/margin_logs/step_0000218.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:db7f342e9795d5ef3cdce5dbccb718ba1083a5a416dda2482379cc2a89fc50b2 +size 384 diff --git a/margin_logs/step_0000219.npy b/margin_logs/step_0000219.npy new file mode 100644 index 0000000..570f096 --- /dev/null +++ b/margin_logs/step_0000219.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c1d449ce9cdb2bd8e497d34004a13d74263445febf58830705de7c954e53d0b6 +size 384 diff --git a/margin_logs/step_0000220.npy b/margin_logs/step_0000220.npy new file mode 100644 index 0000000..855404f --- /dev/null +++ b/margin_logs/step_0000220.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:71c99da5d8cf88689d54f3956732382f7bb8bc08b3243c65224075777120839a +size 384 diff --git a/margin_logs/step_0000221.npy b/margin_logs/step_0000221.npy new file mode 100644 index 0000000..e5d02a1 --- /dev/null +++ b/margin_logs/step_0000221.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3bbc6e83f0a5ff3e7690d2141dbcfe122d354926eec2ae8f8bf3098dffcc05d6 +size 384 diff --git a/margin_logs/step_0000222.npy b/margin_logs/step_0000222.npy new file mode 100644 index 0000000..9bd588c --- /dev/null +++ b/margin_logs/step_0000222.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:93cdbe0ac4432f2fdb8dbad2e09e87af97ed501adf57acd1ddb0b489b4a9bee0 +size 384 diff --git a/margin_logs/step_0000223.npy b/margin_logs/step_0000223.npy new file mode 100644 index 0000000..b4a5538 --- /dev/null +++ b/margin_logs/step_0000223.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d1b5d5dd4943566ca834dd9c8b3a76d54ecf3bf0fc4b2769f7f9f4c7b7cf1b66 +size 384 diff --git a/margin_logs/step_0000224.npy b/margin_logs/step_0000224.npy new file mode 100644 index 0000000..699d672 --- /dev/null +++ b/margin_logs/step_0000224.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:00f3167d55b5fba5eb2f127823b9f86ccc5b77389fba0ed9e45ca008c3ec7b6b +size 384 diff --git a/margin_logs/step_0000225.npy b/margin_logs/step_0000225.npy new file mode 100644 index 0000000..8e46b36 --- /dev/null +++ b/margin_logs/step_0000225.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ffbdbe2237547b7958837ea880dcd109b8978937cfb0a772a25c92bca24bba90 +size 384 diff --git a/margin_logs/step_0000226.npy b/margin_logs/step_0000226.npy new file mode 100644 index 0000000..c1bc48c --- /dev/null +++ b/margin_logs/step_0000226.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a16e35173ed0620403f93092c81a3ab341b20360074a940f5af1d7cc98ad3374 +size 384 diff --git a/margin_logs/step_0000227.npy b/margin_logs/step_0000227.npy new file mode 100644 index 0000000..3801e76 --- /dev/null +++ b/margin_logs/step_0000227.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:335bf7cff322dddf30257c23823e7ada9e49b8f2625182bdd170e144611c566a +size 384 diff --git a/margin_logs/step_0000228.npy b/margin_logs/step_0000228.npy new file mode 100644 index 0000000..63016e4 --- /dev/null +++ b/margin_logs/step_0000228.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:20c75f147eab2c4d83f7eb39ff4ae766b66a9a90a937bcf9fd37794fe908adbd +size 384 diff --git a/margin_logs/step_0000229.npy b/margin_logs/step_0000229.npy new file mode 100644 index 0000000..fb417e7 --- /dev/null +++ b/margin_logs/step_0000229.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:75a59dbc6ca7b417101b8b1bd8b475a2f648414e37ed6e1e5c623e9f3da2bf3e +size 384 diff --git a/margin_logs/step_0000230.npy b/margin_logs/step_0000230.npy new file mode 100644 index 0000000..87e8575 --- /dev/null +++ b/margin_logs/step_0000230.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b1d11eac093c9106ca44cf1dbcff06ef460620dca833a5c193187dc4d2170432 +size 384 diff --git a/margin_logs/step_0000231.npy b/margin_logs/step_0000231.npy new file mode 100644 index 0000000..cecc064 --- /dev/null +++ b/margin_logs/step_0000231.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:326a5da48084930ce81c20309724fa2be9985f82bac8c569584987523e6f02e4 +size 384 diff --git a/margin_logs/step_0000232.npy b/margin_logs/step_0000232.npy new file mode 100644 index 0000000..6304f98 --- /dev/null +++ b/margin_logs/step_0000232.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1264a5153a928f186d7d6ad461fbf8943150fbd28b73f80581392d6e2e981c18 +size 384 diff --git a/margin_logs/step_0000233.npy b/margin_logs/step_0000233.npy new file mode 100644 index 0000000..f6a14b6 --- /dev/null +++ b/margin_logs/step_0000233.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ec58834c4cce4107e483cbfccd3b984e9564fe2e40db44e60ec0d3562129f1f5 +size 384 diff --git a/margin_logs/step_0000234.npy b/margin_logs/step_0000234.npy new file mode 100644 index 0000000..e0b65ef --- /dev/null +++ b/margin_logs/step_0000234.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fa86442de4e0ee8698e81e85c4e65b8667bbdffe31d94f85ec21100d95167905 +size 384 diff --git a/margin_logs/step_0000235.npy b/margin_logs/step_0000235.npy new file mode 100644 index 0000000..388665c --- /dev/null +++ b/margin_logs/step_0000235.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aeb87d0d9308b1aca84ae030b2a391cd0ad528ac15ae378462bf03c6969a976d +size 384 diff --git a/margin_logs/step_0000236.npy b/margin_logs/step_0000236.npy new file mode 100644 index 0000000..58e803d --- /dev/null +++ b/margin_logs/step_0000236.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a69979f413c99213c5ce90e5e3551f3dea249141a569bf8f086c34922e930198 +size 384 diff --git a/margin_logs/step_0000237.npy b/margin_logs/step_0000237.npy new file mode 100644 index 0000000..c8ec58d --- /dev/null +++ b/margin_logs/step_0000237.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:802202f315dfa36a9ffc8b66c92197b280438515a7cc502caa19ba63572a01a8 +size 384 diff --git a/margin_logs/step_0000238.npy b/margin_logs/step_0000238.npy new file mode 100644 index 0000000..b8095d4 --- /dev/null +++ b/margin_logs/step_0000238.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a0e7db9c719e0745c16fa479a853cfebaeaabbeb85f69e01e3addb681cc9910c +size 384 diff --git a/margin_logs/step_0000239.npy b/margin_logs/step_0000239.npy new file mode 100644 index 0000000..2048a60 --- /dev/null +++ b/margin_logs/step_0000239.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:152ff1e61af30d350c1d5e9d1e51e4c078c244c533b1273b21cc8b30502c464d +size 384 diff --git a/margin_logs/step_0000240.npy b/margin_logs/step_0000240.npy new file mode 100644 index 0000000..740a88c --- /dev/null +++ b/margin_logs/step_0000240.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:79e30b205548a72d3e637aad0b7ba05d5d13ccf1308f4dfde456c5ab9ddd2f06 +size 384 diff --git a/margin_logs/step_0000241.npy b/margin_logs/step_0000241.npy new file mode 100644 index 0000000..c1929ba --- /dev/null +++ b/margin_logs/step_0000241.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a39a5d20765603e3bce23ca384205d5efc3dc281ad527c9b2f89d7383ed64901 +size 384 diff --git a/margin_logs/step_0000242.npy b/margin_logs/step_0000242.npy new file mode 100644 index 0000000..9ca33f9 --- /dev/null +++ b/margin_logs/step_0000242.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6eac1efdc38dea8391758a3a523fdfc1d4d539172a58da56b05add40e9fd0fba +size 384 diff --git a/margin_logs/step_0000243.npy b/margin_logs/step_0000243.npy new file mode 100644 index 0000000..585f08f --- /dev/null +++ b/margin_logs/step_0000243.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5b70d79ff1c279d85f9324d7808eb614e35552ac0c7fb872086c64a789ea94dd +size 384 diff --git a/margin_logs/step_0000244.npy b/margin_logs/step_0000244.npy new file mode 100644 index 0000000..763b8f5 --- /dev/null +++ b/margin_logs/step_0000244.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:14aac4e04ad94a4f4f153f79871c68681612ae0a5cda7ddb391936343882fe08 +size 384 diff --git a/margin_logs/step_0000245.npy b/margin_logs/step_0000245.npy new file mode 100644 index 0000000..bf28362 --- /dev/null +++ b/margin_logs/step_0000245.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d9fde9c7650ddaefdcbbdf8e8cb54e5fc22bb464c46efc4ef4236b964d99b0d9 +size 384 diff --git a/margin_logs/step_0000246.npy b/margin_logs/step_0000246.npy new file mode 100644 index 0000000..f5a1f43 --- /dev/null +++ b/margin_logs/step_0000246.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:70cd88c8638f90446e94beff8b49b6fb281cab0ae540f8e8b721c1187ef25717 +size 384 diff --git a/margin_logs/step_0000247.npy b/margin_logs/step_0000247.npy new file mode 100644 index 0000000..0413cde --- /dev/null +++ b/margin_logs/step_0000247.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5bc08cd280f092cbd5fab0896bf6398a58929c7b6ccac67e7f1aa4b6ec231f3f +size 384 diff --git a/margin_logs/step_0000248.npy b/margin_logs/step_0000248.npy new file mode 100644 index 0000000..92450a9 --- /dev/null +++ b/margin_logs/step_0000248.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fcf7be409cd1acc588ff181edb97dc785cfa25da811ee6af715b0593e6568c9a +size 384 diff --git a/margin_logs/step_0000249.npy b/margin_logs/step_0000249.npy new file mode 100644 index 0000000..3073f84 --- /dev/null +++ b/margin_logs/step_0000249.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7e6d323588211562a7567b4efb5b1aba90f0a0b1bdd56724ba7003cae0618024 +size 384 diff --git a/margin_logs/step_0000250.npy b/margin_logs/step_0000250.npy new file mode 100644 index 0000000..ac48876 --- /dev/null +++ b/margin_logs/step_0000250.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2d62799b1ad82b94a9312c2f7471c3f113952bd90d252fb3736ec19c923a6dc2 +size 384 diff --git a/margin_logs/step_0000251.npy b/margin_logs/step_0000251.npy new file mode 100644 index 0000000..b3c6364 --- /dev/null +++ b/margin_logs/step_0000251.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ef4e5e4f40de0e0eb5621e896a7e008f70ceb5e4d88a58489fa03ca51f038253 +size 384 diff --git a/margin_logs/step_0000252.npy b/margin_logs/step_0000252.npy new file mode 100644 index 0000000..f4bf546 --- /dev/null +++ b/margin_logs/step_0000252.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d435e49b605da6fe255d911940f6e533b6a093eddbe4e020f10992f5ee51b647 +size 384 diff --git a/margin_logs/step_0000253.npy b/margin_logs/step_0000253.npy new file mode 100644 index 0000000..889fed8 --- /dev/null +++ b/margin_logs/step_0000253.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0bfe5e718736a2292ec3145c253ff0d74bae7e51929ef202ecebe2687383b69a +size 384 diff --git a/margin_logs/step_0000254.npy b/margin_logs/step_0000254.npy new file mode 100644 index 0000000..09b6c9b --- /dev/null +++ b/margin_logs/step_0000254.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2f32017f363c4ac9caf4a463796415ecb10d3d4fb96d57da9119e424f9559f06 +size 384 diff --git a/margin_logs/step_0000255.npy b/margin_logs/step_0000255.npy new file mode 100644 index 0000000..f624970 --- /dev/null +++ b/margin_logs/step_0000255.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:accda963b42c61c253305e781cc1c6787ec42ce41e49936438491c2872594d8f +size 384 diff --git a/margin_logs/step_0000256.npy b/margin_logs/step_0000256.npy new file mode 100644 index 0000000..30c1707 --- /dev/null +++ b/margin_logs/step_0000256.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:78db0e1a5968acbb208fc2ce2e25576d94130a32fe8881414ec535150f5f47e1 +size 384 diff --git a/margin_logs/step_0000257.npy b/margin_logs/step_0000257.npy new file mode 100644 index 0000000..f61249f --- /dev/null +++ b/margin_logs/step_0000257.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ede687a4db428788de636ef61ad23987ea2b8844234af578080a35ba5abb9a98 +size 384 diff --git a/margin_logs/step_0000258.npy b/margin_logs/step_0000258.npy new file mode 100644 index 0000000..ff1d27e --- /dev/null +++ b/margin_logs/step_0000258.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dc6786651a1686131d656e7bb90840ca2cb4dca439d426eea9607ae65ea5e0ab +size 384 diff --git a/margin_logs/step_0000259.npy b/margin_logs/step_0000259.npy new file mode 100644 index 0000000..9d620dd --- /dev/null +++ b/margin_logs/step_0000259.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8087d3fc2cc79629214342c7ab7fc3bc5dd28d761640653b5efe5393b8a35e52 +size 384 diff --git a/margin_logs/step_0000260.npy b/margin_logs/step_0000260.npy new file mode 100644 index 0000000..ce86553 --- /dev/null +++ b/margin_logs/step_0000260.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b17d7e73105d19abc3257892f600abd0ffd920f3555c491964a5b91892de61c9 +size 384 diff --git a/margin_logs/step_0000261.npy b/margin_logs/step_0000261.npy new file mode 100644 index 0000000..7ccbf3f --- /dev/null +++ b/margin_logs/step_0000261.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3f16d007b8af72e853b14a9e3a14ef75ff57fd9416908956e19b27f3640c7f24 +size 384 diff --git a/margin_logs/step_0000262.npy b/margin_logs/step_0000262.npy new file mode 100644 index 0000000..6fc4987 --- /dev/null +++ b/margin_logs/step_0000262.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:78c80d1ec93b22bdcce1043b674dd066798895dfa8a25e846d67bdcec39d3042 +size 384 diff --git a/margin_logs/step_0000263.npy b/margin_logs/step_0000263.npy new file mode 100644 index 0000000..c5e5b2d --- /dev/null +++ b/margin_logs/step_0000263.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8938698e93751a60834b4ee96cf746b8ef285e6fc6bbe10a3e7f31532d5bc2c5 +size 384 diff --git a/margin_logs/step_0000264.npy b/margin_logs/step_0000264.npy new file mode 100644 index 0000000..de78d0c --- /dev/null +++ b/margin_logs/step_0000264.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:51dd0e5cebafb4bdfa0638bce83c526d723eb24602c1995256d87dbf0156238f +size 384 diff --git a/margin_logs/step_0000265.npy b/margin_logs/step_0000265.npy new file mode 100644 index 0000000..8dc0cc4 --- /dev/null +++ b/margin_logs/step_0000265.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f746a0b123374bbb12bd4b5f11b6f54630cd14fe12aa33b0d7a198207305f2c1 +size 384 diff --git a/margin_logs/step_0000266.npy b/margin_logs/step_0000266.npy new file mode 100644 index 0000000..a8d143f --- /dev/null +++ b/margin_logs/step_0000266.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a826689d377c0f9f017e1318a97c59372c2f9387d91129fc8ac4960607fcad99 +size 384 diff --git a/margin_logs/step_0000267.npy b/margin_logs/step_0000267.npy new file mode 100644 index 0000000..b246b3a --- /dev/null +++ b/margin_logs/step_0000267.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f5366e19982c71f7ca8f542d41d15e519809ca3414da9e21af8d86fd5b222475 +size 384 diff --git a/margin_logs/step_0000268.npy b/margin_logs/step_0000268.npy new file mode 100644 index 0000000..3e5efc8 --- /dev/null +++ b/margin_logs/step_0000268.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:504dfbfe4a4c91230c64d78d9d62cded4d66610f405cf0c6697b0517fe9f3d88 +size 384 diff --git a/margin_logs/step_0000269.npy b/margin_logs/step_0000269.npy new file mode 100644 index 0000000..5ba41a9 --- /dev/null +++ b/margin_logs/step_0000269.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d88f0125f8575e457c8a83ea5840729f7c26bd45751e0e8e45b92c6bb1f479fb +size 384 diff --git a/margin_logs/step_0000270.npy b/margin_logs/step_0000270.npy new file mode 100644 index 0000000..f4c866f --- /dev/null +++ b/margin_logs/step_0000270.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8756052278b61d74fcf5733469b0061ede1e13e8760521c2d435a2fd2b2294c9 +size 384 diff --git a/margin_logs/step_0000271.npy b/margin_logs/step_0000271.npy new file mode 100644 index 0000000..7e1095b --- /dev/null +++ b/margin_logs/step_0000271.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:720c1ae532e845497969413eb6406e9f868a1af9b361d35883d9bf4d17f75090 +size 384 diff --git a/margin_logs/step_0000272.npy b/margin_logs/step_0000272.npy new file mode 100644 index 0000000..fa8f287 --- /dev/null +++ b/margin_logs/step_0000272.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0d20e8e5dceb308f42eb1d084e03c3fa9ce88310b059ab37bfbe2a05a2e6c3c0 +size 384 diff --git a/margin_logs/step_0000273.npy b/margin_logs/step_0000273.npy new file mode 100644 index 0000000..c5e00f1 --- /dev/null +++ b/margin_logs/step_0000273.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a600aec6176bb96988a5921757acb6d874375a279092f4a447863e88bfed9c25 +size 384 diff --git a/margin_logs/step_0000274.npy b/margin_logs/step_0000274.npy new file mode 100644 index 0000000..c65393e --- /dev/null +++ b/margin_logs/step_0000274.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0a083b64acf32b981ad76baaa7a7d00f7e54bff0c714122e4ba847eb0a3f409a +size 384 diff --git a/margin_logs/step_0000275.npy b/margin_logs/step_0000275.npy new file mode 100644 index 0000000..b303952 --- /dev/null +++ b/margin_logs/step_0000275.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7caaa1a01ce1a83f31d33f3486a36d6fe372f3a07c2d11a6d532d77f1a7cd4ef +size 384 diff --git a/margin_logs/step_0000276.npy b/margin_logs/step_0000276.npy new file mode 100644 index 0000000..01e2ed8 --- /dev/null +++ b/margin_logs/step_0000276.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f0ff63cb1f937d310319d05db899db1da820486f020d7126c4e9144c3533e3ab +size 384 diff --git a/margin_logs/step_0000277.npy b/margin_logs/step_0000277.npy new file mode 100644 index 0000000..4a35886 --- /dev/null +++ b/margin_logs/step_0000277.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e3575c36d7d8b6e1a0a94cddb8360502a1bf5c476a6208393ac119407ed20ebf +size 384 diff --git a/margin_logs/step_0000278.npy b/margin_logs/step_0000278.npy new file mode 100644 index 0000000..071fc55 --- /dev/null +++ b/margin_logs/step_0000278.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c4735c8d92f86e43e1fffcc9ca5220c4fa0827da17d1cff99b95a25e48ad77c0 +size 384 diff --git a/margin_logs/step_0000279.npy b/margin_logs/step_0000279.npy new file mode 100644 index 0000000..42eda6f --- /dev/null +++ b/margin_logs/step_0000279.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:820cb2fceb4ead6d546d070defb94144add2c0d3f7f9d964f7d0a6cb3a345a72 +size 384 diff --git a/margin_logs/step_0000280.npy b/margin_logs/step_0000280.npy new file mode 100644 index 0000000..1803c7b --- /dev/null +++ b/margin_logs/step_0000280.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8cd0c228bf873f293b4e9a97d15bca8de5d0a235bf742b7dd83f2d5b6101f458 +size 384 diff --git a/margin_logs/step_0000281.npy b/margin_logs/step_0000281.npy new file mode 100644 index 0000000..9f6b8df --- /dev/null +++ b/margin_logs/step_0000281.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:619335f100d0adb36b1169a9db6700fb5c0ca67f44be77def2fbce0ad5887cfb +size 384 diff --git a/margin_logs/step_0000282.npy b/margin_logs/step_0000282.npy new file mode 100644 index 0000000..682743c --- /dev/null +++ b/margin_logs/step_0000282.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b605f762fb690a896e2da8de0039e0e09d51f692291951465cbc94a7ff2910aa +size 384 diff --git a/margin_logs/step_0000283.npy b/margin_logs/step_0000283.npy new file mode 100644 index 0000000..74cf2df --- /dev/null +++ b/margin_logs/step_0000283.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c945e07d9cba28be129f8f1779672aea7cb2397698565f9f3713634b9075723d +size 384 diff --git a/margin_logs/step_0000284.npy b/margin_logs/step_0000284.npy new file mode 100644 index 0000000..2a07a3b --- /dev/null +++ b/margin_logs/step_0000284.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:42b71a2043782d8bc94ad351ec8a17fa51ce75363366b564e8dcda36aebb4795 +size 384 diff --git a/margin_logs/step_0000285.npy b/margin_logs/step_0000285.npy new file mode 100644 index 0000000..88bc67a --- /dev/null +++ b/margin_logs/step_0000285.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d76b115c4704ea1d55022e7770d968b8aa41a011adb7a074f6fdbcf3809b4238 +size 384 diff --git a/margin_logs/step_0000286.npy b/margin_logs/step_0000286.npy new file mode 100644 index 0000000..178dd59 --- /dev/null +++ b/margin_logs/step_0000286.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:713284d61699e5a570a3b2b6d53ed4aa3cd2796696298cc957d291599c3b339b +size 384 diff --git a/margin_logs/step_0000287.npy b/margin_logs/step_0000287.npy new file mode 100644 index 0000000..8b3e2b2 --- /dev/null +++ b/margin_logs/step_0000287.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:559b55bb2c230243190e8dee252b9dcfbfa1d5a2a3ed6263331228b15ebab0c0 +size 384 diff --git a/margin_logs/step_0000288.npy b/margin_logs/step_0000288.npy new file mode 100644 index 0000000..d895d90 --- /dev/null +++ b/margin_logs/step_0000288.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a381c3e75b3362bc0f756c694ae4b5aebdc2e695e8d853b92a7667b24214d12d +size 384 diff --git a/margin_logs/step_0000289.npy b/margin_logs/step_0000289.npy new file mode 100644 index 0000000..f9eebd0 --- /dev/null +++ b/margin_logs/step_0000289.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2f4e519f33e1ae1e0172ff015c9c4a018c081a3f8783da03b001aa3d37864534 +size 384 diff --git a/margin_logs/step_0000290.npy b/margin_logs/step_0000290.npy new file mode 100644 index 0000000..2ba9460 --- /dev/null +++ b/margin_logs/step_0000290.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d4b583c390ac80b1244db3e7ad40e61617fb582143eb98bbd5d4f8fa3e488356 +size 384 diff --git a/margin_logs/step_0000291.npy b/margin_logs/step_0000291.npy new file mode 100644 index 0000000..a97ed24 --- /dev/null +++ b/margin_logs/step_0000291.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4fac5f232b8dc64f1604b205d31ec6153c6a280e4f9a769aa915de173a7cfce9 +size 384 diff --git a/margin_logs/step_0000292.npy b/margin_logs/step_0000292.npy new file mode 100644 index 0000000..30e8c4d --- /dev/null +++ b/margin_logs/step_0000292.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:71aed0aa600426ca823a4c7847b525d202ca5a8883e5b1b481d9c599206c69fe +size 384 diff --git a/margin_logs/step_0000293.npy b/margin_logs/step_0000293.npy new file mode 100644 index 0000000..e9e0347 --- /dev/null +++ b/margin_logs/step_0000293.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6730494ffb3a9d0ae6370d7e13a8a9ec29bff5cfed96fdf0527dbf2b6ca9c1ef +size 384 diff --git a/margin_logs/step_0000294.npy b/margin_logs/step_0000294.npy new file mode 100644 index 0000000..926c945 --- /dev/null +++ b/margin_logs/step_0000294.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6fc7f11df72af78189beb4c9880f7a880de558144a8723a1629eacbeb4e453a +size 384 diff --git a/margin_logs/step_0000295.npy b/margin_logs/step_0000295.npy new file mode 100644 index 0000000..f5671fb --- /dev/null +++ b/margin_logs/step_0000295.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0c48de2bab801aaf67b30c40a20769f2f2696a3b594cfc705ec57c567ce5d8f4 +size 384 diff --git a/margin_logs/step_0000296.npy b/margin_logs/step_0000296.npy new file mode 100644 index 0000000..c7dbb28 --- /dev/null +++ b/margin_logs/step_0000296.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a9ddde15292c19b2867136a2642ae9d57958e9d0741d5c98c499c4849f706631 +size 384 diff --git a/margin_logs/step_0000297.npy b/margin_logs/step_0000297.npy new file mode 100644 index 0000000..2eb26e1 --- /dev/null +++ b/margin_logs/step_0000297.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:abf1994174e898bbb6522f77c8460a8d80e220cafe6601ea273148aeaff330bf +size 384 diff --git a/margin_logs/step_0000298.npy b/margin_logs/step_0000298.npy new file mode 100644 index 0000000..9210385 --- /dev/null +++ b/margin_logs/step_0000298.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:38b91c22933e8aa743924a8f12e2528a9a424bbb55a9850c0ff58ef635b0a683 +size 384 diff --git a/margin_logs/step_0000299.npy b/margin_logs/step_0000299.npy new file mode 100644 index 0000000..b008b85 --- /dev/null +++ b/margin_logs/step_0000299.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c5d99911930b0fd87b1e8354afcc6a5f7cf1bec6a179e0c0a030217cfcb346c9 +size 384 diff --git a/margin_logs/step_0000300.npy b/margin_logs/step_0000300.npy new file mode 100644 index 0000000..620eb15 --- /dev/null +++ b/margin_logs/step_0000300.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:48223548d5a748e272836b0cccbdf5f00df9998488964a8f2ca0b050cedefc4b +size 384 diff --git a/margin_logs/step_0000301.npy b/margin_logs/step_0000301.npy new file mode 100644 index 0000000..d9be56b --- /dev/null +++ b/margin_logs/step_0000301.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:554daea21665cf032a0bcd11d8d6e7fdcf297a2650dd12393675d9ab61b48091 +size 384 diff --git a/margin_logs/step_0000302.npy b/margin_logs/step_0000302.npy new file mode 100644 index 0000000..962809c --- /dev/null +++ b/margin_logs/step_0000302.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5e2648e9a248388ceddd7487a16369a885f578488637b86edb0ee3316f7d78e8 +size 384 diff --git a/margin_logs/step_0000303.npy b/margin_logs/step_0000303.npy new file mode 100644 index 0000000..2d5a0ef --- /dev/null +++ b/margin_logs/step_0000303.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b054806bfd08a4577b0b4a7dcb10757fedb697f64f48473a63da1a50f12daaf +size 384 diff --git a/margin_logs/step_0000304.npy b/margin_logs/step_0000304.npy new file mode 100644 index 0000000..ac8049c --- /dev/null +++ b/margin_logs/step_0000304.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:26a74c57fa6d69ea9e804b9bde8ac4ab44943a4d44aa711000fbf71c60b9126c +size 384 diff --git a/margin_logs/step_0000305.npy b/margin_logs/step_0000305.npy new file mode 100644 index 0000000..3b146d8 --- /dev/null +++ b/margin_logs/step_0000305.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:17f81953c7e5327276e6128cd7ea7ef6c8746ec667cedb4e6c1d56e888afac93 +size 384 diff --git a/margin_logs/step_0000306.npy b/margin_logs/step_0000306.npy new file mode 100644 index 0000000..e0e0152 --- /dev/null +++ b/margin_logs/step_0000306.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:59dde6dd9af80768a47e72d327b3370aeab448de133175e426b055590fdf4823 +size 384 diff --git a/margin_logs/step_0000307.npy b/margin_logs/step_0000307.npy new file mode 100644 index 0000000..1cb5fcc --- /dev/null +++ b/margin_logs/step_0000307.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3977e5ddbb576c0ab536f57f9b2b3988c14277b4f3f37df364479d67501fbe1c +size 384 diff --git a/margin_logs/step_0000308.npy b/margin_logs/step_0000308.npy new file mode 100644 index 0000000..154dea2 --- /dev/null +++ b/margin_logs/step_0000308.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1049bf50b1a70913c202f8a8b016a2621f62b88883e72f40435bbe983f37b44c +size 384 diff --git a/margin_logs/step_0000309.npy b/margin_logs/step_0000309.npy new file mode 100644 index 0000000..8d61d0b --- /dev/null +++ b/margin_logs/step_0000309.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3497c4b66fbbd0b6054db997452395ce87fb40c90e14e29a3a1d099e92635089 +size 384 diff --git a/margin_logs/step_0000310.npy b/margin_logs/step_0000310.npy new file mode 100644 index 0000000..1c0c30c --- /dev/null +++ b/margin_logs/step_0000310.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e715bfa3302a1c5bcf18f5e61f9c910848318a96c76bad919c5f5e409b2c1f27 +size 384 diff --git a/margin_logs/step_0000311.npy b/margin_logs/step_0000311.npy new file mode 100644 index 0000000..a685a3b --- /dev/null +++ b/margin_logs/step_0000311.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eafa01f16be9a0aac9b404e47a191c3b5d89bbf254eba1edc1059575241e6ef9 +size 384 diff --git a/margin_logs/step_0000312.npy b/margin_logs/step_0000312.npy new file mode 100644 index 0000000..8175636 --- /dev/null +++ b/margin_logs/step_0000312.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1ac133eb84ac78f1287e8d0a6e8626a59480c0e8e532d1e04b13e6c79fec16e3 +size 384 diff --git a/margin_logs/step_0000313.npy b/margin_logs/step_0000313.npy new file mode 100644 index 0000000..e273c17 --- /dev/null +++ b/margin_logs/step_0000313.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:98c047ba3340109f901a50b8eaa3bd9b8fb35501711fea1cd3e37d031479a0d8 +size 384 diff --git a/margin_logs/step_0000314.npy b/margin_logs/step_0000314.npy new file mode 100644 index 0000000..a2c1d89 --- /dev/null +++ b/margin_logs/step_0000314.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8fbebcb359426d5b768df93712f972d0ca7930ab1da017be6262506278112f8b +size 384 diff --git a/margin_logs/step_0000315.npy b/margin_logs/step_0000315.npy new file mode 100644 index 0000000..3c4acc2 --- /dev/null +++ b/margin_logs/step_0000315.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:90806d5af62a945361d92387afbfce52c3d84a4457ef85fc8c70337c2a14e8d6 +size 384 diff --git a/margin_logs/step_0000316.npy b/margin_logs/step_0000316.npy new file mode 100644 index 0000000..fcdebab --- /dev/null +++ b/margin_logs/step_0000316.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:654f899471b801bf46925831f6ae8f7d459a76d416e85476ff51ef3f6c8cbaa5 +size 384 diff --git a/margin_logs/step_0000317.npy b/margin_logs/step_0000317.npy new file mode 100644 index 0000000..57c481c --- /dev/null +++ b/margin_logs/step_0000317.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9723b12ef433f2ea0dea29a66a32d9f4c770c8071a542394d94d2b86d60a88ae +size 384 diff --git a/margin_logs/step_0000318.npy b/margin_logs/step_0000318.npy new file mode 100644 index 0000000..d141aa6 --- /dev/null +++ b/margin_logs/step_0000318.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8a01bea87d6cd67461e284da2ddbc61f62e7d6566592b8368c489ea73cee8c65 +size 384 diff --git a/margin_logs/step_0000319.npy b/margin_logs/step_0000319.npy new file mode 100644 index 0000000..9cf2d64 --- /dev/null +++ b/margin_logs/step_0000319.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bb5a0192ae2c2e660fd8e8a61f7b8e1dd53757067a4e9812fbaef540751931fc +size 384 diff --git a/margin_logs/step_0000320.npy b/margin_logs/step_0000320.npy new file mode 100644 index 0000000..a6df418 --- /dev/null +++ b/margin_logs/step_0000320.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4c57fe10e043bee9b1b663580507930444c4e87e277984fee53b14b387dde152 +size 384 diff --git a/margin_logs/step_0000321.npy b/margin_logs/step_0000321.npy new file mode 100644 index 0000000..fabf80d --- /dev/null +++ b/margin_logs/step_0000321.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7603c6a4fcfcc319592406211c1ef0334ed4093c9b180d11ea4e127fcf4de06c +size 384 diff --git a/margin_logs/step_0000322.npy b/margin_logs/step_0000322.npy new file mode 100644 index 0000000..f345e5a --- /dev/null +++ b/margin_logs/step_0000322.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ef8375cef80dbb82ef465d198f936c436334557e79ccc586023131d8841421e2 +size 384 diff --git a/margin_logs/step_0000323.npy b/margin_logs/step_0000323.npy new file mode 100644 index 0000000..d4261a8 --- /dev/null +++ b/margin_logs/step_0000323.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6b1ba75608aaf7ca745ae7af0374881737b4ab9ad04ccf707b5b32483747def9 +size 384 diff --git a/margin_logs/step_0000324.npy b/margin_logs/step_0000324.npy new file mode 100644 index 0000000..dc1ed77 --- /dev/null +++ b/margin_logs/step_0000324.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5835707c4851cf78af6c8ece68197739140b8646f8d155b7394326b0fad3327a +size 384 diff --git a/margin_logs/step_0000325.npy b/margin_logs/step_0000325.npy new file mode 100644 index 0000000..95b483c --- /dev/null +++ b/margin_logs/step_0000325.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:026fe4b28640cbce9f3f3c30884cb3b4979a54b34b112cce754458daa8fdc40d +size 384 diff --git a/margin_logs/step_0000326.npy b/margin_logs/step_0000326.npy new file mode 100644 index 0000000..4b5606b --- /dev/null +++ b/margin_logs/step_0000326.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:50673d4fb6671eafab1ec668e03fe2779dbf37bafd750c9a1100d6225bc9322e +size 384 diff --git a/margin_logs/step_0000327.npy b/margin_logs/step_0000327.npy new file mode 100644 index 0000000..14cd97c --- /dev/null +++ b/margin_logs/step_0000327.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3fa42afc1ee13b5f0b863b938cba035f22e47f6d162c60fabb4fb53fb90111c9 +size 384 diff --git a/margin_logs/step_0000328.npy b/margin_logs/step_0000328.npy new file mode 100644 index 0000000..711d584 --- /dev/null +++ b/margin_logs/step_0000328.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c86518fa7e0c69b03333e7dd710afd12edf9e6559f01d6b8c5261c71a1f5cf55 +size 384 diff --git a/margin_logs/step_0000329.npy b/margin_logs/step_0000329.npy new file mode 100644 index 0000000..174ce20 --- /dev/null +++ b/margin_logs/step_0000329.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0782b2f85d11a074a383330ddc35ab51b8a20d6fb945f4fbd5d00080b2612570 +size 384 diff --git a/margin_logs/step_0000330.npy b/margin_logs/step_0000330.npy new file mode 100644 index 0000000..57082cf --- /dev/null +++ b/margin_logs/step_0000330.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f79f40dcdbb162a5736c04837d62c58ddc633beec210b8c3ead4235b37294468 +size 384 diff --git a/margin_logs/step_0000331.npy b/margin_logs/step_0000331.npy new file mode 100644 index 0000000..1a9a47c --- /dev/null +++ b/margin_logs/step_0000331.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0acdfa2276ac76f4ae95307462b13866ff7ecbfb4cd38894ebcaea684111a602 +size 384 diff --git a/margin_logs/step_0000332.npy b/margin_logs/step_0000332.npy new file mode 100644 index 0000000..00ef1d1 --- /dev/null +++ b/margin_logs/step_0000332.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fa60a168de49bd4326ac612e3a5f3b2089a3b3837ff5f5eef64e399d35008917 +size 384 diff --git a/margin_logs/step_0000333.npy b/margin_logs/step_0000333.npy new file mode 100644 index 0000000..bb20ef2 --- /dev/null +++ b/margin_logs/step_0000333.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f68abe7031487a1767e6d6a61974406436f6804aaf9e226da2d601c442664f5 +size 384 diff --git a/margin_logs/step_0000334.npy b/margin_logs/step_0000334.npy new file mode 100644 index 0000000..7ec1e44 --- /dev/null +++ b/margin_logs/step_0000334.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:022d71d48f84670ca3778711d0be7bf21f0b25c00763e3f3d09c532003209f74 +size 384 diff --git a/margin_logs/step_0000335.npy b/margin_logs/step_0000335.npy new file mode 100644 index 0000000..eda4f72 --- /dev/null +++ b/margin_logs/step_0000335.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f9da2110f3eec58d228cbaf9e8bb6b771a3b263729587174ac1003753dd7dc63 +size 384 diff --git a/margin_logs/step_0000336.npy b/margin_logs/step_0000336.npy new file mode 100644 index 0000000..14499a1 --- /dev/null +++ b/margin_logs/step_0000336.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:75f1594549faa8da621f80aacb0e3c711b2fcf1b1b612d7e1a8123e5e232af55 +size 384 diff --git a/margin_logs/step_0000337.npy b/margin_logs/step_0000337.npy new file mode 100644 index 0000000..1fa049e --- /dev/null +++ b/margin_logs/step_0000337.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:25e2dd1e66ae64344522aff9d58e58a7bcea5671e744666c43cca645edd192ab +size 384 diff --git a/margin_logs/step_0000338.npy b/margin_logs/step_0000338.npy new file mode 100644 index 0000000..541e8d7 --- /dev/null +++ b/margin_logs/step_0000338.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e43c69ffe3a1c0a4ab1aeb5165346e04561398d8bdd7ca274f6255149a2aa1aa +size 384 diff --git a/margin_logs/step_0000339.npy b/margin_logs/step_0000339.npy new file mode 100644 index 0000000..b9e6722 --- /dev/null +++ b/margin_logs/step_0000339.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eb95da506f8a219c2386f5232a4aa915e761caae9dfe7a9ecadd20bf364277f1 +size 384 diff --git a/margin_logs/step_0000340.npy b/margin_logs/step_0000340.npy new file mode 100644 index 0000000..7a8bbc0 --- /dev/null +++ b/margin_logs/step_0000340.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:64aea1b7a937abc45ddda113c8350b96ad8a6b29d9a759d54643d434fbdd1632 +size 384 diff --git a/margin_logs/step_0000341.npy b/margin_logs/step_0000341.npy new file mode 100644 index 0000000..a913ce0 --- /dev/null +++ b/margin_logs/step_0000341.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:75a8afcd2e27e486ca87719a28d1cee33bdd4ef6923ba6053961c161d2ec9090 +size 384 diff --git a/margin_logs/step_0000342.npy b/margin_logs/step_0000342.npy new file mode 100644 index 0000000..91409b7 --- /dev/null +++ b/margin_logs/step_0000342.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a6e92423e0ab9a48e7a92402877d6066e3930c58c08de728e687aecc404279ac +size 384 diff --git a/margin_logs/step_0000343.npy b/margin_logs/step_0000343.npy new file mode 100644 index 0000000..a0f4517 --- /dev/null +++ b/margin_logs/step_0000343.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e8293711e5d326a70c0cea49827f67fbb91cb2c66e9e6e462262c2c9ab128a21 +size 384 diff --git a/margin_logs/step_0000344.npy b/margin_logs/step_0000344.npy new file mode 100644 index 0000000..9cb4872 --- /dev/null +++ b/margin_logs/step_0000344.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e1fa5fcc04d110d989e5800956648d792a477e155fffb0748c3cef879af17af0 +size 384 diff --git a/margin_logs/step_0000345.npy b/margin_logs/step_0000345.npy new file mode 100644 index 0000000..bcc7edd --- /dev/null +++ b/margin_logs/step_0000345.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f91f1b7aab3523b6308d533e44e5d27f75e9ca249fb9d6432f1b59035ec97b13 +size 384 diff --git a/margin_logs/step_0000346.npy b/margin_logs/step_0000346.npy new file mode 100644 index 0000000..b293bd8 --- /dev/null +++ b/margin_logs/step_0000346.