初始化项目,由ModelHub XC社区提供模型
Model: jackf857/llama-3-8b-base-new-dpo-harmless-s_star0.4-q_t0.4 Source: Original Platform
This commit is contained in:
36
.gitattributes
vendored
Normal file
36
.gitattributes
vendored
Normal file
@@ -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
|
||||
78
README.md
Normal file
78
README.md
Normal file
@@ -0,0 +1,78 @@
|
||||
---
|
||||
library_name: transformers
|
||||
base_model: W-61/llama-3-8b-base-sft-hh-harmless-4xh200
|
||||
tags:
|
||||
- alignment-handbook
|
||||
- new-dpo
|
||||
- generated_from_trainer
|
||||
datasets:
|
||||
- Anthropic/hh-rlhf
|
||||
model-index:
|
||||
- name: llama3-8b-base-new-method-s_star0.4
|
||||
results: []
|
||||
---
|
||||
|
||||
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
||||
should probably proofread and complete it, then remove this comment. -->
|
||||
|
||||
# llama3-8b-base-new-method-s_star0.4
|
||||
|
||||
This model is a fine-tuned version of [W-61/llama-3-8b-base-sft-hh-harmless-4xh200](https://huggingface.co/W-61/llama-3-8b-base-sft-hh-harmless-4xh200) on the Anthropic/hh-rlhf dataset.
|
||||
It achieves the following results on the evaluation set:
|
||||
- Loss: 0.6075
|
||||
- Fcm Dpo/beta: 0.0032
|
||||
- Margin Dpo/margin Mean: 80.9270
|
||||
- Margin Dpo/margin Std: 168.3527
|
||||
- Logps/chosen: -278.6199
|
||||
- Logps/rejected: -364.2365
|
||||
- Logps/ref Chosen: -74.8595
|
||||
- Logps/ref Rejected: -79.5490
|
||||
- Logits/chosen: 0.8914
|
||||
- Logits/rejected: 0.8742
|
||||
|
||||
## 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 | Margin Dpo/margin Mean | Margin Dpo/margin Std | Logps/chosen | Logps/rejected | Logps/ref Chosen | Logps/ref Rejected | Logits/chosen | Logits/rejected |
|
||||
|:-------------:|:------:|:----:|:---------------:|:------------:|:----------------------:|:---------------------:|:------------:|:--------------:|:----------------:|:------------------:|:-------------:|:---------------:|
|
||||
| 1.0917 | 0.3023 | 200 | 0.5616 | 0.0221 | 20.9587 | 36.0271 | -112.0402 | -137.6885 | -74.8595 | -79.5490 | 0.5290 | 0.4788 |
|
||||
| 1.134 | 0.6047 | 400 | 0.5988 | 0.0047 | 61.8184 | 122.3443 | -231.1051 | -297.6130 | -74.8595 | -79.5490 | 0.7731 | 0.7383 |
|
||||
| 1.242 | 0.9070 | 600 | 0.6075 | 0.0032 | 80.9270 | 168.3527 | -278.6199 | -364.2365 | -74.8595 | -79.5490 | 0.8914 | 0.8742 |
|
||||
|
||||
|
||||
### Framework versions
|
||||
|
||||
- Transformers 4.51.0
|
||||
- Pytorch 2.3.1+cu121
|
||||
- Datasets 2.21.0
|
||||
- Tokenizers 0.21.4
|
||||
23
all_results.json
Normal file
23
all_results.json
Normal file
@@ -0,0 +1,23 @@
|
||||
{
|
||||
"epoch": 0.999244142101285,
|
||||
"eval_fcm_dpo/beta": 0.003048304468393326,
|
||||
"eval_logits/chosen": 0.9178658127784729,
|
||||
"eval_logits/rejected": 0.8999158143997192,
|
||||
"eval_logps/chosen": -279.8544006347656,
|
||||
"eval_logps/ref_chosen": -74.85946655273438,
|
||||
"eval_logps/ref_rejected": -79.54898834228516,
|
||||
"eval_logps/rejected": -365.8829650878906,
|
||||
"eval_loss": 0.6095079779624939,
|
||||
"eval_margin_dpo/margin_mean": 81.33904266357422,
|
||||
"eval_margin_dpo/margin_std": 169.26637268066406,
|
||||
"eval_runtime": 38.7977,
|
||||
"eval_samples": 2303,
|
||||
"eval_samples_per_second": 59.359,
|
||||
"eval_steps_per_second": 1.856,
|
||||
"total_flos": 0.0,
|
||||
"train_loss": 1.1812975648311552,
|
||||
"train_runtime": 1809.2515,
|
||||
"train_samples": 42336,
|
||||
"train_samples_per_second": 23.4,
|
||||
"train_steps_per_second": 0.365
|
||||
}
|
||||
29
config.json
Normal file
29
config.json
Normal file
@@ -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
|
||||
}
|
||||
17
eval_results.json
Normal file
17
eval_results.json
Normal file
@@ -0,0 +1,17 @@
|
||||
{
|
||||
"epoch": 0.999244142101285,
|
||||
"eval_fcm_dpo/beta": 0.003048304468393326,
|
||||
"eval_logits/chosen": 0.9178658127784729,
|
||||
"eval_logits/rejected": 0.8999158143997192,
|
||||
"eval_logps/chosen": -279.8544006347656,
|
||||
"eval_logps/ref_chosen": -74.85946655273438,
|
||||
"eval_logps/ref_rejected": -79.54898834228516,
|
||||
"eval_logps/rejected": -365.8829650878906,
|
||||
"eval_loss": 0.6095079779624939,
|
||||
"eval_margin_dpo/margin_mean": 81.33904266357422,
|
||||
"eval_margin_dpo/margin_std": 169.26637268066406,
|
||||
"eval_runtime": 38.7977,
|
||||
"eval_samples": 2303,
|
||||
"eval_samples_per_second": 59.359,
|
||||
"eval_steps_per_second": 1.856
|
||||
}
|
||||
9
generation_config.json
Normal file
9
generation_config.json
Normal file
@@ -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"
|
||||
}
|
||||
661
margin_logs/margins.jsonl
Normal file
661
margin_logs/margins.jsonl
Normal file
@@ -0,0 +1,661 @@
|
||||
{"epoch": 0.0, "step": 1, "batch_size": 64, "mean": -0.0013527870178222656, "std": 0.2564818859100342, "min": -0.736083984375, "p10": -0.3432229995727539, "median": 0.038166046142578125, "p90": 0.29227676391601565, "max": 0.645111083984375, "pos_frac": 0.578125, "sample": [0.1120758056640625, 0.12518310546875, 0.31621551513671875, 0.13765716552734375, -0.12592506408691406, 0.23141098022460938, -0.21887779235839844, 0.21950721740722656, 0.04480743408203125, 0.020877838134765625, 0.0570220947265625, 0.058269500732421875, -0.4338226318359375, -0.030628204345703125, 0.645111083984375, -0.395477294921875, 0.09050941467285156, 0.0007190704345703125, -0.34615325927734375, 0.016077041625976562, -0.33638572692871094, 0.293853759765625, 0.17610931396484375, 0.22386932373046875, 0.21470260620117188, -0.08536529541015625, 0.0907745361328125, -0.03816986083984375, 0.39190101623535156, 0.16336441040039062, 0.08024787902832031, -0.031158447265625, 0.08477020263671875, 0.002460479736328125, -0.242034912109375, 0.07232666015625, -0.60186767578125, 0.20531463623046875, 0.155731201171875, -0.14299774169921875, -0.25698089599609375, 0.12331962585449219, -0.26497650146484375, 0.15140533447265625, -0.0920257568359375, -0.18599319458007812, 0.19028091430664062, 0.2496490478515625, 0.42162322998046875, 0.17873382568359375, -0.1525421142578125, -0.4972076416015625, 0.32010650634765625, -0.10365867614746094, -0.233795166015625, -0.19828224182128906, -0.4018898010253906, -0.13407135009765625, -0.09596633911132812, 0.031524658203125, 0.28859710693359375, -0.192962646484375, -0.736083984375, 0.3026123046875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000001.npy"}
|
||||
{"epoch": 0.0015117157974300832, "step": 2, "batch_size": 64, "mean": 0.03744968771934509, "std": 0.2875921130180359, "min": -0.7604827880859375, "p10": -0.2812448501586914, "median": 0.03963661193847656, "p90": 0.3654294967651367, "max": 0.8134727478027344, "pos_frac": 0.5625, "sample": [0.30594635009765625, -0.24289894104003906, -0.11509323120117188, -0.13417816162109375, 0.06942558288574219, 0.36568641662597656, -0.14640045166015625, 0.1497650146484375, 0.30261993408203125, 0.10124588012695312, 0.13028717041015625, -0.0031890869140625, 0.0361480712890625, 0.5662612915039062, 0.09694290161132812, -0.01091766357421875, 0.1128997802734375, 0.0411834716796875, -0.21860504150390625, -0.1236419677734375, -0.08812713623046875, 0.10360527038574219, 0.1790008544921875, -0.5114288330078125, 0.3056755065917969, -0.14553451538085938, 0.28168487548828125, 0.26990509033203125, 0.1686878204345703, 0.038089752197265625, 0.19541168212890625, -0.10783576965332031, -0.2644004821777344, -0.19707489013671875, -0.140472412109375, 0.1349811553955078, 0.19672012329101562, -0.0714111328125, 0.53369140625, 0.1271820068359375, 0.8134727478027344, 0.2990264892578125, -0.7604827880859375, -0.08274078369140625, 0.05890846252441406, 0.029361724853515625, 0.4510040283203125, -0.1599273681640625, -0.29346656799316406, 0.10005569458007812, -0.27509117126464844, -0.1937713623046875, 0.19167327880859375, 0.28173065185546875, -0.09406471252441406, -0.3380699157714844, -0.29186248779296875, 0.36483001708984375, 0.009979248046875, 0.44391632080078125, -0.126708984375, -0.6550216674804688, 0.6160736083984375, -0.28388214111328125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000002.npy"}
|
||||
{"epoch": 0.0030234315948601664, "step": 3, "batch_size": 64, "mean": 0.03398832678794861, "std": 0.34009286761283875, "min": -0.9418182373046875, "p10": -0.3103809356689453, "median": 0.04606437683105469, "p90": 0.3908416748046877, "max": 1.0906219482421875, "pos_frac": 0.578125, "sample": [0.08205032348632812, 0.030282974243164062, -0.095489501953125, -0.5462646484375, 0.2054290771484375, 0.343353271484375, 0.05162811279296875, 0.0534515380859375, 0.41119384765625, -0.19660186767578125, 0.13656997680664062, -0.25408935546875, 0.327178955078125, 0.09978675842285156, 0.48465728759765625, 0.11382293701171875, -0.3034515380859375, 0.00438690185546875, 0.17412567138671875, -0.052276611328125, -0.9418182373046875, -0.057361602783203125, -0.1786651611328125, -0.11523818969726562, -0.118072509765625, 0.1917743682861328, 0.10346221923828125, 0.6988525390625, 0.040500640869140625, 0.3245086669921875, 0.17124557495117188, -0.09479522705078125, 0.05745697021484375, 0.03023529052734375, -0.079437255859375, -0.002017974853515625, -0.3994598388671875, 0.3274078369140625, 0.1286792755126953, 0.5689620971679688, 0.45458984375, -0.2111053466796875, 0.02001953125, 1.0906219482421875, 0.19634246826171875, 0.22673797607421875, -0.10190200805664062, 0.17009353637695312, -0.037586212158203125, -0.6961822509765625, -0.858184814453125, 0.20313262939453125, 0.12673187255859375, 0.31229591369628906, -0.6481781005859375, 0.31048583984375, 0.2660789489746094, -0.0130615234375, -0.18184661865234375, -0.0838623046875, -0.3133506774902344, 0.5002288818359375, -0.1046905517578125, -0.17811965942382812], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000003.npy"}
|
||||
{"epoch": 0.0045351473922902496, "step": 4, "batch_size": 64, "mean": 0.02435511350631714, "std": 0.3523942530155182, "min": -0.54937744140625, "p10": -0.39495639801025384, "median": -0.04857063293457031, "p90": 0.4349420547485352, "max": 1.28717041015625, "pos_frac": 0.453125, "sample": [-0.099029541015625, 0.0360260009765625, 0.29273223876953125, 0.19994354248046875, -0.1935577392578125, -0.3057708740234375, -0.054229736328125, 0.048389434814453125, 0.17417144775390625, 0.6680374145507812, 0.17127609252929688, -0.044765472412109375, -0.2730255126953125, 0.00872039794921875, -0.3106231689453125, -0.54937744140625, 0.05068206787109375, -0.29552459716796875, 0.479461669921875, -0.41539764404296875, 0.26068115234375, -0.000698089599609375, -0.154632568359375, -0.4123382568359375, -0.17360305786132812, 0.04418182373046875, 0.2972755432128906, 0.43732643127441406, 0.2726097106933594, 0.1160736083984375, 1.28717041015625, -0.31870269775390625, 0.3363914489746094, 0.1729888916015625, -0.49184417724609375, 0.8460845947265625, 0.191619873046875, -0.3422737121582031, 0.6521148681640625, -0.1009521484375, -0.047626495361328125, -0.07928466796875, 0.5465545654296875, -0.3543987274169922, 0.31200408935546875, 0.39017486572265625, -0.231231689453125, -0.17817115783691406, -0.48561859130859375, 0.3086814880371094, 0.384033203125, -0.19313430786132812, -0.050567626953125, -0.18976974487304688, -0.2657928466796875, -0.0495147705078125, -0.2660675048828125, 0.4293785095214844, -0.060359954833984375, -0.4353199005126953, -0.41931915283203125, -0.103973388671875, 0.26334381103515625, -0.17290496826171875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000004.npy"}
|
||||
{"epoch": 0.006046863189720333, "step": 5, "batch_size": 64, "mean": -0.04486989974975586, "std": 0.26818743348121643, "min": -0.9185791015625, "p10": -0.3917705535888672, "median": -0.060726165771484375, "p90": 0.2728904724121095, "max": 0.5258636474609375, "pos_frac": 0.421875, "sample": [-0.15078353881835938, -0.34966278076171875, 0.07441329956054688, -0.16241455078125, -0.4895172119140625, -0.1726531982421875, -0.14418411254882812, 0.22247314453125, 0.18141937255859375, 0.2011871337890625, 0.0535888671875, -0.9185791015625, -0.018688201904296875, 0.009540557861328125, -0.4246673583984375, -0.355499267578125, 0.19945526123046875, -0.1596393585205078, -0.408233642578125, -0.07321929931640625, -0.11059951782226562, -0.026332855224609375, 0.14389801025390625, 0.28472137451171875, 0.167205810546875, -0.507354736328125, -0.4202423095703125, -0.31650352478027344, 0.3296966552734375, 0.11555862426757812, 0.38202667236328125, -0.1234283447265625, -0.34951019287109375, 0.21168899536132812, 0.356903076171875, -0.39307403564453125, -0.08830833435058594, -0.20436859130859375, -0.1354503631591797, 0.13656234741210938, 0.046672821044921875, -0.1291961669921875, -0.2523231506347656, 0.1996002197265625, 0.3882293701171875, -0.18403053283691406, -0.0359344482421875, 0.09310150146484375, 0.2154388427734375, -0.1192626953125, 0.21529006958007812, -0.1552734375, 0.3195610046386719, 0.5258636474609375, 0.0926513671875, -0.07217025756835938, -0.3887290954589844, -0.1473064422607422, -0.12819290161132812, -0.049282073974609375, -0.01564788818359375, 0.2452850341796875, -0.33234405517578125, 0.22890090942382812], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000005.npy"}
|
||||
{"epoch": 0.007558578987150416, "step": 6, "batch_size": 64, "mean": -0.03759393095970154, "std": 0.32241499423980713, "min": -0.6926059722900391, "p10": -0.4644296646118164, "median": -0.04454612731933594, "p90": 0.36993484497070317, "max": 0.88006591796875, "pos_frac": 0.421875, "sample": [0.0067291259765625, -0.04379844665527344, 0.0036773681640625, -0.07091522216796875, 0.17140579223632812, 0.22130775451660156, 0.0073299407958984375, 0.18334197998046875, -0.489410400390625, -0.19968605041503906, 0.14191627502441406, 0.2591552734375, -0.3279609680175781, 0.42382049560546875, -0.08428955078125, -0.0077056884765625, 0.00693511962890625, 0.6381034851074219, -0.10912704467773438, -0.6926059722900391, -0.3355560302734375, -0.12835693359375, -0.26119041442871094, 0.3592643737792969, 0.1699981689453125, -0.1021881103515625, -0.0892486572265625, 0.145904541015625, -0.025392532348632812, 0.46216392517089844, 0.1301746368408203, 0.24829864501953125, -0.00933837890625, 0.09434127807617188, -0.4877471923828125, -0.24964141845703125, -0.016826629638671875, -0.37039947509765625, -0.32498931884765625, -0.3414459228515625, -0.5315093994140625, 0.10642051696777344, -0.19805145263671875, -0.12618255615234375, -0.31058502197265625, -0.15503692626953125, 0.88006591796875, 0.20357513427734375, -0.42883872985839844, -0.04529380798339844, 0.454193115234375, 0.6773452758789062, -0.28702545166015625, -0.24184799194335938, -0.47968292236328125, 0.13680267333984375, -0.1383686065673828, -0.057342529296875, -0.64599609375, -0.41485595703125, -0.54205322265625, 0.3745079040527344, 0.35317230224609375, 0.10452842712402344], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000006.npy"}
|
||||
{"epoch": 0.009070294784580499, "step": 7, "batch_size": 64, "mean": 0.0037850141525268555, "std": 0.27835723757743835, "min": -0.7815322875976562, "p10": -0.36120758056640623, "median": 0.055336952209472656, "p90": 0.32605304718017597, "max": 0.7770538330078125, "pos_frac": 0.5625, "sample": [-0.07359695434570312, -0.41538238525390625, 0.00095367431640625, -0.2547168731689453, 0.16691970825195312, 0.17937088012695312, 0.3438892364501953, 0.137908935546875, -0.23126983642578125, 0.27816009521484375, -0.158935546875, -0.04295921325683594, 0.12919235229492188, 0.08243560791015625, -0.05309867858886719, -0.07719802856445312, 0.020267486572265625, 0.243194580078125, 0.3463096618652344, -0.49764251708984375, 0.2135467529296875, 0.2844352722167969, 0.1075592041015625, 0.4544830322265625, -0.3838958740234375, 0.05061531066894531, 0.17304229736328125, 0.0678558349609375, -0.0176849365234375, 0.06005859375, -0.24817657470703125, -0.3418731689453125, -0.2889995574951172, 0.207855224609375, 0.11482620239257812, 0.42891693115234375, 0.14619827270507812, 0.3791351318359375, 0.03991889953613281, -0.017251968383789062, 0.21590423583984375, -0.07207107543945312, 0.119232177734375, 0.7770538330078125, -0.7815322875976562, -0.3651123046875, -0.2750053405761719, 0.09400749206542969, -0.17302703857421875, 0.08038330078125, 0.20502281188964844, -0.274627685546875, 0.2330341339111328, -0.14168548583984375, -0.16782379150390625, -0.3520965576171875, -0.4056854248046875, -0.439849853515625, 0.081817626953125, 0.5301513671875, -0.3287811279296875, 0.16645050048828125, 0.12034988403320312, -0.1582355499267578], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000007.npy"}
|
||||
{"epoch": 0.010582010582010581, "step": 8, "batch_size": 64, "mean": 0.006621301174163818, "std": 0.29272568225860596, "min": -0.631103515625, "p10": -0.29898300170898434, "median": 0.0071849822998046875, "p90": 0.40882949829101567, "max": 0.917205810546875, "pos_frac": 0.515625, "sample": [0.0166015625, 0.6096038818359375, 0.917205810546875, -0.27816009521484375, -0.631103515625, -0.41143035888671875, -0.139801025390625, 0.41542816162109375, -0.27416038513183594, 0.1343536376953125, -0.520660400390625, -0.16721343994140625, -0.1073455810546875, -0.2699394226074219, 0.6902923583984375, 0.503753662109375, 0.19568634033203125, 0.1520252227783203, -0.2559776306152344, 0.02337646484375, 0.10544586181640625, -0.05219268798828125, -0.39936065673828125, 0.0330657958984375, -0.03127288818359375, -0.1849994659423828, 0.000720977783203125, 0.17017364501953125, -0.10330581665039062, -0.22008705139160156, 0.22185516357421875, 0.14116668701171875, -0.12552452087402344, -0.3079071044921875, 0.116943359375, -0.1968841552734375, 0.08740234375, 0.2351226806640625, 0.5292510986328125, -0.1103057861328125, 0.48540496826171875, -0.17473411560058594, -0.17470169067382812, 0.01364898681640625, 0.023639678955078125, -0.2431793212890625, 0.27680015563964844, -0.26076507568359375, -0.3983268737792969, 0.0387115478515625, 0.1401081085205078, 0.11637687683105469, -0.31658172607421875, -0.23968887329101562, -0.0970001220703125, 0.2639312744140625, 0.04526329040527344, -0.234771728515625, 0.3934326171875, 0.12886428833007812, 0.051349639892578125, 0.205841064453125, -0.02283477783203125, -0.10886764526367188], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000008.npy"}
|
||||
{"epoch": 0.012093726379440665, "step": 9, "batch_size": 64, "mean": 0.03980347514152527, "std": 0.3651173710823059, "min": -0.9029617309570312, "p10": -0.44429244995117184, "median": 0.07248497009277344, "p90": 0.43422317504882824, "max": 1.068084716796875, "pos_frac": 0.609375, "sample": [0.8240966796875, 0.3568267822265625, -0.08730697631835938, 0.0899200439453125, -0.11724472045898438, -0.6024818420410156, -0.0817718505859375, -0.389678955078125, -0.0124359130859375, -0.6773529052734375, 0.22170257568359375, -0.15592384338378906, 0.44689178466796875, -0.086578369140625, 0.201995849609375, -0.5281410217285156, 0.080902099609375, 1.068084716796875, 0.10382080078125, 0.096771240234375, 0.1401214599609375, 0.057460784912109375, 0.5971527099609375, -0.07280731201171875, -0.46363067626953125, 0.138031005859375, 0.12360954284667969, 0.21148681640625, 0.24001121520996094, -0.399169921875, -0.9029617309570312, -0.24738311767578125, -0.076171875, 0.695465087890625, 0.3329124450683594, -0.15445899963378906, 0.4046630859375, -0.2448883056640625, -0.0866241455078125, 0.638946533203125, 0.0622711181640625, -0.571258544921875, 0.0364990234375, 0.11962890625, 0.03809928894042969, 0.20855331420898438, -0.06690788269042969, 0.06589317321777344, 0.15996932983398438, -0.31240081787109375, 0.2521705627441406, 0.2505645751953125, 0.048954010009765625, 0.3046112060546875, 0.3262767791748047, 0.35333251953125, 0.09509658813476562, -0.2151641845703125, 0.011383056640625, 0.07907676696777344, 0.576904296875, -0.71038818359375, -0.35611724853515625, 0.10651397705078125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000009.npy"}