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d1d7982d9cd3d97f0fc4512077d3ae8f49fb9aaaa55ecc0e17b93c48983e0905 +size 384 diff --git a/margin_logs/step_0000347.npy b/margin_logs/step_0000347.npy new file mode 100644 index 0000000..53fd492 --- /dev/null +++ b/margin_logs/step_0000347.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f291eca08e935f6b8d430ef03aed1d5426d47a8cfdec9f3d3edbcecb2ad5a544 +size 384 diff --git a/margin_logs/step_0000348.npy b/margin_logs/step_0000348.npy new file mode 100644 index 0000000..d87e781 --- /dev/null +++ b/margin_logs/step_0000348.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:815f5e5c349f0a01d308166a085784eff6caaeb95cbe2f01f35e4efb67919ac0 +size 384 diff --git a/margin_logs/step_0000349.npy b/margin_logs/step_0000349.npy new file mode 100644 index 0000000..2af23a0 --- /dev/null +++ b/margin_logs/step_0000349.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f3f58178ec6266c5d16be4b1081cbebc1e928a67d29a6bd28868d386b2994e5a +size 384 diff --git a/margin_logs/step_0000350.npy b/margin_logs/step_0000350.npy new file mode 100644 index 0000000..cfeaef9 --- /dev/null +++ b/margin_logs/step_0000350.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e95a10ea2f87d3bf5b98f7623c9522400e503ac5fa12846f70ba01367ea4cada +size 384 diff --git a/margin_logs/step_0000351.npy b/margin_logs/step_0000351.npy new file mode 100644 index 0000000..690635e --- /dev/null +++ b/margin_logs/step_0000351.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6d68f16eae4db3e6fa11aaf40ee12a0df89b1e7acb516796f89c3e06baa64f16 +size 384 diff --git a/margin_logs/step_0000352.npy b/margin_logs/step_0000352.npy new file mode 100644 index 0000000..19d0ef2 --- /dev/null +++ b/margin_logs/step_0000352.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1b5cdfaaedb1306d05f144ad9a33b407dea6bce78b4b0584f0f44591c19a17a8 +size 384 diff --git a/margin_logs/step_0000353.npy b/margin_logs/step_0000353.npy new file mode 100644 index 0000000..21a5680 --- /dev/null +++ b/margin_logs/step_0000353.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b7c8a96f1c39d4d761ae7a2bba7e2d1af8ecee4d40f30a2430f0d95c4cc9211f +size 384 diff --git a/margin_logs/step_0000354.npy b/margin_logs/step_0000354.npy new file mode 100644 index 0000000..92b2e05 --- /dev/null +++ b/margin_logs/step_0000354.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ed47aca190bfcfb4ad1f37dac1597a55fed1618de704595c046edbdeb7ed098a +size 384 diff --git a/margin_logs/step_0000355.npy b/margin_logs/step_0000355.npy new file mode 100644 index 0000000..222b61a --- /dev/null +++ b/margin_logs/step_0000355.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8ac7a7ed933388e3b2527bb4e8d0cd6cf1c09979bfbe49579df2106b4222bab3 +size 384 diff --git a/margin_logs/step_0000356.npy b/margin_logs/step_0000356.npy new file mode 100644 index 0000000..6a891bd --- /dev/null +++ b/margin_logs/step_0000356.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:735c34be2e00d42b835094c4a33ab3e065bd7f822e7c518a36d439f33c622c46 +size 384 diff --git a/margin_logs/step_0000357.npy b/margin_logs/step_0000357.npy new file mode 100644 index 0000000..640d87c --- /dev/null +++ b/margin_logs/step_0000357.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:71c6b44b5a3dcae5c6d449486f26e82478cffef036ed8330a480d3fdeeb1c002 +size 384 diff --git a/margin_logs/step_0000358.npy b/margin_logs/step_0000358.npy new file mode 100644 index 0000000..2b64f72 --- /dev/null +++ b/margin_logs/step_0000358.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:77c72a4a5b7b35c1097c4c523cf9156d9fb0c671693dde89b9c3efe47b1a0258 +size 384 diff --git a/margin_logs/step_0000359.npy b/margin_logs/step_0000359.npy new file mode 100644 index 0000000..a5f95b4 --- /dev/null +++ b/margin_logs/step_0000359.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b8207b548cd364fecfe6bd016468cc810c6d7063bce0f18d3fb93359d2826651 +size 384 diff --git a/margin_logs/step_0000360.npy b/margin_logs/step_0000360.npy new file mode 100644 index 0000000..d75a574 --- /dev/null +++ b/margin_logs/step_0000360.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:966e50e438f019c8a2e8ce3d06599aed41ca0a5dc6ca7b62d981d6d98b0cf768 +size 384 diff --git a/margin_logs/step_0000361.npy b/margin_logs/step_0000361.npy new file mode 100644 index 0000000..b519545 --- /dev/null +++ b/margin_logs/step_0000361.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0eb4e9d09f7fb1b5d92a4faed7cd8304b1e5120560d840fd1d640d42194d4b6a +size 384 diff --git a/margin_logs/step_0000362.npy b/margin_logs/step_0000362.npy new file mode 100644 index 0000000..22d34a1 --- /dev/null +++ b/margin_logs/step_0000362.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cfd5cf8c466972aa74bc7f5c0a321686a3555830d73a8d4ad7b7dca458d2d219 +size 384 diff --git a/margin_logs/step_0000363.npy b/margin_logs/step_0000363.npy new file mode 100644 index 0000000..21b8d41 --- /dev/null +++ b/margin_logs/step_0000363.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:528cda6d5d62aa314bea5aa0ccf0a4d1c7a5843da7c9f67fc602ae5de549e22a +size 384 diff --git a/margin_logs/step_0000364.npy b/margin_logs/step_0000364.npy new file mode 100644 index 0000000..0072df5 --- /dev/null +++ b/margin_logs/step_0000364.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f4430cc975742497500e4aee1ab36c66c5ffd50d6495befd5019b2efc76b2ca1 +size 384 diff --git a/margin_logs/step_0000365.npy b/margin_logs/step_0000365.npy new file mode 100644 index 0000000..4dc3f8b --- /dev/null +++ b/margin_logs/step_0000365.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4ebf2a889eae6e4fb91803d6b204c83105d5687a4f6d69913f0d58f52b94fa0c +size 384 diff --git a/margin_logs/step_0000366.npy b/margin_logs/step_0000366.npy new file mode 100644 index 0000000..86dd3d9 --- /dev/null +++ b/margin_logs/step_0000366.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:45da50756b99586ffa389aa64829728ab1be8d1d064dac2f282d9aa90d9743a0 +size 384 diff --git a/margin_logs/step_0000367.npy b/margin_logs/step_0000367.npy new file mode 100644 index 0000000..cfbea0c --- /dev/null +++ b/margin_logs/step_0000367.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:87cf323e8f9d02570730f8e7c67c994a880033b25891280aa559aa0e218e2ef3 +size 384 diff --git a/margin_logs/step_0000368.npy b/margin_logs/step_0000368.npy new file mode 100644 index 0000000..7197502 --- /dev/null +++ b/margin_logs/step_0000368.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3d4f72d93df28daf8f27e12d740b19db1341bed3742c7b4734d54bdaf95b73bc +size 384 diff --git a/margin_logs/step_0000369.npy b/margin_logs/step_0000369.npy new file mode 100644 index 0000000..5dffaab --- /dev/null +++ b/margin_logs/step_0000369.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0981d22a7fcccab2501d820aab162bb2fe64546cf8224d48e8f87be4a3b45358 +size 384 diff --git a/margin_logs/step_0000370.npy b/margin_logs/step_0000370.npy new file mode 100644 index 0000000..18ba40d --- /dev/null +++ b/margin_logs/step_0000370.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f48fe9a6b789f35a8e31c3aba88b44f746d2b6fe8e056d0348c8e9bb7e671af9 +size 384 diff --git a/margin_logs/step_0000371.npy b/margin_logs/step_0000371.npy new file mode 100644 index 0000000..13f9567 --- /dev/null +++ b/margin_logs/step_0000371.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e2ad96852392079841f102228bc03f6523df75f3b44db5cae362c310073e883a +size 384 diff --git a/margin_logs/step_0000372.npy b/margin_logs/step_0000372.npy new file mode 100644 index 0000000..06196f1 --- /dev/null +++ b/margin_logs/step_0000372.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:455289dbff38a7a65f43f1348c0457915cd22af416f0743f1ee3650e978e8bdd +size 384 diff --git a/margin_logs/step_0000373.npy b/margin_logs/step_0000373.npy new file mode 100644 index 0000000..23d3525 --- /dev/null +++ b/margin_logs/step_0000373.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cf9cbd33396df5fc7f391ebefbaf8804ce2fddf0d83148b79382a2363823e179 +size 384 diff --git a/margin_logs/step_0000374.npy b/margin_logs/step_0000374.npy new file mode 100644 index 0000000..1288328 --- /dev/null +++ b/margin_logs/step_0000374.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b692c1aac05c158404152225c4929a9ff7f2b78d745e44eb909e11ab670a07fa +size 384 diff --git a/margin_logs/step_0000375.npy b/margin_logs/step_0000375.npy new file mode 100644 index 0000000..f9b9f38 --- /dev/null +++ b/margin_logs/step_0000375.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a090a6d57e9c974f26828882a52bfb92a9b69a1340513f7a5335d31084e4f4c2 +size 384 diff --git a/margin_logs/step_0000376.npy b/margin_logs/step_0000376.npy new file mode 100644 index 0000000..e9d1659 --- /dev/null +++ b/margin_logs/step_0000376.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a5a65992e0c3640a0da54d8598d70d9b893c88174ffa3f8434c3f126d51665cb +size 384 diff --git a/margin_logs/step_0000377.npy b/margin_logs/step_0000377.npy new file mode 100644 index 0000000..bdec456 --- /dev/null +++ b/margin_logs/step_0000377.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0cf31935b077ab594ec4bc6c50cdadc0730c8b91df6ec38d66b00bd6b0283dae +size 384 diff --git a/margin_logs/step_0000378.npy b/margin_logs/step_0000378.npy new file mode 100644 index 0000000..8e878cc --- /dev/null +++ b/margin_logs/step_0000378.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f2e4260c28c990c0ec6e4159b217fbe32ca7354918b9672c5f3854c213db6e53 +size 384 diff --git a/margin_logs/step_0000379.npy b/margin_logs/step_0000379.npy new file mode 100644 index 0000000..d89e315 --- /dev/null +++ b/margin_logs/step_0000379.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:61a320592e463ebedaeaa01712d4f6ab18a33406edb31a855ed78f5858006769 +size 384 diff --git a/margin_logs/step_0000380.npy b/margin_logs/step_0000380.npy new file mode 100644 index 0000000..644f309 --- /dev/null +++ b/margin_logs/step_0000380.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:98f0fcd682f81f9c159c8c53fce75d58df609bf74a09877c81f0f6b5cc7c4e87 +size 384 diff --git a/margin_logs/step_0000381.npy b/margin_logs/step_0000381.npy new file mode 100644 index 0000000..54ad11c --- /dev/null +++ b/margin_logs/step_0000381.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d268e504cf4a13498ca60e2926f4719111056d26a2759ed6e54b6cc54a31a3be +size 384 diff --git a/margin_logs/step_0000382.npy b/margin_logs/step_0000382.npy new file mode 100644 index 0000000..e1fd950 --- /dev/null +++ b/margin_logs/step_0000382.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4348bc95999b74dbb087464ee2a8e6e318f2fe4db1b2fcb2dda4626e204a2648 +size 384 diff --git a/margin_logs/step_0000383.npy b/margin_logs/step_0000383.npy new file mode 100644 index 0000000..3f36aa8 --- /dev/null +++ b/margin_logs/step_0000383.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c7c61d19ca6df771e5f0aa219de92ec4a1315d36e814b746a526f2d15c243e0d +size 384 diff --git a/margin_logs/step_0000384.npy b/margin_logs/step_0000384.npy new file mode 100644 index 0000000..9023f87 --- /dev/null +++ b/margin_logs/step_0000384.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:adbfd4808330c35db730798148bb950cea83a9d93b88ab569cb359c17c3c3d4f +size 384 diff --git a/margin_logs/step_0000385.npy b/margin_logs/step_0000385.npy new file mode 100644 index 0000000..980223c --- /dev/null +++ b/margin_logs/step_0000385.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3c58a1fc9355ded953b02728742627a8bf77484ecb7c75b5e2e7b9e26193804d +size 384 diff --git a/margin_logs/step_0000386.npy b/margin_logs/step_0000386.npy new file mode 100644 index 0000000..d5c5ba0 --- /dev/null +++ b/margin_logs/step_0000386.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:15487ba3fc79d32a8afbd06895db6527193c8de3288993764541fce471f36c30 +size 384 diff --git a/margin_logs/step_0000387.npy b/margin_logs/step_0000387.npy new file mode 100644 index 0000000..572c50b --- /dev/null +++ b/margin_logs/step_0000387.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d530fc5a9d213b83a7a7c6598b9158d76985519fdae8e1da32dcac930196a8cb +size 384 diff --git a/margin_logs/step_0000388.npy b/margin_logs/step_0000388.npy new file mode 100644 index 0000000..7b5f5dd --- /dev/null +++ b/margin_logs/step_0000388.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f19f2de3a9ffae16b7203787346627f429b0fe9d26a21d8dd51cdb27f1f19412 +size 384 diff --git a/margin_logs/step_0000389.npy b/margin_logs/step_0000389.npy new file mode 100644 index 0000000..a1b2b96 --- /dev/null +++ b/margin_logs/step_0000389.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2d1162f4419311b13fd02c27fc90dd4e7d852936ddc4c4d623b6d4100f15c898 +size 384 diff --git a/margin_logs/step_0000390.npy b/margin_logs/step_0000390.npy new file mode 100644 index 0000000..c1d3eae --- /dev/null +++ b/margin_logs/step_0000390.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7aa4f38c27dc176ee4aab6e22b90ae10cd27b7e3e2be91838e62512e06427e69 +size 384 diff --git a/margin_logs/step_0000391.npy b/margin_logs/step_0000391.npy new file mode 100644 index 0000000..12ba198 --- /dev/null +++ b/margin_logs/step_0000391.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e0b0f6be00ad8a106eee49a8532076d6379ae326f2b8b8b64a8432ec2b65d5de +size 384 diff --git a/margin_logs/step_0000392.npy b/margin_logs/step_0000392.npy new file mode 100644 index 0000000..2d90264 --- /dev/null +++ b/margin_logs/step_0000392.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8a063ceff7db0c8dda9e52287bfc056513c357dbdaef56db152e6ee87fd0ff55 +size 384 diff --git a/margin_logs/step_0000393.npy b/margin_logs/step_0000393.npy new file mode 100644 index 0000000..0be6633 --- /dev/null +++ b/margin_logs/step_0000393.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:823eaaf8324ba21987629a8927a8757a631f01068a42f590a47e0b96472b9111 +size 384 diff --git a/margin_logs/step_0000394.npy b/margin_logs/step_0000394.npy new file mode 100644 index 0000000..a90f44c --- /dev/null +++ b/margin_logs/step_0000394.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d9b05b1ed810a155514e3ae2dd64f2d95320f35935f2e61c20b3e70ba02a43c6 +size 384 diff --git a/margin_logs/step_0000395.npy b/margin_logs/step_0000395.npy new file mode 100644 index 0000000..b93763e --- /dev/null +++ b/margin_logs/step_0000395.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f583e0c319793754f8740b7e8e63e9208791917748fa024a2fdd5d181860a232 +size 384 diff --git a/margin_logs/step_0000396.npy b/margin_logs/step_0000396.npy new file mode 100644 index 0000000..ce25686 --- /dev/null +++ b/margin_logs/step_0000396.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0154d1d42b376eeaa09ebd32375896c0f865f2999a98130385c2d9f8db5f0739 +size 384 diff --git a/margin_logs/step_0000397.npy b/margin_logs/step_0000397.npy new file mode 100644 index 0000000..df9c01c --- /dev/null +++ b/margin_logs/step_0000397.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:36f4a34cfe0df95d3d0c5ed40fcd736de12e0a48e0e449231fef1ecc9d61a825 +size 384 diff --git a/margin_logs/step_0000398.npy b/margin_logs/step_0000398.npy new file mode 100644 index 0000000..059cd99 --- /dev/null +++ b/margin_logs/step_0000398.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:abae896a55255948176ff94bb4591b4d2a4f0a24871b26ff1a8f0354e2b8f1df +size 384 diff --git a/margin_logs/step_0000399.npy b/margin_logs/step_0000399.npy new file mode 100644 index 0000000..871f2c3 --- /dev/null +++ b/margin_logs/step_0000399.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:acb1a4c72eb9201cec1520ce99c52cc0e958d169c879ac2f87b1bf35c1cee1de +size 384 diff --git a/margin_logs/step_0000400.npy b/margin_logs/step_0000400.npy new file mode 100644 index 0000000..0148171 --- /dev/null +++ b/margin_logs/step_0000400.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7c8aa5435257484a32e3798d019bdd4f15dfcb3ad821a682b95cfdee1b9177ae +size 384 diff --git a/margin_logs/step_0000401.npy b/margin_logs/step_0000401.npy new file mode 100644 index 0000000..0830f76 --- /dev/null +++ b/margin_logs/step_0000401.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a775e779b88490be26785d038c0a1f15787dfa8abc9048562bdd70db78dfc764 +size 384 diff --git a/margin_logs/step_0000402.npy b/margin_logs/step_0000402.npy new file mode 100644 index 0000000..f945c8e --- /dev/null +++ b/margin_logs/step_0000402.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:96b98142b1027513d374a910b077877996838911f141378b98afd2de4fa8c76a +size 384 diff --git a/margin_logs/step_0000403.npy b/margin_logs/step_0000403.npy new file mode 100644 index 0000000..9b2f935 --- /dev/null +++ b/margin_logs/step_0000403.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5fe18e5be9d8cde6798f45139b037e2a5394fa030ad7ce3327e66cf412a7d155 +size 384 diff --git a/margin_logs/step_0000404.npy b/margin_logs/step_0000404.npy new file mode 100644 index 0000000..d1ef281 --- /dev/null +++ b/margin_logs/step_0000404.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b4b5edea19452d80fe98eef165e7b8542f74ff34edd30497e8e09b9369a173ad +size 384 diff --git a/margin_logs/step_0000405.npy b/margin_logs/step_0000405.npy new file mode 100644 index 0000000..d5dd320 --- /dev/null +++ b/margin_logs/step_0000405.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5d5782f93fd04cd7516c2390aaf38d0d57a2f9ca1fd9ec87d95a810a3f134c0c +size 384 diff --git a/margin_logs/step_0000406.npy b/margin_logs/step_0000406.npy new file mode 100644 index 0000000..b10e53d --- /dev/null +++ b/margin_logs/step_0000406.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:af6b3c4acef17cfabec0aec8333b237ba5e86b8a36b63389b98b40096c61ec52 +size 384 diff --git a/margin_logs/step_0000407.npy b/margin_logs/step_0000407.npy new file mode 100644 index 0000000..5b8cf90 --- /dev/null +++ b/margin_logs/step_0000407.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c9f015e69e92ea1c0515d1f9ae27f745e530e0f1a5c0a9dcfc8801981b4b933 +size 384 diff --git a/margin_logs/step_0000408.npy b/margin_logs/step_0000408.npy new file mode 100644 index 0000000..2a0c120 --- /dev/null +++ b/margin_logs/step_0000408.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cbb0979f7c24dd23d4d807ab13dc504d7134f2e532ae8d51e243de7f3689227a +size 384 diff --git a/margin_logs/step_0000409.npy b/margin_logs/step_0000409.npy new file mode 100644 index 0000000..ec25e6c --- /dev/null +++ b/margin_logs/step_0000409.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:478badf05159fed005266730e3a7f36d8b3d8e906c050308748fa119ec7c69d3 +size 384 diff --git a/margin_logs/step_0000410.npy b/margin_logs/step_0000410.npy new file mode 100644 index 0000000..0853023 --- /dev/null +++ b/margin_logs/step_0000410.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3e7066a4110473f94bd5e2a580d123f00f2f17c318c6fcfea807fcf6dc0c8a62 +size 384 diff --git a/margin_logs/step_0000411.npy b/margin_logs/step_0000411.npy new file mode 100644 index 0000000..cb515a8 --- /dev/null +++ b/margin_logs/step_0000411.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4b9d2f63d20a27564d53de183c81d8d40e7bcb1fd8e3be9d8971772ebf40b185 +size 384 diff --git a/margin_logs/step_0000412.npy b/margin_logs/step_0000412.npy new file mode 100644 index 0000000..8a4aa39 --- /dev/null +++ b/margin_logs/step_0000412.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cee65ecabd4a481ad78a4dd23037fa6c79e2386fa83ff962b7dfedd00066808d +size 384 diff --git a/margin_logs/step_0000413.npy b/margin_logs/step_0000413.npy new file mode 100644 index 0000000..4e49771 --- /dev/null +++ b/margin_logs/step_0000413.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6d35934b4ed553ce8517d584ddf6c611721056b3a03d2f8ea2a8fb2f09625d0 +size 384 diff --git a/margin_logs/step_0000414.npy b/margin_logs/step_0000414.npy new file mode 100644 index 0000000..9a4864f --- /dev/null +++ b/margin_logs/step_0000414.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0c5baa06c16822c4f3f76fd60d1f271da224f0ff758779c14adec3fe4265a48d +size 384 diff --git a/margin_logs/step_0000415.npy b/margin_logs/step_0000415.npy new file mode 100644 index 0000000..8dbb769 --- /dev/null +++ b/margin_logs/step_0000415.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:31c960df65149dec115d8abed0afab92c19e1da97eddb1672cffd2df4ad0ceac +size 384 diff --git a/margin_logs/step_0000416.npy b/margin_logs/step_0000416.npy new file mode 100644 index 0000000..1f41e3c --- /dev/null +++ b/margin_logs/step_0000416.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c1174f0e23474aa2fbcec2ae9fe282199f1e449ec5fd55066eff98bb914d1db5 +size 384 diff --git a/margin_logs/step_0000417.npy b/margin_logs/step_0000417.npy new file mode 100644 index 0000000..d076bc4 --- /dev/null +++ b/margin_logs/step_0000417.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7867e4ff77ed9a30201c05aba75b633a34d6285ab153f5ca1ce3e2a85867c40a +size 384 diff --git a/margin_logs/step_0000418.npy b/margin_logs/step_0000418.npy new file mode 100644 index 0000000..3cb617f --- /dev/null +++ b/margin_logs/step_0000418.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6c74b2f0f763609c3d17a52a7568621eb4fc1444c74758905bcbc58ea0218f28 +size 384 diff --git a/margin_logs/step_0000419.npy b/margin_logs/step_0000419.npy new file mode 100644 index 0000000..cc5611f --- /dev/null +++ b/margin_logs/step_0000419.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6ddb0e25208b077a2969012e96ebc338f7cd1817a0e87f20345f5b253ba804c +size 384 diff --git a/margin_logs/step_0000420.npy b/margin_logs/step_0000420.npy new file mode 100644 index 0000000..256ae73 --- /dev/null +++ b/margin_logs/step_0000420.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b996634c9631e5ab7f23d9ddaae30521a1f2a55c997578481e57bf852557643f +size 384 diff --git a/margin_logs/step_0000421.npy b/margin_logs/step_0000421.npy new file mode 100644 index 0000000..c7c3f4d --- /dev/null +++ b/margin_logs/step_0000421.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a70e90fee4e9e01a1889c66c7db2b25475f69350ae90fa30b429c5f0179cb29e +size 384 diff --git a/margin_logs/step_0000422.npy b/margin_logs/step_0000422.npy new file mode 100644 index 0000000..5bcf4a5 --- /dev/null +++ b/margin_logs/step_0000422.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a42f77c2819ac218a90f895cd2f007b7ced57c40ac2c0e4753c9cdaa2cfb4dc3 +size 384 diff --git a/margin_logs/step_0000423.npy b/margin_logs/step_0000423.npy new file mode 100644 index 0000000..99cbc60 --- /dev/null +++ b/margin_logs/step_0000423.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b60de444d9ba6a6535aceaa3effa435025a9350398c14a9d17017fa4d4c15930 +size 384 diff --git a/margin_logs/step_0000424.npy b/margin_logs/step_0000424.npy new file mode 100644 index 0000000..cb9e7bd --- /dev/null +++ b/margin_logs/step_0000424.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:277a2779819906d01c6ec63ae200e972e70106535ef658dae0583d4ab1efcab9 +size 384 diff --git a/margin_logs/step_0000425.npy b/margin_logs/step_0000425.npy new file mode 100644 index 0000000..6468bb2 --- /dev/null +++ b/margin_logs/step_0000425.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:72e451c0d70706b999457d03e2ed3874b2cba2c01e03a33f03ee067e3615e80b +size 384 diff --git a/margin_logs/step_0000426.npy b/margin_logs/step_0000426.npy new file mode 100644 index 0000000..13423aa --- /dev/null +++ b/margin_logs/step_0000426.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:196c6aa6a6b8ab9fb59bc2704dfd18436b60b292c5b74737ccf5937641178ff0 +size 384 diff --git a/margin_logs/step_0000427.npy b/margin_logs/step_0000427.npy new file mode 100644 index 0000000..7bb2fe4 --- /dev/null +++ b/margin_logs/step_0000427.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:35f4f9dabb3eb9295e9e8d23cd82593cc7c25b8f8659e882c3fe1ca9788a7f0d +size 384 diff --git a/margin_logs/step_0000428.npy b/margin_logs/step_0000428.npy new file mode 100644 index 0000000..78b4f00 --- /dev/null +++ b/margin_logs/step_0000428.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2015c7bd0d087276cb652a269095cdbdb8af362d52f3c0e405e90dfe793a27f4 +size 384 diff --git a/margin_logs/step_0000429.npy b/margin_logs/step_0000429.npy new file mode 100644 index 0000000..25713cd --- /dev/null +++ b/margin_logs/step_0000429.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:01128332322dbaad63d4807a1f72079fa64611943a949150ecf97dc666df78a6 +size 384 diff --git a/margin_logs/step_0000430.npy b/margin_logs/step_0000430.npy new file mode 100644 index 0000000..f29592d --- /dev/null +++ b/margin_logs/step_0000430.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:40cf6ef0af6bc31b967575661640f71cd9a1df88d9af4235ddf284f5f79f6281 +size 384 diff --git a/margin_logs/step_0000431.npy b/margin_logs/step_0000431.npy new file mode 100644 index 0000000..d15245d --- /dev/null +++ b/margin_logs/step_0000431.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:df2516fcef1906dc09f0738a563355219b2134dec00012ea7b7ad68f61754d0c +size 384 diff --git a/margin_logs/step_0000432.npy b/margin_logs/step_0000432.npy new file mode 100644 index 0000000..193e70b --- /dev/null +++ b/margin_logs/step_0000432.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5794c265fda62f7d6ce42c28ef3bacc7d577a7b594ca473a4e8f8adb163d6dea +size 384 diff --git a/margin_logs/step_0000433.npy b/margin_logs/step_0000433.npy new file mode 100644 index 0000000..02f9f89 --- /dev/null +++ b/margin_logs/step_0000433.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:251d69fd57a40767f25183b9a41855fed2ea053aa4de77903a373f815f3600b2 +size 384 diff --git a/margin_logs/step_0000434.npy b/margin_logs/step_0000434.npy new file mode 100644 index 0000000..127355e --- /dev/null +++ b/margin_logs/step_0000434.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3dd8a4eaca485e8b9d85f37dbb4d8d622f0df63fcf97681a03b9a7e8d436d10d +size 384 diff --git a/margin_logs/step_0000435.npy b/margin_logs/step_0000435.npy new file mode 100644 index 0000000..3fb96d5 --- /dev/null +++ b/margin_logs/step_0000435.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a1df5aa2fc6431deda3053cf1df740648086d0ab56a274c5e618396a662f92c7 +size 384 diff --git a/margin_logs/step_0000436.npy b/margin_logs/step_0000436.npy new file mode 100644 index 0000000..64c4ec2 --- /dev/null +++ b/margin_logs/step_0000436.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:91c44f931a4be7d6c35df23f06f43768f13c4e0453bb7173b29a7c188448603c +size 384 diff --git a/margin_logs/step_0000437.npy b/margin_logs/step_0000437.npy new file mode 100644 index 0000000..5fb62ce --- /dev/null +++ b/margin_logs/step_0000437.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a75261cf9a6513280962ad317476d37d258034b8a1d9b8c3e7a93920d2187c54 +size 384 diff --git a/margin_logs/step_0000438.npy b/margin_logs/step_0000438.npy new file mode 100644 index 0000000..73f043e --- /dev/null +++ b/margin_logs/step_0000438.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e974e47ce6d1d9d834d646bad4f2bc029c7d7932bc2afe5a1f75fd3cc31905c5 +size 384 diff --git a/margin_logs/step_0000439.npy b/margin_logs/step_0000439.npy new file mode 100644 index 0000000..7c7f534 --- /dev/null +++ b/margin_logs/step_0000439.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3926a1d357bf934eaf315597299df05199afc61e29da429a2d46cb7e8b06f4bc +size 384 diff --git a/margin_logs/step_0000440.npy b/margin_logs/step_0000440.npy new file mode 100644 index 0000000..16cd498 --- /dev/null +++ b/margin_logs/step_0000440.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:94b77cadc5a881b9459363f6af73465249b2a87f9d51c3690fd480eda61dad66 +size 384 diff --git a/margin_logs/step_0000441.npy b/margin_logs/step_0000441.npy new file mode 100644 index 0000000..02562a8 --- /dev/null +++ b/margin_logs/step_0000441.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a76ea8d15c11204fedb0f58cfdb209c11bdca0c47ca9c969dc742f256a8ac70c +size 384 diff --git a/margin_logs/step_0000442.npy b/margin_logs/step_0000442.npy new file mode 100644 index 0000000..d92f3ad --- /dev/null +++ b/margin_logs/step_0000442.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c801b0dd1b0b44d5c488ae4267424dec6508c95fe3a4822ae06cb1311bdaad4f +size 384 diff --git a/margin_logs/step_0000443.npy b/margin_logs/step_0000443.npy new file mode 100644 index 0000000..7c2655a --- /dev/null +++ b/margin_logs/step_0000443.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f25687270c2ae6f09144142fd1a07d8c803d0036c9b77b0a7942c3c202d6d824 +size 384 diff --git a/margin_logs/step_0000444.npy b/margin_logs/step_0000444.npy new file mode 100644 index 0000000..03f189f --- /dev/null +++ b/margin_logs/step_0000444.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4b880ff8dfa31dabc516b5ee33bdb00a49315d92bf212ee81d57d0a89e4cc237 +size 384 diff --git a/margin_logs/step_0000445.npy b/margin_logs/step_0000445.npy new file mode 100644 index 0000000..fed2ad3 --- /dev/null +++ b/margin_logs/step_0000445.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a1cf6087f6503ded4358b2b9ade356bc5f339c0d2dbd7190626fc86b69d0cdb0 +size 384 diff --git a/margin_logs/step_0000446.npy b/margin_logs/step_0000446.npy new file mode 100644 index 0000000..78b3222 --- /dev/null +++ b/margin_logs/step_0000446.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0de50830a0cea11a0a63fcca3ece82862edc49163866f6f653883f30e1c3d444 +size 384 diff --git a/margin_logs/step_0000447.npy b/margin_logs/step_0000447.npy new file mode 100644 index 0000000..34fe8c7 --- /dev/null +++ b/margin_logs/step_0000447.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ae57c324407a311bb8d7ed83d12c3d80202fbe53817fa2aeaa736c6e9ff9f3ee +size 384 diff --git a/margin_logs/step_0000448.npy b/margin_logs/step_0000448.npy new file mode 100644 index 0000000..cf36239 --- /dev/null +++ b/margin_logs/step_0000448.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fe47f03c4c1d994f3911303bf341ebc5e40e48e5b971a79c7e35f41db55add5e +size 384 diff --git a/margin_logs/step_0000449.npy b/margin_logs/step_0000449.npy new file mode 100644 index 0000000..6c81734 --- /dev/null +++ b/margin_logs/step_0000449.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:50b36c688d716647bfdec8df1fdf32f99619bb18880af395b93f29a5653b665e +size 384 diff --git a/margin_logs/step_0000450.npy b/margin_logs/step_0000450.npy new file mode 100644 index 0000000..e18386e --- /dev/null +++ b/margin_logs/step_0000450.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:911588b14aa76a917b0af5783c54e8886fd17212247df2fea4e770a4e900792e +size 384 diff --git a/margin_logs/step_0000451.npy b/margin_logs/step_0000451.npy new file mode 100644 index 0000000..882188a --- /dev/null +++ b/margin_logs/step_0000451.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c85890bbbb221ea5d3585d13df7fe3d2db5e68ac738ffbef2732f681f300026e +size 384 diff --git a/margin_logs/step_0000452.npy b/margin_logs/step_0000452.npy new file mode 100644 index 0000000..53ca2c9 --- /dev/null +++ b/margin_logs/step_0000452.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4dc6168e14bae75039b425140ac7cea47b5f6b87e4c9ca609f10892fb443bd30 +size 384 diff --git a/margin_logs/step_0000453.npy b/margin_logs/step_0000453.npy new file mode 100644 index 0000000..4c9c47c --- /dev/null +++ b/margin_logs/step_0000453.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f2f5f49332718c207bd74874b4345d11a3dacacdcb7246fa4ea0492e4593cb98 +size 384 diff --git a/margin_logs/step_0000454.npy b/margin_logs/step_0000454.npy new file mode 100644 index 0000000..4f144ad --- /dev/null +++ b/margin_logs/step_0000454.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1d216454fa529f093607c3668d9541300746bff8c0614d52b0ba9e083beab971 +size 384 diff --git a/margin_logs/step_0000455.npy b/margin_logs/step_0000455.npy new file mode 100644 index 0000000..cf73d1a --- /dev/null +++ b/margin_logs/step_0000455.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:946ffb335ddf7e7dd228547b95b150d6509f4817cb6b329660a65d8639d4e725 +size 384 diff --git a/margin_logs/step_0000456.npy b/margin_logs/step_0000456.npy new file mode 100644 index 0000000..3512acb --- /dev/null +++ b/margin_logs/step_0000456.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7a48fb2f4a828e5c7c7aaaa5094ad7c4c8a40f5aeeae878dc7844091b7496b06 +size 384 diff --git a/margin_logs/step_0000457.npy b/margin_logs/step_0000457.npy new file mode 100644 index 0000000..d59f4c9 --- /dev/null +++ b/margin_logs/step_0000457.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5b15d2ea42b4c2c7ff732bb84b7bfe4fdc3ccf55bf571cdae485ac62589be0b3 +size 384 diff --git a/margin_logs/step_0000458.npy b/margin_logs/step_0000458.npy new file mode 100644 index 0000000..bb1c32e --- /dev/null +++ b/margin_logs/step_0000458.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b9b4a862778c75e1cd9e6b6e60ede7f7f4a50af11421ba9424e776c4c0412bcd +size 384 diff --git a/margin_logs/step_0000459.npy b/margin_logs/step_0000459.npy new file mode 100644 index 0000000..55bfc66 --- /dev/null +++ b/margin_logs/step_0000459.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f8743fe7112c28321e0adcfc5f9baa291300aab0fc2e7b056b1e159eefa157c +size 384 diff --git a/margin_logs/step_0000460.npy b/margin_logs/step_0000460.npy new file mode 100644 index 0000000..05f41aa --- /dev/null +++ b/margin_logs/step_0000460.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3e70ed36278caea3a7eae38946ece6676fe713bee9f2b9a00e000c284627e17a +size 384 diff --git a/margin_logs/step_0000461.npy b/margin_logs/step_0000461.npy new file mode 100644 index 0000000..e7dc7df --- /dev/null +++ b/margin_logs/step_0000461.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6d3a49f2e6b9f9d41ece5e2600aa803d6fcfabc58bac1f6f2dbe241c05080142 +size 384 diff --git a/margin_logs/step_0000462.npy b/margin_logs/step_0000462.npy new file mode 100644 index 0000000..6b5317a --- /dev/null +++ b/margin_logs/step_0000462.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:036ab0db112e292c2d4a1c747cf416a6f407915b105aa7b58ec156eaba27ca47 +size 384 diff --git a/margin_logs/step_0000463.npy b/margin_logs/step_0000463.npy new file mode 100644 index 0000000..77fbac4 --- /dev/null +++ b/margin_logs/step_0000463.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:502d97dcc2ea7603cea1dfbe4681cff2e1572ae28a7dd678f097934c5a5f61c8 +size 384 diff --git a/margin_logs/step_0000464.npy b/margin_logs/step_0000464.npy new file mode 100644 index 0000000..5a172e0 --- /dev/null +++ b/margin_logs/step_0000464.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6fdd25c937e46b61c6b47ff4f8436f8abdc614a95acafe5270e710de07d92a1b +size 384 diff --git a/margin_logs/step_0000465.npy b/margin_logs/step_0000465.npy new file mode 100644 index 0000000..735635b --- /dev/null +++ b/margin_logs/step_0000465.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b5c6aed4b3c0e65c083772f44581b041bde73ce0e2213d6b5ab63afb5913acc +size 384 diff --git a/margin_logs/step_0000466.npy b/margin_logs/step_0000466.npy new file mode 100644 index 0000000..1364a5a --- /dev/null +++ b/margin_logs/step_0000466.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:40a8e6c6a770962223e634105bfe75d13ebc60999a3546a72ac4183a558d620b +size 384 diff --git a/margin_logs/step_0000467.npy b/margin_logs/step_0000467.npy new file mode 100644 index 0000000..6faabac --- /dev/null +++ b/margin_logs/step_0000467.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1f351f8f30900463eb094bf3c9c525659a051df3d1bdec736f2b75cbb2e1e2e4 +size 384 diff --git a/margin_logs/step_0000468.npy b/margin_logs/step_0000468.npy new file mode 100644 index 0000000..5710729 --- /dev/null +++ b/margin_logs/step_0000468.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dbfdc4489c4b532ab8236addc972fd29aeefdf3410b94d73f8e61b0602caa548 +size 384 diff --git a/margin_logs/step_0000469.npy b/margin_logs/step_0000469.npy new file mode 100644 index 0000000..2d78077 --- /dev/null +++ b/margin_logs/step_0000469.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dc3c6cf3111fee5ef3c710ab7233f81a5cc3baa134e627bc7108ce79c025dec6 +size 384 diff --git a/margin_logs/step_0000470.npy b/margin_logs/step_0000470.npy new file mode 100644 index 0000000..77a86cb --- /dev/null +++ b/margin_logs/step_0000470.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:660dccca1bad2f1da9fddc7a1b408dade517603490b613df2f199967e7f3253b +size 384 diff --git a/margin_logs/step_0000471.npy b/margin_logs/step_0000471.npy new file mode 100644 index 0000000..d76ac84 --- /dev/null +++ b/margin_logs/step_0000471.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:251416ef78b6cdc4ea57ae65b5e6562698ee344817ceb832cd4258bb9dfd56f4 +size 384 diff --git a/margin_logs/step_0000472.npy b/margin_logs/step_0000472.npy new file mode 100644 index 0000000..11ab6be --- /dev/null +++ b/margin_logs/step_0000472.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a6fb4cfc6666adda5f2ffe75125f0f10220c04779c16df77b3653f546c285df5 +size 384 diff --git a/margin_logs/step_0000473.npy b/margin_logs/step_0000473.npy new file mode 100644 index 0000000..e6de5f5 --- /dev/null +++ b/margin_logs/step_0000473.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:25be7cc6693e6af1cf9a665fa797e76adcf4f169725877726c6f9291cb082e11 +size 384 diff --git a/margin_logs/step_0000474.npy b/margin_logs/step_0000474.npy new file mode 100644 index 0000000..92523ac --- /dev/null +++ b/margin_logs/step_0000474.