|
||||
{"epoch": 0.013605442176870748, "step": 10, "batch_size": 64, "mean": -0.015915364027023315, "std": 0.3761499524116516, "min": -0.709259033203125, "p10": -0.47168617248535155, "median": -0.02990436553955078, "p90": 0.4087066650390626, "max": 1.174072265625, "pos_frac": 0.453125, "sample": [0.057952880859375, -0.184722900390625, -0.09820556640625, 0.02606201171875, 0.09976959228515625, -0.0932769775390625, -0.08701133728027344, -0.16912841796875, 0.061248779296875, 0.009613037109375, -0.41834259033203125, -0.2644233703613281, -0.6557083129882812, 0.6440277099609375, 0.1537933349609375, 0.10614013671875, 0.32035064697265625, -0.35748291015625, 0.8702392578125, 0.2656898498535156, -0.008518218994140625, -0.23401832580566406, 0.3504180908203125, 0.37821197509765625, -0.2308502197265625, -0.4636192321777344, -0.16445541381835938, 0.128204345703125, -0.02901458740234375, -0.22010040283203125, 0.033725738525390625, 0.749053955078125, -0.0750885009765625, 0.25484466552734375, -0.10971641540527344, 0.42177581787109375, -0.4521903991699219, 0.2269306182861328, 0.2689628601074219, -0.030794143676757812, -0.32666778564453125, -0.39844322204589844, -0.548095703125, -0.2992095947265625, -0.4751434326171875, 0.12818145751953125, -0.4521827697753906, -0.0085906982421875, 0.6231842041015625, -0.157012939453125, -0.709259033203125, 0.46373748779296875, -0.36470794677734375, 0.06885719299316406, -0.13409423828125, -0.23688507080078125, 1.174072265625, 0.143829345703125, -0.4906463623046875, 0.25763702392578125, 0.34664154052734375, -0.4949188232421875, 0.3482208251953125, -0.55743408203125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000010.npy"}
|
||||
{"epoch": 0.015117157974300832, "step": 11, "batch_size": 64, "mean": 0.015301555395126343, "std": 0.3707212507724762, "min": -1.5084304809570312, "p10": -0.36024036407470705, "median": 0.07547187805175781, "p90": 0.4199352264404298, "max": 0.8996429443359375, "pos_frac": 0.515625, "sample": [-0.15190887451171875, 0.3703575134277344, -0.218780517578125, 0.62286376953125, -0.6685352325439453, 0.1630096435546875, -0.1575946807861328, -0.13389205932617188, -0.3671417236328125, 0.47637939453125, -0.02184295654296875, 0.22875213623046875, 0.13967132568359375, -0.049182891845703125, -0.12104225158691406, 0.4315910339355469, 0.7576141357421875, 0.3318328857421875, 0.0911865234375, -0.11429595947265625, 0.3708381652832031, -0.3615570068359375, 0.14322662353515625, -0.06556320190429688, -0.35716819763183594, -0.11103057861328125, 0.07750320434570312, 0.2774810791015625, 0.11046791076660156, 0.5521087646484375, -0.07598876953125, 0.14885902404785156, 0.2939434051513672, -0.03216552734375, 0.08013343811035156, -0.15668487548828125, -0.19542312622070312, 0.8996429443359375, 0.10601806640625, 0.0734405517578125, -0.4968109130859375, 0.09160614013671875, -0.13736724853515625, 0.2204742431640625, -0.046783447265625, -1.5084304809570312, -0.2614402770996094, 0.07880020141601562, 0.19499969482421875, 0.39273834228515625, 0.077789306640625, -0.9116058349609375, -0.04741859436035156, 0.17770004272460938, -0.43369293212890625, -0.141998291015625, -0.2502880096435547, 0.24836158752441406, -0.18543243408203125, -0.155609130859375, 0.12805938720703125, 0.16663360595703125, -0.1659698486328125, 0.557861328125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000011.npy"}
|
||||
{"epoch": 0.016628873771730914, "step": 12, "batch_size": 64, "mean": -0.02333444356918335, "std": 0.29752811789512634, "min": -0.8015823364257812, "p10": -0.3386341094970703, "median": -0.030788421630859375, "p90": 0.3345066070556641, "max": 0.697418212890625, "pos_frac": 0.453125, "sample": [-0.2642860412597656, -0.31586456298828125, -0.022775650024414062, 0.027387619018554688, -0.1436595916748047, -0.207305908203125, 0.1379547119140625, -0.10860443115234375, 0.13023948669433594, -0.2017364501953125, -0.15676116943359375, -0.16176605224609375, -0.410888671875, 0.697418212890625, 0.29198646545410156, -0.583831787109375, -0.48625946044921875, -0.3303642272949219, -0.315216064453125, 0.2003612518310547, -0.03697967529296875, -0.3421783447265625, -0.0725555419921875, 0.4278106689453125, 0.2841053009033203, -0.15802001953125, -0.11488151550292969, -0.6179084777832031, -0.055660247802734375, 0.43352508544921875, -0.1909637451171875, -0.2429656982421875, -0.8015823364257812, -0.02459716796875, -0.17451858520507812, 0.07328033447265625, -0.32257080078125, 0.3242912292480469, 0.4009590148925781, -0.3756542205810547, -0.20577239990234375, -0.23526763916015625, -0.14273834228515625, -0.1516876220703125, 0.17071533203125, 0.1543560028076172, 0.22802734375, 0.3252754211425781, 0.33846282958984375, 0.23387908935546875, 0.6571388244628906, 0.17090225219726562, 0.2362823486328125, 0.177642822265625, 0.0701446533203125, -0.26918792724609375, -0.3061084747314453, 0.38653564453125, 0.013483047485351562, 0.06046485900878906, -0.004425048828125, 0.172454833984375, 0.2289714813232422, 0.008083343505859375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000012.npy"}
|
||||
{"epoch": 0.018140589569160998, "step": 13, "batch_size": 64, "mean": -0.06307116150856018, "std": 0.2957088053226471, "min": -1.1441650390625, "p10": -0.3820430755615234, "median": -0.06828594207763672, "p90": 0.29508094787597666, "max": 0.5077362060546875, "pos_frac": 0.421875, "sample": [-1.1441650390625, -0.30313873291015625, 0.1591033935546875, -0.0006866455078125, 0.07183837890625, -0.38909149169921875, 0.3668251037597656, -0.032825469970703125, -0.2432861328125, -0.24414825439453125, -0.0796356201171875, 0.009038925170898438, 0.5077362060546875, 0.2511749267578125, -0.07990264892578125, 0.27433013916015625, -0.2529296875, 0.3039741516113281, -0.0040435791015625, 0.144378662109375, -0.2532806396484375, -0.1940460205078125, -0.3150482177734375, -0.03521728515625, -0.3655967712402344, -0.09904098510742188, -0.19977188110351562, 0.1470012664794922, -0.1399383544921875, -0.34703826904296875, 0.1492748260498047, -0.31510162353515625, 0.4752655029296875, -0.3303108215332031, -0.43578338623046875, 0.13606643676757812, 0.15637969970703125, -0.11960983276367188, 0.08592796325683594, -0.4127349853515625, 0.14208221435546875, 0.3450431823730469, -0.2364044189453125, -0.28722381591796875, -0.5749969482421875, 0.17288970947265625, 0.37145233154296875, 0.22653961181640625, -0.309783935546875, 0.00890350341796875, -0.0701751708984375, -0.09177398681640625, -0.15715408325195312, 0.018655776977539062, 0.43083953857421875, -0.5566177368164062, -0.52557373046875, -0.2656230926513672, 0.2279510498046875, 0.16680908203125, 0.240966796875, -0.2180938720703125, 0.06918716430664062, -0.06639671325683594], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000013.npy"}
|
||||
{"epoch": 0.019652305366591082, "step": 14, "batch_size": 64, "mean": 0.11851927638053894, "std": 0.34638702869415283, "min": -0.593902587890625, "p10": -0.17802047729492185, "median": 0.07866668701171875, "p90": 0.5283254623413087, "max": 1.3873977661132812, "pos_frac": 0.640625, "sample": [0.0692138671875, -0.262298583984375, -0.009857177734375, 0.2877960205078125, 0.1745128631591797, 0.41986656188964844, 0.1276092529296875, 0.555450439453125, -0.1436328887939453, -0.21240997314453125, 0.0215606689453125, 0.2863807678222656, -0.13476943969726562, 0.12833404541015625, 0.23160934448242188, -0.1274871826171875, 0.0748291015625, 0.04705810546875, -0.13263702392578125, 0.011676788330078125, -0.12422943115234375, -0.5465164184570312, 0.116790771484375, 0.08777618408203125, 0.0825042724609375, 0.24396514892578125, -0.53656005859375, 0.40026092529296875, -0.005405426025390625, 1.3873977661132812, 0.031841278076171875, 1.119964599609375, 0.0372161865234375, 0.23553466796875, 0.5380649566650391, -0.324188232421875, 0.20743179321289062, 0.09810638427734375, 0.20562362670898438, 0.5055999755859375, 0.205322265625, 0.48615455627441406, -0.044097900390625, 0.990447998046875, 0.00672149658203125, 0.04635429382324219, -0.18773651123046875, 0.5780029296875, 0.6368541717529297, 0.094482421875, -0.1553497314453125, -0.593902587890625, 0.08981704711914062, -0.10857391357421875, 0.10149002075195312, 0.11798095703125, -0.086883544921875, -0.0045680999755859375, -0.14213943481445312, -0.11858177185058594, 0.4431800842285156, -0.1239166259765625, 0.23363494873046875, -0.05344390869140625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000014.npy"}
|
||||
{"epoch": 0.021164021164021163, "step": 15, "batch_size": 64, "mean": 0.03145986795425415, "std": 0.37765005230903625, "min": -1.04608154296875, "p10": -0.41704635620117186, "median": 0.033069610595703125, "p90": 0.43840141296386737, "max": 1.4120941162109375, "pos_frac": 0.546875, "sample": [0.0064373016357421875, -0.2898712158203125, 0.10150909423828125, -0.0580596923828125, -0.16558456420898438, 0.2349700927734375, 0.5283050537109375, -0.12233734130859375, 0.045166015625, -0.4328155517578125, 0.2983131408691406, -0.1384258270263672, -0.117950439453125, -0.0880889892578125, 0.3051605224609375, 0.11713027954101562, 1.4120941162109375, -0.12800979614257812, 0.3473052978515625, -0.2013683319091797, 0.16670989990234375, -0.22744369506835938, -0.08417510986328125, -0.3483619689941406, -0.010438919067382812, 0.5918807983398438, 0.1305084228515625, 0.1162109375, -0.3292999267578125, -0.4209442138671875, 0.7038116455078125, -0.477569580078125, -0.40795135498046875, -0.66162109375, 0.3982429504394531, 0.26877593994140625, 0.5541458129882812, 0.060455322265625, -0.07118988037109375, -0.0851287841796875, 0.0069580078125, 0.10602569580078125, -0.7600173950195312, 0.3377227783203125, 0.15689468383789062, -1.04608154296875, -0.46630859375, -0.03340911865234375, 0.16689491271972656, 0.10223388671875, 0.4556121826171875, 0.11365509033203125, -0.15161514282226562, 0.106231689453125, -0.16063499450683594, 0.388824462890625, 0.232635498046875, -0.18471527099609375, 0.04897117614746094, 0.02097320556640625, 0.8465194702148438, -0.071319580078125, 0.1640777587890625, 0.11280632019042969], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000015.npy"}