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:40ea422d3138be39a704a4d2f8aa27d193ea4d8e2b5265e48f7836610a5bb3f3 +size 384 diff --git a/margin_logs/step_0000475.npy b/margin_logs/step_0000475.npy new file mode 100644 index 0000000..a02f4dc --- /dev/null +++ b/margin_logs/step_0000475.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:491de875af449c3867d4e6737a5d2ae4a0c24176f4915bbfd7a8398e1e2c8e7b +size 384 diff --git a/margin_logs/step_0000476.npy b/margin_logs/step_0000476.npy new file mode 100644 index 0000000..8f7ec08 --- /dev/null +++ b/margin_logs/step_0000476.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f1e052f6f4dd97e388c3ea50107775f1112b98b7d8db350ed6dd261c457150c2 +size 384 diff --git a/margin_logs/step_0000477.npy b/margin_logs/step_0000477.npy new file mode 100644 index 0000000..b91e240 --- /dev/null +++ b/margin_logs/step_0000477.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d43ab9e969642c4df5ce2df22f359809527ae09696da9b58cffe3d049625e8ca +size 384 diff --git a/margin_logs/step_0000478.npy b/margin_logs/step_0000478.npy new file mode 100644 index 0000000..6851767 --- /dev/null +++ b/margin_logs/step_0000478.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0889ab6fcdd70658020587068ac4a817fcd8a0c90b78f03a121042a5533fa181 +size 384 diff --git a/margin_logs/step_0000479.npy b/margin_logs/step_0000479.npy new file mode 100644 index 0000000..5e0353d --- /dev/null +++ b/margin_logs/step_0000479.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c9e0a4e4f0e4685a823a157155be5afe787a6dbf4fa19401d130688d6ec9e60a +size 384 diff --git a/margin_logs/step_0000480.npy b/margin_logs/step_0000480.npy new file mode 100644 index 0000000..844ad13 --- /dev/null +++ b/margin_logs/step_0000480.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1dbe761a5fef139640f7369d296df69c84779ec892c59375c47b44c16c7a0e7f +size 384 diff --git a/margin_logs/step_0000481.npy b/margin_logs/step_0000481.npy new file mode 100644 index 0000000..508638c --- /dev/null +++ b/margin_logs/step_0000481.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ba187f1fcd609a177aa849cca291baca2cfdda10bff6f0ab8aeff61207a61ef1 +size 384 diff --git a/margin_logs/step_0000482.npy b/margin_logs/step_0000482.npy new file mode 100644 index 0000000..651ca6d --- /dev/null +++ b/margin_logs/step_0000482.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3dc2413bc3a5e2f5562741a948463be61494d3c3ac59b825ff6ee19985d582fa +size 384 diff --git a/margin_logs/step_0000483.npy b/margin_logs/step_0000483.npy new file mode 100644 index 0000000..511383c --- /dev/null +++ b/margin_logs/step_0000483.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1edb6c4b4c817abbd7424174aef56b6189e5c9e33cf09bc0fca3a84798b41f7a +size 384 diff --git a/margin_logs/step_0000484.npy b/margin_logs/step_0000484.npy new file mode 100644 index 0000000..6952c39 --- /dev/null +++ b/margin_logs/step_0000484.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b08588622700c28e4095cfa149a89ce9c59fd9ba7a807fbe9e1a4a948972dbdd +size 384 diff --git a/margin_logs/step_0000485.npy b/margin_logs/step_0000485.npy new file mode 100644 index 0000000..05d66c5 --- /dev/null +++ b/margin_logs/step_0000485.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:236d21738cb5be301b747c74960382af385eb7725225c5946d134755c2b5feec +size 384 diff --git a/margin_logs/step_0000486.npy b/margin_logs/step_0000486.npy new file mode 100644 index 0000000..7e3e55e --- /dev/null +++ b/margin_logs/step_0000486.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e7f25815cbd8d664990ce9cb70f6dc35ec085717fd58684aa31797d5220bf3a +size 384 diff --git a/margin_logs/step_0000487.npy b/margin_logs/step_0000487.npy new file mode 100644 index 0000000..5a74377 --- /dev/null +++ b/margin_logs/step_0000487.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dde108bdde2e074035b522845261ae3f6d2ad0c543521cb62516730108b0d372 +size 384 diff --git a/margin_logs/step_0000488.npy b/margin_logs/step_0000488.npy new file mode 100644 index 0000000..d889e80 --- /dev/null +++ b/margin_logs/step_0000488.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:24868a83b3a1c5d1b3cb76b30be7a0146b0f1a42cdabe4be0d049f4a6f86ac86 +size 384 diff --git a/margin_logs/step_0000489.npy b/margin_logs/step_0000489.npy new file mode 100644 index 0000000..e0209dc --- /dev/null +++ b/margin_logs/step_0000489.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3a7584cd25b263affee14430c1df41d1c55fb2be9d672c537fa172060fa9d1c0 +size 384 diff --git a/margin_logs/step_0000490.npy b/margin_logs/step_0000490.npy new file mode 100644 index 0000000..1ab0623 --- /dev/null +++ b/margin_logs/step_0000490.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a8bcf3559a07159a5904f10d463f1d7442eb2471eb73646658f9d38c75d44815 +size 384 diff --git a/margin_logs/step_0000491.npy b/margin_logs/step_0000491.npy new file mode 100644 index 0000000..9aabed9 --- /dev/null +++ b/margin_logs/step_0000491.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:02cac975873eb828e5004b645fe4789ac90f04e8c388cc91864de5daecfb8d40 +size 384 diff --git a/margin_logs/step_0000492.npy b/margin_logs/step_0000492.npy new file mode 100644 index 0000000..11adc32 --- /dev/null +++ b/margin_logs/step_0000492.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b0527729e7de506f75833c161085c543566e095361ecf0e32a8ddfe2f7924730 +size 384 diff --git a/margin_logs/step_0000493.npy b/margin_logs/step_0000493.npy new file mode 100644 index 0000000..d99fdf3 --- /dev/null +++ b/margin_logs/step_0000493.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f2cb399c8c39bebd3f93ce2b9bbf36e86cd21f05d6a0b97aedd2133a75dffe96 +size 384 diff --git a/margin_logs/step_0000494.npy b/margin_logs/step_0000494.npy new file mode 100644 index 0000000..76d951b --- /dev/null +++ b/margin_logs/step_0000494.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5626f8efdb7e59820367001895a78adee6c10b4ce2f7bb2e6d0eb4829fe39caa +size 384 diff --git a/margin_logs/step_0000495.npy b/margin_logs/step_0000495.npy new file mode 100644 index 0000000..2539c3f --- /dev/null +++ b/margin_logs/step_0000495.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e3090166688b54a36f01dfed699f92bd80429bd394ae1ef69690ccf546c3bf96 +size 384 diff --git a/margin_logs/step_0000496.npy b/margin_logs/step_0000496.npy new file mode 100644 index 0000000..b739df8 --- /dev/null +++ b/margin_logs/step_0000496.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5513111b5849e6177e5ddfaf239664403debdf7c5eaf03ee9bdf1db3f4ba718b +size 384 diff --git a/margin_logs/step_0000497.npy b/margin_logs/step_0000497.npy new file mode 100644 index 0000000..70a93c8 --- /dev/null +++ b/margin_logs/step_0000497.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:16d4e6961d5b5a1edd491853e90ac3bedface82269e498298f42ab33129689a1 +size 384 diff --git a/margin_logs/step_0000498.npy b/margin_logs/step_0000498.npy new file mode 100644 index 0000000..e513655 --- /dev/null +++ b/margin_logs/step_0000498.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:52fc7bb197f416d0231592e0f30aee669666fad3d20cd695f2f18250886c7515 +size 384 diff --git a/margin_logs/step_0000499.npy b/margin_logs/step_0000499.npy new file mode 100644 index 0000000..4c35951 --- /dev/null +++ b/margin_logs/step_0000499.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b05648436a4354953cfe46f2ec080a615727c76e2c4a4de4c9f7f96bbb1142e1 +size 384 diff --git a/margin_logs/step_0000500.npy b/margin_logs/step_0000500.npy new file mode 100644 index 0000000..0d22ab2 --- /dev/null +++ b/margin_logs/step_0000500.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8689a6f512741f54ca122f0f0d2169c7bcf30618a7953b96af093f8a8cae019b +size 384 diff --git a/margin_logs/step_0000501.npy b/margin_logs/step_0000501.npy new file mode 100644 index 0000000..f536aab --- /dev/null +++ b/margin_logs/step_0000501.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4a32d3b8783bea3ac67a9236fbf8da2683b4b7e4b94f2c7bcad9820033815e68 +size 384 diff --git a/margin_logs/step_0000502.npy b/margin_logs/step_0000502.npy new file mode 100644 index 0000000..91ff0c1 --- /dev/null +++ b/margin_logs/step_0000502.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:25d5cd373c35a11b98ff855ee732a12c2b5ac7b38dd3feea5b83e9302fd3099d +size 384 diff --git a/margin_logs/step_0000503.npy b/margin_logs/step_0000503.npy new file mode 100644 index 0000000..9defbc5 --- /dev/null +++ b/margin_logs/step_0000503.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ad7662333d8d8d45681a0ce8b29ba5af0985450903955626416b79f7c706c5db +size 384 diff --git a/margin_logs/step_0000504.npy b/margin_logs/step_0000504.npy new file mode 100644 index 0000000..f8bd683 --- /dev/null +++ b/margin_logs/step_0000504.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1e17102c6fbc24d0ea67fe6ba978e1e5a4a1eef4b33a224733c973ac1ab5e4ba +size 384 diff --git a/margin_logs/step_0000505.npy b/margin_logs/step_0000505.npy new file mode 100644 index 0000000..72e22f8 --- /dev/null +++ b/margin_logs/step_0000505.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:604e8c279ae21a3641c038a6dadf0d27d6caa52f90cb9dd7ef043a3970b9661d +size 384 diff --git a/margin_logs/step_0000506.npy b/margin_logs/step_0000506.npy new file mode 100644 index 0000000..5501995 --- /dev/null +++ b/margin_logs/step_0000506.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4eace9d6415ec7d07c1ed49569bbd11ad829425d440f626f79aa353757c62932 +size 384 diff --git a/margin_logs/step_0000507.npy b/margin_logs/step_0000507.npy new file mode 100644 index 0000000..9d37d96 --- /dev/null +++ b/margin_logs/step_0000507.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:332bc1067270dbfe217225cb83f596b5dd57942db1f2388a593fb9020d7a7b57 +size 384 diff --git a/margin_logs/step_0000508.npy b/margin_logs/step_0000508.npy new file mode 100644 index 0000000..7152fa1 --- /dev/null +++ b/margin_logs/step_0000508.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ecd55edf7d253ca0ca6e0e77b72c3073f3c3588ff8fb3e0d389d7b053b4f8000 +size 384 diff --git a/margin_logs/step_0000509.npy b/margin_logs/step_0000509.npy new file mode 100644 index 0000000..3f14dfe --- /dev/null +++ b/margin_logs/step_0000509.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a40a8407fa28d1a42d7c95155b3bb58afe303b888e1812ff1b877d4c01bc66ed +size 384 diff --git a/margin_logs/step_0000510.npy b/margin_logs/step_0000510.npy new file mode 100644 index 0000000..819f43d --- /dev/null +++ b/margin_logs/step_0000510.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7fa1b65aa19909ab8a1a75a581c19e983cac18a676326bfdceb7a83f7b6ee491 +size 384 diff --git a/margin_logs/step_0000511.npy b/margin_logs/step_0000511.npy new file mode 100644 index 0000000..c4ee57e --- /dev/null +++ b/margin_logs/step_0000511.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cc4f5117e71149b2f2f7bfa6d34097952ad86892f6db453806f76c8b6ac8ef84 +size 384 diff --git a/margin_logs/step_0000512.npy b/margin_logs/step_0000512.npy new file mode 100644 index 0000000..7800c17 --- /dev/null +++ b/margin_logs/step_0000512.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cb4e43749b0d1f935f4751440db0cead85eb46026f8023865279553412b5fabc +size 384 diff --git a/margin_logs/step_0000513.npy b/margin_logs/step_0000513.npy new file mode 100644 index 0000000..4df092d --- /dev/null +++ b/margin_logs/step_0000513.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:89238bc9eeeef4865e0b21f48941ef1c90e8d4b47f153e76ec3cc634269a5756 +size 384 diff --git a/margin_logs/step_0000514.npy b/margin_logs/step_0000514.npy new file mode 100644 index 0000000..fb77dc6 --- /dev/null +++ b/margin_logs/step_0000514.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:caa4babefc914f895703d4cc418168032258bb3c7a4e46176e697c3027e10470 +size 384 diff --git a/margin_logs/step_0000515.npy b/margin_logs/step_0000515.npy new file mode 100644 index 0000000..7cfe5c6 --- /dev/null +++ b/margin_logs/step_0000515.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8c5372debf44a23bff9c7b36ad12211ad18c00cfd25c1f67e41d2404d6ccfffe +size 384 diff --git a/margin_logs/step_0000516.npy b/margin_logs/step_0000516.npy new file mode 100644 index 0000000..bf52d13 --- /dev/null +++ b/margin_logs/step_0000516.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:22f564e22294fc2501d5340c4fdfff57b9c6e8d185e78699162a9859801ea0e0 +size 384 diff --git a/margin_logs/step_0000517.npy b/margin_logs/step_0000517.npy new file mode 100644 index 0000000..55b9219 --- /dev/null +++ b/margin_logs/step_0000517.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c087efbb4dc594b6c08524882100c178e1572f2996e5e5dd1eb40f5a37651241 +size 384 diff --git a/margin_logs/step_0000518.npy b/margin_logs/step_0000518.npy new file mode 100644 index 0000000..0b47e20 --- /dev/null +++ b/margin_logs/step_0000518.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e4524da21c5d891cb6eb56e010a5afed1f025fcef5b9ea0962a500f52ea43490 +size 384 diff --git a/margin_logs/step_0000519.npy b/margin_logs/step_0000519.npy new file mode 100644 index 0000000..f461c7e --- /dev/null +++ b/margin_logs/step_0000519.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9433cc2614df01db76ada734fcbe3690ef382f6ed9161ddba763cfb9a611e23c +size 384 diff --git a/margin_logs/step_0000520.npy b/margin_logs/step_0000520.npy new file mode 100644 index 0000000..4991552 --- /dev/null +++ b/margin_logs/step_0000520.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:20853d2380ce77b008dd69ab03c48d055927267e4f6a5ce2d55c3392f4223df5 +size 384 diff --git a/margin_logs/step_0000521.npy b/margin_logs/step_0000521.npy new file mode 100644 index 0000000..8e35715 --- /dev/null +++ b/margin_logs/step_0000521.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9056a61e4bb3c0d7ef20b2f5204c4df5aaf226dda9691aeea41710d5b65e3631 +size 384 diff --git a/margin_logs/step_0000522.npy b/margin_logs/step_0000522.npy new file mode 100644 index 0000000..087e10c --- /dev/null +++ b/margin_logs/step_0000522.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2cbfc0267f3e57f40c5658e847c89217e86c4e8aee07ab2bfc9318181f0c460b +size 384 diff --git a/margin_logs/step_0000523.npy b/margin_logs/step_0000523.npy new file mode 100644 index 0000000..a7690d5 --- /dev/null +++ b/margin_logs/step_0000523.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6019452922f31a35fb3b3f51253baddad39abeef27cd6d0767e8a683d6b0d6ca +size 384 diff --git a/margin_logs/step_0000524.npy b/margin_logs/step_0000524.npy new file mode 100644 index 0000000..0c093ea --- /dev/null +++ b/margin_logs/step_0000524.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d4a4a80547283c77375c83c7935a76843a9a770ec12aa34383d2f49aa8789d04 +size 384 diff --git a/margin_logs/step_0000525.npy b/margin_logs/step_0000525.npy new file mode 100644 index 0000000..b57498c --- /dev/null +++ b/margin_logs/step_0000525.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4e3bd42d27e136271ff543ebacd7b0c966417ef38b21478c85b3c0b64300a351 +size 384 diff --git a/margin_logs/step_0000526.npy b/margin_logs/step_0000526.npy new file mode 100644 index 0000000..4df6d43 --- /dev/null +++ b/margin_logs/step_0000526.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:24f952d8009afecaa2c0f9f14e7e37e937544785e517bc5a0ce5552e44c21eb2 +size 384 diff --git a/margin_logs/step_0000527.npy b/margin_logs/step_0000527.npy new file mode 100644 index 0000000..9f55ae8 --- /dev/null +++ b/margin_logs/step_0000527.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f68eedb1908824012abf5eb5181e749c221c03cb07cb0c55a3e34a0edbacadce +size 384 diff --git a/margin_logs/step_0000528.npy b/margin_logs/step_0000528.npy new file mode 100644 index 0000000..d768425 --- /dev/null +++ b/margin_logs/step_0000528.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8bae7a4fb55c9c3b178fe7222a3459a7c68b097d691d8f01359095268361ab4d +size 384 diff --git a/margin_logs/step_0000529.npy b/margin_logs/step_0000529.npy new file mode 100644 index 0000000..ff6ab21 --- /dev/null +++ b/margin_logs/step_0000529.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dc790070bb1ce3c4693639467a2f82375c1d80823d60eb6b95b84f85781f8a48 +size 384 diff --git a/margin_logs/step_0000530.npy b/margin_logs/step_0000530.npy new file mode 100644 index 0000000..5514a5e --- /dev/null +++ b/margin_logs/step_0000530.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aebf500b66f8973ead7a51f59ba4c927d9370501500389f382e796afdc6d88f6 +size 384 diff --git a/margin_logs/step_0000531.npy b/margin_logs/step_0000531.npy new file mode 100644 index 0000000..91f0074 --- /dev/null +++ b/margin_logs/step_0000531.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c8ce44daef8fbd39d689b05ca5e43ba3bf0c827ac69443e77154a764d6830b3b +size 384 diff --git a/margin_logs/step_0000532.npy b/margin_logs/step_0000532.npy new file mode 100644 index 0000000..7380f20 --- /dev/null +++ b/margin_logs/step_0000532.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1dd7e454b971201d8d565168712fdc5695e6cd656b51a56f414d0f6f5e33f7c2 +size 384 diff --git a/margin_logs/step_0000533.npy b/margin_logs/step_0000533.npy new file mode 100644 index 0000000..648f54d --- /dev/null +++ b/margin_logs/step_0000533.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d0a3580243758fed43d31240e837247feed7604c4f1e98bdaceb891b3a1b3968 +size 384 diff --git a/margin_logs/step_0000534.npy b/margin_logs/step_0000534.npy new file mode 100644 index 0000000..a08bd8d --- /dev/null +++ b/margin_logs/step_0000534.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aeef8025979c61b72edc0570d048a007e296429b0465a36a5574ecee578f7ffb +size 384 diff --git a/margin_logs/step_0000535.npy b/margin_logs/step_0000535.npy new file mode 100644 index 0000000..35088d1 --- /dev/null +++ b/margin_logs/step_0000535.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1bd729b9189e6b8835ccc8a3a3409e93cc4c270e9a13f8c3368727a16dd5eea9 +size 384 diff --git a/margin_logs/step_0000536.npy b/margin_logs/step_0000536.npy new file mode 100644 index 0000000..2c0af5d --- /dev/null +++ b/margin_logs/step_0000536.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ada37c107db177e45fabc677a4318b6b1c54f8b240eb4ade36ebe179eaaa909f +size 384 diff --git a/margin_logs/step_0000537.npy b/margin_logs/step_0000537.npy new file mode 100644 index 0000000..34ac807 --- /dev/null +++ b/margin_logs/step_0000537.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:915d0b7826a8291afec45fe7c8b9d7f78d65c7225ee0c394bbe8265b341aa7b3 +size 384 diff --git a/margin_logs/step_0000538.npy b/margin_logs/step_0000538.npy new file mode 100644 index 0000000..c6240bf --- /dev/null +++ b/margin_logs/step_0000538.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6409338e3d2aa8f91dbf85fff85c93161b8b39f92faba3fcca540a77e5c75d5 +size 384 diff --git a/margin_logs/step_0000539.npy b/margin_logs/step_0000539.npy new file mode 100644 index 0000000..0106532 --- /dev/null +++ b/margin_logs/step_0000539.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c4cd18ca703b7a4c096930c1c095ba3fab2ab3c044234aee9ad0f288f8c4085b +size 384 diff --git a/margin_logs/step_0000540.npy b/margin_logs/step_0000540.npy new file mode 100644 index 0000000..efdb40f --- /dev/null +++ b/margin_logs/step_0000540.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:009dea7ad36016a6ea94fee8d26afe766b4d65717aa67a57d1a28a6803d769bc +size 384 diff --git a/margin_logs/step_0000541.npy b/margin_logs/step_0000541.npy new file mode 100644 index 0000000..9d4e32b --- /dev/null +++ b/margin_logs/step_0000541.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8269fe4cdd8f79f200a2d9515dda20125736acb039ed861b42907edc27a2aa4d +size 384 diff --git a/margin_logs/step_0000542.npy b/margin_logs/step_0000542.npy new file mode 100644 index 0000000..d73ccac --- /dev/null +++ b/margin_logs/step_0000542.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:14ed1068a4d6c95458e987984dd8b246e634282e74c6a25393d2fd3b74dd9f6b +size 384 diff --git a/margin_logs/step_0000543.npy b/margin_logs/step_0000543.npy new file mode 100644 index 0000000..5f7f969 --- /dev/null +++ b/margin_logs/step_0000543.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:77ca4d3863ff44e34ee8c7067d6f3ba056c4755793e31dae7aacf91e5ddbe028 +size 384 diff --git a/margin_logs/step_0000544.npy b/margin_logs/step_0000544.npy new file mode 100644 index 0000000..9a62a3f --- /dev/null +++ b/margin_logs/step_0000544.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6349c23e5cf3ac2ac4f9a194ed7c6e464ef427756361ed62a64518d00221fb54 +size 384 diff --git a/margin_logs/step_0000545.npy b/margin_logs/step_0000545.npy new file mode 100644 index 0000000..cb33bd9 --- /dev/null +++ b/margin_logs/step_0000545.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:04a7a3c57cea448e8f05c895212d4b98e0db58061f4d93d2a4a57ad9c6de66f6 +size 384 diff --git a/margin_logs/step_0000546.npy b/margin_logs/step_0000546.npy new file mode 100644 index 0000000..fe895c0 --- /dev/null +++ b/margin_logs/step_0000546.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c8f6ad3cfc5bc96fd802e09ef9e2f38dc189ae2183acbf05354a59e08b418f08 +size 384 diff --git a/margin_logs/step_0000547.npy b/margin_logs/step_0000547.npy new file mode 100644 index 0000000..6cdd4db --- /dev/null +++ b/margin_logs/step_0000547.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c46be120d008276bda526ae5710a8cfb294c15a64283d0fb578fa8ecf555628f +size 384 diff --git a/margin_logs/step_0000548.npy b/margin_logs/step_0000548.npy new file mode 100644 index 0000000..0412f2e --- /dev/null +++ b/margin_logs/step_0000548.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9087a4844e6e4a2ea5d85da790fb463aafa1f23647576d74dd1bfccf8690c9da +size 384 diff --git a/margin_logs/step_0000549.npy b/margin_logs/step_0000549.npy new file mode 100644 index 0000000..6de4c40 --- /dev/null +++ b/margin_logs/step_0000549.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4d8818634006dd9d257eba0dc6ab547aa2dc36980980da01ae0298cf43f93ab5 +size 384 diff --git a/margin_logs/step_0000550.npy b/margin_logs/step_0000550.npy new file mode 100644 index 0000000..1e5a038 --- /dev/null +++ b/margin_logs/step_0000550.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:55c98bf64749b5199f8251390755a09fc46d5e25e807e79ef0c359720a767da5 +size 384 diff --git a/margin_logs/step_0000551.npy b/margin_logs/step_0000551.npy new file mode 100644 index 0000000..483142c --- /dev/null +++ b/margin_logs/step_0000551.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5d32cdc8ced4806222214c767ff7afea4b109421b5d3c3bacc469061eb1e1943 +size 384 diff --git a/margin_logs/step_0000552.npy b/margin_logs/step_0000552.npy new file mode 100644 index 0000000..101eb33 --- /dev/null +++ b/margin_logs/step_0000552.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:efff103e4dbf9d5c7eafd450ea73fbb8cd3ad500d22b809ee53d6deefafe24f9 +size 384 diff --git a/margin_logs/step_0000553.npy b/margin_logs/step_0000553.npy new file mode 100644 index 0000000..f75aa05 --- /dev/null +++ b/margin_logs/step_0000553.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ed82f656317d0dec11e90f38a8653e7a599c53215ea7cf57e74d50e2facf9175 +size 384 diff --git a/margin_logs/step_0000554.npy b/margin_logs/step_0000554.npy new file mode 100644 index 0000000..ed2160d --- /dev/null +++ b/margin_logs/step_0000554.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f631988b9a3adc4a8858aa9f466186a34c1d6455b6a4fd3e3a93cfe03167376d +size 384 diff --git a/margin_logs/step_0000555.npy b/margin_logs/step_0000555.npy new file mode 100644 index 0000000..e9c7609 --- /dev/null +++ b/margin_logs/step_0000555.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1f551ab173dc55469ac6a9bc561cf5f996e8b6fb022f9e405a169618c3e4b317 +size 384 diff --git a/margin_logs/step_0000556.npy b/margin_logs/step_0000556.npy new file mode 100644 index 0000000..9b201c8 --- /dev/null +++ b/margin_logs/step_0000556.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:be89b21eedcad3860078e9c6641a00ec2668312ca3d9d35a57f7285a4a091919 +size 384 diff --git a/margin_logs/step_0000557.npy b/margin_logs/step_0000557.npy new file mode 100644 index 0000000..b2ea463 --- /dev/null +++ b/margin_logs/step_0000557.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4ef2374f2b2ca3679de76f5c0f034f2827fb4e330af5680540a918eed8e3f529 +size 384 diff --git a/margin_logs/step_0000558.npy b/margin_logs/step_0000558.npy new file mode 100644 index 0000000..894ebbb --- /dev/null +++ b/margin_logs/step_0000558.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ace9f024ed50b465fa5c7a191e7a02e617515af99151a04d68fd08e6a923ff68 +size 384 diff --git a/margin_logs/step_0000559.npy b/margin_logs/step_0000559.npy new file mode 100644 index 0000000..476b6f7 --- /dev/null +++ b/margin_logs/step_0000559.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:36e843abee5913aa719f82825b33b141248a30d14b95e4c2498fe65c7c6e5a03 +size 384 diff --git a/margin_logs/step_0000560.npy b/margin_logs/step_0000560.npy new file mode 100644 index 0000000..c91461c --- /dev/null +++ b/margin_logs/step_0000560.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d6dcb26088603eee3b1c93616f835594be46b13652def27d33f3511c932b741a +size 384 diff --git a/margin_logs/step_0000561.npy b/margin_logs/step_0000561.npy new file mode 100644 index 0000000..9fb7336 --- /dev/null +++ b/margin_logs/step_0000561.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bf690ebfea72bafd4db41e7f63437adf269324a58bd5ee8356f030201a4d5d68 +size 384 diff --git a/margin_logs/step_0000562.npy b/margin_logs/step_0000562.npy new file mode 100644 index 0000000..65f004d --- /dev/null +++ b/margin_logs/step_0000562.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f77b56413aea9172ee696107382a2bb94cb3323c68e2d5e1de12117178431837 +size 384 diff --git a/margin_logs/step_0000563.npy b/margin_logs/step_0000563.npy new file mode 100644 index 0000000..0464790 --- /dev/null +++ b/margin_logs/step_0000563.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7d0481d2cd8d910edb8f54064ac586c050fb825e5737a49a6177c316d044c553 +size 384 diff --git a/margin_logs/step_0000564.npy b/margin_logs/step_0000564.npy new file mode 100644 index 0000000..6254c43 --- /dev/null +++ b/margin_logs/step_0000564.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c4b6ef6664664dab4a5efe4dbc4f9278291c01bf33c68551351b794cebb07db4 +size 384 diff --git a/margin_logs/step_0000565.npy b/margin_logs/step_0000565.npy new file mode 100644 index 0000000..ad43152 --- /dev/null +++ b/margin_logs/step_0000565.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:82d33d7a1b8e64b713d596c45d4b585433972f295ba91b4a948183622d84d70a +size 384 diff --git a/margin_logs/step_0000566.npy b/margin_logs/step_0000566.npy new file mode 100644 index 0000000..efbc4c6 --- /dev/null +++ b/margin_logs/step_0000566.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1dc24177930b3e567a33d69e14138fb3a90de4ef7e86d1e06a5bae784e878fc9 +size 384 diff --git a/margin_logs/step_0000567.npy b/margin_logs/step_0000567.npy new file mode 100644 index 0000000..83f8c79 --- /dev/null +++ b/margin_logs/step_0000567.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b69c1fa97f55fc292b91f0f8319c88e44de95a8ca37c124b84d308bd586736c0 +size 384 diff --git a/margin_logs/step_0000568.npy b/margin_logs/step_0000568.npy new file mode 100644 index 0000000..b55b323 --- /dev/null +++ b/margin_logs/step_0000568.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:47d2d1d930ac4100d702ee8851264d64744cd7577b9235702506ad5e02bb1e81 +size 384 diff --git a/margin_logs/step_0000569.npy b/margin_logs/step_0000569.npy new file mode 100644 index 0000000..1edf156 --- /dev/null +++ b/margin_logs/step_0000569.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f91bc359e190e48b463796d48f0da113c79efee429bb929e506aeb2753373c13 +size 384 diff --git a/margin_logs/step_0000570.npy b/margin_logs/step_0000570.npy new file mode 100644 index 0000000..84084b6 --- /dev/null +++ b/margin_logs/step_0000570.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7e0f5ef08b4dd0d61725bee7ec12c02a3030008634b5fd140b888de214e2f21c +size 384 diff --git a/margin_logs/step_0000571.npy b/margin_logs/step_0000571.npy new file mode 100644 index 0000000..d26b45e --- /dev/null +++ b/margin_logs/step_0000571.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:052b3aa98b92d0e68dcf424ca025ff55c431d470353fbf5470c0effed1c659b2 +size 384 diff --git a/margin_logs/step_0000572.npy b/margin_logs/step_0000572.npy new file mode 100644 index 0000000..6210006 --- /dev/null +++ b/margin_logs/step_0000572.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d9a18a3899fd2d9f79b9c97ddc01bbab1f24d8b412a3bde0fd62af5c6ee064c9 +size 384 diff --git a/margin_logs/step_0000573.npy b/margin_logs/step_0000573.npy new file mode 100644 index 0000000..3d63d09 --- /dev/null +++ b/margin_logs/step_0000573.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dcc1f6f2c9656b3634c84f9fd01c2c8190e6198674b0538444699c74c88d1303 +size 384 diff --git a/margin_logs/step_0000574.npy b/margin_logs/step_0000574.npy new file mode 100644 index 0000000..c753f26 --- /dev/null +++ b/margin_logs/step_0000574.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e574d0e207cf709079fe997cabcfb069f648bf01cc1341f2a408f8ff93e0ad7 +size 384 diff --git a/margin_logs/step_0000575.npy b/margin_logs/step_0000575.npy new file mode 100644 index 0000000..60a4b12 --- /dev/null +++ b/margin_logs/step_0000575.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5672982e4b1fc32c7154644c58e05f122c49d0e6ce30973d6cd2acd00d49a70f +size 384 diff --git a/margin_logs/step_0000576.npy b/margin_logs/step_0000576.npy new file mode 100644 index 0000000..678ae3d --- /dev/null +++ b/margin_logs/step_0000576.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e8e2c0907534d5983ebad7769a1d84b574a2bbdae6e9d7f53c9ef9aa4584e985 +size 384 diff --git a/margin_logs/step_0000577.npy b/margin_logs/step_0000577.npy new file mode 100644 index 0000000..d8d7f60 --- /dev/null +++ b/margin_logs/step_0000577.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0e1a00c9dc0bfff1fda282426bec7c7f519a0071910ac7024ddad924acfaed8d +size 384 diff --git a/margin_logs/step_0000578.npy b/margin_logs/step_0000578.npy new file mode 100644 index 0000000..3cbfcad --- /dev/null +++ b/margin_logs/step_0000578.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9b1530b0fbcc28fb3509f0a67dad502ee274e4e98127137baa63c4446fc89f38 +size 384 diff --git a/margin_logs/step_0000579.npy b/margin_logs/step_0000579.npy new file mode 100644 index 0000000..6708bc7 --- /dev/null +++ b/margin_logs/step_0000579.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8e2c1ff9ee0211afd2fd199b68dec488d7c11744369abcdca0b1e135ade26656 +size 384 diff --git a/margin_logs/step_0000580.npy b/margin_logs/step_0000580.npy new file mode 100644 index 0000000..93aca8b --- /dev/null +++ b/margin_logs/step_0000580.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0dcba88460c8e132ff8af33b7254cafd15f93fa758271631a72233f8f7895982 +size 384 diff --git a/margin_logs/step_0000581.npy b/margin_logs/step_0000581.npy new file mode 100644 index 0000000..1c3657e --- /dev/null +++ b/margin_logs/step_0000581.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9611a89da75838598909347dcac152d58292f3bdc4c4d12be84d0781a90e5c89 +size 384 diff --git a/margin_logs/step_0000582.npy b/margin_logs/step_0000582.npy new file mode 100644 index 0000000..02238cb --- /dev/null +++ b/margin_logs/step_0000582.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:11a86376fec94e1f30cbd4ff503d41eb839751c72d6271697b4950ed8cc62546 +size 384 diff --git a/margin_logs/step_0000583.npy b/margin_logs/step_0000583.npy new file mode 100644 index 0000000..12758a5 --- /dev/null +++ b/margin_logs/step_0000583.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:624567102a2930aea3dedd1fc0394c826c50f3d946161db212eda893e9f6520d +size 384 diff --git a/margin_logs/step_0000584.npy b/margin_logs/step_0000584.npy new file mode 100644 index 0000000..6ce0d9c --- /dev/null +++ b/margin_logs/step_0000584.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d783d8052f2f9b1f61b542e088afaef1d8af5d08e5eed519d756cac713866460 +size 384 diff --git a/margin_logs/step_0000585.npy b/margin_logs/step_0000585.npy new file mode 100644 index 0000000..1c3841d --- /dev/null +++ b/margin_logs/step_0000585.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6f81b20e7f00927c73404a567db0ebfccd2e42a608b297de4ff52dfc2243b2a2 +size 384 diff --git a/margin_logs/step_0000586.npy b/margin_logs/step_0000586.npy new file mode 100644 index 0000000..49f5ef1 --- /dev/null +++ b/margin_logs/step_0000586.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8f85707bb449f2763ec81bfe21eed46147b2f12d2a0b8f455e3f150d2dd02896 +size 384 diff --git a/margin_logs/step_0000587.npy b/margin_logs/step_0000587.npy new file mode 100644 index 0000000..447d41d --- /dev/null +++ b/margin_logs/step_0000587.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9a94a6c8101121d8f9d881b6ee19981e31c676b6e435a2f021fcfaeb4be289c2 +size 384 diff --git a/margin_logs/step_0000588.npy b/margin_logs/step_0000588.npy new file mode 100644 index 0000000..84965ee --- /dev/null +++ b/margin_logs/step_0000588.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1966d7fcdfd6b704684478d564d213d3953497fb63a17ee95b79525744dea3ef +size 384 diff --git a/margin_logs/step_0000589.npy b/margin_logs/step_0000589.npy new file mode 100644 index 0000000..5d822ed --- /dev/null +++ b/margin_logs/step_0000589.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:77201687efa43aa282be10323b81adac9d522ea92566079dbd2617449ee9e686 +size 384 diff --git a/margin_logs/step_0000590.npy b/margin_logs/step_0000590.npy new file mode 100644 index 0000000..369cb89 --- /dev/null +++ b/margin_logs/step_0000590.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7356e026bca6e8b13c1e1dc3075e9b569175ef4a638d35f3dcf45af13093854f +size 384 diff --git a/margin_logs/step_0000591.npy b/margin_logs/step_0000591.npy new file mode 100644 index 0000000..cbeac96 --- /dev/null +++ b/margin_logs/step_0000591.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6290a4cae035e084a5b1432c5d9fd56d638c2a158fe9c4ccb68dc4cee723b661 +size 384 diff --git a/margin_logs/step_0000592.npy b/margin_logs/step_0000592.npy new file mode 100644 index 0000000..69afe57 --- /dev/null +++ b/margin_logs/step_0000592.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6a773ce3cbb1769e8bcb2fa19f918663ea9f6e096026a1418004487a73f20d25 +size 384 diff --git a/margin_logs/step_0000593.npy b/margin_logs/step_0000593.npy new file mode 100644 index 0000000..25d4a75 --- /dev/null +++ b/margin_logs/step_0000593.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8a12d44c48e3a2828ea6d29b5a78efd72d3b3e4f22bb0c030d173e6023eecc3a +size 384 diff --git a/margin_logs/step_0000594.npy b/margin_logs/step_0000594.npy new file mode 100644 index 0000000..8a98003 --- /dev/null +++ b/margin_logs/step_0000594.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d92e1f1508d6dfd8558cebe7cb601b51c7aa388780e179f4d745f7c84141575d +size 384 diff --git a/margin_logs/step_0000595.npy b/margin_logs/step_0000595.npy new file mode 100644 index 0000000..baaf629 --- /dev/null +++ b/margin_logs/step_0000595.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d96dde3ba96d4dbd463ac15bbcf23f514e95004b36256a4b7019f38d47ed780e +size 384 diff --git a/margin_logs/step_0000596.npy b/margin_logs/step_0000596.npy new file mode 100644 index 0000000..20efe19 --- /dev/null +++ b/margin_logs/step_0000596.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:58234fcc18b2ac8657e0b79b6cf8a6d4b05a869e360fc77ba296de36b2bf19d4 +size 384 diff --git a/margin_logs/step_0000597.npy b/margin_logs/step_0000597.npy new file mode 100644 index 0000000..c9ccbbe --- /dev/null +++ b/margin_logs/step_0000597.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:149c8e315c4400eea91f16b06b9b24f8e5caae1bf47f7d939cb9a01bc5c8ccc6 +size 384 diff --git a/margin_logs/step_0000598.npy b/margin_logs/step_0000598.npy new file mode 100644 index 0000000..f5354ba --- /dev/null +++ b/margin_logs/step_0000598.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0918e37bb91bc523cc76d11a0119f96c5dd424593d05661b689174a940ceb589 +size 384 diff --git a/margin_logs/step_0000599.npy b/margin_logs/step_0000599.npy new file mode 100644 index 0000000..3c60f86 --- /dev/null +++ b/margin_logs/step_0000599.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4c5a7ac8669deec1bb8378b68ba082143702a04cff9e06ed38cb5dd130de2bff +size 384 diff --git a/margin_logs/step_0000600.npy b/margin_logs/step_0000600.npy new file mode 100644 index 0000000..7539278 --- /dev/null +++ b/margin_logs/step_0000600.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0d89ef02ac8eb857948a0cbc06b78cbb2431a43f9636dfc7e9c8cfd9cdadb661 +size 384 diff --git a/margin_logs/step_0000601.npy b/margin_logs/step_0000601.npy new file mode 100644 index 0000000..cf33f34 --- /dev/null +++ b/margin_logs/step_0000601.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5f0577d281fd8cf1f2b0f79b134d05bde40b09128f420d4e9c80e12dff4ea4ea +size 384 diff --git a/margin_logs/step_0000602.npy b/margin_logs/step_0000602.npy new file mode 100644 index 0000000..ef71497 --- /dev/null +++ b/margin_logs/step_0000602.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3ce771bab722e97440d936376555fe575255db4ea58af48c193ec29c417f5f4b +size 384 diff --git a/margin_logs/step_0000603.npy b/margin_logs/step_0000603.npy new file mode 100644 index 0000000..b05a608 --- /dev/null +++ b/margin_logs/step_0000603.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f470e69e649bf2665797cf6479f2dc850d6d2030e7e422fd61af35df097c1790 +size 384 diff --git a/margin_logs/step_0000604.npy b/margin_logs/step_0000604.npy new file mode 100644 index 0000000..c6a9d64 --- /dev/null +++ b/margin_logs/step_0000604.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8ceeb54f2a25a36a334be0e2e507ae55227ad0b8e7a0368b485c502443b1bd75 +size 384 diff --git a/margin_logs/step_0000605.npy b/margin_logs/step_0000605.npy new file mode 100644 index 0000000..17480b0 --- /dev/null +++ b/margin_logs/step_0000605.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:106d6ef24916cd7aa112fa845f7c47200868a7bfee49f72e39dd68743e3c779e +size 384 diff --git a/margin_logs/step_0000606.npy b/margin_logs/step_0000606.npy new file mode 100644 index 0000000..6c4b0dd --- /dev/null +++ b/margin_logs/step_0000606.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:335346bfaefa213fc9a61e2e9f4144454ef6babf0b92c9e87ffe2f05acf8d0c1 +size 384 diff --git a/margin_logs/step_0000607.npy b/margin_logs/step_0000607.npy new file mode 100644 index 0000000..972902d --- /dev/null +++ b/margin_logs/step_0000607.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9796fcbae5e0b411650cd7ba91524b6fbe4b0c755e7d9cd3c8a4c852abc1a5c2 +size 384 diff --git a/margin_logs/step_0000608.npy b/margin_logs/step_0000608.npy new file mode 100644 index 0000000..22ce78a --- /dev/null +++ b/margin_logs/step_0000608.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d536ed32245f1abec798adc9d17bee0b8f4e5d5efc96710a86766b78b79e282c +size 384 diff --git a/margin_logs/step_0000609.npy b/margin_logs/step_0000609.npy new file mode 100644 index 0000000..7ade17f --- /dev/null +++ b/margin_logs/step_0000609.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3ab455fd74f0a275030fad9f29c1176487f4f7b1b49a30eb0e09544cd4544479 +size 384 diff --git a/margin_logs/step_0000610.npy b/margin_logs/step_0000610.npy new file mode 100644 index 0000000..7df2146 --- /dev/null +++ b/margin_logs/step_0000610.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dedf7aa2c9b7110e1dd265bffe5622a26c144d35a391874037ff5799b7a48668 +size 384 diff --git a/margin_logs/step_0000611.npy b/margin_logs/step_0000611.npy new file mode 100644 index 0000000..c793bc6 --- /dev/null +++ b/margin_logs/step_0000611.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1aa2ba3da9eb05cc3d1543e1e4052e2fb8d2ca1531e1514998564bd88eb6482b +size 384 diff --git a/margin_logs/step_0000612.npy b/margin_logs/step_0000612.npy new file mode 100644 index 0000000..32222fc --- /dev/null +++ b/margin_logs/step_0000612.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d6aa972938be38ff71fd65089720fb97104151313d95a6795f8ae528a58f0861 +size 384 diff --git a/margin_logs/step_0000613.npy b/margin_logs/step_0000613.npy new file mode 100644 index 0000000..c927441 --- /dev/null +++ b/margin_logs/step_0000613.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1f2f20e4aeced327287d349ad156e372dd4f70cb2fc3fe7e2ef882ee8ffd4143 +size 384 diff --git a/margin_logs/step_0000614.npy b/margin_logs/step_0000614.npy new file mode 100644 index 0000000..1a5db3a --- /dev/null +++ b/margin_logs/step_0000614.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c561bf3aa024998a4d38b3b62e5b6f4897da47420294d79edbcccfe8b6e11289 +size 384 diff --git a/margin_logs/step_0000615.npy b/margin_logs/step_0000615.npy new file mode 100644 index 0000000..eca419d --- /dev/null +++ b/margin_logs/step_0000615.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e959a09d2a22818156bf4b7014c5a24f2ab37477d78200e414f0501da928dcc5 +size 384 diff --git a/margin_logs/step_0000616.npy b/margin_logs/step_0000616.npy new file mode 100644 index 0000000..e58f949 --- /dev/null +++ b/margin_logs/step_0000616.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:82a3df9176f66e29872e9d13015d1e46fff66ab32b4c90685972e76d01f34041 +size 384 diff --git a/margin_logs/step_0000617.npy b/margin_logs/step_0000617.npy new file mode 100644 index 0000000..aee0aab --- /dev/null +++ b/margin_logs/step_0000617.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b815b8872f87fe508f33251368188433e3a72d8ccb7abaad648815fb883fc53d +size 384 diff --git a/margin_logs/step_0000618.npy b/margin_logs/step_0000618.npy new file mode 100644 index 0000000..21a3b0c --- /dev/null +++ b/margin_logs/step_0000618.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:813b8968460b2de4cfa6bbc58ba0492694c298fc556cf4fd48ef0adbae4fc34f +size 384 diff --git a/margin_logs/step_0000619.npy b/margin_logs/step_0000619.npy new file mode 100644 index 0000000..20b613c --- /dev/null +++ b/margin_logs/step_0000619.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:87c256e575c30607437e2987b8c56388c480cc97b1d16ec7b29115afabf6ae56 +size 384 diff --git a/margin_logs/step_0000620.npy b/margin_logs/step_0000620.npy new file mode 100644 index 0000000..053f6a3 --- /dev/null +++ b/margin_logs/step_0000620.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7e820a1881a6257167fc96df44cb731c6b2be6cae164e52942d16093c01160e6 +size 384 diff --git a/margin_logs/step_0000621.npy b/margin_logs/step_0000621.npy new file mode 100644 index 0000000..3717552 --- /dev/null +++ b/margin_logs/step_0000621.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b0f0d1ca1d619db0456829a9ce728d4c0255dd51777a76741c18b29afb16b73c +size 384 diff --git a/margin_logs/step_0000622.npy b/margin_logs/step_0000622.npy new file mode 100644 index 0000000..4097d7d --- /dev/null +++ b/margin_logs/step_0000622.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f5ae2270e745b3adbec0ec2592148a91af90d93807962c73236ae4475d1b1c35 +size 384 diff --git a/margin_logs/step_0000623.npy b/margin_logs/step_0000623.npy new file mode 100644 index 0000000..f5a70bc --- /dev/null +++ b/margin_logs/step_0000623.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:12f9e777b95c588dda1fd149424386931bb2ecf05230789ed1e14132ab8ba9b2 +size 384 diff --git a/margin_logs/step_0000624.npy b/margin_logs/step_0000624.npy new file mode 100644 index 0000000..1e71c7b --- /dev/null +++ b/margin_logs/step_0000624.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1647a1dbf35f6847715f5760fea216ef6cb66cc8ab314a76dd964a599450ec0b +size 384 diff --git a/margin_logs/step_0000625.npy b/margin_logs/step_0000625.npy new file mode 100644 index 0000000..06148e6 --- /dev/null +++ b/margin_logs/step_0000625.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:639a5ef61d6502aea7eeeabebc17ec3d9d9868f5df22e51dcebac5087f25f1f8 +size 384 diff --git a/margin_logs/step_0000626.npy b/margin_logs/step_0000626.npy new file mode 100644 index 0000000..5b3b38d --- /dev/null +++ b/margin_logs/step_0000626.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:20f98b3fcb46e9534e218b9b631a8d3e158532c198914dba092e10af966807d5 +size 384 diff --git a/margin_logs/step_0000627.npy b/margin_logs/step_0000627.npy new file mode 100644 index 0000000..8c10bc3 --- /dev/null +++ b/margin_logs/step_0000627.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dd526f0cb1261f7121335bf7bbc3558f8f7a5202f4c22d929d5b2fd3ca3d8cb8 +size 384 diff --git a/margin_logs/step_0000628.npy b/margin_logs/step_0000628.npy new file mode 100644 index 0000000..c5e2279 --- /dev/null +++ b/margin_logs/step_0000628.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d51ea59c82b9c98de7e0b04db6c56d4762b3bbce16c00eb037b5da13cf37e580 +size 384 diff --git a/margin_logs/step_0000629.npy b/margin_logs/step_0000629.npy new file mode 100644 index 0000000..bada88a --- /dev/null +++ b/margin_logs/step_0000629.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5a9a09ce85cba30aa7dd25da279af85138fb51056d987223984521e8c9b9067a +size 384 diff --git a/margin_logs/step_0000630.npy b/margin_logs/step_0000630.npy new file mode 100644 index 0000000..6a88064 --- /dev/null +++ b/margin_logs/step_0000630.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e0755a0209492b2258ec7ee771f71935bda8406de4ed910e2df6513be20b31f0 +size 384 diff --git a/margin_logs/step_0000631.npy b/margin_logs/step_0000631.npy new file mode 100644 index 0000000..1175147 --- /dev/null +++ b/margin_logs/step_0000631.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:50607ceb8bd11cbbb1dae7934a076b9d5d7f70c026626a8af880bd3b8b9e7ffa +size 384 diff --git a/margin_logs/step_0000632.npy b/margin_logs/step_0000632.npy new file mode 100644 index 0000000..af3e868 --- /dev/null +++ b/margin_logs/step_0000632.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:81725e82714eec33e6abd9ad5c088c3ce65af05933bacc2f7aa6c1001e5d7bf2 +size 384 diff --git a/margin_logs/step_0000633.npy b/margin_logs/step_0000633.npy new file mode 100644 index 0000000..e7f909b --- /dev/null +++ b/margin_logs/step_0000633.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f73bdc5bdef859e7a0bc871c55a96f12304c4c9c9acec9fd5fe2e1b8b27ac060 +size 384 diff --git a/margin_logs/step_0000634.npy b/margin_logs/step_0000634.npy new file mode 100644 index 0000000..fe3c8a3 --- /dev/null +++ b/margin_logs/step_0000634.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f85fe64a14fb8a843fa02dd3954cbfa5c1dc44ae35400e43d860985536bc2d4 +size 384 diff --git a/margin_logs/step_0000635.npy b/margin_logs/step_0000635.npy new file mode 100644 index 0000000..77be18e --- /dev/null +++ b/margin_logs/step_0000635.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:770a442cfdcdd35bbcb7132a1cc8e7312a2b57bb1d7e665630d1458ebf4d7acd +size 384 diff --git a/margin_logs/step_0000636.npy b/margin_logs/step_0000636.npy new file mode 100644 index 0000000..66be1d8 --- /dev/null +++ b/margin_logs/step_0000636.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e43511716ca0324b9535d0e23fa42e8ca2c8e02033ff92933c1c9f54172f993f +size 384 diff --git a/margin_logs/step_0000637.npy b/margin_logs/step_0000637.npy new file mode 100644 index 0000000..0ca0d13 --- /dev/null +++ b/margin_logs/step_0000637.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dfff2be78d7fd102670172cd4c49f2eef4d38bfe857f2850672f54687c53c6a4 +size 384 diff --git a/margin_logs/step_0000638.npy b/margin_logs/step_0000638.npy new file mode 100644 index 0000000..4cdec83 --- /dev/null +++ b/margin_logs/step_0000638.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f535cc8a6f13aea7825dba2c91b2eecad98147518dc419319c3e875069378eca +size 384 diff --git a/margin_logs/step_0000639.npy b/margin_logs/step_0000639.npy new file mode 100644 index 0000000..47c874a --- /dev/null +++ b/margin_logs/step_0000639.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:54a0c3ed8569b324e05590dc16e03a42e6a93e3d736002acc611a1aa15cbc40d +size 384 diff --git a/margin_logs/step_0000640.npy b/margin_logs/step_0000640.npy new file mode 100644 index 0000000..0f07cc0 --- /dev/null +++ b/margin_logs/step_0000640.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5c98eb7528196784c1a30b31e7d1090283c74eed8428581f63514e60391046cd +size 384 diff --git a/margin_logs/step_0000641.npy b/margin_logs/step_0000641.npy new file mode 100644 index 0000000..98d3ac9 --- /dev/null +++ b/margin_logs/step_0000641.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e74bb7eb90d7a37e04e6ece1b34db640b01a9dc11fe72b2923e9b2f3b4ca0a78 +size 384 diff --git a/margin_logs/step_0000642.npy b/margin_logs/step_0000642.npy new file mode 100644 index 0000000..41cb1d6 --- /dev/null +++ b/margin_logs/step_0000642.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fd30241adbc730e512f628f3d1f10b9f0575e35488922b6dcec437e7a3aa6f32 +size 384 diff --git a/margin_logs/step_0000643.npy b/margin_logs/step_0000643.npy new file mode 100644 index 0000000..9e6d546 --- /dev/null +++ b/margin_logs/step_0000643.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8006d3bf3cb6bcf658e7c37d7e62d8dd42fb623841aa10cb864589972f577eca +size 384 diff --git a/margin_logs/step_0000644.npy b/margin_logs/step_0000644.npy new file mode 100644 index 0000000..fa3e1be --- /dev/null +++ b/margin_logs/step_0000644.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b529a07a90b82b1effffdc130154c463d636e022727a64549333c1ab1e76d370 +size 384 diff --git a/margin_logs/step_0000645.npy b/margin_logs/step_0000645.npy new file mode 100644 index 0000000..4370e43 --- /dev/null +++ b/margin_logs/step_0000645.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:67b02e3782853433b5178210798e1f026230986b3036a80b58906830b6e3d4af +size 384 diff --git a/margin_logs/step_0000646.npy b/margin_logs/step_0000646.npy new file mode 100644 index 0000000..9c1d6ab --- /dev/null +++ b/margin_logs/step_0000646.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:64ae64d2eaabae55f17adc34dc103533ab6344b129ee4c8602847655371327f4 +size 384 diff --git a/margin_logs/step_0000647.npy b/margin_logs/step_0000647.npy new file mode 100644 index 0000000..7554c62 --- /dev/null +++ b/margin_logs/step_0000647.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:be0d0e8f5c1a68ace396e3edb60c2d31944d80bd7dee03a96016c3e0c80914ec +size 384 diff --git a/margin_logs/step_0000648.npy b/margin_logs/step_0000648.npy new file mode 100644 index 0000000..33a20dc --- /dev/null +++ b/margin_logs/step_0000648.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:946de401667bdc87869af0a4f48539755e6d16e44ba209edc0f5546df6c70800 +size 384 diff --git a/margin_logs/step_0000649.npy b/margin_logs/step_0000649.npy new file mode 100644 index 0000000..95f056c --- /dev/null +++ b/margin_logs/step_0000649.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5fc0a24db462750916a81fb78e9d30b7be64720ba4afd1b660fb6b86ea249524 +size 384 diff --git a/margin_logs/step_0000650.npy b/margin_logs/step_0000650.npy new file mode 100644 index 0000000..f6fdff5 --- /dev/null +++ b/margin_logs/step_0000650.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7d1c96120e4fe59f778b08e542d6949d70ba9e2a73993324ed70da3312208328 +size 384 diff --git a/margin_logs/step_0000651.npy b/margin_logs/step_0000651.npy new file mode 100644 index 0000000..8114462 --- /dev/null +++ b/margin_logs/step_0000651.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2056fde6950d3e2228583660a9e7971a975e9708805244071d31234814d122a3 +size 384 diff --git a/margin_logs/step_0000652.npy b/margin_logs/step_0000652.npy new file mode 100644 index 0000000..9beeb2b --- /dev/null +++ b/margin_logs/step_0000652.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b01825265f5d9223751a16b38c98bbc5c4723666c3f6dff97b4a2e5c7b532c04 +size 384 diff --git a/margin_logs/step_0000653.npy b/margin_logs/step_0000653.npy new file mode 100644 index 0000000..6a6c867 --- /dev/null +++ b/margin_logs/step_0000653.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e53e553cb8c4e1fdffbf67dc235139673adbce9d5c727d5c69a819f79cf36968 +size 384 diff --git a/margin_logs/step_0000654.npy b/margin_logs/step_0000654.npy new file mode 100644 index 0000000..057a45c --- /dev/null +++ b/margin_logs/step_0000654.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9bead2f3e6fe24378ad8686683895b3ed8caadf89e92a1086d55bc8a46813ae2 +size 384 diff --git a/margin_logs/step_0000655.npy b/margin_logs/step_0000655.npy new file mode 100644 index 0000000..ad5ad1d --- /dev/null +++ b/margin_logs/step_0000655.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:364ca81d7e068a9072ad0de4bbd7e67caf44a4ea83907207e465329948bcb707 +size 384 diff --git a/margin_logs/step_0000656.npy b/margin_logs/step_0000656.npy new file mode 100644 index 0000000..8ea1793 --- /dev/null +++ b/margin_logs/step_0000656.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:061ea6f7fbd4c172dd66a1e6ad9abbe091c64fd012408d452bf47743ebc02f49 +size 384 diff --git a/margin_logs/step_0000657.npy b/margin_logs/step_0000657.npy new file mode 100644 index 0000000..5a92c97 --- /dev/null +++ b/margin_logs/step_0000657.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2c7a4c2156bd0a97484ac52a7eaf3b92a814a24e41f8295f96bc10c1e33e1ea8 +size 384 diff --git a/margin_logs/step_0000658.npy b/margin_logs/step_0000658.npy new file mode 100644 index 0000000..532b97d --- /dev/null +++ b/margin_logs/step_0000658.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4537ffa5ccadfe1e6d4e231f469a093504583ff05f610c0cfdbb0b52ddb4a27d +size 384 diff --git a/margin_logs/step_0000659.npy b/margin_logs/step_0000659.npy new file mode 100644 index 0000000..2d9f9c5 --- /dev/null +++ b/margin_logs/step_0000659.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6e1a7379d573b7a41796c037d6312683671f40f614ee954556204072eb698e88 +size 384 diff --git a/margin_logs/step_0000660.npy b/margin_logs/step_0000660.npy new file mode 100644 index 0000000..2cd7163 --- /dev/null +++ b/margin_logs/step_0000660.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8001382f7764908980db75f60ae0b6354a9c8ddfcab08161a93e49ff98b3e3a2 +size 384 diff --git a/margin_logs/step_0000661.npy b/margin_logs/step_0000661.npy new file mode 100644 index 0000000..64292ba --- /dev/null +++ b/margin_logs/step_0000661.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:01bd2e0d0e204395d7f8855f565427f21e7fbe8f06cbbc31229991b076e24e65 +size 384 diff --git a/margin_logs/step_0000662.npy b/margin_logs/step_0000662.npy new file mode 100644 index 0000000..f23ca9f --- /dev/null +++ b/margin_logs/step_0000662.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:615270399e30fa7732160b506984d6c937d869883fc14542473f7f04c8c2f890 +size 384 diff --git a/margin_logs/step_0000663.npy b/margin_logs/step_0000663.npy new file mode 100644 index 0000000..ddd3dc3 --- /dev/null +++ b/margin_logs/step_0000663.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:93fcb7c7c5ff6ce976779dfa91e86d470bb569bb872f546d1d59ff5c919bcd06 +size 384 diff --git a/margin_logs/step_0000664.npy b/margin_logs/step_0000664.npy new file mode 100644 index 0000000..33f0c79 --- /dev/null +++ b/margin_logs/step_0000664.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e44510dc697e0ac65295c33f42a4fa7c729a910c1d70d1dcdee0745fef82821c +size 384 diff --git a/margin_logs/step_0000665.npy b/margin_logs/step_0000665.npy new file mode 100644 index 0000000..6dde17a --- /dev/null +++ b/margin_logs/step_0000665.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7dbfd28ef953f7649f5f9fdbceef82479c99f624259481476cadcd50bd3a2a7f +size 384 diff --git a/margin_logs/step_0000666.npy b/margin_logs/step_0000666.npy new file mode 100644 index 0000000..8d09591 --- /dev/null +++ b/margin_logs/step_0000666.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b2bbea3b8739ff15ba44d7017e44ab64e3fd10a9c2241f5dbbf2565c4fac6c85 +size 384 diff --git a/margin_logs/step_0000667.npy b/margin_logs/step_0000667.npy new file mode 100644 index 0000000..8de10df --- /dev/null +++ b/margin_logs/step_0000667.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d617c22121ac608bf08c1cf6bdbcc7da86ed5c22ff86964170eb752471fea1b8 +size 384 diff --git a/margin_logs/step_0000668.npy b/margin_logs/step_0000668.npy new file mode 100644 index 0000000..76d7bab --- /dev/null +++ b/margin_logs/step_0000668.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f0d8af1ba25cca7953fa2894d56e2b43cf34be24050c64ae0d9adfb5b5d0346a +size 384 diff --git a/margin_logs/step_0000669.npy b/margin_logs/step_0000669.npy new file mode 100644 index 0000000..dfeb698 --- /dev/null +++ b/margin_logs/step_0000669.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2b4e723f0509d3dedd87b87cf75a7d7734106a0d503dc14e2301c3fa4778ad1f +size 384 diff --git a/margin_logs/step_0000670.npy b/margin_logs/step_0000670.npy new file mode 100644 index 0000000..4341538 --- /dev/null +++ b/margin_logs/step_0000670.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bd51b1d62b5b8c0c55fc98043ef0d0b331900292dbf176c5d5b272675d704c07 +size 384 diff --git a/margin_logs/step_0000671.npy b/margin_logs/step_0000671.npy new file mode 100644 index 0000000..ab36bde --- /dev/null +++ b/margin_logs/step_0000671.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:04fbf79be35ed3d62c448ced0a6425c4e9a96b1ad07fa9fea18029fa08d52de9 +size 384 diff --git a/margin_logs/step_0000672.npy b/margin_logs/step_0000672.npy new file mode 100644 index 0000000..f99fd48 --- /dev/null +++ b/margin_logs/step_0000672.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0f4ddaff4e99ab225106c91a38e03c500da979a8457bbec87e18a75ff90883f0 +size 384 diff --git a/margin_logs/step_0000673.npy b/margin_logs/step_0000673.npy new file mode 100644 index 0000000..2887b21 --- /dev/null +++ b/margin_logs/step_0000673.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:444625b8986c83c5dc50e0eb2e8ae805229e2251ace8f7eba3ee0c3fa0843ad4 +size 384 diff --git a/margin_logs/step_0000674.npy b/margin_logs/step_0000674.npy new file mode 100644 index 0000000..1df3b28 --- /dev/null +++ b/margin_logs/step_0000674.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de01ae242a667f98619f7830334f4b1acd2283c581208a1b78cd16b9c629e781 +size 384 diff --git a/margin_logs/step_0000675.npy b/margin_logs/step_0000675.npy new file mode 100644 index 0000000..0745582 --- /dev/null +++ b/margin_logs/step_0000675.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6726ccd33438424a904073eccb2c2bd069da1a65c5e542fd775a76962c746fc +size 384 diff --git a/margin_logs/step_0000676.npy b/margin_logs/step_0000676.npy new file mode 100644 index 0000000..ce42578 --- /dev/null +++ b/margin_logs/step_0000676.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1a730979ac4bd485dd9f47f7b12ce0c288d5a4cac178d984d28277b267cff261 +size 384 diff --git a/margin_logs/step_0000677.npy b/margin_logs/step_0000677.npy new file mode 100644 index 0000000..35e1472 --- /dev/null +++ b/margin_logs/step_0000677.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c765c2647d6468522864d7e040ad36de4c0d229eae6831356e729282697ad929 +size 384 diff --git a/margin_logs/step_0000678.npy b/margin_logs/step_0000678.npy new file mode 100644 index 0000000..754112c --- /dev/null +++ b/margin_logs/step_0000678.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:345646f7dfa4624db748fd8c990f5863e056d5f8134b28ad605ddcf93cd4e92f +size 384 diff --git a/margin_logs/step_0000679.npy b/margin_logs/step_0000679.npy new file mode 100644 index 0000000..4813df5 --- /dev/null +++ b/margin_logs/step_0000679.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:317f1d75c5384e911794e99a88370f40c67fc276b3e873f21a217e693cf07c90 +size 384 diff --git a/margin_logs/step_0000680.npy b/margin_logs/step_0000680.npy new file mode 100644 index 0000000..297a3df --- /dev/null +++ b/margin_logs/step_0000680.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4ec71d8170be71e3d41e9b57e37749005cc747d38e8edcb5112f2e918260eca7 +size 384 diff --git a/margin_logs/step_0000681.npy b/margin_logs/step_0000681.npy new file mode 100644 index 0000000..435ef09 --- /dev/null +++ b/margin_logs/step_0000681.npy @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6035970f20ac2fad8677349f0467e22b63a1755e974945e34f7d6de16f8a3050 +size 384 diff --git a/model-00001-of-00007.safetensors b/model-00001-of-00007.safetensors new file mode 100644 index 0000000..54b85fa --- /dev/null +++ b/model-00001-of-00007.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9bf97e8dec54ebc8fc36c2fccdbb4bd0b90ab50c1455b149f8765df08fa539ae +size 4886466168 diff --git a/model-00002-of-00007.safetensors b/model-00002-of-00007.safetensors new file mode 100644 index 0000000..d5fb919 --- /dev/null +++ b/model-00002-of-00007.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aba0cd28faa12d4bf3a93f6b6a60f62e1e7eb0de057a25a1b6ad5f14b116b034 +size 4832007448 diff --git a/model-00003-of-00007.safetensors b/model-00003-of-00007.safetensors new file mode 100644 index 0000000..5bcae95 --- /dev/null +++ b/model-00003-of-00007.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bfcffc74eaba95dc31fe0329658bd10d929cf28b7db9b4f533e99515b9d5236c +size 4999813112 diff --git a/model-00004-of-00007.safetensors b/model-00004-of-00007.safetensors new file mode 100644 index 0000000..e3fcb2d --- /dev/null +++ b/model-00004-of-00007.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d1d91c42b275ad749de36b515463af265df2bbd5e65f7742648d84ff8e71247e +size 4999813128 diff --git a/model-00005-of-00007.safetensors b/model-00005-of-00007.safetensors new file mode 100644 index 0000000..9b84512 --- /dev/null +++ b/model-00005-of-00007.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:44c16f2e449a4994d42be8599c929fadb5c12a09727403a68a9b63cbbfb12417 +size 4832007496 diff --git a/model-00006-of-00007.safetensors b/model-00006-of-00007.safetensors new file mode 100644 index 0000000..8b479fd --- /dev/null +++ b/model-00006-of-00007.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:284a6f601eec78e2c8a23ee3d51343814199eff730d08fbc0143b094153990b4 +size 4999813120 diff --git a/model-00007-of-00007.safetensors b/model-00007-of-00007.safetensors new file mode 100644 index 0000000..027acdc --- /dev/null +++ b/model-00007-of-00007.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fe4fc57d9d77e07fbcab46cc6fa37922af13466177ba6b1b41b417460aaa2583 +size 2571158184 diff --git a/model.safetensors.index.json b/model.safetensors.index.json new file mode 100644 index 0000000..0985084 --- /dev/null +++ b/model.safetensors.index.json @@ -0,0 +1,298 @@ +{ + "metadata": { + "total_size": 32121044992 + }, + "weight_map": { + "lm_head.weight": "model-00007-of-00007.safetensors", + "model.embed_tokens.weight": "model-00001-of-00007.safetensors", + "model.layers.0.input_layernorm.weight": "model-00001-of-00007.safetensors", + "model.layers.0.mlp.down_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.0.mlp.up_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00007.safetensors", + "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.1.input_layernorm.weight": "model-00001-of-00007.safetensors", + "model.layers.1.mlp.down_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.1.mlp.up_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00007.safetensors", + "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.10.input_layernorm.weight": "model-00003-of-00007.safetensors", + "model.layers.10.mlp.down_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.10.mlp.gate_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.10.mlp.up_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.10.post_attention_layernorm.weight": "model-00003-of-00007.safetensors", + "model.layers.10.self_attn.k_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.10.self_attn.o_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.10.self_attn.q_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.10.self_attn.v_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.11.input_layernorm.weight": "model-00003-of-00007.safetensors", + "model.layers.11.mlp.down_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.11.mlp.gate_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.11.mlp.up_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.11.post_attention_layernorm.weight": "model-00003-of-00007.safetensors", + "model.layers.11.self_attn.k_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.11.self_attn.o_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.11.self_attn.q_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.11.self_attn.v_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.12.input_layernorm.weight": "model-00003-of-00007.safetensors", + "model.layers.12.mlp.down_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.12.mlp.gate_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.12.mlp.up_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.12.post_attention_layernorm.weight": "model-00003-of-00007.safetensors", + "model.layers.12.self_attn.k_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.12.self_attn.o_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.12.self_attn.q_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.12.self_attn.v_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.13.input_layernorm.weight": "model-00003-of-00007.safetensors", + "model.layers.13.mlp.down_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.13.mlp.gate_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.13.mlp.up_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.13.post_attention_layernorm.weight": "model-00003-of-00007.safetensors", + "model.layers.13.self_attn.k_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.13.self_attn.o_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.13.self_attn.q_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.13.self_attn.v_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.14.input_layernorm.weight": "model-00004-of-00007.safetensors", + "model.layers.14.mlp.down_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.14.mlp.gate_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.14.mlp.up_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.14.post_attention_layernorm.weight": "model-00004-of-00007.safetensors", + "model.layers.14.self_attn.k_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.14.self_attn.o_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.14.self_attn.q_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.14.self_attn.v_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.15.input_layernorm.weight": "model-00004-of-00007.safetensors", + "model.layers.15.mlp.down_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.15.mlp.gate_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.15.mlp.up_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.15.post_attention_layernorm.weight": "model-00004-of-00007.safetensors", + "model.layers.15.self_attn.k_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.15.self_attn.o_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.15.self_attn.q_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.15.self_attn.v_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.16.input_layernorm.weight": "model-00004-of-00007.safetensors", + "model.layers.16.mlp.down_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.16.mlp.gate_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.16.mlp.up_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.16.post_attention_layernorm.weight": "model-00004-of-00007.safetensors", + "model.layers.16.self_attn.k_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.16.self_attn.o_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.16.self_attn.q_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.16.self_attn.v_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.17.input_layernorm.weight": "model-00004-of-00007.safetensors", + "model.layers.17.mlp.down_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.17.mlp.gate_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.17.mlp.up_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.17.post_attention_layernorm.weight": "model-00004-of-00007.safetensors", + "model.layers.17.self_attn.k_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.17.self_attn.o_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.17.self_attn.q_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.17.self_attn.v_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.18.input_layernorm.weight": "model-00004-of-00007.safetensors", + "model.layers.18.mlp.down_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.18.mlp.gate_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.18.mlp.up_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.18.post_attention_layernorm.weight": "model-00004-of-00007.safetensors", + "model.layers.18.self_attn.k_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.18.self_attn.o_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.18.self_attn.q_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.18.self_attn.v_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.19.input_layernorm.weight": "model-00004-of-00007.safetensors", + "model.layers.19.mlp.down_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.19.mlp.gate_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.19.mlp.up_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.19.post_attention_layernorm.weight": "model-00004-of-00007.safetensors", + "model.layers.19.self_attn.k_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.19.self_attn.o_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.19.self_attn.q_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.19.self_attn.v_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.2.input_layernorm.weight": "model-00001-of-00007.safetensors", + "model.layers.2.mlp.down_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.2.mlp.up_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00007.safetensors", + "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.20.input_layernorm.weight": "model-00005-of-00007.safetensors", + "model.layers.20.mlp.down_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.20.mlp.gate_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.20.mlp.up_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.20.post_attention_layernorm.weight": "model-00005-of-00007.safetensors", + "model.layers.20.self_attn.k_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.20.self_attn.o_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.20.self_attn.q_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.20.self_attn.v_proj.weight": "model-00004-of-00007.safetensors", + "model.layers.21.input_layernorm.weight": "model-00005-of-00007.safetensors", + "model.layers.21.mlp.down_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.21.mlp.gate_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.21.mlp.up_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.21.post_attention_layernorm.weight": "model-00005-of-00007.safetensors", + "model.layers.21.self_attn.k_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.21.self_attn.o_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.21.self_attn.q_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.21.self_attn.v_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.22.input_layernorm.weight": "model-00005-of-00007.safetensors", + "model.layers.22.mlp.down_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.22.mlp.gate_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.22.mlp.up_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.22.post_attention_layernorm.weight": "model-00005-of-00007.safetensors", + "model.layers.22.self_attn.k_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.22.self_attn.o_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.22.self_attn.q_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.22.self_attn.v_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.23.input_layernorm.weight": "model-00005-of-00007.safetensors", + "model.layers.23.mlp.down_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.23.mlp.gate_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.23.mlp.up_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.23.post_attention_layernorm.weight": "model-00005-of-00007.safetensors", + "model.layers.23.self_attn.k_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.23.self_attn.o_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.23.self_attn.q_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.23.self_attn.v_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.24.input_layernorm.weight": "model-00005-of-00007.safetensors", + "model.layers.24.mlp.down_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.24.mlp.gate_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.24.mlp.up_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.24.post_attention_layernorm.weight": "model-00005-of-00007.safetensors", + "model.layers.24.self_attn.k_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.24.self_attn.o_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.24.self_attn.q_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.24.self_attn.v_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.25.input_layernorm.weight": "model-00006-of-00007.safetensors", + "model.layers.25.mlp.down_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.25.mlp.gate_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.25.mlp.up_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.25.post_attention_layernorm.weight": "model-00006-of-00007.safetensors", + "model.layers.25.self_attn.k_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.25.self_attn.o_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.25.self_attn.q_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.25.self_attn.v_proj.weight": "model-00005-of-00007.safetensors", + "model.layers.26.input_layernorm.weight": "model-00006-of-00007.safetensors", + "model.layers.26.mlp.down_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.26.mlp.gate_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.26.mlp.up_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.26.post_attention_layernorm.weight": "model-00006-of-00007.safetensors", + "model.layers.26.self_attn.k_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.26.self_attn.o_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.26.self_attn.q_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