|
||||
{"epoch": 0.022675736961451247, "step": 16, "batch_size": 64, "mean": -0.02647252380847931, "std": 0.29655349254608154, "min": -0.848480224609375, "p10": -0.4314727783203125, "median": -0.00054168701171875, "p90": 0.30241203308105474, "max": 0.5575942993164062, "pos_frac": 0.5, "sample": [-0.0066585540771484375, 0.0371551513671875, -0.569915771484375, 0.04289817810058594, 0.5575942993164062, 0.10055160522460938, 0.0021572113037109375, -0.44140625, -0.1517791748046875, 0.06034278869628906, 0.00787353515625, 0.27063751220703125, 0.051006317138671875, -0.22915267944335938, 0.29113197326660156, -0.131561279296875, -0.06305503845214844, 0.11835670471191406, -0.20575714111328125, -0.408294677734375, -0.14444732666015625, -0.09972763061523438, -0.0247650146484375, 0.5139007568359375, 0.1764049530029297, -0.354736328125, 0.17159652709960938, 0.1667633056640625, 0.12418556213378906, -0.7904891967773438, 0.2940673828125, -0.14204025268554688, 0.148406982421875, 0.3059883117675781, 0.0759429931640625, 0.27927398681640625, 0.4211273193359375, -0.0542755126953125, 0.06963729858398438, -0.15715408325195312, 0.0057659149169921875, -0.12384223937988281, -0.244232177734375, 0.3821296691894531, 0.4340667724609375, 0.22478485107421875, 0.450775146484375, -0.3449859619140625, -0.07862663269042969, -0.2167530059814453, -0.0032405853271484375, -0.095733642578125, -0.047453880310058594, -0.5716171264648438, -0.4710693359375, 0.0543212890625, -0.07483673095703125, -0.6760406494140625, -0.848480224609375, 0.03833961486816406, -0.06518173217773438, -0.10733795166015625, 0.2620391845703125, 0.11118316650390625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000016.npy"}
|
||||
{"epoch": 0.02418745275888133, "step": 17, "batch_size": 64, "mean": -0.01597180962562561, "std": 0.26538151502609253, "min": -0.7306976318359375, "p10": -0.35429725646972654, "median": -0.001140594482421875, "p90": 0.291108512878418, "max": 0.5415916442871094, "pos_frac": 0.5, "sample": [0.08532905578613281, 0.2539215087890625, -0.10550308227539062, -0.17901229858398438, -0.298248291015625, 0.13212966918945312, 0.02567291259765625, -0.3030357360839844, -0.09612464904785156, -0.35459136962890625, 0.1381072998046875, -0.043491363525390625, -0.3163318634033203, 0.08760452270507812, -0.2259674072265625, -0.10269927978515625, -0.1283416748046875, -0.24102401733398438, 0.06544685363769531, -0.663177490234375, 0.26031494140625, 0.4178142547607422, 0.026611328125, -0.185546875, 0.0474395751953125, -0.3658714294433594, -0.138275146484375, -0.06781578063964844, 0.5415916442871094, 0.350494384765625, -0.10468673706054688, -0.40118408203125, 0.21416473388671875, -0.16093063354492188, 0.33562469482421875, 0.200408935546875, 0.19658279418945312, 0.27069091796875, -0.08341217041015625, -0.037933349609375, -0.7306976318359375, -0.16844558715820312, 0.28142738342285156, 0.44629669189453125, 0.06411552429199219, 0.251617431640625, -0.3919219970703125, 0.295257568359375, 0.35802650451660156, 0.2056884765625, -0.3475608825683594, -0.4561004638671875, -0.3536109924316406, -0.20521163940429688, 0.03057098388671875, 0.12879180908203125, 0.1538829803466797, 0.2446880340576172, 0.15267181396484375, -0.06455230712890625, 0.07927322387695312, 0.14167022705078125, -0.15686416625976562, -0.0279541015625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000017.npy"}
|
||||
{"epoch": 0.025699168556311415, "step": 18, "batch_size": 64, "mean": -0.0012346208095550537, "std": 0.3209212124347687, "min": -0.8498992919921875, "p10": -0.4012580871582031, "median": 0.021697998046875, "p90": 0.38300495147705077, "max": 0.9290847778320312, "pos_frac": 0.5625, "sample": [0.18573760986328125, 0.37483978271484375, -0.026906967163085938, 0.04090690612792969, 0.0198974609375, 0.40874481201171875, 0.18314743041992188, 0.25335693359375, 0.38488006591796875, 0.02349853515625, 0.14484596252441406, -0.4794464111328125, 0.08056068420410156, 0.0273895263671875, -0.25391387939453125, 0.9290847778320312, 0.15152931213378906, -0.7124862670898438, 0.17419052124023438, -0.2931365966796875, 0.3839073181152344, 0.49603271484375, -0.225830078125, 0.10794830322265625, -0.008480072021484375, -0.28264617919921875, -0.41595458984375, 0.18079185485839844, -0.13484573364257812, 0.041187286376953125, 0.06115913391113281, 0.16741943359375, -0.03607940673828125, 0.1491851806640625, -0.038909912109375, -0.006103515625, 0.03309440612792969, -0.10395240783691406, -0.5982894897460938, -0.362274169921875, -0.13361358642578125, -0.8498992919921875, -0.249786376953125, -0.30181884765625, 0.04885673522949219, 0.38089942932128906, 0.010068893432617188, 0.0068836212158203125, -0.13018798828125, -0.36696624755859375, 0.21019744873046875, 0.1685924530029297, -0.6803188323974609, -0.21891021728515625, 0.3159942626953125, -0.12447738647460938, 0.0039844512939453125, -0.015077590942382812, 0.623992919921875, 0.04369354248046875, -0.4186363220214844, 0.0706787109375, 0.6106834411621094, -0.10792922973632812], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000018.npy"}
|
||||
{"epoch": 0.027210884353741496, "step": 19, "batch_size": 64, "mean": 0.025185495615005493, "std": 0.3136102557182312, "min": -0.877685546875, "p10": -0.32975082397460936, "median": 0.03704357147216797, "p90": 0.41109695434570326, "max": 0.6450881958007812, "pos_frac": 0.5625, "sample": [0.379913330078125, -0.293975830078125, 0.029293060302734375, -0.780242919921875, 0.12015533447265625, 0.2168426513671875, -0.68231201171875, -0.22056961059570312, -0.2566680908203125, 0.147918701171875, 0.12446975708007812, -0.20611572265625, 0.5981369018554688, 0.14633941650390625, -0.04012298583984375, 0.06389808654785156, 0.1666107177734375, -0.06926345825195312, -0.3620452880859375, 0.2917613983154297, 0.293792724609375, 0.269073486328125, -0.22501373291015625, -0.3863525390625, 0.48321533203125, 0.3672637939453125, -0.04052734375, -0.09751129150390625, 0.10852813720703125, -0.12267112731933594, 0.31363677978515625, -0.07463836669921875, 0.0062103271484375, 0.2721595764160156, -0.877685546875, -0.2548675537109375, -0.12353897094726562, 0.1360015869140625, 0.349609375, 0.0048770904541015625, 0.09847640991210938, -0.4190406799316406, 0.0099639892578125, 0.10726547241210938, 0.0678863525390625, 0.3064441680908203, -0.197265625, -0.3103485107421875, -0.06319236755371094, 0.4289703369140625, 0.5625152587890625, -0.00568389892578125, 0.08514404296875, 0.04479408264160156, -0.33806610107421875, 0.5897293090820312, -0.1882781982421875, -0.2011871337890625, -0.14105224609375, 0.42446136474609375, -0.10151290893554688, 0.17360687255859375, 0.6450881958007812, 0.257568359375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000019.npy"}
|
||||
{"epoch": 0.02872260015117158, "step": 20, "batch_size": 64, "mean": -0.005217447876930237, "std": 0.2737097442150116, "min": -0.7575912475585938, "p10": -0.3490425109863281, "median": -0.015951156616210938, "p90": 0.3224458694458009, "max": 0.9113388061523438, "pos_frac": 0.484375, "sample": [0.19950103759765625, -0.20782470703125, -0.3780479431152344, -0.16605377197265625, -0.2027111053466797, -0.32244110107421875, 0.08633804321289062, 0.066680908203125, -0.07202911376953125, -0.7575912475585938, 0.2685394287109375, 0.24912643432617188, -0.220947265625, 0.20094680786132812, -0.38873291015625, 0.0604705810546875, -0.47780609130859375, 0.08893013000488281, 0.07729339599609375, -0.4129829406738281, 0.18853759765625, -0.067535400390625, 0.08851814270019531, 0.02529144287109375, -0.013835906982421875, -0.1099395751953125, 0.2908496856689453, -0.09760856628417969, 0.1421966552734375, 0.11929988861083984, -0.360443115234375, -0.245941162109375, 0.00264739990234375, -0.04923057556152344, -0.15826034545898438, 0.3429832458496094, 0.011449813842773438, 0.074249267578125, -0.051723480224609375, 0.3359870910644531, -0.25191497802734375, -0.3121795654296875, 0.12001800537109375, -0.065216064453125, 0.39885711669921875, -0.2398681640625, 0.5155563354492188, -0.28603363037109375, 0.22764205932617188, -0.12041473388671875, -0.41141510009765625, -0.01806640625, 0.359954833984375, 0.9113388061523438, -0.12371063232421875, -0.018518447875976562, 0.5620536804199219, 0.16148757934570312, 0.1995697021484375, 0.08188438415527344, -0.03038787841796875, -0.15197372436523438, -0.030364990234375, 0.029634475708007812], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000020.npy"}
|
||||
{"epoch": 0.030234315948601664, "step": 21, "batch_size": 64, "mean": -0.01130935549736023, "std": 0.3377159535884857, "min": -0.6185455322265625, "p10": -0.388677978515625, "median": -0.05753898620605469, "p90": 0.3527580261230469, "max": 1.237457275390625, "pos_frac": 0.4375, "sample": [0.2583637237548828, -0.3262367248535156, -0.17304611206054688, -0.4412078857421875, 0.290740966796875, -0.1354217529296875, 0.23415184020996094, 0.01705169677734375, -0.3293724060058594, -0.4727497100830078, 0.3140602111816406, -0.06527328491210938, 0.04498291015625, 0.163238525390625, 0.33416748046875, -0.01371002197265625, 0.20301055908203125, 0.9012832641601562, -0.2396697998046875, 0.03170013427734375, -0.520172119140625, 0.2303333282470703, 0.07733154296875, -0.3583221435546875, 0.117767333984375, -0.6185455322265625, -0.2668628692626953, 0.36200714111328125, -0.0498046875, -0.20648765563964844, 0.41033935546875, -0.1838245391845703, 0.6142578125, -0.3606147766113281, -0.567779541015625, 0.3588714599609375, 1.237457275390625, 0.1306610107421875, -0.5217437744140625, -0.14423179626464844, 0.2621498107910156, -0.01334381103515625, -0.3900146484375, 0.33849334716796875, 0.2832183837890625, -0.09571647644042969, -0.04857635498046875, 0.26306915283203125, -0.2753486633300781, 0.3763313293457031, 0.00373077392578125, -0.0733642578125, -0.20717239379882812, -0.38555908203125, -0.1995391845703125, -0.10992431640625, -0.1432018280029297, -0.10402679443359375, -0.09916114807128906, 0.1009063720703125, -0.33730316162109375, -0.1121826171875, 0.0409698486328125, -0.1349334716796875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000021.npy"}