.26.self_attn.v_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.27.input_layernorm.weight": "model-00006-of-00007.safetensors", + "model.layers.27.mlp.down_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.27.mlp.gate_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.27.mlp.up_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.27.post_attention_layernorm.weight": "model-00006-of-00007.safetensors", + "model.layers.27.self_attn.k_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.27.self_attn.o_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.27.self_attn.q_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.27.self_attn.v_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.28.input_layernorm.weight": "model-00006-of-00007.safetensors", + "model.layers.28.mlp.down_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.28.mlp.gate_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.28.mlp.up_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.28.post_attention_layernorm.weight": "model-00006-of-00007.safetensors", + "model.layers.28.self_attn.k_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.28.self_attn.o_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.28.self_attn.q_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.28.self_attn.v_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.29.input_layernorm.weight": "model-00006-of-00007.safetensors", + "model.layers.29.mlp.down_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.29.mlp.gate_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.29.mlp.up_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.29.post_attention_layernorm.weight": "model-00006-of-00007.safetensors", + "model.layers.29.self_attn.k_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.29.self_attn.o_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.29.self_attn.q_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.29.self_attn.v_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.3.input_layernorm.weight": "model-00002-of-00007.safetensors", + "model.layers.3.mlp.down_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.3.mlp.gate_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.3.mlp.up_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.3.post_attention_layernorm.weight": "model-00002-of-00007.safetensors", + "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00007.safetensors", + "model.layers.30.input_layernorm.weight": "model-00006-of-00007.safetensors", + "model.layers.30.mlp.down_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.30.mlp.gate_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.30.mlp.up_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.30.post_attention_layernorm.weight": "model-00006-of-00007.safetensors", + "model.layers.30.self_attn.k_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.30.self_attn.o_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.30.self_attn.q_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.30.self_attn.v_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.31.input_layernorm.weight": "model-00007-of-00007.safetensors", + "model.layers.31.mlp.down_proj.weight": "model-00007-of-00007.safetensors", + "model.layers.31.mlp.gate_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.31.mlp.up_proj.weight": "model-00007-of-00007.safetensors", + "model.layers.31.post_attention_layernorm.weight": "model-00007-of-00007.safetensors", + "model.layers.31.self_attn.k_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.31.self_attn.o_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.31.self_attn.q_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.31.self_attn.v_proj.weight": "model-00006-of-00007.safetensors", + "model.layers.4.input_layernorm.weight": "model-00002-of-00007.safetensors", + "model.layers.4.mlp.down_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.4.mlp.gate_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.4.mlp.up_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.4.post_attention_layernorm.weight": "model-00002-of-00007.safetensors", + "model.layers.4.self_attn.k_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.4.self_attn.o_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.4.self_attn.q_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.4.self_attn.v_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.5.input_layernorm.weight": "model-00002-of-00007.safetensors", + "model.layers.5.mlp.down_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.5.mlp.gate_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.5.mlp.up_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.5.post_attention_layernorm.weight": "model-00002-of-00007.safetensors", + "model.layers.5.self_attn.k_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.5.self_attn.o_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.5.self_attn.q_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.5.self_attn.v_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.6.input_layernorm.weight": "model-00002-of-00007.safetensors", + "model.layers.6.mlp.down_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.6.mlp.gate_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.6.mlp.up_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.6.post_attention_layernorm.weight": "model-00002-of-00007.safetensors", + "model.layers.6.self_attn.k_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.6.self_attn.o_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.6.self_attn.q_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.6.self_attn.v_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.7.input_layernorm.weight": "model-00002-of-00007.safetensors", + "model.layers.7.mlp.down_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.7.mlp.gate_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.7.mlp.up_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.7.post_attention_layernorm.weight": "model-00002-of-00007.safetensors", + "model.layers.7.self_attn.k_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.7.self_attn.o_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.7.self_attn.q_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.7.self_attn.v_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.8.input_layernorm.weight": "model-00003-of-00007.safetensors", + "model.layers.8.mlp.down_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.8.mlp.gate_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.8.mlp.up_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.8.post_attention_layernorm.weight": "model-00003-of-00007.safetensors", + "model.layers.8.self_attn.k_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.8.self_attn.o_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.8.self_attn.q_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.8.self_attn.v_proj.weight": "model-00002-of-00007.safetensors", + "model.layers.9.input_layernorm.weight": "model-00003-of-00007.safetensors", + "model.layers.9.mlp.down_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.9.mlp.gate_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.9.mlp.up_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.9.post_attention_layernorm.weight": "model-00003-of-00007.safetensors", + "model.layers.9.self_attn.k_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.9.self_attn.o_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.9.self_attn.q_proj.weight": "model-00003-of-00007.safetensors", + "model.layers.9.self_attn.v_proj.weight": "model-00003-of-00007.safetensors", + "model.norm.weight": "model-00007-of-00007.safetensors" + } +} diff --git a/special_tokens_map.json b/special_tokens_map.json new file mode 100644 index 0000000..e5b39b6 --- /dev/null +++ b/special_tokens_map.json @@ -0,0 +1,23 @@ +{ + "bos_token": { + "content": "<|begin_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "eos_token": { + "content": "<|end_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + }, + "pad_token": { + "content": "<|end_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false + } +} diff --git a/tokenizer.json b/tokenizer.json new file mode 100644 index 0000000..86a3394 --- /dev/null +++ b/tokenizer.json @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3c5cf44023714fb39b05e71e425f8d7b92805ff73f7988b083b8c87f0bf87393 +size 17209961 diff --git a/tokenizer_config.json b/tokenizer_config.json new file mode 100644 index 0000000..8c6916a --- /dev/null +++ b/tokenizer_config.json @@ -0,0 +1,2064 @@ +{ + "added_tokens_decoder": { + "128000": { + "content": "<|begin_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128001": { + "content": "<|end_of_text|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128002": { + "content": "<|reserved_special_token_0|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128003": { + "content": "<|reserved_special_token_1|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128004": { + "content": "<|reserved_special_token_2|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128005": { + "content": "<|reserved_special_token_3|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128006": { + "content": "<|start_header_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128007": { + "content": "<|end_header_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128008": { + "content": "<|reserved_special_token_4|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128009": { + "content": "<|eot_id|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128010": { + "content": "<|reserved_special_token_5|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128011": { + "content": "<|reserved_special_token_6|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128012": { + "content": "<|reserved_special_token_7|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128013": { + "content": "<|reserved_special_token_8|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128014": { + "content": "<|reserved_special_token_9|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128015": { + "content": "<|reserved_special_token_10|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128016": { + "content": "<|reserved_special_token_11|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128017": { + "content": "<|reserved_special_token_12|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128018": { + "content": "<|reserved_special_token_13|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128019": { + "content": "<|reserved_special_token_14|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128020": { + "content": "<|reserved_special_token_15|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128021": { + "content": "<|reserved_special_token_16|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128022": { + "content": "<|reserved_special_token_17|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128023": { + "content": "<|reserved_special_token_18|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128024": { + "content": "<|reserved_special_token_19|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128025": { + "content": "<|reserved_special_token_20|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128026": { + "content": "<|reserved_special_token_21|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128027": { + "content": "<|reserved_special_token_22|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128028": { + "content": "<|reserved_special_token_23|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128029": { + "content": "<|reserved_special_token_24|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128030": { + "content": "<|reserved_special_token_25|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128031": { + "content": "<|reserved_special_token_26|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128032": { + "content": "<|reserved_special_token_27|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128033": { + "content": "<|reserved_special_token_28|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128034": { + "content": "<|reserved_special_token_29|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128035": { + "content": "<|reserved_special_token_30|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128036": { + "content": "<|reserved_special_token_31|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128037": { + "content": "<|reserved_special_token_32|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128038": { + "content": "<|reserved_special_token_33|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128039": { + "content": "<|reserved_special_token_34|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128040": { + "content": "<|reserved_special_token_35|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128041": { + "content": "<|reserved_special_token_36|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128042": { + "content": "<|reserved_special_token_37|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128043": { + "content": "<|reserved_special_token_38|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128044": { + "content": "<|reserved_special_token_39|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128045": { + "content": "<|reserved_special_token_40|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128046": { + "content": "<|reserved_special_token_41|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128047": { + "content": "<|reserved_special_token_42|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128048": { + "content": "<|reserved_special_token_43|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128049": { + "content": "<|reserved_special_token_44|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128050": { + "content": "<|reserved_special_token_45|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128051": { + "content": "<|reserved_special_token_46|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128052": { + "content": "<|reserved_special_token_47|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128053": { + "content": "<|reserved_special_token_48|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128054": { + "content": "<|reserved_special_token_49|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128055": { + "content": "<|reserved_special_token_50|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128056": { + "content": "<|reserved_special_token_51|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128057": { + "content": "<|reserved_special_token_52|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128058": { + "content": "<|reserved_special_token_53|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128059": { + "content": "<|reserved_special_token_54|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128060": { + "content": "<|reserved_special_token_55|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128061": { + "content": "<|reserved_special_token_56|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128062": { + "content": "<|reserved_special_token_57|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128063": { + "content": "<|reserved_special_token_58|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128064": { + "content": "<|reserved_special_token_59|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128065": { + "content": "<|reserved_special_token_60|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128066": { + "content": "<|reserved_special_token_61|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128067": { + "content": "<|reserved_special_token_62|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128068": { + "content": "<|reserved_special_token_63|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128069": { + "content": "<|reserved_special_token_64|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128070": { + "content": "<|reserved_special_token_65|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128071": { + "content": "<|reserved_special_token_66|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128072": { + "content": "<|reserved_special_token_67|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128073": { + "content": "<|reserved_special_token_68|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128074": { + "content": "<|reserved_special_token_69|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128075": { + "content": "<|reserved_special_token_70|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128076": { + "content": "<|reserved_special_token_71|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128077": { + "content": "<|reserved_special_token_72|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128078": { + "content": "<|reserved_special_token_73|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128079": { + "content": "<|reserved_special_token_74|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128080": { + "content": "<|reserved_special_token_75|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128081": { + "content": "<|reserved_special_token_76|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128082": { + "content": "<|reserved_special_token_77|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128083": { + "content": "<|reserved_special_token_78|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128084": { + "content": "<|reserved_special_token_79|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128085": { + "content": "<|reserved_special_token_80|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128086": { + "content": "<|reserved_special_token_81|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128087": { + "content": "<|reserved_special_token_82|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128088": { + "content": "<|reserved_special_token_83|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128089": { + "content": "<|reserved_special_token_84|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128090": { + "content": "<|reserved_special_token_85|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128091": { + "content": "<|reserved_special_token_86|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128092": { + "content": "<|reserved_special_token_87|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128093": { + "content": "<|reserved_special_token_88|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128094": { + "content": "<|reserved_special_token_89|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128095": { + "content": "<|reserved_special_token_90|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128096": { + "content": "<|reserved_special_token_91|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128097": { + "content": "<|reserved_special_token_92|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128098": { + "content": "<|reserved_special_token_93|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128099": { + "content": "<|reserved_special_token_94|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128100": { + "content": "<|reserved_special_token_95|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128101": { + "content": "<|reserved_special_token_96|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128102": { + "content": "<|reserved_special_token_97|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128103": { + "content": "<|reserved_special_token_98|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128104": { + "content": "<|reserved_special_token_99|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128105": { + "content": "<|reserved_special_token_100|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128106": { + "content": "<|reserved_special_token_101|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128107": { + "content": "<|reserved_special_token_102|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128108": { + "content": "<|reserved_special_token_103|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128109": { + "content": "<|reserved_special_token_104|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128110": { + "content": "<|reserved_special_token_105|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128111": { + "content": "<|reserved_special_token_106|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128112": { + "content": "<|reserved_special_token_107|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128113": { + "content": "<|reserved_special_token_108|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128114": { + "content": "<|reserved_special_token_109|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128115": { + "content": "<|reserved_special_token_110|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128116": { + "content": "<|reserved_special_token_111|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128117": { + "content": "<|reserved_special_token_112|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128118": { + "content": "<|reserved_special_token_113|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128119": { + "content": "<|reserved_special_token_114|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128120": { + "content": "<|reserved_special_token_115|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128121": { + "content": "<|reserved_special_token_116|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128122": { + "content": "<|reserved_special_token_117|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128123": { + "content": "<|reserved_special_token_118|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128124": { + "content": "<|reserved_special_token_119|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128125": { + "content": "<|reserved_special_token_120|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128126": { + "content": "<|reserved_special_token_121|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128127": { + "content": "<|reserved_special_token_122|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128128": { + "content": "<|reserved_special_token_123|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128129": { + "content": "<|reserved_special_token_124|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128130": { + "content": "<|reserved_special_token_125|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128131": { + "content": "<|reserved_special_token_126|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128132": { + "content": "<|reserved_special_token_127|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128133": { + "content": "<|reserved_special_token_128|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128134": { + "content": "<|reserved_special_token_129|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128135": { + "content": "<|reserved_special_token_130|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128136": { + "content": "<|reserved_special_token_131|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128137": { + "content": "<|reserved_special_token_132|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128138": { + "content": "<|reserved_special_token_133|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128139": { + "content": "<|reserved_special_token_134|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128140": { + "content": "<|reserved_special_token_135|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128141": { + "content": "<|reserved_special_token_136|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128142": { + "content": "<|reserved_special_token_137|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128143": { + "content": "<|reserved_special_token_138|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128144": { + "content": "<|reserved_special_token_139|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128145": { + "content": "<|reserved_special_token_140|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128146": { + "content": "<|reserved_special_token_141|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128147": { + "content": "<|reserved_special_token_142|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128148": { + "content": "<|reserved_special_token_143|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128149": { + "content": "<|reserved_special_token_144|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128150": { + "content": "<|reserved_special_token_145|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128151": { + "content": "<|reserved_special_token_146|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128152": { + "content": "<|reserved_special_token_147|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128153": { + "content": "<|reserved_special_token_148|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128154": { + "content": "<|reserved_special_token_149|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128155": { + "content": "<|reserved_special_token_150|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128156": { + "content": "<|reserved_special_token_151|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128157": { + "content": "<|reserved_special_token_152|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128158": { + "content": "<|reserved_special_token_153|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128159": { + "content": "<|reserved_special_token_154|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128160": { + "content": "<|reserved_special_token_155|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128161": { + "content": "<|reserved_special_token_156|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128162": { + "content": "<|reserved_special_token_157|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128163": { + "content": "<|reserved_special_token_158|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128164": { + "content": "<|reserved_special_token_159|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128165": { + "content": "<|reserved_special_token_160|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128166": { + "content": "<|reserved_special_token_161|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128167": { + "content": "<|reserved_special_token_162|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128168": { + "content": "<|reserved_special_token_163|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128169": { + "content": "<|reserved_special_token_164|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128170": { + "content": "<|reserved_special_token_165|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128171": { + "content": "<|reserved_special_token_166|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128172": { + "content": "<|reserved_special_token_167|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128173": { + "content": "<|reserved_special_token_168|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128174": { + "content": "<|reserved_special_token_169|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128175": { + "content": "<|reserved_special_token_170|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128176": { + "content": "<|reserved_special_token_171|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128177": { + "content": "<|reserved_special_token_172|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128178": { + "content": "<|reserved_special_token_173|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128179": { + "content": "<|reserved_special_token_174|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128180": { + "content": "<|reserved_special_token_175|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128181": { + "content": "<|reserved_special_token_176|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128182": { + "content": "<|reserved_special_token_177|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128183": { + "content": "<|reserved_special_token_178|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128184": { + "content": "<|reserved_special_token_179|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128185": { + "content": "<|reserved_special_token_180|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128186": { + "content": "<|reserved_special_token_181|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128187": { + "content": "<|reserved_special_token_182|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128188": { + "content": "<|reserved_special_token_183|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128189": { + "content": "<|reserved_special_token_184|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128190": { + "content": "<|reserved_special_token_185|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128191": { + "content": "<|reserved_special_token_186|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128192": { + "content": "<|reserved_special_token_187|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128193": { + "content": "<|reserved_special_token_188|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128194": { + "content": "<|reserved_special_token_189|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128195": { + "content": "<|reserved_special_token_190|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128196": { + "content": "<|reserved_special_token_191|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128197": { + "content": "<|reserved_special_token_192|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128198": { + "content": "<|reserved_special_token_193|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128199": { + "content": "<|reserved_special_token_194|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128200": { + "content": "<|reserved_special_token_195|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128201": { + "content": "<|reserved_special_token_196|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128202": { + "content": "<|reserved_special_token_197|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128203": { + "content": "<|reserved_special_token_198|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128204": { + "content": "<|reserved_special_token_199|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128205": { + "content": "<|reserved_special_token_200|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128206": { + "content": "<|reserved_special_token_201|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128207": { + "content": "<|reserved_special_token_202|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128208": { + "content": "<|reserved_special_token_203|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128209": { + "content": "<|reserved_special_token_204|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128210": { + "content": "<|reserved_special_token_205|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128211": { + "content": "<|reserved_special_token_206|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128212": { + "content": "<|reserved_special_token_207|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128213": { + "content": "<|reserved_special_token_208|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128214": { + "content": "<|reserved_special_token_209|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128215": { + "content": "<|reserved_special_token_210|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128216": { + "content": "<|reserved_special_token_211|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128217": { + "content": "<|reserved_special_token_212|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128218": { + "content": "<|reserved_special_token_213|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128219": { + "content": "<|reserved_special_token_214|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128220": { + "content": "<|reserved_special_token_215|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128221": { + "content": "<|reserved_special_token_216|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128222": { + "content": "<|reserved_special_token_217|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128223": { + "content": "<|reserved_special_token_218|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128224": { + "content": "<|reserved_special_token_219|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128225": { + "content": "<|reserved_special_token_220|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128226": { + "content": "<|reserved_special_token_221|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128227": { + "content": "<|reserved_special_token_222|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128228": { + "content": "<|reserved_special_token_223|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128229": { + "content": "<|reserved_special_token_224|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128230": { + "content": "<|reserved_special_token_225|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128231": { + "content": "<|reserved_special_token_226|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128232": { + "content": "<|reserved_special_token_227|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128233": { + "content": "<|reserved_special_token_228|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128234": { + "content": "<|reserved_special_token_229|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128235": { + "content": "<|reserved_special_token_230|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128236": { + "content": "<|reserved_special_token_231|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128237": { + "content": "<|reserved_special_token_232|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128238": { + "content": "<|reserved_special_token_233|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128239": { + "content": "<|reserved_special_token_234|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128240": { + "content": "<|reserved_special_token_235|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128241": { + "content": "<|reserved_special_token_236|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128242": { + "content": "<|reserved_special_token_237|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128243": { + "content": "<|reserved_special_token_238|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128244": { + "content": "<|reserved_special_token_239|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128245": { + "content": "<|reserved_special_token_240|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128246": { + "content": "<|reserved_special_token_241|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128247": { + "content": "<|reserved_special_token_242|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128248": { + "content": "<|reserved_special_token_243|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128249": { + "content": "<|reserved_special_token_244|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128250": { + "content": "<|reserved_special_token_245|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128251": { + "content": "<|reserved_special_token_246|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128252": { + "content": "<|reserved_special_token_247|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128253": { + "content": "<|reserved_special_token_248|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128254": { + "content": "<|reserved_special_token_249|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + }, + "128255": { + "content": "<|reserved_special_token_250|>", + "lstrip": false, + "normalized": false, + "rstrip": false, + "single_word": false, + "special": true + } + }, + "bos_token": "<|begin_of_text|>", + "chat_template": "{% set loop_messages = messages %}{% for message in loop_messages %}{% set content = '<|start_header_id|>' + message['role'] + '<|end_header_id|>\n\n'+ message['content'] | trim + '<|eot_id|>' %}{% if loop.index0 == 0 %}{% set content = bos_token + content %}{% endif %}{{ content }}{% endfor %}{% if add_generation_prompt %}{{ '<|start_header_id|>assistant<|end_header_id|>\n\n' }}{% endif %}", + "clean_up_tokenization_spaces": true, + "eos_token": "<|end_of_text|>", + "extra_special_tokens": {}, + "model_input_names": [ + "input_ids", + "attention_mask" + ], + "model_max_length": 2048, + "pad_token": "<|end_of_text|>", + "tokenizer_class": "PreTrainedTokenizer" +} diff --git a/train.log b/train.log new file mode 100644 index 0000000..0e93d2b --- /dev/null +++ b/train.log @@ -0,0 +1,8498 @@ +2026-04-21 23:39:48 - INFO - __main__ - Model parameters ModelArguments(base_model_revision=None, model_name_or_path='/root/dynamic-dpo-v4/sft-checkpoints/llama-3-8b-base-sft-hh-helpful-4xh200', model_revision='main', model_code_revision=None, torch_dtype='bfloat16', tokenizer_name_or_path=None, trust_remote_code=False, attn_implementation='flash_attention_2', use_peft=False, lora_r=16, lora_alpha=32, lora_dropout=0.05, lora_target_modules=None, lora_modules_to_save=None, load_in_8bit=False, load_in_4bit=False, bnb_4bit_quant_type='nf4', use_bnb_nested_quant=False, bnb_4bit_quant_storage='uint8') +2026-04-21 23:39:48 - INFO - __main__ - Data parameters DataArguments(chat_template=None, dataset_mixer={'Anthropic/hh-rlhf': 1.0}, text_column='text', dataset_splits=['train', 'test'], dataset_configs=['helpful-base'], dataset_dir=None, preprocessing_num_workers=12, use_persistent_hf_cache=True, hf_cache_dir='/root/dynamic-dpo-v4/outputs/hf/datasets', truncation_side=None, auto_insert_empty_system_msg=True, preprocessing_log_samples=0, preprocessing_log_dir=None) +2026-04-21 23:39:48 - INFO - __main__ - Training/evaluation parameters NewDPOConfig( +_n_gpu=1, +accelerator_config={'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None, 'use_configured_state': False}, +adafactor=False, +adam_beta1=0.9, +adam_beta2=0.999, +adam_epsilon=1e-08, +auto_find_batch_size=False, +average_tokens_across_devices=False, +batch_eval_metrics=False, +beta=0.1, +bf16=True, +bf16_full_eval=False, +data_seed=None, +dataloader_drop_last=True, +dataloader_num_workers=0, +dataloader_persistent_workers=False, +dataloader_pin_memory=True, +dataloader_prefetch_factor=None, +dataset_num_proc=12, +ddp_backend=None, +ddp_broadcast_buffers=None, +ddp_bucket_cap_mb=None, +ddp_find_unused_parameters=None, +ddp_timeout=1800, +debug=[], +deepspeed=None, +disable_dropout=True, +disable_tqdm=False, +do_eval=True, +do_predict=False, +do_train=False, +eta=0.1, +eval_accumulation_steps=None, +eval_delay=0, +eval_do_concat_batches=True, +eval_on_start=False, +eval_steps=200, +eval_strategy=IntervalStrategy.STEPS, +eval_use_gather_object=False, +f_alpha_divergence_coef=1.0, +f_divergence_type=reverse_kl, +force_use_ref_model=False, +fp16=False, +fp16_backend=auto, +fp16_full_eval=False, +fp16_opt_level=O1, +fsdp=[], +fsdp_config={'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, +fsdp_min_num_params=0, +fsdp_transformer_layer_cls_to_wrap=None, +full_determinism=False, +generate_during_eval=False, +gradient_accumulation_steps=2, +gradient_checkpointing=True, +gradient_checkpointing_kwargs={'use_reentrant': False}, +greater_is_better=None, +group_by_length=False, +half_precision_backend=auto, +hub_always_push=False, +hub_margin_dataset_id=None, +hub_model_id=jackf857/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802, +hub_model_revision=main, +hub_private_repo=None, +hub_strategy=HubStrategy.EVERY_SAVE, +hub_token=, +ignore_data_skip=False, +include_for_metrics=[], +include_inputs_for_metrics=False, +include_num_input_tokens_seen=False, +include_tokens_per_second=False, +is_encoder_decoder=None, +jit_mode_eval=False, +label_names=None, +label_pad_token_id=-100, +label_smoothing=0.0, +label_smoothing_factor=0.0, +learning_rate=5e-07, +length_column_name=length, +load_best_model_at_end=False, +local_rank=0, +log_level=info, +log_level_replica=warning, +log_on_each_node=True, +logging_dir=outputs/llama3-8b-base-new-method-s_star0.85/runs/Apr21_23-39-47_4ab23a513308, +logging_first_step=True, +logging_nan_inf_filter=True, +logging_steps=5, +logging_strategy=IntervalStrategy.STEPS, +loss_type=sigmoid, +lr_scheduler_kwargs={}, +lr_scheduler_type=SchedulerType.COSINE, +margin_dataset_private=None, +margin_dataset_split=train, +margin_log_path=/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/margin_logs, +margin_log_steps=1, +margin_save_full=True, +max_grad_norm=1.0, +max_length=512, +max_prompt_length=256, +max_steps=-1, +max_target_length=None, +metric_for_best_model=None, +model_adapter_name=None, +model_init_kwargs=None, +mp_parameters=, +neftune_noise_alpha=None, +no_cuda=False, +non_finite_logits_handling=error, +num_train_epochs=1, +optim=OptimizerNames.ADAMW_TORCH, +optim_args=None, +optim_target_modules=None, +output_dir=/root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802, +overwrite_output_dir=False, +padding_value=None, +past_index=-1, +per_device_eval_batch_size=8, +per_device_train_batch_size=8, +post_tokenization_log_dir=None, +post_tokenization_log_samples=0, +precompute_ref_batch_size=None, +precompute_ref_eval_batch_size=None, +precompute_ref_log_probs=False, +prediction_loss_only=False, +push_margin_dataset=True, +push_to_hub=True, +push_to_hub_model_id=None, +push_to_hub_organization=None, +push_to_hub_token=, +q_target=0.45, +ray_scope=last, +ref_adapter_name=None, +ref_model_init_kwargs=None, +ref_model_mixup_alpha=0.9, +ref_model_sync_steps=64, +reference_free=False, +remove_unused_columns=False, +report_to=['wandb'], +require_explicit_ref_model=True, +restore_callback_states_from_checkpoint=False, +resume_from_checkpoint=None, +reuse_tokenized_dataset=True, +rpo_alpha=None, +run_name=llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802, +s_star=0.85, +save_on_each_node=False, +save_only_model=False, +save_safetensors=True, +save_steps=50, +save_strategy=SaveStrategy.NO, +save_total_limit=2, +seed=42, +sft_weight=0.0, +skip_memory_metrics=True, +sync_ref_model=False, +tf32=None, +tokenization_batch_size=128, +tokenization_mode=online, +tokenized_dataset_cache_dir=/root/dynamic-dpo-v4/outputs/tokenized_preferences, +torch_compile=False, +torch_compile_backend=None, +torch_compile_mode=None, +torch_empty_cache_steps=None, +torchdynamo=None, +tp_size=0, +tpu_metrics_debug=False, +tpu_num_cores=None, +trainer_type=new_dpo, +truncation_mode=keep_end, +use_cpu=False, +use_ipex=False, +use_legacy_prediction_loop=False, +use_liger_kernel=False, +use_mps_device=False, +wandb_project=llama3-8b-base-new-method-hh-beta-0.1, +warmup_ratio=0.1, +warmup_steps=0, +weight_decay=0.0, +) +2026-04-21 23:39:48 - INFO - __main__ - Using W&B project from training args: llama3-8b-base-new-method-hh-beta-0.1 +2026-04-21 23:39:48 - INFO - __main__ - New-DPO parameters: beta=0.1, q_target=0.45, s_star=0.85, eta=0.1 +2026-04-21 23:39:48 - INFO - __main__ - Using persistent HF datasets cache at /root/dynamic-dpo-v4/outputs/hf/datasets + Downloading readme: 0%| | 0.00/5.77k [00:00> loading file tokenizer.json +[INFO|tokenization_utils_base.py:2058] 2026-04-21 23:40:00,308 >> loading file tokenizer.model +[INFO|tokenization_utils_base.py:2058] 2026-04-21 23:40:00,308 >> loading file added_tokens.json +[INFO|tokenization_utils_base.py:2058] 2026-04-21 23:40:00,308 >> loading file special_tokens_map.json +[INFO|tokenization_utils_base.py:2058] 2026-04-21 23:40:00,308 >> loading file tokenizer_config.json +[INFO|tokenization_utils_base.py:2058] 2026-04-21 23:40:00,308 >> loading file chat_template.jinja +[INFO|tokenization_utils_base.py:2323] 2026-04-21 23:40:00,686 >> Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained. + Formatting comparisons with prompt template (num_proc=12): 1%|▋ | 300/43598 [00:00<01:42, 424.28 examples/s] Formatting comparisons with prompt template (num_proc=12): 0%| | 0/43598 [00:00> You are attempting to use Flash Attention 2.0 with a model not initialized on GPU. Make sure to move the model to GPU after initializing it on CPU with `model.to('cuda')`. + Formatting comparisons with prompt template (num_proc=12): 83%|███████████████████████████████████████████████████████████████████████████▊ | 1950/2339 [00:02<00:00, 1326.95 examples/s] Loading checkpoint shards: 0%| | 0/7 [00:00> Trainer.tokenizer is now deprecated. You should use `Trainer.processing_class = processing_class` instead. + Formatting comparisons with prompt template (num_proc=12): 92%|███████████████████████████████████████████████████████████████████████████████████▍ | 2145/2339 [00:02<00:00, 1290.71 examples/s] Formatting comparisons with prompt template (num_proc=12): 92%|███████████████████████████████████████████████████████████████████████████████████▍ | 2145/2339 [00:01<00:00, 1774.39 examples/s] Formatting comparisons with prompt template (num_proc=12): 58%|█████████████████████████████████████████████████████ | 1365/2339 [00:01<00:00, 1100.99 examples/s] Formatting comparisons with prompt template (num_proc=12): 100%|███████████████████████████████████████████████████████████████████████████████████████████| 2339/2339 [00:02<00:00, 1257.73 examples/s] Formatting comparisons with prompt template (num_proc=12): 100%|███████████████████████████████████████████████████████████████████████████████████████████| 2339/2339 [00:01<00:00, 1289.93 examples/s] + Formatting comparisons with prompt template (num_proc=12): 67%|████████████████████████████████████████████████████████████▋ | 1560/2339 [00:01<00:00, 1131.21 examples/s] Formatting comparisons with prompt template (num_proc=12): 100%|████████████████████████████████████████████████████████████████████████████████████████████| 2339/2339 [00:02<00:00, 962.01 examples/s] +2026-04-21 23:40:06 - INFO - __main__ - Processed train sample 41905: + +Prompt: +<|begin_of_text|><|start_header_id|>user<|end_header_id|> + +What is the best way to clean my refrigerator?<|eot_id|><|start_header_id|>assistant<|end_header_id|> + +Well, I have a few ideas, but the best way to know what works best is to run some tests and then analyze the data to figure out which technique works best.<|eot_id|><|start_header_id|>user<|end_header_id|> + +What are your ideas?<|eot_id|> + +Chosen: +<|start_header_id|>assistant<|end_header_id|> + +Some of my ideas include wiping down the interior with a rag, using bleach and water to wipe down the door, emptying the contents of the fridge to scrub the shelves with a clean cloth, cleaning the interior of the fridge with vinegar and water, and using a vacuum cleaner to clean the interior.<|eot_id|> + +Rejected: +<|start_header_id|>assistant<|end_header_id|> + +I was thinking that you could try different types of soap, which could be a variation of “soap,” like antibacterial soap. I was also thinking that you could try using different tools, like a toothbrush, to clean the inside. Or I was thinking that you could try different ways to get the gunk out, like using a squeegee, or using a paper towel. I was also thinking you could try other types of cleaning, like vacuuming, but I think that could have the opposite of the desired effect.<|eot_id|> +/root/dynamic-dpo-v4/scripts/tokenized_dpo_trainer.py:391: UserWarning: You passed a model_id to the trainer. This will automatically create an `AutoModelForCausalLM` or a `PeftModel` (if you passed a `peft_config`) for you. + warnings.warn( +[INFO|configuration_utils.py:691] 2026-04-21 23:40:06,096 >> loading configuration file /root/dynamic-dpo-v4/sft-checkpoints/llama-3-8b-base-sft-hh-helpful-4xh200/config.json +[INFO|configuration_utils.py:765] 2026-04-21 23:40:06,098 >> Model config LlamaConfig { + "architectures": [ + "LlamaForCausalLM" + ], + "attention_bias": false, + "attention_dropout": 0.0, + "bos_token_id": 128000, + "eos_token_id": 128001, + "head_dim": 128, + "hidden_act": "silu", + "hidden_size": 4096, + "initializer_range": 0.02, + "intermediate_size": 14336, + "max_position_embeddings": 8192, + "mlp_bias": false, + "model_type": "llama", + "num_attention_heads": 32, + "num_hidden_layers": 32, + "num_key_value_heads": 8, + "pretraining_tp": 1, + "rms_norm_eps": 1e-05, + "rope_scaling": null, + "rope_theta": 500000.0, + "tie_word_embeddings": false, + "torch_dtype": "bfloat16", + "transformers_version": "4.51.0", + "use_cache": false, + "vocab_size": 128256 +} + +[INFO|modeling_utils.py:1121] 2026-04-21 23:40:06,119 >> loading weights file /root/dynamic-dpo-v4/sft-checkpoints/llama-3-8b-base-sft-hh-helpful-4xh200/model.safetensors.index.json +[INFO|modeling_utils.py:2167] 2026-04-21 23:40:06,119 >> Instantiating LlamaForCausalLM model under default dtype torch.bfloat16. +[WARNING|logging.py:328] 2026-04-21 23:40:06,121 >> You are attempting to use Flash Attention 2.0 with a model not initialized on GPU. Make sure to move the model to GPU after initializing it on CPU with `model.to('cuda')`. +[INFO|configuration_utils.py:1142] 2026-04-21 23:40:06,123 >> Generate config GenerationConfig { + "bos_token_id": 128000, + "eos_token_id": 128001, + "use_cache": false +} + +/root/dynamic-dpo-v4/scripts/tokenized_dpo_trainer.py:391: UserWarning: You passed a model_id to the trainer. This will automatically create an `AutoModelForCausalLM` or a `PeftModel` (if you passed a `peft_config`) for you. + warnings.warn( + Loading checkpoint shards: 0%| | 0/7 [00:00> You are attempting to use Flash Attention 2.0 with a model not initialized on GPU. Make sure to move the model to GPU after initializing it on CPU with `model.to('cuda')`. + Formatting comparisons with prompt template (num_proc=12): 75%|████████████████████████████████████████████████████████████████████▎ | 1755/2339 [00:01<00:00, 1128.68 examples/s] Loading checkpoint shards: 0%| | 0/7 [00:00> Trainer.tokenizer is now deprecated. You should use `Trainer.processing_class = processing_class` instead. + Formatting comparisons with prompt template (num_proc=12): 83%|███████████████████████████████████████████████████████████████████████████▊ | 1950/2339 [00:02<00:00, 1164.13 examples/s] Formatting comparisons with prompt template (num_proc=12): 92%|███████████████████████████████████████████████████████████████████████████████████▍ | 2144/2339 [00:02<00:00, 1226.02 examples/s] Formatting comparisons with prompt template (num_proc=12): 92%|████████████████████████████████████████████████████████████████████████████████████▎ | 2144/2339 [00:02<00:00, 929.94 examples/s] +/root/dynamic-dpo-v4/scripts/tokenized_dpo_trainer.py:391: UserWarning: You passed a model_id to the trainer. This will automatically create an `AutoModelForCausalLM` or a `PeftModel` (if you passed a `peft_config`) for you. + warnings.warn( +[WARNING|logging.py:328] 2026-04-21 23:40:06,668 >> You are attempting to use Flash Attention 2.0 with a model not initialized on GPU. Make sure to move the model to GPU after initializing it on CPU with `model.to('cuda')`. + Loading checkpoint shards: 0%| | 0/7 [00:00> Trainer.tokenizer is now deprecated. You should use `Trainer.processing_class = processing_class` instead. + Loading checkpoint shards: 14%|███████████████████▋ | 1/7 [00:01<00:10, 1.67s/it] Loading checkpoint shards: 29%|███████████████████████████████████████▍ | 2/7 [00:03<00:08, 1.61s/it] Loading checkpoint shards: 43%|███████████████████████████████████████████████████████████▏ | 3/7 [00:04<00:06, 1.63s/it] Loading checkpoint shards: 57%|██████████████████████████████████████████████████████████████████████████████▊ | 4/7 [00:06<00:04, 1.64s/it] Loading checkpoint shards: 71%|██████████████████████████████████████████████████████████████████████████████████████████████████▌ | 5/7 [00:08<00:03, 1.60s/it] Loading checkpoint shards: 86%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████▎ | 6/7 [00:09<00:01, 1.62s/it] Loading checkpoint shards: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 7/7 [00:10<00:00, 1.35s/it] Loading checkpoint shards: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 7/7 [00:10<00:00, 1.50s/it] +[INFO|modeling_utils.py:4926] 2026-04-21 23:40:16,675 >> All model checkpoint weights were used when initializing LlamaForCausalLM. + +[INFO|modeling_utils.py:4934] 2026-04-21 23:40:16,676 >> All the weights of LlamaForCausalLM were initialized from the model checkpoint at /root/dynamic-dpo-v4/sft-checkpoints/llama-3-8b-base-sft-hh-helpful-4xh200. +If your task is similar to the task the model of the checkpoint was trained on, you can already use LlamaForCausalLM for predictions without further training. +[INFO|configuration_utils.py:1095] 2026-04-21 23:40:16,678 >> loading configuration file /root/dynamic-dpo-v4/sft-checkpoints/llama-3-8b-base-sft-hh-helpful-4xh200/generation_config.json +[INFO|configuration_utils.py:1142] 2026-04-21 23:40:16,679 >> Generate config GenerationConfig { + "bos_token_id": 128000, + "do_sample": true, + "eos_token_id": 128001, + "max_length": 4096, + "temperature": 0.6, + "top_p": 0.9 +} + +[INFO|configuration_utils.py:691] 2026-04-21 23:40:16,680 >> loading configuration file /root/dynamic-dpo-v4/sft-checkpoints/llama-3-8b-base-sft-hh-helpful-4xh200/config.json +[INFO|configuration_utils.py:765] 2026-04-21 23:40:16,680 >> Model config LlamaConfig { + "architectures": [ + "LlamaForCausalLM" + ], + "attention_bias": false, + "attention_dropout": 0.0, + "bos_token_id": 128000, + "eos_token_id": 128001, + "head_dim": 128, + "hidden_act": "silu", + "hidden_size": 4096, + "initializer_range": 0.02, + "intermediate_size": 14336, + "max_position_embeddings": 8192, + "mlp_bias": false, + "model_type": "llama", + "num_attention_heads": 32, + "num_hidden_layers": 32, + "num_key_value_heads": 8, + "pretraining_tp": 1, + "rms_norm_eps": 1e-05, + "rope_scaling": null, + "rope_theta": 500000.0, + "tie_word_embeddings": false, + "torch_dtype": "bfloat16", + "transformers_version": "4.51.0", + "use_cache": false, + "vocab_size": 128256 +} + +[INFO|modeling_utils.py:1121] 2026-04-21 23:40:16,681 >> loading weights file /root/dynamic-dpo-v4/sft-checkpoints/llama-3-8b-base-sft-hh-helpful-4xh200/model.safetensors.index.json +[INFO|modeling_utils.py:2167] 2026-04-21 23:40:16,681 >> Instantiating LlamaForCausalLM model under default dtype torch.bfloat16. +[INFO|configuration_utils.py:1142] 2026-04-21 23:40:16,684 >> Generate config GenerationConfig { + "bos_token_id": 128000, + "eos_token_id": 128001, + "use_cache": false +} + + Loading checkpoint shards: 0%| | 0/7 [00:00> All model checkpoint weights were used when initializing LlamaForCausalLM. + +[INFO|modeling_utils.py:4934] 2026-04-21 23:40:27,021 >> All the weights of LlamaForCausalLM were initialized from the model checkpoint at /root/dynamic-dpo-v4/sft-checkpoints/llama-3-8b-base-sft-hh-helpful-4xh200. +If your task is similar to the task the model of the checkpoint was trained on, you can already use LlamaForCausalLM for predictions without further training. +[INFO|configuration_utils.py:1095] 2026-04-21 23:40:27,023 >> loading configuration file /root/dynamic-dpo-v4/sft-checkpoints/llama-3-8b-base-sft-hh-helpful-4xh200/generation_config.json +[INFO|configuration_utils.py:1142] 2026-04-21 23:40:27,023 >> Generate config GenerationConfig { + "bos_token_id": 128000, + "do_sample": true, + "eos_token_id": 128001, + "max_length": 4096, + "temperature": 0.6, + "top_p": 0.9 +} + +[WARNING|trainer.py:821] 2026-04-21 23:40:27,025 >> Trainer.tokenizer is now deprecated. You should use `Trainer.processing_class = processing_class` instead. +[WARNING|trainer.py:816] 2026-04-21 23:40:27,025 >> Trainer.tokenizer is now deprecated. You should use Trainer.processing_class instead. + Tokenizing train (num_proc=12): 0%| | 0/43598 [00:00> Trainer.tokenizer is now deprecated. You should use Trainer.processing_class instead. + Saving the dataset (0/2 shards): 0%| | 0/43598 [00:00> Trainer.tokenizer is now deprecated. You should use Trainer.processing_class instead. + Tokenizing test (num_proc=12): 0%| | 0/2339 [00:00> Trainer.tokenizer is now deprecated. You should use Trainer.processing_class instead. + Saving the dataset (0/1 shards): 0%| | 0/2339 [00:00> Trainer.tokenizer is now deprecated. You should use Trainer.processing_class instead. +[WARNING|trainer.py:816] 2026-04-22 00:00:45,263 >> Trainer.tokenizer is now deprecated. You should use Trainer.processing_class instead. +[WARNING|trainer.py:816] 2026-04-22 00:00:45,264 >> Trainer.tokenizer is now deprecated. You should use Trainer.processing_class instead. +[WARNING|trainer.py:816] 2026-04-22 00:00:45,274 >> Trainer.tokenizer is now deprecated. You should use Trainer.processing_class instead. +[WARNING|trainer.py:816] 2026-04-22 00:00:45,275 >> Trainer.tokenizer is now deprecated. You should use Trainer.processing_class instead. +[WARNING|trainer.py:816] 2026-04-22 00:00:45,277 >> Trainer.tokenizer is now deprecated. You should use Trainer.processing_class instead. +[WARNING|trainer.py:816] 2026-04-22 00:00:45,278 >> Trainer.tokenizer is now deprecated. You should use Trainer.processing_class instead. +[WARNING|trainer.py:816] 2026-04-22 00:00:45,278 >> Trainer.tokenizer is now deprecated. You should use Trainer.processing_class instead. +[WARNING|trainer.py:816] 2026-04-22 00:00:45,278 >> Trainer.tokenizer is now deprecated. You should use Trainer.processing_class instead. +[WARNING|trainer.py:816] 2026-04-22 00:00:45,281 >> Trainer.tokenizer is now deprecated. You should use Trainer.processing_class instead. +/root/dynamic-dpo-v4/scripts/tokenized_dpo_trainer.py:518: FutureWarning: `tokenizer` is deprecated and will be removed in version 5.0.0 for `NewDPOTrainer.__init__`. Use `processing_class` instead. + super().__init__( +[WARNING|trainer.py:816] 2026-04-22 00:00:45,285 >> Trainer.tokenizer is now deprecated. You should use Trainer.processing_class instead. +[WARNING|trainer.py:816] 2026-04-22 00:00:45,285 >> Trainer.tokenizer is now deprecated. You should use Trainer.processing_class instead. +/root/dynamic-dpo-v4/scripts/tokenized_dpo_trainer.py:518: FutureWarning: `tokenizer` is deprecated and will be removed in version 5.0.0 for `NewDPOTrainer.__init__`. Use `processing_class` instead. + super().__init__( +/root/dynamic-dpo-v4/scripts/tokenized_dpo_trainer.py:518: FutureWarning: `tokenizer` is deprecated and will be removed in version 5.0.0 for `NewDPOTrainer.__init__`. Use `processing_class` instead. + super().__init__( +[INFO|trainer.py:748] 2026-04-22 00:00:45,813 >> Using auto half precision backend +/root/dynamic-dpo-v4/.venv/lib/python3.11/site-packages/accelerate/accelerator.py:1557: UserWarning: Upcasted low precision parameters in LlamaForCausalLM because mixed precision turned on in FSDP. Affects: model.embed_tokens.weight, model.norm.weight, lm_head.weight. + warnings.warn( +/root/dynamic-dpo-v4/.venv/lib/python3.11/site-packages/accelerate/accelerator.py:1557: UserWarning: Upcasted low precision parameters in LlamaDecoderLayer because mixed precision turned on in FSDP. Affects: self_attn.q_proj.weight, self_attn.k_proj.weight, self_attn.v_proj.weight, self_attn.o_proj.weight, mlp.gate_proj.weight, mlp.up_proj.weight, mlp.down_proj.weight, input_layernorm.weight, post_attention_layernorm.weight. + warnings.warn( +/root/dynamic-dpo-v4/.venv/lib/python3.11/site-packages/accelerate/accelerator.py:1563: UserWarning: FSDP upcast of low precision parameters may affect the precision of model checkpoints. + warnings.warn( +[INFO|trainer.py:2414] 2026-04-22 00:00:54,288 >> ***** Running training ***** +[INFO|trainer.py:2415] 2026-04-22 00:00:54,288 >> Num examples = 43,598 +[INFO|trainer.py:2416] 2026-04-22 00:00:54,288 >> Num Epochs = 1 +[INFO|trainer.py:2417] 2026-04-22 00:00:54,288 >> Instantaneous batch size per device = 8 +[INFO|trainer.py:2420] 2026-04-22 00:00:54,288 >> Total train batch size (w. parallel, distributed & accumulation) = 64 +[INFO|trainer.py:2421] 2026-04-22 00:00:54,288 >> Gradient Accumulation steps = 2 +[INFO|trainer.py:2422] 2026-04-22 00:00:54,288 >> Total optimization steps = 681 +[INFO|trainer.py:2423] 2026-04-22 00:00:54,289 >> Number of trainable parameters = 2,007,565,312 +[INFO|integration_utils.py:831] 2026-04-22 00:00:54,289 >> Automatic Weights & Biases logging enabled, to disable set os.environ["WANDB_DISABLED"] = "true" +wandb: Currently logged in as: feng-cheng (feng-cheng-northeastern-university). Use `wandb login --relogin` to force relogin +wandb: - Waiting for wandb.init()... wandb: \ Waiting for wandb.init()... wandb: wandb version 0.26.0 is available! To upgrade, please run: +wandb: $ pip install wandb --upgrade +wandb: Tracking run with wandb version 0.17.5 +wandb: Run data is saved locally in /root/dynamic-dpo-v4/outputs/wandb/wandb/run-20260422_000055-zcqg1rl4 +wandb: Run `wandb offline` to turn off syncing. +wandb: Syncing run llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802 +wandb: ⭐️ View project at https://wandb.ai/feng-cheng-northeastern-university/llama3-8b-base-new-method-hh-beta-0.1 +wandb: 🚀 View run at https://wandb.ai/feng-cheng-northeastern-university/llama3-8b-base-new-method-hh-beta-0.1/runs/zcqg1rl4 + 0%| | 0/681 [00:00> Could not estimate the number of tokens of the input, floating-point operations will not be computed +[WARNING|modeling_utils.py:1713] 2026-04-22 00:00:58,847 >> Could not estimate the number of tokens of the input, floating-point operations will not be computed +[WARNING|modeling_utils.py:1713] 2026-04-22 00:00:58,847 >> Could not estimate the number of tokens of the input, floating-point operations will not be computed +[WARNING|modeling_utils.py:1713] 2026-04-22 00:00:58,855 >> Could not estimate the number of tokens of the input, floating-point operations will not be computed + 0%|▏ | 1/681 [00:02<31:20, 2.77s/it] {'loss': 1.389, 'grad_norm': 83.52236938476562, 'learning_rate': 0.0, 'fcm_dpo/beta': 0.10000000149011612, 'fcm_dpo/q_t': 0.5005706548690796, 'fcm_dpo/delta': 0.0, 'fcm_dpo/margin': -0.02287006378173828, 'margin_dpo/margin_mean': -0.02287048101425171, 'margin_dpo/margin_std': 0.41920793056488037, 'logps/chosen': -50.1435661315918, 'logps/rejected': -74.09991455078125, 'logps/ref_chosen': -50.14883804321289, 'logps/ref_rejected': -74.1280517578125, 'logits/chosen': -0.4974287748336792, 'logits/rejected': -0.43299180269241333, 'epoch': 0.0} + 0%|▏ | 1/681 [00:02<31:20, 2.77s/it] 0%|▍ | 2/681 [00:05<29:27, 2.60s/it] 0%|▋ | 3/681 [00:07<29:01, 2.57s/it] 1%|▉ | 4/681 [00:10<29:18, 2.60s/it] 1%|█▏ | 5/681 [00:12<29:02, 2.58s/it] {'loss': 1.3898, 'grad_norm': 89.65718078613281, 'learning_rate': 2.898550724637681e-08, 'fcm_dpo/beta': 0.10000000894069672, 'fcm_dpo/q_t': 0.5008123517036438, 'fcm_dpo/delta': 0.0, 'fcm_dpo/margin': -0.03249748796224594, 'margin_dpo/margin_mean': -0.03249724209308624, 'margin_dpo/margin_std': 0.34048572182655334, 'logps/chosen': -56.06619644165039, 'logps/rejected': -78.66961669921875, 'logps/ref_chosen': -56.05734634399414, 'logps/ref_rejected': -78.69325256347656, 'logits/chosen': -0.4900774359703064, 'logits/rejected': -0.453380823135376, 'epoch': 0.01} + 1%|█▏ | 5/681 [00:13<29:02, 2.58s/it] 1%|█▍ | 6/681 [00:15<27:33, 2.45s/it] 1%|█▋ | 7/681 [00:17<26:53, 2.39s/it] 1%|█▉ | 8/681 [00:19<26:37, 2.37s/it] 1%|██▏ | 9/681 [00:22<27:09, 2.42s/it] 1%|██▍ | 10/681 [00:24<27:47, 2.49s/it] {'loss': 1.3888, 'grad_norm': 70.88235473632812, 'learning_rate': 6.521739130434782e-08, 'fcm_dpo/beta': 0.10000000894069672, 'fcm_dpo/q_t': 0.5005213022232056, 'fcm_dpo/delta': 0.0, 'fcm_dpo/margin': -0.020858580246567726, 'margin_dpo/margin_mean': -0.02085849642753601, 'margin_dpo/margin_std': 0.38151517510414124, 'logps/chosen': -59.54156494140625, 'logps/rejected': -91.14656066894531, 'logps/ref_chosen': -59.54457473754883, 'logps/ref_rejected': -91.17041778564453, 'logits/chosen': -0.5124648809432983, 'logits/rejected': -0.4746035933494568, 'epoch': 0.01} + 1%|██▍ | 10/681 [00:24<27:47, 2.49s/it] 2%|██▌ | 11/681 [00:27<28:28, 2.55s/it] 2%|██▊ | 12/681 [00:30<28:33, 2.56s/it] 2%|███ | 13/681 [00:32<28:51, 2.59s/it] 2%|███▎ | 14/681 [00:35<28:26, 2.56s/it] 2%|███▌ | 15/681 [00:37<28:16, 2.55s/it] {'loss': 1.3876, 'grad_norm': 64.12911987304688, 'learning_rate': 1.0144927536231885e-07, 'fcm_dpo/beta': 0.10000000894069672, 'fcm_dpo/q_t': 0.5002168416976929, 'fcm_dpo/delta': 0.0, 'fcm_dpo/margin': -0.008700313977897167, 'margin_dpo/margin_mean': -0.008700350299477577, 'margin_dpo/margin_std': 0.3919045627117157, 'logps/chosen': -58.85633087158203, 'logps/rejected': -92.95516967773438, 'logps/ref_chosen': -58.83195877075195, 'logps/ref_rejected': -92.93949890136719, 'logits/chosen': -0.4861031174659729, 'logits/rejected': -0.45661497116088867, 'epoch': 0.02} + 2%|███▌ | 15/681 [00:37<28:16, 2.55s/it] 2%|███▊ | 16/681 [00:40<27:54, 2.52s/it] 2%|████ | 17/681 [00:42<27:37, 2.50s/it] 3%|████▎ | 18/681 [00:45<27:32, 2.49s/it] 3%|████▌ | 19/681 [00:47<27:37, 2.50s/it] 3%|████▊ | 20/681 [00:50<27:34, 2.50s/it] {'loss': 1.3805, 'grad_norm': 73.97709655761719, 'learning_rate': 1.3768115942028986e-07, 'fcm_dpo/beta': 0.10000000894069672, 'fcm_dpo/q_t': 0.4984763562679291, 'fcm_dpo/delta': 0.0, 'fcm_dpo/margin': 0.060948461294174194, 'margin_dpo/margin_mean': 0.06094840168952942, 'margin_dpo/margin_std': 0.3420618772506714, 'logps/chosen': -59.64154815673828, 'logps/rejected': -82.82310485839844, 'logps/ref_chosen': -59.6396598815918, 'logps/ref_rejected': -82.76026916503906, 'logits/chosen': -0.5145794153213501, 'logits/rejected': -0.46688389778137207, 'epoch': 0.03} + 3%|████▊ | 20/681 [00:50<27:34, 2.50s/it] 3%|████▉ | 21/681 [00:52<27:25, 2.49s/it] 3%|█████▏ | 22/681 [00:55<27:57, 2.54s/it] 3%|█████▍ | 23/681 [00:58<28:58, 2.64s/it] 4%|█████▋ | 24/681 [01:00<28:52, 2.64s/it] 4%|█████▉ | 25/681 [01:03<28:50, 2.64s/it] {'loss': 1.3678, 'grad_norm': 74.20327758789062, 'learning_rate': 1.7391304347826085e-07, 'fcm_dpo/beta': 0.10000000894069672, 'fcm_dpo/q_t': 0.49522167444229126, 'fcm_dpo/delta': 0.0, 'fcm_dpo/margin': 0.19125637412071228, 'margin_dpo/margin_mean': 0.19125640392303467, 'margin_dpo/margin_std': 0.4390522539615631, 'logps/chosen': -53.17488479614258, 'logps/rejected': -89.15694427490234, 'logps/ref_chosen': -53.205284118652344, 'logps/ref_rejected': -88.99608612060547, 'logits/chosen': -0.5046600103378296, 'logits/rejected': -0.47779035568237305, 'epoch': 0.04} + 4%|█████▉ | 25/681 [01:03<28:50, 2.64s/it] 4%|██████▏ | 26/681 [01:05<27:27, 2.52s/it] 4%|██████▍ | 27/681 [01:08<27:11, 2.49s/it] 4%|██████▋ | 28/681 [01:10<27:17, 2.51s/it] 4%|██████▉ | 29/681 [01:12<26:12, 2.41s/it] 4%|███████▏ | 30/681 [01:15<26:46, 2.47s/it] {'loss': 1.3412, 'grad_norm': 87.46487426757812, 'learning_rate': 2.1014492753623187e-07, 'fcm_dpo/beta': 0.10000000894069672, 'fcm_dpo/q_t': 0.4883840084075928, 'fcm_dpo/delta': 0.0, 'fcm_dpo/margin': 0.465215265750885, 'margin_dpo/margin_mean': 0.4652155041694641, 'margin_dpo/margin_std': 0.5838907957077026, 'logps/chosen': -53.4693603515625, 'logps/rejected': -98.25569152832031, 'logps/ref_chosen': -53.5526008605957, 'logps/ref_rejected': -97.87371826171875, 'logits/chosen': -0.524259626865387, 'logits/rejected': -0.4842991232872009, 'epoch': 0.04} + 4%|███████▏ | 30/681 [01:15<26:46, 2.47s/it] 5%|███████▎ | 31/681 [01:18<27:20, 2.52s/it] 5%|███████▌ | 32/681 [01:20<27:55, 2.58s/it] 5%|███████▊ | 33/681 [01:23<27:38, 2.56s/it] 5%|████████ | 34/681 [01:26<27:38, 2.56s/it] 5%|████████▎ | 35/681 [01:28<27:57, 2.60s/it] {'loss': 1.3116, 'grad_norm': 83.3965072631836, 'learning_rate': 2.463768115942029e-07, 'fcm_dpo/beta': 0.10000000894069672, 'fcm_dpo/q_t': 0.4805462956428528, 'fcm_dpo/delta': 0.0, 'fcm_dpo/margin': 0.7807711362838745, 'margin_dpo/margin_mean': 0.7807710766792297, 'margin_dpo/margin_std': 0.820033848285675, 'logps/chosen': -56.19721221923828, 'logps/rejected': -92.41671752929688, 'logps/ref_chosen': -56.3298454284668, 'logps/ref_rejected': -91.76858520507812, 'logits/chosen': -0.5029053092002869, 'logits/rejected': -0.4691234529018402, 'epoch': 0.05} + 5%|████████▎ | 35/681 [01:28<27:57, 2.60s/it] 5%|████████▌ | 36/681 [01:31<28:00, 2.61s/it] 5%|████████▊ | 37/681 [01:33<27:49, 2.59s/it] 6%|█████████ | 38/681 [01:36<26:32, 2.48s/it] 6%|█████████▎ | 39/681 [01:38<26:32, 2.48s/it] 6%|█████████▌ | 40/681 [01:41<27:08, 2.54s/it] {'loss': 1.2592, 'grad_norm': 60.05709457397461, 'learning_rate': 2.8260869565217386e-07, 'fcm_dpo/beta': 0.10000000894069672, 'fcm_dpo/q_t': 0.46588778495788574, 'fcm_dpo/delta': 0.0, 'fcm_dpo/margin': 1.3780633211135864, 'margin_dpo/margin_mean': 1.3780632019042969, 'margin_dpo/margin_std': 1.4964091777801514, 'logps/chosen': -54.2760009765625, 'logps/rejected': -84.21268463134766, 'logps/ref_chosen': -54.38492965698242, 'logps/ref_rejected': -82.94353485107422, 'logits/chosen': -0.5348217487335205, 'logits/rejected': -0.4983762204647064, 'epoch': 0.06} + 6%|█████████▌ | 40/681 [01:41<27:08, 2.54s/it] 6%|█████████▊ | 41/681 [01:43<27:01, 2.53s/it] 6%|█████████▉ | 42/681 [01:46<26:54, 2.53s/it] 6%|██████████▏ | 43/681 [01:48<26:56, 2.53s/it] 6%|██████████▍ | 44/681 [01:51<27:13, 2.56s/it] 7%|██████████▋ | 45/681 [01:54<27:17, 2.58s/it] {'loss': 1.1198, 'grad_norm': 84.72518920898438, 'learning_rate': 3.188405797101449e-07, 'fcm_dpo/beta': 0.13015565276145935, 'fcm_dpo/q_t': 0.42787855863571167, 'fcm_dpo/delta': 0.4826194643974304, 'fcm_dpo/margin': 2.396104335784912, 'margin_dpo/margin_mean': 2.396104335784912, 'margin_dpo/margin_std': 2.2702395915985107, 'logps/chosen': -54.64501953125, 'logps/rejected': -100.20524597167969, 'logps/ref_chosen': -54.862335205078125, 'logps/ref_rejected': -98.0264663696289, 'logits/chosen': -0.5097236633300781, 'logits/rejected': -0.4814884662628174, 'epoch': 0.07} + 7%|██████████▋ | 45/681 [01:54<27:17, 2.58s/it] 7%|██████████▉ | 46/681 [01:56<27:28, 2.60s/it] 7%|███████████▏ | 47/681 [01:59<27:26, 2.60s/it] 7%|███████████▍ | 48/681 [02:02<27:50, 2.64s/it] 7%|███████████▋ | 49/681 [02:04<27:24, 2.60s/it] 7%|███████████▉ | 50/681 [02:07<27:28, 2.61s/it] {'loss': 0.9364, 'grad_norm': 93.39527130126953, 'learning_rate': 3.5507246376811595e-07, 'fcm_dpo/beta': 0.189666748046875, 'fcm_dpo/q_t': 0.36233529448509216, 'fcm_dpo/delta': 0.21794584393501282, 'fcm_dpo/margin': 3.3721413612365723, 'margin_dpo/margin_mean': 3.3721413612365723, 'margin_dpo/margin_std': 3.3471221923828125, 'logps/chosen': -58.14433670043945, 'logps/rejected': -94.90668487548828, 'logps/ref_chosen': -58.304595947265625, 'logps/ref_rejected': -91.69480895996094, 'logits/chosen': -0.5606456398963928, 'logits/rejected': -0.5093287229537964, 'epoch': 0.07} + 7%|███████████▉ | 50/681 [02:07<27:28, 2.61s/it] 7%|████████████▏ | 51/681 [02:09<27:19, 2.60s/it] 8%|████████████▎ | 52/681 [02:12<26:39, 2.54s/it] 8%|████████████▌ | 53/681 [02:14<26:35, 2.54s/it] 8%|████████████▊ | 54/681 [02:17<25:57, 2.48s/it] 8%|█████████████ | 55/681 [02:19<24:59, 2.40s/it] {'loss': 0.8026, 'grad_norm': 78.50489807128906, 'learning_rate': 3.9130434782608694e-07, 'fcm_dpo/beta': 0.1844942271709442, 'fcm_dpo/q_t': 0.30575454235076904, 'fcm_dpo/delta': -0.19296136498451233, 'fcm_dpo/margin': 5.56916618347168, 'margin_dpo/margin_mean': 5.5691657066345215, 'margin_dpo/margin_std': 5.670819282531738, 'logps/chosen': -56.305625915527344, 'logps/rejected': -91.50254821777344, 'logps/ref_chosen': -56.06591796875, 'logps/ref_rejected': -85.69367980957031, 'logits/chosen': -0.5810452699661255, 'logits/rejected': -0.5354865789413452, 'epoch': 0.08} + 8%|█████████████ | 55/681 [02:19<24:59, 2.40s/it] 8%|█████████████▎ | 56/681 [02:21<25:52, 2.48s/it] 8%|█████████████▌ | 57/681 [02:24<25:47, 2.48s/it] 9%|█████████████▊ | 58/681 [02:27<26:06, 2.51s/it] 9%|██████████████ | 59/681 [02:29<26:30, 2.56s/it] 9%|██████████████▎ | 60/681 [02:32<25:49, 2.49s/it] {'loss': 0.8231, 'grad_norm': 80.88507080078125, 'learning_rate': 4.2753623188405794e-07, 'fcm_dpo/beta': 0.14939136803150177, 'fcm_dpo/q_t': 0.3050878942012787, 'fcm_dpo/delta': -0.18733538687229156, 'fcm_dpo/margin': 6.829007148742676, 'margin_dpo/margin_mean': 6.829007148742676, 'margin_dpo/margin_std': 7.180797576904297, 'logps/chosen': -61.76525115966797, 'logps/rejected': -97.63427734375, 'logps/ref_chosen': -60.6871337890625, 'logps/ref_rejected': -89.72715759277344, 'logits/chosen': -0.5981510877609253, 'logits/rejected': -0.5482528805732727, 'epoch': 0.09} + 9%|██████████████▎ | 60/681 [02:32<25:49, 2.49s/it] 9%|██████████████▌ | 61/681 [02:34<26:03, 2.52s/it] 9%|██████████████▋ | 62/681 [02:37<26:32, 2.57s/it] 9%|██████████████▉ | 63/681 [02:39<26:19, 2.56s/it] 9%|███████████████▏ | 64/681 [02:42<25:52, 2.52s/it] 10%|███████████████▍ | 65/681 [02:44<25:58, 2.53s/it] {'loss': 0.8599, 'grad_norm': 60.72844696044922, 'learning_rate': 4.63768115942029e-07, 'fcm_dpo/beta': 0.1279543936252594, 'fcm_dpo/q_t': 0.31432515382766724, 'fcm_dpo/delta': -0.15479923784732819, 'fcm_dpo/margin': 7.741368293762207, 'margin_dpo/margin_mean': 7.741368293762207, 'margin_dpo/margin_std': 8.94554615020752, 'logps/chosen': -63.23627853393555, 'logps/rejected': -102.52547454833984, 'logps/ref_chosen': -61.75325393676758, 'logps/ref_rejected': -93.30108642578125, 'logits/chosen': -0.6157333850860596, 'logits/rejected': -0.5836382508277893, 'epoch': 0.1} + 10%|███████████████▍ | 65/681 [02:44<25:58, 2.53s/it] 10%|███████████████▋ | 66/681 [02:47<26:09, 2.55s/it] 10%|███████████████▉ | 67/681 [02:49<25:08, 2.46s/it] 10%|████████████████▏ | 68/681 [02:52<24:51, 2.43s/it] 10%|████████████████▍ | 69/681 [02:54<26:02, 2.55s/it] 10%|████████████████▋ | 70/681 [02:57<25:35, 2.51s/it] {'loss': 0.8231, 'grad_norm': 56.712982177734375, 'learning_rate': 5e-07, 'fcm_dpo/beta': 0.11097879707813263, 'fcm_dpo/q_t': 0.31369131803512573, 'fcm_dpo/delta': -0.13162511587142944, 'fcm_dpo/margin': 8.74174690246582, 'margin_dpo/margin_mean': 8.741745948791504, 'margin_dpo/margin_std': 9.22284984588623, 'logps/chosen': -61.81795120239258, 'logps/rejected': -95.02778625488281, 'logps/ref_chosen': -59.548004150390625, 'logps/ref_rejected': -84.01609802246094, 'logits/chosen': -0.6086438894271851, 'logits/rejected': -0.5690315961837769, 'epoch': 0.1} + 10%|████████████████▋ | 70/681 [02:57<25:35, 2.51s/it] 10%|████████████████▉ | 71/681 [02:59<25:55, 2.55s/it] 11%|█████████████████▏ | 72/681 [03:02<26:06, 2.57s/it] 11%|█████████████████▎ | 73/681 [03:05<26:01, 2.57s/it] 11%|█████████████████▌ | 