|
||||
{"epoch": 0.031746031746031744, "step": 22, "batch_size": 64, "mean": 0.032350122928619385, "std": 0.24960029125213623, "min": -0.6911773681640625, "p10": -0.2767299652099609, "median": 0.02424144744873047, "p90": 0.326068115234375, "max": 0.5693206787109375, "pos_frac": 0.5625, "sample": [-0.23863983154296875, -0.48941802978515625, -0.06455230712890625, -0.031673431396484375, -0.18599891662597656, 0.27679443359375, 0.2866935729980469, -0.2198333740234375, 0.149993896484375, 0.19421958923339844, 0.15375518798828125, -0.0756378173828125, -0.0070285797119140625, 0.008253097534179688, 0.2503509521484375, -0.32180023193359375, -0.15185546875, 0.003322601318359375, -0.1146392822265625, -0.40522003173828125, 0.43390655517578125, -0.318084716796875, 0.18537139892578125, 0.138885498046875, 0.0245513916015625, -0.169525146484375, 0.106842041015625, -0.08759689331054688, 0.023931503295898438, 0.05857658386230469, 0.0845489501953125, 0.108642578125, 0.032684326171875, -0.05682373046875, 0.07744789123535156, -0.08592987060546875, 0.45350074768066406, 0.14410400390625, -0.3962860107421875, 0.31927490234375, 0.25351524353027344, -0.6911773681640625, -0.26110076904296875, -0.0079193115234375, -0.038524627685546875, 0.2939300537109375, 0.0233154296875, -0.15557861328125, -0.109466552734375, 0.3289794921875, -0.17609024047851562, 0.5693206787109375, 0.20690155029296875, 0.5353317260742188, 0.205230712890625, 0.2674083709716797, -0.04506683349609375, 0.2028636932373047, 0.345611572265625, 0.3937339782714844, 0.12068939208984375, -0.2834281921386719, 0.20226478576660156, -0.2054443359375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000022.npy"}
|
||||
{"epoch": 0.03325774754346183, "step": 23, "batch_size": 64, "mean": 0.018759459257125854, "std": 0.3408527970314026, "min": -0.847442626953125, "p10": -0.33164329528808595, "median": 0.0277862548828125, "p90": 0.42942810058593767, "max": 0.8914794921875, "pos_frac": 0.515625, "sample": [-0.7067718505859375, 0.046905517578125, 0.30844879150390625, -0.847442626953125, -0.1477832794189453, 0.301177978515625, -0.06995010375976562, -0.347625732421875, -0.128875732421875, 0.4451751708984375, -0.1336822509765625, -0.11831283569335938, 0.05838775634765625, -0.11086082458496094, 0.1357440948486328, 0.884307861328125, -0.5228042602539062, 0.18112945556640625, -0.3156776428222656, 0.26019287109375, 0.3926849365234375, 0.17641448974609375, 0.1743144989013672, -0.022319793701171875, 0.4943199157714844, 0.511993408203125, -0.235443115234375, -0.08723640441894531, 0.3442230224609375, -0.0663299560546875, -0.77325439453125, 0.2368927001953125, 0.24695587158203125, 0.08789634704589844, 0.7589797973632812, -0.08662796020507812, 0.124542236328125, 0.8914794921875, 0.12352371215820312, -0.11547660827636719, -0.48319244384765625, 0.023834228515625, -0.24815750122070312, -0.2485198974609375, 0.28208160400390625, -0.182342529296875, 0.18470001220703125, 0.11433792114257812, 0.15560150146484375, -0.3384857177734375, -0.18017578125, -0.1395587921142578, -0.14929962158203125, -0.1422271728515625, -0.21582794189453125, 0.16113853454589844, 0.03173828125, 0.18452835083007812, -0.262420654296875, 0.18047332763671875, 0.489776611328125, -0.2680625915527344, 0.089385986328125, -0.18793487548828125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000023.npy"}
|
||||
{"epoch": 0.03476946334089191, "step": 24, "batch_size": 64, "mean": -0.023777365684509277, "std": 0.27760303020477295, "min": -0.8871726989746094, "p10": -0.33075065612792964, "median": -0.02881622314453125, "p90": 0.26615524291992193, "max": 0.8246688842773438, "pos_frac": 0.40625, "sample": [-0.24538421630859375, -0.03360748291015625, -0.4250946044921875, 0.21108055114746094, 0.108367919921875, -0.14513015747070312, 0.04918861389160156, 0.03833770751953125, -0.186431884765625, -0.45133209228515625, -0.1531200408935547, 0.09160232543945312, 0.07025718688964844, -0.2898712158203125, 0.20578765869140625, 0.170440673828125, -0.16046142578125, -0.10107231140136719, -0.00366973876953125, 0.08575630187988281, -0.128570556640625, 0.8246688842773438, -0.2856292724609375, -0.3462677001953125, 0.23392486572265625, -0.3769035339355469, 0.1127777099609375, -0.15842819213867188, -0.61456298828125, 0.49127197265625, -0.1251201629638672, -0.16808319091796875, -0.8871726989746094, -0.1320934295654297, 0.24768447875976562, -0.2588958740234375, 0.15689849853515625, -0.018482208251953125, -0.007114410400390625, 0.487762451171875, 0.41046142578125, 0.23546600341796875, 0.2722663879394531, 0.03647804260253906, 0.2408599853515625, -0.2598114013671875, 0.44347381591796875, -0.26371002197265625, -0.14304351806640625, -0.021942138671875, -0.2945442199707031, 0.12039947509765625, -0.06401824951171875, -0.02402496337890625, -0.366912841796875, 0.2518959045410156, 0.2976703643798828, -0.06793212890625, -0.1096954345703125, -0.07480621337890625, -0.135406494140625, 0.17633056640625, -0.0555877685546875, -0.0089263916015625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000024.npy"}
|
||||
{"epoch": 0.036281179138321996, "step": 25, "batch_size": 64, "mean": 0.0020164549350738525, "std": 0.3268199861049652, "min": -0.8709716796875, "p10": -0.3897062301635742, "median": -0.02817821502685547, "p90": 0.3633361816406251, "max": 1.02490234375, "pos_frac": 0.4375, "sample": [0.2808380126953125, -0.01871490478515625, -0.15152359008789062, 0.324951171875, 0.209259033203125, -0.3834857940673828, 0.6609725952148438, -0.17133331298828125, -0.127349853515625, -0.16141128540039062, -0.1877288818359375, -0.027357101440429688, -0.473541259765625, -0.1140899658203125, 0.5435333251953125, -0.06864547729492188, 0.24542999267578125, -0.15459060668945312, 0.04700469970703125, 0.3717041015625, -0.12462234497070312, 0.008436203002929688, -0.173095703125, 0.5181732177734375, -0.07209396362304688, 0.14023399353027344, -0.005908966064453125, 1.02490234375, -0.11925125122070312, -0.1290740966796875, -0.05584716796875, 0.19342041015625, -0.0703887939453125, 0.2543144226074219, 0.31246185302734375, -0.10032463073730469, -0.623626708984375, -0.21831130981445312, -0.00965118408203125, -0.7530059814453125, -0.8709716796875, -0.3085441589355469, 0.10345649719238281, 0.04720497131347656, 0.20413589477539062, -0.08501434326171875, 0.1398773193359375, 0.2094268798828125, 0.2041015625, -0.6373214721679688, -0.19332504272460938, -0.06024360656738281, 0.11695098876953125, -0.055866241455078125, -0.02899932861328125, 0.2069377899169922, 0.44409942626953125, 0.09266281127929688, 0.433807373046875, -0.5115852355957031, -0.39237213134765625, 0.34381103515625, -0.07815933227539062, 0.1643218994140625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000025.npy"}
|
||||
{"epoch": 0.03779289493575208, "step": 26, "batch_size": 64, "mean": -0.01955801248550415, "std": 0.34907516837120056, "min": -0.953125, "p10": -0.4066307067871094, "median": -0.04245948791503906, "p90": 0.3903335571289064, "max": 0.7889862060546875, "pos_frac": 0.46875, "sample": [-0.0918731689453125, -0.332550048828125, -0.2097625732421875, -0.14662933349609375, 0.3522605895996094, 0.28011512756347656, 0.3589630126953125, 0.045562744140625, -0.4221534729003906, -0.027553558349609375, -0.3221321105957031, 0.69232177734375, 0.0569610595703125, 0.047637939453125, -0.3707733154296875, -0.3952178955078125, -0.2810401916503906, 0.28471946716308594, 0.08050537109375, 0.22042274475097656, 0.7889862060546875, -0.13687515258789062, -0.375946044921875, -0.18134117126464844, -0.2564964294433594, -0.25251007080078125, 0.403778076171875, -0.062091827392578125, 0.09132194519042969, -0.16283035278320312, -0.11803245544433594, 0.7784595489501953, 0.10791778564453125, -0.16936874389648438, -0.41152191162109375, -0.953125, 0.21087265014648438, 0.26950836181640625, -0.1570587158203125, 0.67852783203125, -0.22974014282226562, 0.05687713623046875, -0.10983657836914062, 0.19482421875, 0.5134925842285156, -0.6400299072265625, -0.3869476318359375, 0.2075176239013672, 0.04754638671875, -0.08979225158691406, 0.197845458984375, -0.48828125, 0.07602691650390625, 0.7284698486328125, -0.5324859619140625, -0.2298431396484375, -0.38105010986328125, -0.05736541748046875, 0.2786865234375, 0.027830123901367188, -0.42266845703125, 0.045063018798828125, 0.03569602966308594, -0.0055065155029296875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000026.npy"}
|
||||