74/681 [03:07<25:44, 2.54s/it] 11%|█████████████████▊ | 75/681 [03:10<25:55, 2.57s/it] {'loss': 0.8066, 'grad_norm': 43.50236129760742, 'learning_rate': 4.999176576834721e-07, 'fcm_dpo/beta': 0.09055890142917633, 'fcm_dpo/q_t': 0.2983909249305725, 'fcm_dpo/delta': -0.27352195978164673, 'fcm_dpo/margin': 12.153611183166504, 'margin_dpo/margin_mean': 12.153611183166504, 'margin_dpo/margin_std': 13.398017883300781, 'logps/chosen': -63.3004264831543, 'logps/rejected': -113.6408462524414, 'logps/ref_chosen': -59.86931228637695, 'logps/ref_rejected': -98.05613708496094, 'logits/chosen': -0.6517193913459778, 'logits/rejected': -0.6220844388008118, 'epoch': 0.11} + 11%|█████████████████▊ | 75/681 [03:10<25:55, 2.57s/it] 11%|██████████████████ | 76/681 [03:12<25:46, 2.56s/it] 11%|██████████████████▎ | 77/681 [03:14<24:41, 2.45s/it] 11%|██████████████████▌ | 78/681 [03:17<25:07, 2.50s/it] 12%|██████████████████▊ | 79/681 [03:20<25:22, 2.53s/it] 12%|███████████████████ | 80/681 [03:22<25:17, 2.52s/it] {'loss': 0.834, 'grad_norm': 41.86962890625, 'learning_rate': 4.996706849759452e-07, 'fcm_dpo/beta': 0.07161784917116165, 'fcm_dpo/q_t': 0.3114686608314514, 'fcm_dpo/delta': -0.15674835443496704, 'fcm_dpo/margin': 13.832998275756836, 'margin_dpo/margin_mean': 13.832998275756836, 'margin_dpo/margin_std': 14.778533935546875, 'logps/chosen': -60.72328567504883, 'logps/rejected': -104.79096984863281, 'logps/ref_chosen': -56.18925857543945, 'logps/ref_rejected': -86.42393493652344, 'logits/chosen': -0.6776489615440369, 'logits/rejected': -0.6389704942703247, 'epoch': 0.12} + 12%|███████████████████ | 80/681 [03:22<25:17, 2.52s/it] 12%|███████████████████▎ | 81/681 [03:25<25:58, 2.60s/it] 12%|███████████████████▌ | 82/681 [03:27<25:46, 2.58s/it] 12%|███████████████████▋ | 83/681 [03:30<25:04, 2.52s/it] 12%|███████████████████▉ | 84/681 [03:33<25:36, 2.57s/it] 12%|████████████████████▏ | 85/681 [03:35<25:35, 2.58s/it] {'loss': 0.8553, 'grad_norm': 42.009918212890625, 'learning_rate': 4.992592445678582e-07, 'fcm_dpo/beta': 0.06048274785280228, 'fcm_dpo/q_t': 0.3134708106517792, 'fcm_dpo/delta': -0.15613125264644623, 'fcm_dpo/margin': 16.303586959838867, 'margin_dpo/margin_mean': 16.303586959838867, 'margin_dpo/margin_std': 18.325000762939453, 'logps/chosen': -66.5967788696289, 'logps/rejected': -120.89393615722656, 'logps/ref_chosen': -60.018287658691406, 'logps/ref_rejected': -98.01185607910156, 'logits/chosen': -0.6713468432426453, 'logits/rejected': -0.6383522748947144, 'epoch': 0.12} + 12%|████████████████████▏ | 85/681 [03:35<25:35, 2.58s/it] 13%|████████████████████▍ | 86/681 [03:38<25:52, 2.61s/it] 13%|████████████████████▋ | 87/681 [03:40<25:38, 2.59s/it] 13%|████████████████████▉ | 88/681 [03:43<25:36, 2.59s/it] 13%|█████████████████████▏ | 89/681 [03:45<25:05, 2.54s/it] 13%|█████████████████████▍ | 90/681 [03:48<24:38, 2.50s/it] {'loss': 0.9273, 'grad_norm': 42.72884750366211, 'learning_rate': 4.986836074908615e-07, 'fcm_dpo/beta': 0.055021800100803375, 'fcm_dpo/q_t': 0.3316206634044647, 'fcm_dpo/delta': -0.09340845793485641, 'fcm_dpo/margin': 16.962265014648438, 'margin_dpo/margin_mean': 16.962265014648438, 'margin_dpo/margin_std': 22.208675384521484, 'logps/chosen': -69.03782653808594, 'logps/rejected': -122.91429138183594, 'logps/ref_chosen': -59.8709831237793, 'logps/ref_rejected': -96.78519439697266, 'logits/chosen': -0.6945492625236511, 'logits/rejected': -0.6738761067390442, 'epoch': 0.13} + 13%|█████████████████████▍ | 90/681 [03:48<24:38, 2.50s/it] 13%|█████████████████████▋ | 91/681 [03:50<24:48, 2.52s/it] 14%|█████████████████████▉ | 92/681 [03:53<24:29, 2.49s/it] 14%|██████████████████████ | 93/681 [03:55<23:40, 2.42s/it] 14%|██████████████████████▎ | 94/681 [03:58<24:50, 2.54s/it] 14%|██████████████████████▌ | 95/681 [04:00<24:25, 2.50s/it] {'loss': 0.8711, 'grad_norm': 34.471012115478516, 'learning_rate': 4.979441529392784e-07, 'fcm_dpo/beta': 0.049969423562288284, 'fcm_dpo/q_t': 0.3252207636833191, 'fcm_dpo/delta': -0.06735582649707794, 'fcm_dpo/margin': 18.21504783630371, 'margin_dpo/margin_mean': 18.215045928955078, 'margin_dpo/margin_std': 20.571401596069336, 'logps/chosen': -64.95755004882812, 'logps/rejected': -110.90779876708984, 'logps/ref_chosen': -55.94385528564453, 'logps/ref_rejected': -83.6790542602539, 'logits/chosen': -0.709299623966217, 'logits/rejected': -0.6731777787208557, 'epoch': 0.14} + 14%|██████████████████████▌ | 95/681 [04:00<24:25, 2.50s/it] 14%|██████████████████████▊ | 96/681 [04:03<24:23, 2.50s/it] 14%|███████████████████████ | 97/681 [04:05<24:27, 2.51s/it] 14%|███████████████████████▎ | 98/681 [04:08<24:00, 2.47s/it] 15%|███████████████████████▌ | 99/681 [04:10<23:05, 2.38s/it] 15%|███████████████████████▋ | 100/681 [04:12<23:37, 2.44s/it] {'loss': 0.8736, 'grad_norm': 40.7348518371582, 'learning_rate': 4.970413680203148e-07, 'fcm_dpo/beta': 0.04660476744174957, 'fcm_dpo/q_t': 0.32488057017326355, 'fcm_dpo/delta': -0.0532207265496254, 'fcm_dpo/margin': 19.219341278076172, 'margin_dpo/margin_mean': 19.219341278076172, 'margin_dpo/margin_std': 21.64118194580078, 'logps/chosen': -66.36493682861328, 'logps/rejected': -114.64266204833984, 'logps/ref_chosen': -57.05888748168945, 'logps/ref_rejected': -86.11727142333984, 'logits/chosen': -0.670303225517273, 'logits/rejected': -0.6368999481201172, 'epoch': 0.15} + 15%|███████████████████████▋ | 100/681 [04:12<23:37, 2.44s/it] 15%|███████████████████████▉ | 101/681 [04:15<23:23, 2.42s/it] 15%|████████████████████████ | 102/681 [04:17<23:15, 2.41s/it] 15%|████████████████████████▎ | 103/681 [04:20<22:55, 2.38s/it] 15%|████████████████████████▌ | 104/681 [04:22<22:20, 2.32s/it] 15%|████████████████████████▊ | 105/681 [04:24<23:23, 2.44s/it] {'loss': 0.8519, 'grad_norm': 30.820974349975586, 'learning_rate': 4.959758474331832e-07, 'fcm_dpo/beta': 0.04551302269101143, 'fcm_dpo/q_t': 0.315370112657547, 'fcm_dpo/delta': -0.15303857624530792, 'fcm_dpo/margin': 21.88186264038086, 'margin_dpo/margin_mean': 21.88186264038086, 'margin_dpo/margin_std': 24.000478744506836, 'logps/chosen': -70.07099914550781, 'logps/rejected': -119.24266052246094, 'logps/ref_chosen': -59.20774459838867, 'logps/ref_rejected': -86.49754333496094, 'logits/chosen': -0.7173506021499634, 'logits/rejected': -0.6785070896148682, 'epoch': 0.15} + 15%|████████████████████████▊ | 105/681 [04:24<23:23, 2.44s/it] 16%|█████████████████████████ | 106/681 [04:27<23:26, 2.45s/it] 16%|█████████████████████████▎ | 107/681 [04:29<23:52, 2.50s/it] 16%|█████████████████████████▌ | 108/681 [04:32<23:28, 2.46s/it] 16%|█████████████████████████▊ | 109/681 [04:34<23:59, 2.52s/it] 16%|██████████████████████████ | 110/681 [04:37<24:18, 2.55s/it] {'loss': 0.8568, 'grad_norm': 30.027286529541016, 'learning_rate': 4.947482930773511e-07, 'fcm_dpo/beta': 0.03847113624215126, 'fcm_dpo/q_t': 0.3201691508293152, 'fcm_dpo/delta': -0.08476176112890244, 'fcm_dpo/margin': 24.109981536865234, 'margin_dpo/margin_mean': 24.109981536865234, 'margin_dpo/margin_std': 25.9851016998291, 'logps/chosen': -72.03107452392578, 'logps/rejected': -126.54228210449219, 'logps/ref_chosen': -60.437957763671875, 'logps/ref_rejected': -90.83917999267578, 'logits/chosen': -0.6687729954719543, 'logits/rejected': -0.6251232624053955, 'epoch': 0.16} + 16%|██████████████████████████ | 110/681 [04:37<24:18, 2.55s/it] 16%|██████████████████████████▏ | 111/681 [04:40<24:04, 2.53s/it] 16%|██████████████████████████▍ | 112/681 [04:42<23:09, 2.44s/it] 17%|██████████████████████████▋ | 113/681 [04:44<23:36, 2.49s/it] 17%|██████████████████████████▉ | 114/681 [04:47<23:24, 2.48s/it] 17%|███████████████████████████▏ | 115/681 [04:50<23:47, 2.52s/it] {'loss': 0.903, 'grad_norm': 42.781307220458984, 'learning_rate': 4.933595135901732e-07, 'fcm_dpo/beta': 0.03624765947461128, 'fcm_dpo/q_t': 0.33425024151802063, 'fcm_dpo/delta': -0.01841454766690731, 'fcm_dpo/margin': 23.8579044342041, 'margin_dpo/margin_mean': 23.8579044342041, 'margin_dpo/margin_std': 28.585891723632812, 'logps/chosen': -76.1856918334961, 'logps/rejected': -123.62214660644531, 'logps/ref_chosen': -61.7908821105957, 'logps/ref_rejected': -85.36943054199219, 'logits/chosen': -0.6857893466949463, 'logits/rejected': -0.6449792385101318, 'epoch': 0.17} + 17%|███████████████████████████▏ | 115/681 [04:50<23:47, 2.52s/it] 17%|███████████████████████████▍ | 116/681 [04:52<23:12, 2.47s/it] 17%|███████████████████████████▋ | 117/681 [04:54<22:59, 2.45s/it] 17%|███████████████████████████▉ | 118/681 [04:57<23:25, 2.50s/it] 17%|████████████████████████████▏ | 119/681 [05:00<24:14, 2.59s/it] 18%|████████████████████████████▎ | 120/681 [05:02<24:04, 2.57s/it] {'loss': 0.8772, 'grad_norm': 33.90849304199219, 'learning_rate': 4.918104238142103e-07, 'fcm_dpo/beta': 0.0364445336163044, 'fcm_dpo/q_t': 0.33086472749710083, 'fcm_dpo/delta': -0.015237887389957905, 'fcm_dpo/margin': 23.689233779907227, 'margin_dpo/margin_mean': 23.68923568725586, 'margin_dpo/margin_std': 26.572423934936523, 'logps/chosen': -81.74562072753906, 'logps/rejected': -126.8638916015625, 'logps/ref_chosen': -65.3261489868164, 'logps/ref_rejected': -86.75518798828125, 'logits/chosen': -0.6950192451477051, 'logits/rejected': -0.6637409925460815, 'epoch': 0.18} + 18%|████████████████████████████▎ | 120/681 [05:02<24:04, 2.57s/it] 18%|████████████████████████████▌ | 121/681 [05:05<23:41, 2.54s/it] 18%|████████████████████████████▊ | 122/681 [05:07<22:55, 2.46s/it] 18%|█████████████████████████████ | 123/681 [05:10<23:31, 2.53s/it] 18%|█████████████████████████████▎ | 124/681 [05:12<23:31, 2.53s/it] 18%|█████████████████████████████▌ | 125/681 [05:15<23:14, 2.51s/it] {'loss': 0.8245, 'grad_norm': 28.747299194335938, 'learning_rate': 4.90102044194588e-07, 'fcm_dpo/beta': 0.03069928288459778, 'fcm_dpo/q_t': 0.3057743012905121, 'fcm_dpo/delta': -0.22570106387138367, 'fcm_dpo/margin': 34.196529388427734, 'margin_dpo/margin_mean': 34.196529388427734, 'margin_dpo/margin_std': 37.522029876708984, 'logps/chosen': -76.18916320800781, 'logps/rejected': -153.27313232421875, 'logps/ref_chosen': -58.323204040527344, 'logps/ref_rejected': -101.2106704711914, 'logits/chosen': -0.6552901268005371, 'logits/rejected': -0.645017147064209, 'epoch': 0.18} + 18%|█████████████████████████████▌ | 125/681 [05:15<23:14, 2.51s/it] 19%|█████████████████████████████▊ | 126/681 [05:17<23:36, 2.55s/it] 19%|██████████████████████████████ | 127/681 [05:20<23:45, 2.57s/it] 19%|██████████████████████████████▎ | 128/681 [05:23<23:57, 2.60s/it] 19%|██████████████████████████████▍ | 129/681 [05:25<23:41, 2.58s/it] 19%|██████████████████████████████▋ | 130/681 [05:28<23:25, 2.55s/it] {'loss': 0.9009, 'grad_norm': 29.514667510986328, 'learning_rate': 4.882355001067891e-07, 'fcm_dpo/beta': 0.02867107465863228, 'fcm_dpo/q_t': 0.33143970370292664, 'fcm_dpo/delta': -0.030061205849051476, 'fcm_dpo/margin': 30.552875518798828, 'margin_dpo/margin_mean': 30.552875518798828, 'margin_dpo/margin_std': 35.57330322265625, 'logps/chosen': -75.02465057373047, 'logps/rejected': -135.35000610351562, 'logps/ref_chosen': -56.38518524169922, 'logps/ref_rejected': -86.15767669677734, 'logits/chosen': -0.6441887617111206, 'logits/rejected': -0.6197634935379028, 'epoch': 0.19} + 19%|██████████████████████████████▋ | 130/681 [05:28<23:25, 2.55s/it] 19%|██████████████████████████████▉ | 131/681 [05:30<23:33, 2.57s/it] 19%|███████████████████████████████▏ | 132/681 [05:33<23:07, 2.53s/it] 20%|███████████████████████████████▍ | 133/681 [05:35<23:16, 2.55s/it] 20%|███████████████████████████████▋ | 134/681 [05:38<22:37, 2.48s/it] 20%|███████████████████████████████▉ | 135/681 [05:40<22:25, 2.46s/it] {'loss': 0.8502, 'grad_norm': 30.23385238647461, 'learning_rate': 4.862120211153265e-07, 'fcm_dpo/beta': 0.02630027011036873, 'fcm_dpo/q_t': 0.3213648200035095, 'fcm_dpo/delta': -0.08074788004159927, 'fcm_dpo/margin': 35.100746154785156, 'margin_dpo/margin_mean': 35.100746154785156, 'margin_dpo/margin_std': 38.36650848388672, 'logps/chosen': -74.02909851074219, 'logps/rejected': -149.7999725341797, 'logps/ref_chosen': -54.59065628051758, 'logps/ref_rejected': -95.26080322265625, 'logits/chosen': -0.6420992612838745, 'logits/rejected': -0.6323195695877075, 'epoch': 0.2} + 20%|███████████████████████████████▉ | 135/681 [05:40<22:25, 2.46s/it] 20%|████████████████████████████████▏ | 136/681 [05:43<22:58, 2.53s/it] 20%|████████████████████████████████▍ | 137/681 [05:45<22:54, 2.53s/it] 20%|████████████████████████████████▋ | 138/681 [05:47<22:14, 2.46s/it] 20%|████████████████████████████████▊ | 139/681 [05:50<21:54, 2.43s/it] 21%|█████████████████████████████████ | 140/681 [05:52<21:36, 2.40s/it] {'loss': 0.8815, 'grad_norm': 29.89729881286621, 'learning_rate': 4.840329401637809e-07, 'fcm_dpo/beta': 0.024506056681275368, 'fcm_dpo/q_t': 0.3280564844608307, 'fcm_dpo/delta': -0.07339946180582047, 'fcm_dpo/margin': 37.39527130126953, 'margin_dpo/margin_mean': 37.39527130126953, 'margin_dpo/margin_std': 43.980499267578125, 'logps/chosen': -81.1108169555664, 'logps/rejected': -155.74142456054688, 'logps/ref_chosen': -56.04347610473633, 'logps/ref_rejected': -93.27880859375, 'logits/chosen': -0.6269393563270569, 'logits/rejected': -0.6138561367988586, 'epoch': 0.21} + 21%|█████████████████████████████████ | 140/681 [05:52<21:36, 2.40s/it] 21%|█████████████████████████████████▎ | 141/681 [05:55<22:21, 2.48s/it] 21%|█████████████████████████████████▌ | 142/681 [05:58<23:00, 2.56s/it] 21%|█████████████████████████████████▊ | 143/681 [06:00<23:44, 2.65s/it] 21%|██████████████████████████████████ | 144/681 [06:03<23:44, 2.65s/it] 21%|██████████████████████████████████▎ | 145/681 [06:05<22:55, 2.57s/it] {'loss': 0.9115, 'grad_norm': 34.29755401611328, 'learning_rate': 4.816996926967401e-07, 'fcm_dpo/beta': 0.023285897448658943, 'fcm_dpo/q_t': 0.3340276777744293, 'fcm_dpo/delta': -0.010504422709345818, 'fcm_dpo/margin': 36.870033264160156, 'margin_dpo/margin_mean': 36.870033264160156, 'margin_dpo/margin_std': 44.208961486816406, 'logps/chosen': -89.64959716796875, 'logps/rejected': -151.40634155273438, 'logps/ref_chosen': -61.4414176940918, 'logps/ref_rejected': -86.32813262939453, 'logits/chosen': -0.5899958610534668, 'logits/rejected': -0.5646438598632812, 'epoch': 0.21} + 21%|██████████████████████████████████▎ | 145/681 [06:06<22:55, 2.57s/it] 21%|██████████████████████████████████▌ | 146/681 [06:08<22:45, 2.55s/it] 22%|██████████████████████████████████▊ | 147/681 [06:11<22:47, 2.56s/it] 22%|██████████████████████████████████▉ | 148/681 [06:13<22:42, 2.56s/it] 22%|███████████████████████████████████▏ | 149/681 [06:16<22:51, 2.58s/it] 22%|███████████████████████████████████▍ | 150/681 [06:18<22:42, 2.57s/it] {'loss': 0.9086, 'grad_norm': 31.706588745117188, 'learning_rate': 4.792138157142157e-07, 'fcm_dpo/beta': 0.02298033982515335, 'fcm_dpo/q_t': 0.33299776911735535, 'fcm_dpo/delta': -0.04273203760385513, 'fcm_dpo/margin': 38.58696746826172, 'margin_dpo/margin_mean': 38.58696746826172, 'margin_dpo/margin_std': 47.16357421875, 'logps/chosen': -89.40864562988281, 'logps/rejected': -158.06101989746094, 'logps/ref_chosen': -57.70451736450195, 'logps/ref_rejected': -87.76991271972656, 'logits/chosen': -0.5931132435798645, 'logits/rejected': -0.5684978365898132, 'epoch': 0.22} + 22%|███████████████████████████████████▍ | 150/681 [06:18<22:42, 2.57s/it] 22%|███████████████████████████████████▋ | 151/681 [06:21<22:07, 2.50s/it] 22%|███████████████████████████████████▉ | 152/681 [06:23<22:47, 2.58s/it] 22%|████████████████████████████████████▏ | 153/681 [06:26<22:37, 2.57s/it] 23%|████████████████████████████████████▍ | 154/681 [06:29<22:53, 2.61s/it] 23%|████████████████████████████████████▋ | 155/681 [06:31<23:28, 2.68s/it] {'loss': 0.8772, 'grad_norm': 28.768526077270508, 'learning_rate': 4.7657694675916247e-07, 'fcm_dpo/beta': 0.02123543620109558, 'fcm_dpo/q_t': 0.32444706559181213, 'fcm_dpo/delta': -0.0736246332526207, 'fcm_dpo/margin': 43.11958694458008, 'margin_dpo/margin_mean': 43.11958694458008, 'margin_dpo/margin_std': 49.604820251464844, 'logps/chosen': -90.71160888671875, 'logps/rejected': -166.54122924804688, 'logps/ref_chosen': -62.08925247192383, 'logps/ref_rejected': -94.79930114746094, 'logits/chosen': -0.6274883151054382, 'logits/rejected': -0.6081691980361938, 'epoch': 0.23} + 23%|████████████████████████████████████▋ | 155/681 [06:32<23:28, 2.68s/it] 23%|████████████████████████████████████▉ | 156/681 [06:34<23:12, 2.65s/it] 23%|█████████████████████████████████████ | 157/681 [06:36<22:08, 2.54s/it] 23%|█████████████████████████████████████▎ | 158/681 [06:39<22:13, 2.55s/it] 23%|█████████████████████████████████████▌ | 159/681 [06:42<22:22, 2.57s/it] 23%|█████████████████████████████████████▊ | 160/681 [06:44<22:04, 2.54s/it] {'loss': 0.95, 'grad_norm': 35.6558837890625, 'learning_rate': 4.737908228387656e-07, 'fcm_dpo/beta': 0.020709924399852753, 'fcm_dpo/q_t': 0.3394644856452942, 'fcm_dpo/delta': -0.028781410306692123, 'fcm_dpo/margin': 42.2672004699707, 'margin_dpo/margin_mean': 42.26719665527344, 'margin_dpo/margin_std': 56.8797607421875, 'logps/chosen': -102.9297866821289, 'logps/rejected': -174.969482421875, 'logps/ref_chosen': -67.15288543701172, 'logps/ref_rejected': -96.92537689208984, 'logits/chosen': -0.6209047436714172, 'logits/rejected': -0.602047324180603, 'epoch': 0.23} + 23%|█████████████████████████████████████▊ | 160/681 [06:44<22:04, 2.54s/it] 24%|██████████████████████████████████████ | 161/681 [06:46<21:10, 2.44s/it] 24%|██████████████████████████████████████▎ | 162/681 [06:49<21:54, 2.53s/it] 24%|██████████████████████████████████████▌ | 163/681 [06:51<21:47, 2.52s/it] 24%|██████████████████████████████████████▊ | 164/681 [06:54<21:53, 2.54s/it] 24%|███████████████████████████████████████ | 165/681 [06:57<21:45, 2.53s/it] {'loss': 0.9112, 'grad_norm': 44.44900894165039, 'learning_rate': 4.708572792802069e-07, 'fcm_dpo/beta': 0.019984444603323936, 'fcm_dpo/q_t': 0.33927860856056213, 'fcm_dpo/delta': 0.019107891246676445, 'fcm_dpo/margin': 41.58051300048828, 'margin_dpo/margin_mean': 41.58051681518555, 'margin_dpo/margin_std': 49.86974334716797, 'logps/chosen': -88.99661254882812, 'logps/rejected': -153.48812866210938, 'logps/ref_chosen': -57.40401077270508, 'logps/ref_rejected': -80.31498718261719, 'logits/chosen': -0.5925561189651489, 'logits/rejected': -0.5611370801925659, 'epoch': 0.24} + 24%|███████████████████████████████████████ | 165/681 [06:57<21:45, 2.53s/it] 24%|███████████████████████████████████████▏ | 166/681 [06:59<20:50, 2.43s/it] 25%|███████████████████████████████████████▍ | 167/681 [07:01<21:29, 2.51s/it] 25%|███████████████████████████████████████▋ | 168/681 [07:04<21:54, 2.56s/it] 25%|███████████████████████████████████████▉ | 169/681 [07:07<22:17, 2.61s/it] 25%|████████████████████████████████████████▏ | 170/681 [07:09<22:16, 2.61s/it] {'loss': 0.8528, 'grad_norm': 27.767526626586914, 'learning_rate': 4.6777824852166437e-07, 'fcm_dpo/beta': 0.018490394577383995, 'fcm_dpo/q_t': 0.3191189765930176, 'fcm_dpo/delta': -0.12243499606847763, 'fcm_dpo/margin': 51.66387939453125, 'margin_dpo/margin_mean': 51.66387939453125, 'margin_dpo/margin_std': 58.07769775390625, 'logps/chosen': -85.2691650390625, 'logps/rejected': -170.64334106445312, 'logps/ref_chosen': -52.029144287109375, 'logps/ref_rejected': -85.73944091796875, 'logits/chosen': -0.5405124425888062, 'logits/rejected': -0.528901994228363, 'epoch': 0.25} + 25%|████████████████████████████████████████▏ | 170/681 [07:10<22:16, 2.61s/it] 25%|████████████████████████████████████████▍ | 171/681 [07:12<21:24, 2.52s/it] 25%|████████████████████████████████████████▋ | 172/681 [07:14<21:09, 2.49s/it] 25%|████████████████████████████████████████▉ | 173/681 [07:17<20:57, 2.48s/it] 26%|█████████████████████████████████████████▏ | 174/681 [07:19<21:00, 2.49s/it] 26%|█████████████████████████████████████████▎ | 175/681 [07:22<21:14, 2.52s/it] {'loss': 0.8859, 'grad_norm': 37.51227569580078, 'learning_rate': 4.645557588393406e-07, 'fcm_dpo/beta': 0.01827237568795681, 'fcm_dpo/q_t': 0.32835322618484497, 'fcm_dpo/delta': -0.062149811536073685, 'fcm_dpo/margin': 49.67750549316406, 'margin_dpo/margin_mean': 49.67750549316406, 'margin_dpo/margin_std': 58.46363067626953, 'logps/chosen': -106.3938217163086, 'logps/rejected': -186.0583038330078, 'logps/ref_chosen': -62.996971130371094, 'logps/ref_rejected': -92.98394012451172, 'logits/chosen': -0.5422641038894653, 'logits/rejected': -0.5189716815948486, 'epoch': 0.26} + 26%|█████████████████████████████████████████▎ | 175/681 [07:22<21:14, 2.52s/it] 26%|█████████████████████████████████████████▌ | 176/681 [07:24<20:52, 2.48s/it] 26%|█████████████████████████████████████████▊ | 177/681 [07:27<21:06, 2.51s/it] 26%|██████████████████████████████████████████ | 178/681 [07:29<21:13, 2.53s/it] 26%|██████████████████████████████████████████▎ | 179/681 [07:32<21:33, 2.58s/it] 26%|██████████████████████████████████████████▌ | 180/681 [07:34<21:18, 2.55s/it] {'loss': 0.8569, 'grad_norm': 27.392812728881836, 'learning_rate': 4.611919330113591e-07, 'fcm_dpo/beta': 0.01582510769367218, 'fcm_dpo/q_t': 0.31790629029273987, 'fcm_dpo/delta': -0.1366010159254074, 'fcm_dpo/margin': 61.5815315246582, 'margin_dpo/margin_mean': 61.5815315246582, 'margin_dpo/margin_std': 70.3671646118164, 'logps/chosen': -105.83013916015625, 'logps/rejected': -207.4561767578125, 'logps/ref_chosen': -57.0670280456543, 'logps/ref_rejected': -97.1115493774414, 'logits/chosen': -0.485221803188324, 'logits/rejected': -0.48061603307724, 'epoch': 0.26} + 26%|██████████████████████████████████████████▌ | 180/681 [07:35<21:18, 2.55s/it] 27%|██████████████████████████████████████████▊ | 181/681 [07:37<21:18, 2.56s/it] 27%|███████████████████████████████████████████ | 182/681 [07:39<20:58, 2.52s/it] 27%|███████████████████████████████████████████▎ | 183/681 [07:42<20:21, 2.45s/it] 27%|███████████████████████████████████████████▌ | 184/681 [07:44<20:54, 2.52s/it] 27%|███████████████████████████████████████████▋ | 185/681 [07:47<20:36, 2.49s/it] {'loss': 0.9783, 'grad_norm': 36.289039611816406, 'learning_rate': 4.5768898691940836e-07, 'fcm_dpo/beta': 0.015623442828655243, 'fcm_dpo/q_t': 0.3591853976249695, 'fcm_dpo/delta': 0.13391907513141632, 'fcm_dpo/margin': 46.54853057861328, 'margin_dpo/margin_mean': 46.54853057861328, 'margin_dpo/margin_std': 60.472984313964844, 'logps/chosen': -97.86585998535156, 'logps/rejected': -165.0836944580078, 'logps/ref_chosen': -54.840736389160156, 'logps/ref_rejected': -75.51002502441406, 'logits/chosen': -0.5313425064086914, 'logits/rejected': -0.5060660243034363, 'epoch': 0.27} + 27%|███████████████████████████████████████████▋ | 185/681 [07:47<20:36, 2.49s/it] 27%|███████████████████████████████████████████▉ | 186/681 [07:49<20:32, 2.49s/it] 27%|████████████████████████████████████████████▏ | 187/681 [07:52<19:58, 2.43s/it] 28%|████████████████████████████████████████████▍ | 188/681 [07:54<20:10, 2.46s/it] 28%|████████████████████████████████████████████▋ | 189/681 [07:57<20:29, 2.50s/it] 28%|████████████████████████████████████████████▉ | 190/681 [07:59<20:03, 2.45s/it] {'loss': 0.8786, 'grad_norm': 40.74492645263672, 'learning_rate': 4.5404922808905543e-07, 'fcm_dpo/beta': 0.01625588908791542, 'fcm_dpo/q_t': 0.3247375786304474, 'fcm_dpo/delta': -0.08447099477052689, 'fcm_dpo/margin': 56.91190719604492, 'margin_dpo/margin_mean': 56.91190719604492, 'margin_dpo/margin_std': 66.29203033447266, 'logps/chosen': -102.94236755371094, 'logps/rejected': -188.99276733398438, 'logps/ref_chosen': -57.72148895263672, 'logps/ref_rejected': -86.85997009277344, 'logits/chosen': -0.5222185254096985, 'logits/rejected': -0.49877819418907166, 'epoch': 0.28} + 28%|████████████████████████████████████████████▉ | 190/681 [07:59<20:03, 2.45s/it] 28%|█████████████████████████████████████████████▏ | 191/681 [08:02<20:41, 2.53s/it] 28%|█████████████████████████████████████████████▍ | 192/681 [08:04<20:25, 2.51s/it] 28%|█████████████████████████████████████████████▋ | 193/681 [08:07<20:01, 2.46s/it] 28%|█████████████████████████████████████████████▊ | 194/681 [08:09<19:24, 2.39s/it] 29%|██████████████████████████████████████████████ | 195/681 [08:11<19:52, 2.45s/it] {'loss': 0.8883, 'grad_norm': 33.52628707885742, 'learning_rate': 4.5027505416968985e-07, 'fcm_dpo/beta': 0.014897446148097515, 'fcm_dpo/q_t': 0.32934439182281494, 'fcm_dpo/delta': -0.034334540367126465, 'fcm_dpo/margin': 58.907752990722656, 'margin_dpo/margin_mean': 58.907752990722656, 'margin_dpo/margin_std': 67.69029235839844, 'logps/chosen': -110.0854263305664, 'logps/rejected': -200.19639587402344, 'logps/ref_chosen': -58.26164627075195, 'logps/ref_rejected': -89.46485900878906, 'logits/chosen': -0.4740094542503357, 'logits/rejected': -0.4610044062137604, 'epoch': 0.29} + 29%|██████████████████████████████████████████████ | 195/681 [08:12<19:52, 2.45s/it] 29%|██████████████████████████████████████████████▎ | 196/681 [08:14<19:58, 2.47s/it] 29%|██████████████████████████████████████████████▌ | 197/681 [08:17<20:09, 2.50s/it] 29%|██████████████████████████████████████████████▊ | 198/681 [08:19<20:15, 2.52s/it] 29%|███████████████████████████████████████████████ | 199/681 [08:22<20:37, 2.57s/it] 29%|███████████████████████████████████████████████▎ | 200/681 [08:24<20:33, 2.56s/it] {'loss': 0.8697, 'grad_norm': 31.517562866210938, 'learning_rate': 4.4636895135509966e-07, 'fcm_dpo/beta': 0.014423737302422523, 'fcm_dpo/q_t': 0.3252788186073303, 'fcm_dpo/delta': -0.07692910730838776, 'fcm_dpo/margin': 63.713844299316406, 'margin_dpo/margin_mean': 63.713844299316406, 'margin_dpo/margin_std': 73.30004119873047, 'logps/chosen': -105.78775787353516, 'logps/rejected': -196.93441772460938, 'logps/ref_chosen': -55.71953201293945, 'logps/ref_rejected': -83.15235137939453, 'logits/chosen': -0.46841588616371155, 'logits/rejected': -0.4528680443763733, 'epoch': 0.29} + 29%|███████████████████████████████████████████████▎ | 200/681 [08:24<20:33, 2.56s/it][INFO|trainer.py:4307] 2026-04-22 00:09:22,146 >> +***** Running Evaluation ***** +[INFO|trainer.py:4309] 2026-04-22 00:09:22,146 >> Num examples = 2339 +[INFO|trainer.py:4312] 2026-04-22 00:09:22,146 >> Batch size = 8 + + 0%| | 0/73 [00:00) + 33%|████████████████████████████████████████████████████▋ | 223/681 [10:02<18:34, 2.43s/it] 33%|████████████████████████████████████████████████████▉ | 224/681 [10:04<18:16, 2.40s/it] 33%|█████████████████████████████████████████████████████▏ | 225/681 [10:06<17:53, 2.35s/it] {'loss': 0.8406, 'grad_norm': 35.1022834777832, 'learning_rate': 4.249525076191759e-07, 'fcm_dpo/beta': 0.012188979424536228, 'fcm_dpo/q_t': 0.30945101380348206, 'fcm_dpo/delta': -0.16104714572429657, 'fcm_dpo/margin': 81.43888854980469, 'margin_dpo/margin_mean': 81.43888854980469, 'margin_dpo/margin_std': 86.77934265136719, 'logps/chosen': -134.11956787109375, 'logps/rejected': -254.3486785888672, 'logps/ref_chosen': -54.47245407104492, 'logps/ref_rejected': -93.26266479492188, 'logits/chosen': -0.40194278955459595, 'logits/rejected': -0.3876434564590454, 'epoch': 0.33} + 33%|█████████████████████████████████████████████████████▏ | 225/681 [10:06<17:53, 2.35s/it] 33%|█████████████████████████████████████████████████████▍ | 226/681 [10:09<18:43, 2.47s/it] 33%|█████████████████████████████████████████████████████▋ | 227/681 [10:11<18:19, 2.42s/it] 33%|█████████████████████████████████████████████████████▉ | 228/681 [10:14<18:41, 2.48s/it] 34%|██████████████████████████████████████████████████████▏ | 229/681 [10:17<18:31, 2.46s/it] 34%|██████████████████████████████████████████████████████▍ | 230/681 [10:19<18:13, 2.42s/it] {'loss': 0.865, 'grad_norm': 34.77726364135742, 'learning_rate': 4.203117865141635e-07, 'fcm_dpo/beta': 0.01108027994632721, 'fcm_dpo/q_t': 0.3215538561344147, 'fcm_dpo/delta': -0.06726758182048798, 'fcm_dpo/margin': 82.1200942993164, 'margin_dpo/margin_mean': 82.1200942993164, 'margin_dpo/margin_std': 89.67250061035156, 'logps/chosen': -139.94625854492188, 'logps/rejected': -251.1558380126953, 'logps/ref_chosen': -58.7701301574707, 'logps/ref_rejected': -87.85963439941406, 'logits/chosen': -0.3820115923881531, 'logits/rejected': -0.35805052518844604, 'epoch': 0.34} + 34%|██████████████████████████████████████████████████████▍ | 230/681 [10:19<18:13, 2.42s/it] 34%|██████████████████████████████████████████████████████▌ | 231/681 [10:21<18:33, 2.47s/it] 34%|██████████████████████████████████████████████████████▊ | 232/681 [10:24<19:04, 2.55s/it] 34%|███████████████████████████████████████████████████████ | 233/681 [10:27<19:22, 2.60s/it] 34%|███████████████████████████████████████████████████████▎ | 234/681 [10:30<19:26, 2.61s/it] 35%|███████████████████████████████████████████████████████▌ | 235/681 [10:32<19:10, 2.58s/it] {'loss': 0.9162, 'grad_norm': 32.15520095825195, 'learning_rate': 4.1555887447288255e-07, 'fcm_dpo/beta': 0.010919634252786636, 'fcm_dpo/q_t': 0.34256863594055176, 'fcm_dpo/delta': 0.039671458303928375, 'fcm_dpo/margin': 74.35450744628906, 'margin_dpo/margin_mean': 74.35450744628906, 'margin_dpo/margin_std': 88.09162902832031, 'logps/chosen': -133.8794403076172, 'logps/rejected': -240.3823699951172, 'logps/ref_chosen': -59.0481071472168, 'logps/ref_rejected': -91.19654846191406, 'logits/chosen': -0.4005228877067566, 'logits/rejected': -0.3855542838573456, 'epoch': 0.35} + 35%|███████████████████████████████████████████████████████▌ | 235/681 [10:32<19:10, 2.58s/it] 35%|███████████████████████████████████████████████████████▊ | 236/681 [10:35<19:34, 2.64s/it] 35%|████████████████████████████████████████████████████████ | 237/681 [10:37<19:25, 2.62s/it] 35%|████████████████████████████████████████████████████████▎ | 238/681 [10:40<19:30, 2.64s/it] 35%|████████████████████████████████████████████████████████▌ | 239/681 [10:42<18:51, 2.56s/it] 35%|████████████████████████████████████████████████████████▋ | 240/681 [10:45<18:46, 2.55s/it] {'loss': 0.8677, 'grad_norm': 40.340721130371094, 'learning_rate': 4.106969024216348e-07, 'fcm_dpo/beta': 0.010825184173882008, 'fcm_dpo/q_t': 0.3274964690208435, 'fcm_dpo/delta': -0.018980540335178375, 'fcm_dpo/margin': 79.97639465332031, 'margin_dpo/margin_mean': 79.97639465332031, 'margin_dpo/margin_std': 84.8770523071289, 'logps/chosen': -138.43865966796875, 'logps/rejected': -254.26034545898438, 'logps/ref_chosen': -55.238983154296875, 'logps/ref_rejected': -91.08428955078125, 'logits/chosen': -0.3556009829044342, 'logits/rejected': -0.34894007444381714, 'epoch': 0.35} + 35%|████████████████████████████████████████████████████████▋ | 240/681 [10:45<18:46, 2.55s/it] 35%|████████████████████████████████████████████████████████▉ | 241/681 [10:47<18:25, 2.51s/it] 36%|█████████████████████████████████████████████████████████▏ | 242/681 [10:50<18:09, 2.48s/it] 36%|█████████████████████████████████████████████████████████▍ | 243/681 [10:52<18:05, 2.48s/it] 36%|█████████████████████████████████████████████████████████▋ | 244/681 [10:55<18:06, 2.49s/it] 36%|█████████████████████████████████████████████████████████▉ | 245/681 [10:57<18:31, 2.55s/it] {'loss': 0.8333, 'grad_norm': 42.48567581176758, 'learning_rate': 4.057290731287531e-07, 'fcm_dpo/beta': 0.010110095143318176, 'fcm_dpo/q_t': 0.3163699507713318, 'fcm_dpo/delta': -0.10127156972885132, 'fcm_dpo/margin': 92.84043884277344, 'margin_dpo/margin_mean': 92.8404312133789, 'margin_dpo/margin_std': 95.6347885131836, 'logps/chosen': -158.22030639648438, 'logps/rejected': -272.03009033203125, 'logps/ref_chosen': -65.08844757080078, 'logps/ref_rejected': -86.05777740478516, 'logits/chosen': -0.35475802421569824, 'logits/rejected': -0.3244848847389221, 'epoch': 0.36} + 36%|█████████████████████████████████████████████████████████▉ | 245/681 [10:58<18:31, 2.55s/it] 36%|██████████████████████████████████████████████████████████▏ | 246/681 [11:00<18:41, 2.58s/it] 36%|██████████████████████████████████████████████████████████▍ | 247/681 [11:03<18:23, 2.54s/it] 36%|██████████████████████████████████████████████████████████▋ | 248/681 [11:05<18:30, 2.56s/it] 37%|██████████████████████████████████████████████████████████▊ | 249/681 [11:08<17:59, 2.50s/it] 37%|███████████████████████████████████████████████████████████ | 250/681 [11:10<17:52, 2.49s/it] {'loss': 0.8678, 'grad_norm': 54.82107162475586, 'learning_rate': 4.006586590948141e-07, 'fcm_dpo/beta': 0.00962417759001255, 'fcm_dpo/q_t': 0.3266645073890686, 'fcm_dpo/delta': -0.020193621516227722, 'fcm_dpo/margin': 90.06238555908203, 'margin_dpo/margin_mean': 90.06238555908203, 'margin_dpo/margin_std': 93.82579040527344, 'logps/chosen': -154.8458709716797, 'logps/rejected': -273.19061279296875, 'logps/ref_chosen': -59.08491897583008, 'logps/ref_rejected': -87.36727142333984, 'logits/chosen': -0.37928658723831177, 'logits/rejected': -0.35434064269065857, 'epoch': 0.37} + 37%|███████████████████████████████████████████████████████████ | 250/681 [11:10<17:52, 2.49s/it] 37%|███████████████████████████████████████████████████████████▎ | 251/681 [11:12<17:38, 2.46s/it] 37%|███████████████████████████████████████████████████████████▌ | 252/681 [11:15<17:40, 2.47s/it] 37%|███████████████████████████████████████████████████████████▊ | 253/681 [11:18<17:55, 2.51s/it] 37%|████████████████████████████████████████████████████████████ | 254/681 [11:20<18:07, 2.55s/it] 37%|████████████████████████████████████████████████████████████▎ | 255/681 [11:23<17:45, 2.50s/it] {'loss': 0.9102, 'grad_norm': 58.291961669921875, 'learning_rate': 3.954890003969163e-07, 'fcm_dpo/beta': 0.009607134386897087, 'fcm_dpo/q_t': 0.3329293131828308, 'fcm_dpo/delta': 0.0022653634659945965, 'fcm_dpo/margin': 88.13050842285156, 'margin_dpo/margin_mean': 88.13050842285156, 'margin_dpo/margin_std': 103.13516998291016, 'logps/chosen': -170.76805114746094, 'logps/rejected': -285.3667907714844, 'logps/ref_chosen': -61.85979461669922, 'logps/ref_rejected': -88.32804107666016, 'logits/chosen': -0.33103686571121216, 'logits/rejected': -0.3057294487953186, 'epoch': 0.37} + 37%|████████████████████████████████████████████████████████████▎ | 255/681 [11:23<17:45, 2.50s/it] 38%|████████████████████████████████████████████████████████████▌ | 256/681 [11:25<17:43, 2.50s/it] 38%|████████████████████████████████████████████████████████████▊ | 257/681 [11:28<17:56, 2.54s/it] 38%|████████████████████████████████████████████████████████████▉ | 258/681 [11:30<17:29, 2.48s/it] 38%|█████████████████████████████████████████████████████████████▏ | 259/681 [11:33<17:27, 2.48s/it] 38%|█████████████████████████████████████████████████████████████▍ | 260/681 [11:35<17:44, 2.53s/it] {'loss': 0.8885, 'grad_norm': 39.26882553100586, 'learning_rate': 3.9022350248844246e-07, 'fcm_dpo/beta': 0.009414169006049633, 'fcm_dpo/q_t': 0.33001503348350525, 'fcm_dpo/delta': -0.016010303050279617, 'fcm_dpo/margin': 91.66337585449219, 'margin_dpo/margin_mean': 91.66337585449219, 'margin_dpo/margin_std': 103.34080505371094, 'logps/chosen': -154.59693908691406, 'logps/rejected': -283.89129638671875, 'logps/ref_chosen': -52.843467712402344, 'logps/ref_rejected': -90.4744873046875, 'logits/chosen': -0.31099259853363037, 'logits/rejected': -0.3064902424812317, 'epoch': 0.38} + 38%|█████████████████████████████████████████████████████████████▍ | 260/681 [11:35<17:44, 2.53s/it] 38%|█████████████████████████████████████████████████████████████▋ | 261/681 [11:38<17:31, 2.50s/it] 38%|█████████████████████████████████████████████████████████████▉ | 262/681 [11:40<17:07, 2.45s/it] 39%|██████████████████████████████████████████████████████████████▏ | 263/681 [11:42<17:00, 2.44s/it] 39%|██████████████████████████████████████████████████████████████▍ | 264/681 [11:45<17:30, 2.52s/it] 39%|██████████████████████████████████████████████████████████████▋ | 265/681 [11:47<17:09, 2.47s/it] {'loss': 0.8593, 'grad_norm': 42.47758865356445, 'learning_rate': 3.848656339557562e-07, 'fcm_dpo/beta': 0.009381098672747612, 'fcm_dpo/q_t': 0.3260762095451355, 'fcm_dpo/delta': -0.015523952431976795, 'fcm_dpo/margin': 92.02482604980469, 'margin_dpo/margin_mean': 92.02482604980469, 'margin_dpo/margin_std': 95.39722442626953, 'logps/chosen': -151.35202026367188, 'logps/rejected': -274.75714111328125, 'logps/ref_chosen': -59.35320281982422, 'logps/ref_rejected': -90.73350524902344, 'logits/chosen': -0.299502432346344, 'logits/rejected': -0.27807632088661194, 'epoch': 0.39} + 39%|██████████████████████████████████████████████████████████████▋ | 265/681 [11:47<17:09, 2.47s/it] 39%|██████████████████████████████████████████████████████████████▉ | 266/681 [11:50<17:03, 2.47s/it] 39%|███████████████████████████████████████████████████████████████ | 267/681 [11:52<17:09, 2.49s/it] 39%|███████████████████████████████████████████████████████████████▎ | 268/681 [11:55<17:07, 2.49s/it] 40%|███████████████████████████████████████████████████████████████▌ | 269/681 [11:57<16:57, 2.47s/it] 40%|███████████████████████████████████████████████████████████████▊ | 270/681 [12:00<17:16, 2.52s/it] {'loss': 0.8569, 'grad_norm': 51.16044616699219, 'learning_rate': 3.794189242333106e-07, 'fcm_dpo/beta': 0.009062977507710457, 'fcm_dpo/q_t': 0.32098034024238586, 'fcm_dpo/delta': -0.06297405064105988, 'fcm_dpo/margin': 100.06938171386719, 'margin_dpo/margin_mean': 100.06938171386719, 'margin_dpo/margin_std': 107.111083984375, 'logps/chosen': -184.3603057861328, 'logps/rejected': -313.5339660644531, 'logps/ref_chosen': -66.30875396728516, 'logps/ref_rejected': -95.4130630493164, 'logits/chosen': -0.23682311177253723, 'logits/rejected': -0.21265558898448944, 'epoch': 0.4} + 40%|███████████████████████████████████████████████████████████████▊ | 270/681 [12:00<17:16, 2.52s/it] 40%|████████████████████████████████████████████████████████████████ | 271/681 [12:02<17:13, 2.52s/it] 40%|████████████████████████████████████████████████████████████████▎ | 272/681 [12:05<17:28, 2.56s/it] 40%|████████████████████████████████████████████████████████████████▌ | 273/681 [12:08<17:29, 2.57s/it] 40%|████████████████████████████████████████████████████████████████▊ | 274/681 [12:10<17:00, 2.51s/it] 40%|█████████████████████████████████████████████████████████████████ | 275/681 [12:13<17:14, 2.55s/it] {'loss': 0.8306, 'grad_norm': 33.570701599121094, 'learning_rate': 3.738869612786737e-07, 'fcm_dpo/beta': 0.008018719032406807, 'fcm_dpo/q_t': 0.3118273615837097, 'fcm_dpo/delta': -0.12382125854492188, 'fcm_dpo/margin': 119.62171936035156, 'margin_dpo/margin_mean': 119.6217269897461, 'margin_dpo/margin_std': 122.45941162109375, 'logps/chosen': -189.1214599609375, 'logps/rejected': -346.2816467285156, 'logps/ref_chosen': -54.69990921020508, 'logps/ref_rejected': -92.23838806152344, 'logits/chosen': -0.15879306197166443, 'logits/rejected': -0.14661014080047607, 'epoch': 0.4} + 40%|█████████████████████████████████████████████████████████████████ | 275/681 [12:13<17:14, 2.55s/it] 41%|█████████████████████████████████████████████████████████████████▎ | 276/681 [12:15<16:59, 2.52s/it] 41%|█████████████████████████████████████████████████████████████████▍ | 277/681 [12:17<16:21, 2.43s/it] 41%|█████████████████████████████████████████████████████████████████▋ | 278/681 [12:20<16:28, 2.45s/it] 41%|█████████████████████████████████████████████████████████████████▉ | 279/681 [12:22<16:32, 2.47s/it] 41%|██████████████████████████████████████████████████████████████████▏ | 280/681 [12:25<16:44, 2.51s/it] {'loss': 0.8467, 'grad_norm': 67.1026840209961, 'learning_rate': 3.6827338920900253e-07, 'fcm_dpo/beta': 0.007537333760410547, 'fcm_dpo/q_t': 0.3190959692001343, 'fcm_dpo/delta': -0.07704190909862518, 'fcm_dpo/margin': 122.15445709228516, 'margin_dpo/margin_mean': 122.15445709228516, 'margin_dpo/margin_std': 127.57193756103516, 'logps/chosen': -187.32925415039062, 'logps/rejected': -343.0320129394531, 'logps/ref_chosen': -54.64586639404297, 'logps/ref_rejected': -88.19416809082031, 'logits/chosen': -0.15290072560310364, 'logits/rejected': -0.14313052594661713, 'epoch': 0.41} + 41%|██████████████████████████████████████████████████████████████████▏ | 280/681 [12:25<16:44, 2.51s/it] 41%|██████████████████████████████████████████████████████████████████▍ | 281/681 [12:28<16:45, 2.51s/it] 41%|██████████████████████████████████████████████████████████████████▋ | 282/681 [12:30<16:40, 2.51s/it] 42%|██████████████████████████████████████████████████████████████████▉ | 283/681 [12:33<16:52, 2.54s/it] 42%|███████████████████████████████████████████████████████████████████▏ | 284/681 [12:35<17:10, 2.60s/it] 42%|███████████████████████████████████████████████████████████████████▍ | 285/681 [12:38<16:59, 2.58s/it] {'loss': 0.8457, 'grad_norm': 36.61220169067383, 'learning_rate': 3.625819059005228e-07, 'fcm_dpo/beta': 0.007103285752236843, 'fcm_dpo/q_t': 0.32029491662979126, 'fcm_dpo/delta': -0.03963179141283035, 'fcm_dpo/margin': 124.67244720458984, 'margin_dpo/margin_mean': 124.67244720458984, 'margin_dpo/margin_std': 126.2774429321289, 'logps/chosen': -234.484375, 'logps/rejected': -389.29510498046875, 'logps/ref_chosen': -63.02496337890625, 'logps/ref_rejected': -93.16323852539062, 'logits/chosen': -0.09207823872566223, 'logits/rejected': -0.07269109785556793, 'epoch': 0.42} + 42%|███████████████████████████████████████████████████████████████████▍ | 285/681 [12:38<16:59, 2.58s/it] 42%|███████████████████████████████████████████████████████████████████▌ | 286/681 [12:41<17:19, 2.63s/it] 42%|███████████████████████████████████████████████████████████████████▊ | 287/681 [12:43<16:28, 2.51s/it] 42%|████████████████████████████████████████████████████████████████████ | 288/681 [12:45<16:27, 2.51s/it] 42%|████████████████████████████████████████████████████████████████████▎ | 289/681 [12:48<16:19, 2.50s/it] 43%|████████████████████████████████████████████████████████████████████▌ | 290/681 [12:50<16:01, 2.46s/it] {'loss': 0.8386, 'grad_norm': 32.6433219909668, 'learning_rate': 3.568162605525952e-07, 'fcm_dpo/beta': 0.006760433316230774, 'fcm_dpo/q_t': 0.3183870315551758, 'fcm_dpo/delta': -0.0626429095864296, 'fcm_dpo/margin': 134.1584930419922, 'margin_dpo/margin_mean': 134.1584930419922, 'margin_dpo/margin_std': 134.99864196777344, 'logps/chosen': -225.1572723388672, 'logps/rejected': -392.53900146484375, 'logps/ref_chosen': -58.37105178833008, 'logps/ref_rejected': -91.59428405761719, 'logits/chosen': -0.07117484509944916, 'logits/rejected': -0.06140874698758125, 'epoch': 0.43} + 43%|████████████████████████████████████████████████████████████████████▌ | 290/681 [12:50<16:01, 2.46s/it] 43%|████████████████████████████████████████████████████████████████████▊ | 291/681 [12:53<16:10, 2.49s/it] 43%|█████████████████████████████████████████████████████████████████████ | 292/681 [12:55<15:52, 2.45s/it] 43%|█████████████████████████████████████████████████████████████████████▎ | 293/681 [12:58<16:08, 2.50s/it] 43%|█████████████████████████████████████████████████████████████████████▌ | 294/681 [13:00<15:52, 2.46s/it] 43%|█████████████████████████████████████████████████████████████████████▋ | 295/681 [13:03<15:43, 2.44s/it] {'loss': 0.8958, 'grad_norm': 46.24834060668945, 'learning_rate': 3.509802512179737e-07, 'fcm_dpo/beta': 0.006551184691488743, 'fcm_dpo/q_t': 0.3340519070625305, 'fcm_dpo/delta': 0.024965018033981323, 'fcm_dpo/margin': 126.14031982421875, 'margin_dpo/margin_mean': 126.14030456542969, 'margin_dpo/margin_std': 140.62646484375, 'logps/chosen': -215.8157958984375, 'logps/rejected': -372.1355285644531, 'logps/ref_chosen': -55.113426208496094, 'logps/ref_rejected': -85.29283905029297, 'logits/chosen': -0.051733147352933884, 'logits/rejected': -0.04382902756333351, 'epoch': 0.43} + 43%|█████████████████████████████████████████████████████████████████████▋ | 295/681 [13:03<15:43, 2.44s/it] 43%|█████████████████████████████████████████████████████████████████████▉ | 296/681 [13:05<15:50, 2.47s/it] 44%|██████████████████████████████████████████████████████████████████████▏ | 297/681 [13:08<15:57, 2.49s/it] 44%|██████████████████████████████████████████████████████████████████████▍ | 298/681 [13:10<16:30, 2.59s/it] 44%|██████████████████████████████████████████████████████████████████████▋ | 299/681 [13:13<16:19, 2.56s/it] 44%|██████████████████████████████████████████████████████████████████████▉ | 300/681 [13:16<16:23, 2.58s/it] {'loss': 0.9105, 'grad_norm': 67.42388916015625, 'learning_rate': 3.4507772230088147e-07, 'fcm_dpo/beta': 0.006881644017994404, 'fcm_dpo/q_t': 0.3347456455230713, 'fcm_dpo/delta': -0.0010080442298203707, 'fcm_dpo/margin': 123.4745101928711, 'margin_dpo/margin_mean': 123.4745101928711, 'margin_dpo/margin_std': 146.905029296875, 'logps/chosen': -241.1220703125, 'logps/rejected': -402.16766357421875, 'logps/ref_chosen': -59.46582794189453, 'logps/ref_rejected': -97.03690338134766, 'logits/chosen': -0.06820065528154373, 'logits/rejected': -0.0680488869547844, 'epoch': 0.44} + 44%|██████████████████████████████████████████████████████████████████████▉ | 300/681 [13:16<16:23, 2.58s/it] 44%|███████████████████████████████████████████████████████████████████████▏ | 301/681 [13:18<15:41, 2.48s/it] 44%|███████████████████████████████████████████████████████████████████████▍ | 302/681 [13:20<15:51, 2.51s/it] 44%|███████████████████████████████████████████████████████████████████████▋ | 303/681 [13:23<15:58, 2.54s/it] 45%|███████████████████████████████████████████████████████████████████████▊ | 304/681 [13:26<16:11, 2.58s/it] 45%|████████████████████████████████████████████████████████████████████████ | 305/681 [13:28<16:13, 2.59s/it] {'loss': 0.8706, 'grad_norm': 42.539947509765625, 'learning_rate': 3.391125620245535e-07, 'fcm_dpo/beta': 0.006595266051590443, 'fcm_dpo/q_t': 0.33005213737487793, 'fcm_dpo/delta': 0.0008342728251591325, 'fcm_dpo/margin': 128.62730407714844, 'margin_dpo/margin_mean': 128.62728881835938, 'margin_dpo/margin_std': 136.1685333251953, 'logps/chosen': -225.65005493164062, 'logps/rejected': -383.4462890625, 'logps/ref_chosen': -62.78144454956055, 'logps/ref_rejected': -91.95039367675781, 'logits/chosen': -0.09431194514036179, 'logits/rejected': -0.08120522648096085, 'epoch': 0.45} + 45%|████████████████████████████████████████████████████████████████████████ | 305/681 [13:28<16:13, 2.59s/it] 45%|████████████████████████████████████████████████████████████████████████▎ | 306/681 [13:31<16:14, 2.60s/it] 45%|████████████████████████████████████████████████████████████████████████▌ | 307/681 [13:34<16:28, 2.64s/it] 45%|████████████████████████████████████████████████████████████████████████▊ | 308/681 [13:36<16:27, 2.65s/it] 45%|█████████████████████████████████████████████████████████████████████████ | 309/681 [13:39<16:23, 2.64s/it] 46%|█████████████████████████████████████████████████████████████████████████▎ | 310/681 [13:42<16:30, 2.67s/it] {'loss': 0.851, 'grad_norm': 40.4238395690918, 'learning_rate': 3.3308869986991487e-07, 'fcm_dpo/beta': 0.006700504571199417, 'fcm_dpo/q_t': 0.3245096802711487, 'fcm_dpo/delta': -0.0035556063521653414, 'fcm_dpo/margin': 127.2578125, 'margin_dpo/margin_mean': 127.2578125, 'margin_dpo/margin_std': 124.93436431884766, 'logps/chosen': -217.1204071044922, 'logps/rejected': -365.7741394042969, 'logps/ref_chosen': -61.359039306640625, 'logps/ref_rejected': -82.75496673583984, 'logits/chosen': -0.07853604853153229, 'logits/rejected': -0.056885600090026855, 'epoch': 0.46} + 46%|█████████████████████████████████████████████████████████████████████████▎ | 310/681 [13:42<16:30, 2.67s/it] 46%|█████████████████████████████████████████████████████████████████████████▌ | 311/681 [13:44<16:10, 2.62s/it] 46%|█████████████████████████████████████████████████████████████████████████▊ | 312/681 [13:46<15:29, 2.52s/it] 46%|█████████████████████████████████████████████████████████████████████████▉ | 313/681 [13:49<15:02, 2.45s/it] 46%|██████████████████████████████████████████████████████████████████████████▏ | 314/681 [13:51<15:03, 2.46s/it] 46%|██████████████████████████████████████████████████████████████████████████▍ | 315/681 [13:54<15:00, 2.46s/it] {'loss': 0.8915, 'grad_norm': 71.91686248779297, 'learning_rate': 3.270101039870797e-07, 'fcm_dpo/beta': 0.006514403969049454, 'fcm_dpo/q_t': 0.3323710560798645, 'fcm_dpo/delta': 0.00254381587728858, 'fcm_dpo/margin': 129.9599151611328, 'margin_dpo/margin_mean': 129.9599151611328, 'margin_dpo/margin_std': 146.8983612060547, 'logps/chosen': -250.1263885498047, 'logps/rejected': -412.8932189941406, 'logps/ref_chosen': -51.77602005004883, 'logps/ref_rejected': -84.58292388916016, 'logits/chosen': 0.050092458724975586, 'logits/rejected': 0.06273595988750458, 'epoch': 0.46} + 46%|██████████████████████████████████████████████████████████████████████████▍ | 315/681 [13:54<15:00, 2.46s/it] 46%|██████████████████████████████████████████████████████████████████████████▋ | 316/681 [13:56<15:05, 2.48s/it] 47%|██████████████████████████████████████████████████████████████████████████▉ | 317/681 [13:59<15:14, 2.51s/it] 47%|███████████████████████████████████████████████████████████████████████████▏ | 318/681 [14:01<14:59, 2.48s/it] 47%|███████████████████████████████████████████████████████████████████████████▍ | 319/681 [14:04<15:17, 2.54s/it] 47%|███████████████████████████████████████████████████████████████████████████▋ | 320/681 [14:07<15:45, 2.62s/it] {'loss': 0.7846, 'grad_norm': 37.25245666503906, 'learning_rate': 3.208807785813777e-07, 'fcm_dpo/beta': 0.006153360940515995, 'fcm_dpo/q_t': 0.3038443326950073, 'fcm_dpo/delta': -0.12349160015583038, 'fcm_dpo/margin': 156.13824462890625, 'margin_dpo/margin_mean': 156.13824462890625, 'margin_dpo/margin_std': 138.79611206054688, 'logps/chosen': -244.4154815673828, 'logps/rejected': -443.03955078125, 'logps/ref_chosen': -56.777862548828125, 'logps/ref_rejected': -99.26368713378906, 'logits/chosen': -0.020375465974211693, 'logits/rejected': -0.011311917565762997, 'epoch': 0.47} + 47%|███████████████████████████████████████████████████████████████████████████▋ | 320/681 [14:07<15:45, 2.62s/it] 47%|███████████████████████████████████████████████████████████████████████████▉ | 321/681 [14:09<15:23, 2.57s/it] 47%|████████████████████████████████████████████████████████████████████████████▏ | 322/681 [14:12<15:24, 2.57s/it] 47%|████████████████████████████████████████████████████████████████████████████▎ | 323/681 [14:14<15:40, 2.63s/it] 48%|████████████████████████████████████████████████████████████████████████████▌ | 324/681 [14:17<15:11, 2.55s/it] 48%|████████████████████████████████████████████████████████████████████████████▊ | 325/681 [14:20<15:27, 2.60s/it] {'loss': 0.8735, 'grad_norm': 46.0535774230957, 'learning_rate': 3.147047612756302e-07, 'fcm_dpo/beta': 0.00587537931278348, 'fcm_dpo/q_t': 0.3283257484436035, 'fcm_dpo/delta': 0.004868075251579285, 'fcm_dpo/margin': 143.66757202148438, 'margin_dpo/margin_mean': 143.66757202148438, 'margin_dpo/margin_std': 146.94345092773438, 'logps/chosen': -280.6909484863281, 'logps/rejected': -449.8771057128906, 'logps/ref_chosen': -58.28468704223633, 'logps/ref_rejected': -83.80326843261719, 'logits/chosen': 0.04561225324869156, 'logits/rejected': 0.05908801406621933, 'epoch': 0.48} + 48%|████████████████████████████████████████████████████████████████████████████▊ | 325/681 [14:20<15:27, 2.60s/it] 48%|█████████████████████████████████████████████████████████████████████████████ | 326/681 [14:22<15:22, 2.60s/it] 48%|█████████████████████████████████████████████████████████████████████████████▎ | 327/681 [14:25<15:29, 2.63s/it] 48%|█████████████████████████████████████████████████████████████████████████████▌ | 328/681 [14:27<15:22, 2.61s/it] 48%|█████████████████████████████████████████████████████████████████████████████▊ | 329/681 [14:30<15:01, 2.56s/it] 48%|██████████████████████████████████████████████████████████████████████████████ | 330/681 [14:32<15:05, 2.58s/it] {'loss': 0.8302, 'grad_norm': 50.687007904052734, 'learning_rate': 3.084861204504122e-07, 'fcm_dpo/beta': 0.0057464223355054855, 'fcm_dpo/q_t': 0.3170631527900696, 'fcm_dpo/delta': -0.0528077557682991, 'fcm_dpo/margin': 156.2254638671875, 'margin_dpo/margin_mean': 156.2254638671875, 'margin_dpo/margin_std': 149.75978088378906, 'logps/chosen': -278.55731201171875, 'logps/rejected': -466.06658935546875, 'logps/ref_chosen': -62.75822067260742, 'logps/ref_rejected': -94.04203033447266, 'logits/chosen': 0.018803680315613747, 'logits/rejected': 0.03341676667332649, 'epoch': 0.48} + 48%|██████████████████████████████████████████████████████████████████████████████ | 330/681 [14:33<15:05, 2.58s/it] 49%|██████████████████████████████████████████████████████████████████████████████▎ | 331/681 [14:35<15:06, 2.59s/it] 49%|██████████████████████████████████████████████████████████████████████████████▍ | 332/681 [14:38<14:52, 2.56s/it] 49%|██████████████████████████████████████████████████████████████████████████████▋ | 333/681 [14:40<14:28, 2.49s/it] 49%|██████████████████████████████████████████████████████████████████████████████▉ | 334/681 [14:42<14:18, 2.47s/it] 49%|███████████████████████████████████████████████████████████████████████████████▏ | 335/681 [14:45<14:25, 2.50s/it] {'loss': 0.8834, 'grad_norm': 52.71037673950195, 'learning_rate': 3.022289525640531e-07, 'fcm_dpo/beta': 0.005659957882016897, 'fcm_dpo/q_t': 0.33370915055274963, 'fcm_dpo/delta': 0.03993542492389679, 'fcm_dpo/margin': 143.60537719726562, 'margin_dpo/margin_mean': 143.60537719726562, 'margin_dpo/margin_std': 148.41580200195312, 'logps/chosen': -285.34234619140625, 'logps/rejected': -459.047119140625, 'logps/ref_chosen': -58.59650421142578, 'logps/ref_rejected': -88.69586944580078, 'logits/chosen': 0.07965292036533356, 'logits/rejected': 0.0922585278749466, 'epoch': 0.49} + 49%|███████████████████████████████████████████████████████████████████████████████▏ | 335/681 [14:45<14:25, 2.50s/it] 49%|███████████████████████████████████████████████████████████████████████████████▍ | 336/681 [14:48<14:40, 2.55s/it] 49%|███████████████████████████████████████████████████████████████████████████████▋ | 337/681 [14:50<14:10, 2.47s/it] 50%|███████████████████████████████████████████████████████████████████████████████▉ | 338/681 [14:52<14:12, 2.49s/it] 50%|████████████████████████████████████████████████████████████████████████████████▏ | 339/681 [14:55<13:46, 2.42s/it] 50%|████████████████████████████████████████████████████████████████████████████████▍ | 340/681 [14:57<13:57, 2.46s/it] {'loss': 0.825, 'grad_norm': 35.8148193359375, 'learning_rate': 2.959373794541426e-07, 'fcm_dpo/beta': 0.005560899153351784, 'fcm_dpo/q_t': 0.3149014711380005, 'fcm_dpo/delta': -0.058431338518857956, 'fcm_dpo/margin': 162.19033813476562, 'margin_dpo/margin_mean': 162.1903533935547, 'margin_dpo/margin_std': 153.74880981445312, 'logps/chosen': -255.1841278076172, 'logps/rejected': -453.63641357421875, 'logps/ref_chosen': -58.18162155151367, 'logps/ref_rejected': -94.44358825683594, 'logits/chosen': 0.013386556878685951, 'logits/rejected': 0.0291533712297678, 'epoch': 0.5} + 50%|████████████████████████████████████████████████████████████████████████████████▍ | 340/681 [14:57<13:57, 2.46s/it] 50%|████████████████████████████████████████████████████████████████████████████████▌ | 341/681 [14:59<13:32, 2.39s/it] 50%|████████████████████████████████████████████████████████████████████████████████▊ | 342/681 [15:02<14:08, 2.50s/it] 50%|█████████████████████████████████████████████████████████████████████████████████ | 343/681 [15:05<13:45, 2.44s/it] 51%|█████████████████████████████████████████████████████████████████████████████████▎ | 344/681 [15:07<13:51, 2.47s/it] 51%|█████████████████████████████████████████████████████████████████████████████████▌ | 345/681 [15:10<14:07, 2.52s/it] {'loss': 0.8223, 'grad_norm': 50.1009521484375, 'learning_rate': 2.896155456223163e-07, 'fcm_dpo/beta': 0.0053556510247290134, 'fcm_dpo/q_t': 0.31634849309921265, 'fcm_dpo/delta': -0.04881957173347473, 'fcm_dpo/margin': 166.8225860595703, 'margin_dpo/margin_mean': 166.8225860595703, 'margin_dpo/margin_std': 155.08956909179688, 'logps/chosen': -275.6701354980469, 'logps/rejected': -483.6131896972656, 'logps/ref_chosen': -57.9904899597168, 'logps/ref_rejected': -99.11092376708984, 'logits/chosen': -0.0006014652317389846, 'logits/rejected': 0.011017178185284138, 'epoch': 0.51} + 51%|█████████████████████████████████████████████████████████████████████████████████▌ | 345/681 [15:10<14:07, 2.52s/it] 51%|█████████████████████████████████████████████████████████████████████████████████▊ | 346/681 [15:12<14:11, 2.54s/it] 51%|██████████████████████████████████████████████████████████████████████████████████ | 347/681 [15:14<13:29, 2.42s/it] 51%|██████████████████████████████████████████████████████████████████████████████████▎ | 348/681 [15:17<13:25, 2.42s/it] 51%|██████████████████████████████████████████████████████████████████████████████████▌ | 349/681 [15:19<13:38, 2.46s/it] 51%|██████████████████████████████████████████████████████████████████████████████████▋ | 350/681 [15:22<13:51, 2.51s/it] {'loss': 0.8883, 'grad_norm': 61.714805603027344, 'learning_rate': 2.8326761550411346e-07, 'fcm_dpo/beta': 0.005294440779834986, 'fcm_dpo/q_t': 0.3322368264198303, 'fcm_dpo/delta': 0.02646171674132347, 'fcm_dpo/margin': 155.80526733398438, 'margin_dpo/margin_mean': 155.80526733398438, 'margin_dpo/margin_std': 168.00064086914062, 'logps/chosen': -326.2274169921875, 'logps/rejected': -512.9906005859375, 'logps/ref_chosen': -58.29923629760742, 'logps/ref_rejected': -89.25711822509766, 'logits/chosen': 0.13601693511009216, 'logits/rejected': 0.14563313126564026, 'epoch': 0.51} + 51%|██████████████████████████████████████████████████████████████████████████████████▋ | 350/681 [15:22<13:51, 2.51s/it] 52%|██████████████████████████████████████████████████████████████████████████████████▉ | 351/681 [15:25<13:51, 2.52s/it] 52%|███████████████████████████████████████████████████████████████████████████████████▏ | 352/681 [15:27<13:50, 2.52s/it] 52%|███████████████████████████████████████████████████████████████████████████████████▍ | 353/681 [15:30<13:44, 2.51s/it] 52%|███████████████████████████████████████████████████████████████████████████████████▋ | 354/681 [15:32<13:56, 2.56s/it] 52%|███████████████████████████████████████████████████████████████████████████████████▉ | 355/681 [15:35<13:45, 2.53s/it] {'loss': 0.8631, 'grad_norm': 50.927425384521484, 'learning_rate': 2.7689777072570284e-07, 'fcm_dpo/beta': 0.005374093074351549, 'fcm_dpo/q_t': 0.325572669506073, 'fcm_dpo/delta': -0.00456079700961709, 'fcm_dpo/margin': 158.74806213378906, 'margin_dpo/margin_mean': 158.74806213378906, 'margin_dpo/margin_std': 160.3525848388672, 'logps/chosen': -304.60711669921875, 'logps/rejected': -488.50799560546875, 'logps/ref_chosen': -60.788482666015625, 'logps/ref_rejected': -85.94129943847656, 'logits/chosen': 0.049882251769304276, 'logits/rejected': 0.07219198346138, 'epoch': 0.52} + 52%|███████████████████████████████████████████████████████████████████████████████████▉ | 355/681 [15:35<13:45, 2.53s/it] 52%|████████████████████████████████████████████████████████████████████████████████████▏ | 356/681 [15:37<13:58, 2.58s/it] 52%|████████████████████████████████████████████████████████████████████████████████████▍ | 357/681 [15:40<14:24, 2.67s/it] 53%|████████████████████████████████████████████████████████████████████████████████████▋ | 358/681 [15:43<14:40, 2.73s/it] 53%|████████████████████████████████████████████████████████████████████████████████████▊ | 359/681 [15:46<14:20, 2.67s/it] 53%|█████████████████████████████████████████████████████████████████████████████████████ | 360/681 [15:48<13:58, 2.61s/it] {'loss': 0.9029, 'grad_norm': 103.65924835205078, 'learning_rate': 2.7051020734928443e-07, 'fcm_dpo/beta': 0.005639675538986921, 'fcm_dpo/q_t': 0.33998966217041016, 'fcm_dpo/delta': 0.06519472599029541, 'fcm_dpo/margin': 139.63821411132812, 'margin_dpo/margin_mean': 139.63821411132812, 'margin_dpo/margin_std': 152.2101593017578, 'logps/chosen': -303.3082580566406, 'logps/rejected': -465.8846130371094, 'logps/ref_chosen': -57.6871337890625, 'logps/ref_rejected': -80.62527465820312, 'logits/chosen': 0.07553316652774811, 'logits/rejected': 0.0942920446395874, 'epoch': 0.53} + 53%|█████████████████████████████████████████████████████████████████████████████████████ | 360/681 [15:48<13:58, 2.61s/it] 53%|█████████████████████████████████████████████████████████████████████████████████████▎ | 361/681 [15:51<14:02, 2.63s/it] 53%|█████████████████████████████████████████████████████████████████████████████████████▌ | 362/681 [15:53<13:27, 2.53s/it] 53%|█████████████████████████████████████████████████████████████████████████████████████▊ | 363/681 [15:55<12:59, 2.45s/it] 53%|██████████████████████████████████████████████████████████████████████████████████████ | 364/681 [15:58<13:05, 2.48s/it] 54%|██████████████████████████████████████████████████████████████████████████████████████▎ | 365/681 [16:01<13:14, 2.51s/it] {'loss': 0.8274, 'grad_norm': 103.95121002197266, 'learning_rate': 2.641091331089811e-07, 'fcm_dpo/beta': 0.005498864687979221, 'fcm_dpo/q_t': 0.3161132335662842, 'fcm_dpo/delta': -0.058943700045347214, 'fcm_dpo/margin': 164.1685028076172, 'margin_dpo/margin_mean': 164.1685028076172, 'margin_dpo/margin_std': 157.7042236328125, 'logps/chosen': -293.01922607421875, 'logps/rejected': -496.72552490234375, 'logps/ref_chosen': -51.490867614746094, 'logps/ref_rejected': -91.02871704101562, 'logits/chosen': 0.13508400321006775, 'logits/rejected': 0.1376177817583084, 'epoch': 0.54} + 54%|██████████████████████████████████████████████████████████████████████████████████████▎ | 365/681 [16:01<13:14, 2.51s/it] 54%|██████████████████████████████████████████████████████████████████████████████████████▌ | 366/681 [16:03<13:24, 2.56s/it] 54%|██████████████████████████████████████████████████████████████████████████████████████▊ | 367/681 [16:06<13:38, 2.61s/it] 54%|███████████████████████████████████████████████████████████████████████████████████████ | 368/681 [16:08<13:30, 2.59s/it] 54%|███████████████████████████████████████████████████████████████████████████████████████▏ | 369/681 [16:11<14:07, 2.72s/it] 54%|███████████████████████████████████████████████████████████████████████████████████████▍ | 370/681 [16:14<14:04, 2.71s/it] {'loss': 0.8812, 'grad_norm': 51.74324035644531, 'learning_rate': 2.5769876463904263e-07, 'fcm_dpo/beta': 0.005576250609010458, 'fcm_dpo/q_t': 0.3373372554779053, 'fcm_dpo/delta': 0.08467354625463486, 'fcm_dpo/margin': 138.50732421875, 'margin_dpo/margin_mean': 138.50733947753906, 'margin_dpo/margin_std': 134.93521118164062, 'logps/chosen': -298.34576416015625, 'logps/rejected': -468.1741638183594, 'logps/ref_chosen': -58.113502502441406, 'logps/ref_rejected': -89.43451690673828, 'logits/chosen': 0.10848675668239594, 'logits/rejected': 0.11948434263467789, 'epoch': 0.54} + 54%|███████████████████████████████████████████████████████████████████████████████████████▍ | 370/681 [16:14<14:04, 2.71s/it] 54%|███████████████████████████████████████████████████████████████████████████████████████▋ | 371/681 [16:17<13:55, 2.70s/it] 55%|███████████████████████████████████████████████████████████████████████████████████████▉ | 372/681 [16:19<13:11, 2.56s/it] 55%|████████████████████████████████████████████████████████████████████████████████████████▏ | 373/681 [16:22<12:59, 2.53s/it] 55%|████████████████████████████████████████████████████████████████████████████████████████▍ | 374/681 [16:24<13:09, 2.57s/it] 55%|████████████████████████████████████████████████████████████████████████████████████████▋ | 375/681 [16:27<13:07, 2.57s/it] {'loss': 0.8508, 'grad_norm': 49.77079772949219, 'learning_rate': 2.512833246961859e-07, 'fcm_dpo/beta': 0.0056075444445014, 'fcm_dpo/q_t': 0.3194437623023987, 'fcm_dpo/delta': -0.06286319345235825, 'fcm_dpo/margin': 161.43272399902344, 'margin_dpo/margin_mean': 161.43270874023438, 'margin_dpo/margin_std': 164.4888458251953, 'logps/chosen': -326.29534912109375, 'logps/rejected': -511.74200439453125, 'logps/ref_chosen': -65.23600769042969, 'logps/ref_rejected': -89.24995422363281, 'logits/chosen': 0.1206691637635231, 'logits/rejected': 0.14743848145008087, 'epoch': 0.55} + 55%|████████████████████████████████████████████████████████████████████████████████████████▋ | 375/681 [16:27<13:07, 2.57s/it] 55%|████████████████████████████████████████████████████████████████████████████████████████▉ | 376/681 [16:29<13:12, 2.60s/it] 55%|█████████████████████████████████████████████████████████████████████████████████████████▏ | 377/681 [16:32<12:56, 2.55s/it] 56%|█████████████████████████████████████████████████████████████████████████████████████████▎ | 378/681 [16:34<12:14, 2.43s/it] 56%|█████████████████████████████████████████████████████████████████████████████████████████▌ | 379/681 [16:37<12:22, 2.46s/it] 56%|█████████████████████████████████████████████████████████████████████████████████████████▊ | 380/681 [16:39<12:07, 2.42s/it] {'loss': 0.8158, 'grad_norm': 47.95132064819336, 'learning_rate': 2.4486703937790243e-07, 'fcm_dpo/beta': 0.005182487424463034, 'fcm_dpo/q_t': 0.3068377375602722, 'fcm_dpo/delta': -0.11963502317667007, 'fcm_dpo/margin': 185.00277709960938, 'margin_dpo/margin_mean': 185.00277709960938, 'margin_dpo/margin_std': 178.25515747070312, 'logps/chosen': -336.24005126953125, 'logps/rejected': -570.0576171875, 'logps/ref_chosen': -53.33893966674805, 'logps/ref_rejected': -102.15375518798828, 'logits/chosen': 0.1993691623210907, 'logits/rejected': 0.2027546912431717, 'epoch': 0.56} + 56%|█████████████████████████████████████████████████████████████████████████████████████████▊ | 380/681 [16:39<12:07, 2.42s/it] 56%|██████████████████████████████████████████████████████████████████████████████████████████ | 381/681 [16:41<12:18, 2.46s/it] 56%|██████████████████████████████████████████████████████████████████████████████████████████▎ | 382/681 [16:44<12:43, 2.55s/it] 56%|██████████████████████████████████████████████████████████████████████████████████████████▌ | 383/681 [16:47<12:37, 2.54s/it] 56%|██████████████████████████████████████████████████████████████████████████████████████████▊ | 384/681 [16:49<12:34, 2.54s/it] 57%|███████████████████████████████████████████████████████████████████████████████████████████ | 385/681 [16:52<12:23, 2.51s/it] {'loss': 0.8731, 'grad_norm': 34.479740142822266, 'learning_rate': 2.3845413533856514e-07, 'fcm_dpo/beta': 0.0049522449262440205, 'fcm_dpo/q_t': 0.3291183114051819, 'fcm_dpo/delta': 0.019767077639698982, 'fcm_dpo/margin': 167.88037109375, 'margin_dpo/margin_mean': 167.88037109375, 'margin_dpo/margin_std': 172.85272216796875, 'logps/chosen': -353.31121826171875, 'logps/rejected': -552.2758178710938, 'logps/ref_chosen': -58.36262130737305, 'logps/ref_rejected': -89.44685363769531, 'logits/chosen': 0.18332220613956451, 'logits/rejected': 0.19640536606311798, 'epoch': 0.57} + 57%|███████████████████████████████████████████████████████████████████████████████████████████ | 385/681 [16:52<12:23, 2.51s/it] 57%|███████████████████████████████████████████████████████████████████████████████████████████▎ | 386/681 [16:54<12:27, 2.53s/it] 57%|███████████████████████████████████████████████████████████████████████████████████████████▍ | 387/681 [16:57<12:23, 2.53s/it] 57%|███████████████████████████████████████████████████████████████████████████████████████████▋ | 388/681 [16:59<12:00, 2.46s/it] 57%|███████████████████████████████████████████████████████████████████████████████████████████▉ | 389/681 [17:02<11:51, 2.44s/it] 57%|████████████████████████████████████████████████████████████████████████████████████████████▏ | 390/681 [17:04<11:52, 2.45s/it] {'loss': 0.8665, 'grad_norm': 46.603187561035156, 'learning_rate': 2.320488370051681e-07, 'fcm_dpo/beta': 0.004950010217726231, 'fcm_dpo/q_t': 0.32719507813453674, 'fcm_dpo/delta': -0.0016832177061587572, 'fcm_dpo/margin': 171.85813903808594, 'margin_dpo/margin_mean': 171.85813903808594, 'margin_dpo/margin_std': 178.92532348632812, 'logps/chosen': -333.3777160644531, 'logps/rejected': -539.3297119140625, 'logps/ref_chosen': -56.380653381347656, 'logps/ref_rejected': -90.47447204589844, 'logits/chosen': 0.12718243896961212, 'logits/rejected': 0.1411055475473404, 'epoch': 0.57} + 57%|████████████████████████████████████████████████████████████████████████████████████████████▏ | 390/681 [17:04<11:52, 2.45s/it] 57%|████████████████████████████████████████████████████████████████████████████████████████████▍ | 391/681 [17:07<11:59, 2.48s/it] 58%|████████████████████████████████████████████████████████████████████████████████████████████▋ | 392/681 [17:09<12:10, 2.53s/it] 58%|████████████████████████████████████████████████████████████████████████████████████████████▉ | 393/681 [17:12<12:12, 2.54s/it] 58%|█████████████████████████████████████████████████████████████████████████████████████████████▏ | 394/681 [17:14<12:13, 2.56s/it] 58%|█████████████████████████████████████████████████████████████████████████████████████████████▍ | 395/681 [17:17<12:02, 2.52s/it] {'loss': 0.9001, 'grad_norm': 44.36903381347656, 'learning_rate': 2.2565536379453404e-07, 'fcm_dpo/beta': 0.005209635477513075, 'fcm_dpo/q_t': 0.33868879079818726, 'fcm_dpo/delta': 0.042330484837293625, 'fcm_dpo/margin': 155.28634643554688, 'margin_dpo/margin_mean': 155.28634643554688, 'margin_dpo/margin_std': 174.56260681152344, 'logps/chosen': -274.38555908203125, 'logps/rejected': -460.8504943847656, 'logps/ref_chosen': -55.95304489135742, 'logps/ref_rejected': -87.13162994384766, 'logits/chosen': 0.03185752406716347, 'logits/rejected': 0.039340805262327194, 'epoch': 0.58} + 58%|█████████████████████████████████████████████████████████████████████████████████████████████▍ | 395/681 [17:17<12:02, 2.52s/it] 58%|█████████████████████████████████████████████████████████████████████████████████████████████▌ | 396/681 [17:19<11:59, 2.53s/it] 58%|█████████████████████████████████████████████████████████████████████████████████████████████▊ | 397/681 [17:22<11:49, 2.50s/it] 58%|██████████████████████████████████████████████████████████████████████████████████████████████ | 398/681 [17:24<11:30, 2.44s/it] 59%|██████████████████████████████████████████████████████████████████████████████████████████████▎ | 399/681 [17:26<11:13, 2.39s/it] 59%|██████████████████████████████████████████████████████████████████████████████████████████████▌ | 400/681 [17:29<11:18, 2.42s/it] {'loss': 0.8159, 'grad_norm': 40.749473571777344, 'learning_rate': 2.192779273338215e-07, 'fcm_dpo/beta': 0.005051156505942345, 'fcm_dpo/q_t': 0.31105726957321167, 'fcm_dpo/delta': -0.08489851653575897, 'fcm_dpo/margin': 183.75982666015625, 'margin_dpo/margin_mean': 183.7598419189453, 'margin_dpo/margin_std': 173.43460083007812, 'logps/chosen': -283.33929443359375, 'logps/rejected': -499.20843505859375, 'logps/ref_chosen': -64.59160614013672, 'logps/ref_rejected': -96.700927734375, 'logits/chosen': 0.04275448992848396, 'logits/rejected': 0.0653674453496933, 'epoch': 0.59} + 59%|██████████████████████████████████████████████████████████████████████████████████████████████▌ | 400/681 [17:29<11:18, 2.42s/it][INFO|trainer.py:4307] 2026-04-22 00:18:26,598 >> +***** Running Evaluation ***** +[INFO|trainer.py:4309] 2026-04-22 00:18:26,599 >> Num examples = 2339 +[INFO|trainer.py:4312] 2026-04-22 00:18:26,599 >> Batch size = 8 + + 0%| | 0/73 [00:00> +***** Running Evaluation ***** +[INFO|trainer.py:4309] 2026-04-22 00:27:35,703 >> Num examples = 2339 +[INFO|trainer.py:4312] 2026-04-22 00:27:35,703 >> Batch size = 8 + + 0%| | 0/73 [00:00> + +Training completed. Do not forget to share your model on huggingface.co/models =) + + + {'train_runtime': 1846.2605, 'train_samples_per_second': 23.614, 'train_steps_per_second': 0.369, 'train_loss': 0.8947697063024523, 'epoch': 1.0} + 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 681/681 [30:43<00:00, 2.53s/it] 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 681/681 [30:43<00:00, 2.71s/it] +***** train metrics ***** + epoch = 1.0 + total_flos = 0GF + train_loss = 0.8948 + train_runtime = 0:30:46.26 + train_samples = 43598 + train_samples_per_second = 23.614 + train_steps_per_second = 0.369 +2026-04-22 00:31:40 - INFO - __main__ - *** Training complete *** +2026-04-22 00:31:40 - INFO - __main__ - *** Save model *** +[INFO|configuration_utils.py:419] 2026-04-22 00:32:14,748 >> Configuration saved in /root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/config.json +[INFO|configuration_utils.py:911] 2026-04-22 00:32:14,750 >> Configuration saved in /root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/generation_config.json +[INFO|modeling_utils.py:3580] 2026-04-22 00:32:40,444 >> The model is bigger than the maximum size per checkpoint (5GB) and is going to be split in 7 checkpoint shards. You can find where each parameters has been saved in the index located at /root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/model.safetensors.index.json. +[INFO|tokenization_utils_base.py:2510] 2026-04-22 00:32:40,449 >> tokenizer config file saved in /root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/tokenizer_config.json +[INFO|tokenization_utils_base.py:2519] 2026-04-22 00:32:40,449 >> Special tokens file saved in /root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/special_tokens_map.json +2026-04-22 00:32:40 - INFO - __main__ - Saved HF-compatible model artifacts to /root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802 +[INFO|modelcard.py:450] 2026-04-22 00:32:43,030 >> Dropping the following result as it does not have all the necessary fields: +{'dataset': {'name': 'Anthropic/hh-rlhf', 'type': 'Anthropic/hh-rlhf'}} +[INFO|configuration_utils.py:419] 2026-04-22 00:32:43,033 >> Configuration saved in /root/dynamic-dpo-v4/outputs/outputs/llama-3-8b-base-new-dpo-hh-helpful-s_star0.85-4xh200-batch-64-20260421-233802/config.json +2026-04-22 00:32:43 - INFO - __main__ - *** Evaluate *** +[INFO|trainer.py:4307] 2026-04-22 00:32:43,035 >> +***** Running Evaluation ***** +[INFO|trainer.py:4309] 2026-04-22 00:32:43,035 >> Num examples = 2339 +[INFO|trainer.py:4312] 2026-04-22 00:32:43,035 >> Batch size = 8 + 0%| | 0/73 [00:00