{"epoch": 0.039304610733182165, "step": 27, "batch_size": 64, "mean": 0.029484301805496216, "std": 0.2778511643409729, "min": -0.6304244995117188, "p10": -0.30720748901367184, "median": 0.027912139892578125, "p90": 0.41223907470703136, "max": 0.7670211791992188, "pos_frac": 0.546875, "sample": [0.173065185546875, 0.21836471557617188, -0.135528564453125, -0.0850830078125, -0.05718803405761719, 0.00681304931640625, -0.145172119140625, -0.6304244995117188, 0.11806488037109375, 0.1703948974609375, 0.575286865234375, -0.4516143798828125, 0.3337593078613281, 0.15874862670898438, 0.021305084228515625, -0.1922454833984375, -0.1616687774658203, -0.267913818359375, 0.13298797607421875, -0.22235870361328125, 0.05730438232421875, -0.062530517578125, 0.429046630859375, 0.7670211791992188, -0.5488433837890625, -0.237945556640625, 0.01390838623046875, 0.2545166015625, 0.10827255249023438, -0.38691139221191406, -0.3222198486328125, 0.09216690063476562, -0.10556221008300781, 0.163818359375, 0.2093944549560547, -0.27190399169921875, 0.18886375427246094, 0.42327117919921875, 0.45999908447265625, -0.06539154052734375, 0.38649749755859375, -0.099365234375, -0.2638740539550781, -0.27217864990234375, 0.3161773681640625, 0.36746978759765625, -0.17353057861328125, 0.20755767822265625, -0.146759033203125, 0.07359695434570312, 0.034519195556640625, -0.3450469970703125, -0.031049728393554688, 0.521392822265625, 0.19283294677734375, 0.06551551818847656, -0.0613555908203125, -0.051082611083984375, -0.06597900390625, -0.3386249542236328, 0.14381027221679688, 0.0570831298828125, 0.14449691772460938, 0.4990234375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000027.npy"}
|
||||
{"epoch": 0.04081632653061224, "step": 28, "batch_size": 64, "mean": -0.004229158163070679, "std": 0.27890264987945557, "min": -0.7429962158203125, "p10": -0.3697463989257812, "median": -0.016817092895507812, "p90": 0.36661663055419924, "max": 0.520599365234375, "pos_frac": 0.46875, "sample": [0.3097381591796875, -0.384033203125, 0.520599365234375, 0.2271270751953125, -0.3364105224609375, -0.07648086547851562, 0.0130615234375, 0.3332366943359375, 0.3691577911376953, 0.360687255859375, -0.247802734375, -0.02498626708984375, -0.2863006591796875, -0.06317138671875, -0.264251708984375, -0.01303863525390625, -0.08351707458496094, -0.39853477478027344, 0.013416290283203125, 0.1801910400390625, 0.40842437744140625, -0.2739982604980469, 0.3485374450683594, 0.24607086181640625, 0.03282737731933594, 0.30692481994628906, -0.0020961761474609375, -0.047412872314453125, -0.43747711181640625, 0.14640045166015625, -0.42890167236328125, 0.37317657470703125, 0.01657867431640625, 0.4320526123046875, 0.3499298095703125, -0.4783172607421875, 0.11698532104492188, -0.21332168579101562, -0.2565460205078125, -0.480865478515625, -0.09064102172851562, -0.12726593017578125, -0.7429962158203125, 0.30043792724609375, 0.4730720520019531, -0.03544807434082031, -0.020595550537109375, -0.15222930908203125, 0.13810157775878906, 0.12583541870117188, 0.0271453857421875, 0.11470794677734375, 0.06221199035644531, -0.10790252685546875, 0.51580810546875, -0.0656280517578125, -0.10689544677734375, -0.17806243896484375, -0.1870880126953125, -0.28206634521484375, -0.1064605712890625, 0.11644744873046875, 0.08163833618164062, -0.33045196533203125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000028.npy"}
|
||||
{"epoch": 0.042328042328042326, "step": 29, "batch_size": 64, "mean": -0.04260393977165222, "std": 0.352817565202713, "min": -1.0908584594726562, "p10": -0.45645370483398434, "median": -0.018876075744628906, "p90": 0.4331722259521486, "max": 0.6273193359375, "pos_frac": 0.5, "sample": [0.16677093505859375, -0.4618072509765625, 0.590850830078125, -1.0908584594726562, 0.3600921630859375, -0.06656265258789062, -0.44396209716796875, 0.07053184509277344, -0.7496185302734375, 0.6273193359375, 0.16008567810058594, 0.20943450927734375, -0.48474884033203125, -0.22201156616210938, 0.3890190124511719, 0.202789306640625, -0.943115234375, -0.224700927734375, -0.2185821533203125, 0.5727996826171875, -0.05892753601074219, -0.3432159423828125, 0.170166015625, -0.2592201232910156, 0.10485076904296875, 0.45209503173828125, 0.0702362060546875, -0.531402587890625, 0.11224365234375, -0.07819747924804688, 0.4945259094238281, -0.2428436279296875, -0.255767822265625, 0.1803264617919922, -0.257537841796875, -0.12097549438476562, -0.29024505615234375, 0.09499740600585938, -0.08360099792480469, -0.252532958984375, -0.284088134765625, 0.06385421752929688, -0.32513427734375, -0.3546791076660156, -0.26651954650878906, -0.26917266845703125, 0.119720458984375, -0.145477294921875, 0.0657196044921875, 0.11284446716308594, 0.296051025390625, -0.05157661437988281, 0.28981781005859375, 0.014162063598632812, -0.5761871337890625, 0.031658172607421875, -0.17437744140625, 0.08647918701171875, 0.2019805908203125, 0.5843658447265625, 0.013824462890625, -0.181854248046875, 0.15788650512695312, 0.515350341796875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000029.npy"}
|
||||
{"epoch": 0.04383975812547241, "step": 30, "batch_size": 64, "mean": -0.013216495513916016, "std": 0.30666711926460266, "min": -0.796142578125, "p10": -0.36239776611328123, "median": -0.02662181854248047, "p90": 0.30652694702148436, "max": 0.950164794921875, "pos_frac": 0.453125, "sample": [-0.06266403198242188, -0.06037712097167969, -0.147369384765625, -0.6318817138671875, -0.012353897094726562, -0.14207839965820312, 0.20602035522460938, 0.3070831298828125, -0.33026885986328125, -0.37645721435546875, -0.00809478759765625, 0.30522918701171875, -0.3656158447265625, 0.34842681884765625, 0.20438385009765625, -0.2640228271484375, -0.03250694274902344, 0.92694091796875, -0.7544021606445312, 0.248504638671875, 0.2360382080078125, 0.027557373046875, -0.08457565307617188, -0.09867668151855469, -0.354888916015625, -0.16956329345703125, -0.0810394287109375, -0.41475677490234375, -0.04116058349609375, -0.21657180786132812, 0.231231689453125, 0.1888580322265625, 0.1172027587890625, 0.3609161376953125, 0.950164794921875, 0.05517578125, -0.13813018798828125, -0.1352672576904297, 0.07065582275390625, -0.11375617980957031, -0.1174163818359375, -0.0436248779296875, 0.30072021484375, 0.0381011962890625, -0.1397705078125, 0.170989990234375, 0.019805908203125, -0.08275985717773438, 0.3295135498046875, -0.045623779296875, 0.007823944091796875, -0.796142578125, 0.11006355285644531, 0.024921417236328125, -0.3082160949707031, 0.05085182189941406, -0.050079345703125, 0.0557708740234375, -0.588592529296875, -0.10989189147949219, -0.0207366943359375, 0.1518707275390625, 0.01653289794921875, 0.43212318420410156], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000030.npy"}
|
||||
{"epoch": 0.045351473922902494, "step": 31, "batch_size": 64, "mean": 0.021683931350708008, "std": 0.35979074239730835, "min": -0.921661376953125, "p10": -0.49632949829101564, "median": 0.01987171173095703, "p90": 0.45199279785156254, "max": 1.08221435546875, "pos_frac": 0.515625, "sample": [0.1498565673828125, -0.16234588623046875, -0.6212005615234375, 0.44408607482910156, -0.23751449584960938, -0.5259552001953125, -0.2977752685546875, 0.24287796020507812, 0.08054351806640625, -0.1367950439453125, -0.03739738464355469, 0.53277587890625, 0.4553813934326172, 0.218994140625, -0.02342987060546875, 0.972686767578125, 0.32868194580078125, -0.00099945068359375, 0.5214767456054688, -0.5017356872558594, 0.2797088623046875, 0.16321945190429688, 0.03960418701171875, -0.2244873046875, 0.08745765686035156, 0.24517822265625, 0.5616455078125, -0.39962005615234375, 0.06948089599609375, -0.013153076171875, 0.181396484375, -0.23646163940429688, 0.3177032470703125, 0.0073719024658203125, -0.1373577117919922, 0.03237152099609375, -0.267059326171875, -0.5085716247558594, -0.4837150573730469, -0.14081764221191406, 0.381622314453125, -0.5772476196289062, 0.051296234130859375, 0.0573883056640625, 1.08221435546875, -0.029994964599609375, -0.07574653625488281, -0.48134422302246094, 0.054492950439453125, 0.3646240234375, -0.921661376953125, -0.017246246337890625, -0.035186767578125, -0.03429603576660156, 0.062469482421875, 0.11200332641601562, 0.11085128784179688, -0.506195068359375, -0.07840347290039062, 0.09398651123046875, 0.5408706665039062, -0.02884674072265625, 0.2900199890136719, -0.00400543212890625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000031.npy"}
|
||||
{"epoch": 0.04686318972033258, "step": 32, "batch_size": 64, "mean": -0.03526049852371216, "std": 0.3221476972103119, "min": -0.8321762084960938, "p10": -0.39241561889648435, "median": -0.07816219329833984, "p90": 0.3434183120727539, "max": 0.912384033203125, "pos_frac": 0.40625, "sample": [-0.281219482421875, -0.45787811279296875, -0.5967330932617188, -0.0456390380859375, -0.1452178955078125, 0.09066390991210938, -0.1833648681640625, -0.5397109985351562, 0.47296142578125, -0.06385040283203125, 0.38512420654296875, 0.2607421875, 0.10850143432617188, 0.17794418334960938, 0.34456443786621094, -0.348907470703125, -0.3057403564453125, 0.2687721252441406, -0.27680206298828125, -0.15161514282226562, -0.19075965881347656, -0.019601821899414062, -0.10885810852050781, 0.3407440185546875, 0.10617828369140625, -0.06776237487792969, 0.14337158203125, -0.1681365966796875, -0.12109375, 0.912384033203125, -0.10882759094238281, 0.2644462585449219, 0.13881683349609375, 0.0111236572265625, -0.00662994384765625, -0.2086029052734375, -0.08856201171875, -0.2625541687011719, 0.366363525390625, 0.6353836059570312, 0.27547264099121094, 0.3153800964355469, -0.57421875, -0.3135185241699219, -0.02854156494140625, 0.12976837158203125, -0.1063079833984375, -0.8321762084960938, -0.25916290283203125, 0.23698997497558594, -0.718658447265625, 0.3369407653808594, -0.18596267700195312, -0.11817359924316406, -0.4054298400878906, -0.188446044921875, -0.27301025390625, 0.259918212890625, -0.3620491027832031, 0.1630687713623047, -0.1810150146484375, 0.17755889892578125, 0.36871337890625, -0.2538299560546875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000032.npy"}
|
||||
{"epoch": 0.04837490551776266, "step": 33, "batch_size": 64, "mean": -0.014193922281265259, "std": 0.3771737217903137, "min": -1.0253524780273438, "p10": -0.48163833618164065, "median": 0.03386878967285156, "p90": 0.35488719940185554, "max": 1.1548614501953125, "pos_frac": 0.546875, "sample": [-0.3147735595703125, -0.7468414306640625, 0.281341552734375, -0.5304412841796875, 0.07842826843261719, -0.24408721923828125, -0.445465087890625, 0.34131813049316406, 0.29134368896484375, 0.29048919677734375, -0.0955047607421875, 0.12144088745117188, -0.27606201171875, -0.3049297332763672, 0.37180328369140625, -0.4832725524902344, -0.8383636474609375, 0.23154830932617188, 0.1569976806640625, 0.095001220703125, 0.3607025146484375, 1.1548614501953125, 0.18268585205078125, 0.26425743103027344, -1.0253524780273438, 0.5277252197265625, 0.08313751220703125, -0.04901123046875, -0.1464996337890625, -0.0168304443359375, 0.1626415252685547, -0.04522514343261719, 0.19145965576171875, 0.036159515380859375, -0.2201213836669922, 0.04283714294433594, 0.15165328979492188, 0.31847381591796875, 0.017131805419921875, -0.2711372375488281, 0.231658935546875, 0.03157806396484375, -0.0840606689453125, 0.0939788818359375, -0.14734649658203125, -0.5351409912109375, 0.4716033935546875, 0.10399818420410156, 0.09906768798828125, -0.034717559814453125, -0.16420936584472656, 0.006031036376953125, 0.22652816772460938, -0.4778251647949219, -0.16048049926757812, -0.0398712158203125, 0.04579925537109375, 0.38039398193359375, -0.885955810546875, -0.4031219482421875, -0.4281959533691406, 0.29810333251953125, -0.028348922729492188, 0.7926025390625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000033.npy"}
|
||||
{"epoch": 0.049886621315192746, "step": 34, "batch_size": 64, "mean": 0.08382615447044373, "std": 0.25981155037879944, "min": -0.5959930419921875, "p10": -0.2872859954833984, "median": 0.07547283172607422, "p90": 0.4051456451416016, "max": 0.6269245147705078, "pos_frac": 0.671875, "sample": [-0.0281829833984375, 0.072479248046875, 0.45352935791015625, 0.3913002014160156, 0.10052871704101562, 0.3148651123046875, -0.037570953369140625, -0.2668876647949219, 0.19496536254882812, 0.04724693298339844, 0.4234771728515625, -0.5959930419921875, 0.002716064453125, -0.0762939453125, -0.2566986083984375, 0.151763916015625, 0.41107940673828125, 0.315216064453125, 0.1869373321533203, 0.2913055419921875, -0.13748550415039062, 0.07846641540527344, 0.3617668151855469, 0.014984130859375, -0.3018951416015625, 0.19649505615234375, 0.362701416015625, 0.11214828491210938, -0.00766754150390625, -0.2253265380859375, -0.16478729248046875, 0.18371963500976562, -0.055522918701171875, 0.03992462158203125, 0.2290496826171875, 0.21270751953125, -0.3434906005859375, 0.5227985382080078, 0.19488143920898438, 0.2857093811035156, 0.25890350341796875, -0.053730010986328125, 0.03478813171386719, 0.048770904541015625, -0.117645263671875, 0.56170654296875, 0.04213714599609375, 0.199127197265625, 0.10849380493164062, 0.6269245147705078, 0.20617103576660156, -0.3155059814453125, 0.37012481689453125, 0.6154022216796875, -0.4187812805175781, -0.3775596618652344, -0.08954620361328125, -0.12036895751953125, 0.008089065551757812, 0.05408287048339844, 0.11342620849609375, 0.00698089599609375, 0.24394989013671875, -0.29602813720703125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000034.npy"}
|
||||
{"epoch": 0.05139833711262283, "step": 35, "batch_size": 64, "mean": 0.009958773851394653, "std": 0.32427510619163513, "min": -0.6395950317382812, "p10": -0.37060852050781246, "median": 0.06895637512207031, "p90": 0.3670246124267578, "max": 1.0732269287109375, "pos_frac": 0.5625, "sample": [0.1115264892578125, -0.23289108276367188, 0.5588359832763672, 0.0788116455078125, -0.20445823669433594, 1.0732269287109375, -0.2117919921875, 0.146636962890625, -0.6395950317382812, -0.42440223693847656, -0.097320556640625, 0.09001922607421875, 0.23943710327148438, 0.1548290252685547, -0.4412078857421875, 0.08834075927734375, 0.718170166015625, -0.12713623046875, 0.47692108154296875, -0.5473899841308594, 0.07416152954101562, 0.390777587890625, 0.2972431182861328, -0.31396484375, -0.10068893432617188, -0.08934402465820312, -0.6292877197265625, -0.117218017578125, 0.02796173095703125, -0.2739715576171875, -0.33892822265625, 0.3597412109375, -0.3361663818359375, -0.2962532043457031, -0.27158355712890625, -0.2643585205078125, 0.3307323455810547, -0.601348876953125, 0.13147544860839844, 0.3609123229980469, -0.2680511474609375, 0.16707801818847656, 0.3694038391113281, 0.11675643920898438, 0.16358566284179688, 0.38079071044921875, 0.17639923095703125, -0.06085968017578125, 0.063751220703125, -0.0306243896484375, 0.044826507568359375, -0.33940887451171875, 0.013088226318359375, 0.36147308349609375, -0.22844696044921875, 0.2176055908203125, 0.194854736328125, -0.38397979736328125, 0.17481613159179688, 0.1092681884765625, 0.10252952575683594, 0.14501953125, 0.09050559997558594, -0.09347343444824219], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000035.npy"}
|
||||
{"epoch": 0.05291005291005291, "step": 36, "batch_size": 64, "mean": -0.0021797120571136475, "std": 0.36727437376976013, "min": -1.4093017578125, "p10": -0.32947845458984376, "median": -0.05300140380859375, "p90": 0.47447128295898444, "max": 0.9419975280761719, "pos_frac": 0.40625, "sample": [-0.2009735107421875, 0.7129669189453125, 0.45948028564453125, -0.2792835235595703, -0.29096221923828125, -0.13023757934570312, -0.23413848876953125, 0.07375717163085938, 0.19905471801757812, -0.0471038818359375, -0.39998626708984375, -0.20755386352539062, 0.2543182373046875, -0.265167236328125, -0.04131317138671875, 0.082611083984375, 0.01845550537109375, -0.24222564697265625, -0.3374671936035156, 0.7382011413574219, -0.09490776062011719, -0.3482666015625, 0.1992645263671875, -0.0149688720703125, -0.10460662841796875, -0.371368408203125, 0.06521224975585938, 0.29979705810546875, -0.33465576171875, -1.4093017578125, 0.0069427490234375, -0.03840065002441406, -0.061065673828125, 0.2945404052734375, 0.6255569458007812, 0.2563743591308594, -0.21221160888671875, -0.3088226318359375, -0.5577735900878906, -0.1850738525390625, -0.05889892578125, 0.39794921875, 0.48089599609375, -0.2113971710205078, -0.00072479248046875, -0.0048046112060546875, 0.3741111755371094, 0.6862602233886719, 0.01357269287109375, 0.9419975280761719, 0.01943206787109375, -0.12726211547851562, 0.10301971435546875, 0.24224090576171875, -0.15283584594726562, -0.3173980712890625, -0.12646484375, -0.09027862548828125, -0.05914306640625, -0.18058013916015625, -0.22165298461914062, 0.7902374267578125, -0.3029136657714844, 0.09643936157226562], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000036.npy"}
|
||||
{"epoch": 0.05442176870748299, "step": 37, "batch_size": 64, "mean": 0.10792455077171326, "std": 0.3912637531757355, "min": -1.2143402099609375, "p10": -0.3118282318115234, "median": 0.08628082275390625, "p90": 0.5646957397460939, "max": 1.37799072265625, "pos_frac": 0.609375, "sample": [0.21379852294921875, 0.4566650390625, 0.6186981201171875, -0.14382171630859375, 0.2838134765625, 0.26982879638671875, -0.10281181335449219, -0.15747451782226562, 0.020486831665039062, -0.0128326416015625, 0.57562255859375, 0.17741775512695312, 0.1110687255859375, 0.60015869140625, -0.19916152954101562, -0.11496734619140625, 0.07647705078125, 0.32611846923828125, 0.205902099609375, 0.481597900390625, -0.29283714294433594, -0.31996726989746094, -0.507965087890625, 0.3155059814453125, -0.16267013549804688, 0.2773323059082031, 0.11927032470703125, -0.017702102661132812, 0.23458099365234375, 0.017242431640625, -0.2135009765625, -0.536468505859375, 0.3798065185546875, 0.7601547241210938, 0.83697509765625, 0.46015167236328125, 0.02153778076171875, 0.14648056030273438, -0.141632080078125, 0.06397628784179688, 1.37799072265625, -0.1214752197265625, -0.2729644775390625, 0.0960845947265625, -1.2143402099609375, -0.23365402221679688, 0.3361053466796875, -0.0096435546875, 0.3133392333984375, -0.3633880615234375, 0.3112525939941406, -0.1396942138671875, 0.029409408569335938, 0.12440109252929688, 0.036163330078125, 0.37871551513671875, 0.338531494140625, 0.28806304931640625, -0.321258544921875, -0.38488006591796875, 0.5391998291015625, -0.04769134521484375, -0.2056560516357422, 0.9257049560546875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000037.npy"}
|
||||
{"epoch": 0.055933484504913075, "step": 38, "batch_size": 64, "mean": 0.018855631351470947, "std": 0.3156125843524933, "min": -0.6601486206054688, "p10": -0.4031532287597656, "median": 0.033379554748535156, "p90": 0.36588592529296887, "max": 1.0634193420410156, "pos_frac": 0.53125, "sample": [0.03872108459472656, 0.09036445617675781, 0.15612030029296875, 1.0634193420410156, -0.14625167846679688, 0.3397064208984375, 0.09105682373046875, 0.4549598693847656, -0.0208282470703125, 0.2440338134765625, 0.2522735595703125, -0.3141746520996094, 0.09018898010253906, 0.011976242065429688, 0.05838775634765625, -0.21734619140625, 0.1751556396484375, 0.08243560791015625, 0.12383270263671875, -0.21245574951171875, -0.1131439208984375, 0.4589996337890625, -0.6601486206054688, -0.4053802490234375, 0.09502410888671875, 0.223785400390625, -0.03534698486328125, -0.5628166198730469, -0.27117919921875, -0.0751800537109375, -0.6160659790039062, -0.42420196533203125, -0.009799957275390625, -0.21107864379882812, 0.32598114013671875, 0.2389678955078125, -0.09522247314453125, -0.5616264343261719, -0.11701202392578125, -0.05068206787109375, 0.02803802490234375, -0.410675048828125, 0.26940155029296875, -0.18304443359375, 0.14198684692382812, -0.1276092529296875, -0.20641708374023438, -0.2884864807128906, 0.5640068054199219, 0.52978515625, 0.532012939453125, 0.377105712890625, 0.09253692626953125, -0.19415283203125, 0.22748374938964844, 0.29783058166503906, -0.01802825927734375, -0.25217437744140625, -0.39795684814453125, 0.0688323974609375, 0.31050872802734375, 0.22723388671875, 0.2644481658935547, -0.1413555145263672], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000038.npy"}
|
||||
{"epoch": 0.05744520030234316, "step": 39, "batch_size": 64, "mean": 0.08394494652748108, "std": 0.3936171531677246, "min": -1.160125732421875, "p10": -0.35462036132812497, "median": 0.0870513916015625, "p90": 0.5681427001953125, "max": 0.934295654296875, "pos_frac": 0.625, "sample": [0.60430908203125, 0.19525146484375, 0.12377166748046875, 0.2197265625, 0.0948028564453125, -0.0550537109375, 0.5923309326171875, 0.23101043701171875, 0.7357215881347656, 0.36794471740722656, -0.13401031494140625, 0.11919975280761719, 0.4750556945800781, 0.025760650634765625, -0.3694610595703125, 0.10745048522949219, -0.2115459442138672, 0.21560096740722656, 0.5587940216064453, 0.0636444091796875, -0.8346290588378906, -0.4808349609375, -0.2063426971435547, -0.0697479248046875, -0.007537841796875, -0.08611679077148438, -1.160125732421875, 0.21177291870117188, 0.16849517822265625, -0.1859588623046875, 0.1493206024169922, -0.2249908447265625, -0.0230865478515625, 0.21389389038085938, 0.030895233154296875, 0.2956390380859375, -0.09095001220703125, 0.5703125, 0.3251075744628906, -0.705596923828125, 0.934295654296875, -0.90838623046875, 0.051830291748046875, 0.5306854248046875, 0.563079833984375, -0.39534759521484375, 0.2559051513671875, 0.473663330078125, -0.08842849731445312, -0.028942108154296875, 0.73236083984375, -0.3199920654296875, 0.011707305908203125, -0.019975662231445312, 0.031084060668945312, 0.487152099609375, 0.3118858337402344, 0.0792999267578125, 0.4571533203125, -0.3071556091308594, 0.5919647216796875, 0.10260963439941406, 0.06510543823242188, -0.08890151977539062], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000039.npy"}
|
||||
{"epoch": 0.05895691609977324, "step": 40, "batch_size": 64, "mean": 0.07392571866512299, "std": 0.3782210350036621, "min": -1.0976104736328125, "p10": -0.2720659255981445, "median": 0.09836578369140625, "p90": 0.4786277770996094, "max": 0.9683151245117188, "pos_frac": 0.578125, "sample": [0.29265594482421875, 0.8278121948242188, 0.23112106323242188, 0.2761821746826172, -0.3775463104248047, -0.18523025512695312, 0.30957794189453125, 0.48320770263671875, 0.0358734130859375, 0.5265312194824219, 0.36260223388671875, 0.16550254821777344, 0.18790435791015625, -0.16114425659179688, 0.3184051513671875, -0.0724639892578125, 0.2647857666015625, 0.6403045654296875, 0.6251049041748047, -0.0670013427734375, 0.20699310302734375, 0.4679412841796875, -0.0468292236328125, 0.19648361206054688, 0.4654045104980469, -0.085662841796875, -0.033954620361328125, -0.029510498046875, 0.9683151245117188, -0.03574562072753906, 0.33478546142578125, -0.503692626953125, 0.8013916015625, -0.17549896240234375, 0.17856597900390625, 0.14035701751708984, -0.16173553466796875, 0.4188270568847656, -0.7810745239257812, -0.1558074951171875, -0.25876808166503906, 0.0999755859375, 0.24651527404785156, -1.0976104736328125, -0.263885498046875, 0.20577621459960938, 0.08182525634765625, -0.26564788818359375, -0.18737411499023438, -0.02874755859375, 0.40865325927734375, 0.329681396484375, -0.10840225219726562, 0.1687774658203125, -0.0495452880859375, -0.7456169128417969, -0.759552001953125, 0.027496337890625, 0.0377655029296875, -0.18842315673828125, 0.22616195678710938, -0.27481651306152344, 0.0967559814453125, 0.176513671875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000040.npy"}
|
||||
{"epoch": 0.06046863189720333, "step": 41, "batch_size": 64, "mean": 0.11904790997505188, "std": 0.40857017040252686, "min": -0.912322998046875, "p10": -0.28691520690917965, "median": 0.07705116271972656, "p90": 0.606488037109375, "max": 1.1236114501953125, "pos_frac": 0.5625, "sample": [-0.16049957275390625, 0.345916748046875, 0.002338409423828125, -0.6947021484375, -0.2309417724609375, 0.35552978515625, -0.08431053161621094, 0.5036544799804688, 0.10404205322265625, -0.08346939086914062, 0.443878173828125, 0.22278976440429688, -0.05532073974609375, 0.540069580078125, 0.16085052490234375, 0.5459136962890625, -0.2043628692626953, 0.3730010986328125, -0.0776519775390625, -0.09804344177246094, 0.6471405029296875, -0.624908447265625, -0.12085723876953125, 0.341827392578125, 0.1871776580810547, -0.23957061767578125, 0.0537872314453125, 0.3234233856201172, -0.0506591796875, 0.1790313720703125, -0.2790374755859375, 0.152801513671875, 0.836273193359375, 0.6101455688476562, -0.2796440124511719, 0.15810394287109375, 0.5696678161621094, -0.4776115417480469, 0.3761138916015625, -0.03960418701171875, -0.0166473388671875, -0.912322998046875, -0.20123672485351562, -0.124725341796875, 1.1236114501953125, -0.3944110870361328, 0.4268379211425781, 0.09065628051757812, 0.7972259521484375, 0.8626861572265625, -0.29003143310546875, -0.828460693359375, -0.04102325439453125, -0.04376983642578125, 0.3975257873535156, 0.7682037353515625, 0.2570037841796875, 0.00040435791015625, 0.5606975555419922, -0.0684814453125, 0.385894775390625, 0.5979537963867188, -0.02425384521484375, 0.063446044921875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000041.npy"}
|
||||
{"epoch": 0.06198034769463341, "step": 42, "batch_size": 64, "mean": 0.053393036127090454, "std": 0.44279181957244873, "min": -1.0146102905273438, "p10": -0.39068527221679683, "median": 0.040065765380859375, "p90": 0.545355224609375, "max": 1.625213623046875, "pos_frac": 0.578125, "sample": [-0.0053253173828125, 1.625213623046875, -0.24353790283203125, 0.1114044189453125, 0.017232894897460938, -0.509796142578125, -0.07386016845703125, 0.019300460815429688, -0.21160888671875, 0.15562820434570312, -0.3414154052734375, 0.05218505859375, -0.18406295776367188, -0.34696197509765625, -0.23198699951171875, 0.5363082885742188, 0.986419677734375, 0.61895751953125, 0.2303619384765625, 0.5531082153320312, 0.49489593505859375, 0.44043731689453125, 0.26229095458984375, 0.19852828979492188, -0.5596199035644531, -0.39868927001953125, 0.14421844482421875, 0.440582275390625, -0.2144927978515625, -0.3553943634033203, 0.13900375366210938, 0.5510883331298828, 0.10193252563476562, -0.1630859375, 0.8045082092285156, 0.5492324829101562, 0.32228851318359375, 0.02794647216796875, -0.8135452270507812, -0.13558197021484375, -0.37200927734375, -0.26316070556640625, -0.7396392822265625, -0.5628280639648438, 0.0053558349609375, -0.1123504638671875, 0.15460968017578125, -0.32135009765625, -0.3434867858886719, -0.2772674560546875, 0.1955585479736328, 0.395751953125, -0.27325439453125, 0.5159072875976562, 0.3966522216796875, 0.28143310546875, -1.0146102905273438, 0.0937042236328125, 0.4543914794921875, 0.49008941650390625, 0.13344383239746094, 0.28227806091308594, 0.00603485107421875, -0.30220794677734375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000042.npy"}
|
||||
{"epoch": 0.06349206349206349, "step": 43, "batch_size": 64, "mean": 0.04106879234313965, "std": 0.3837270736694336, "min": -1.1148452758789062, "p10": -0.42020092010498045, "median": 0.06722068786621094, "p90": 0.5349891662597657, "max": 0.927581787109375, "pos_frac": 0.578125, "sample": [0.09682464599609375, 0.20436859130859375, 0.0665283203125, -0.08708763122558594, -0.22457504272460938, -0.3546314239501953, -0.040836334228515625, 0.044315338134765625, 0.6597442626953125, -0.7556304931640625, 0.11967849731445312, -0.352691650390625, -0.434814453125, -0.45931243896484375, 0.927581787109375, 0.5933990478515625, 0.08438873291015625, -0.157470703125, 0.1406230926513672, 0.45482635498046875, 0.27171897888183594, 0.18079376220703125, -0.21192169189453125, -0.040210723876953125, 0.14247894287109375, -0.06362342834472656, 0.06610679626464844, 0.5365524291992188, 0.08746337890625, 0.154205322265625, 0.186859130859375, 0.0626373291015625, 0.002841949462890625, -0.44191741943359375, 0.06791305541992188, 0.531341552734375, 0.9081497192382812, -0.3577594757080078, 0.6369552612304688, 0.304718017578125, -0.03572845458984375, -0.056427001953125, -0.02532196044921875, -1.1148452758789062, 0.12750816345214844, 0.48101806640625, -0.1331634521484375, 0.36626625061035156, -0.1337261199951172, 0.40296173095703125, 0.12479972839355469, 0.104034423828125, 0.1184539794921875, -0.36600494384765625, 0.36475372314453125, -0.0649261474609375, -0.561981201171875, 0.4580726623535156, 0.5721435546875, 0.0711669921875, -0.70849609375, -0.1856842041015625, -0.38610267639160156, -0.3409004211425781], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000043.npy"}
|
||||
{"epoch": 0.06500377928949358, "step": 44, "batch_size": 64, "mean": 0.03525695204734802, "std": 0.4844135344028473, "min": -1.0936660766601562, "p10": -0.611016845703125, "median": 0.059253692626953125, "p90": 0.5516830444335942, "max": 1.285369873046875, "pos_frac": 0.515625, "sample": [-0.44145965576171875, -0.035533905029296875, -0.2034912109375, 0.06628036499023438, 0.7785110473632812, -1.0469970703125, 0.0776824951171875, -0.5768585205078125, -0.6952018737792969, -0.7460098266601562, 0.21892547607421875, 1.2141952514648438, -0.474090576171875, 0.34316253662109375, 0.3036155700683594, -0.12113189697265625, -0.000438690185546875, 0.076690673828125, 0.052227020263671875, 0.18283653259277344, 0.09075546264648438, 0.3044281005859375, 0.2736644744873047, -0.7816696166992188, -0.13867950439453125, -0.0841064453125, -0.25946044921875, 0.15535736083984375, -0.3315582275390625, 0.4234161376953125, 0.77349853515625, -0.2909431457519531, -0.07859039306640625, -0.08470916748046875, -0.10126495361328125, 0.09642982482910156, 1.0657196044921875, 0.192138671875, 0.336517333984375, -0.799713134765625, -0.1405048370361328, 0.10766220092773438, -0.1837615966796875, -0.09749603271484375, 0.42946624755859375, 0.255523681640625, 0.191680908203125, 0.5928955078125, -0.12258148193359375, 0.11993026733398438, -0.6256561279296875, 0.4555206298828125, 0.3288726806640625, -0.0032958984375, -0.1754913330078125, 0.3233489990234375, 0.9430007934570312, 1.285369873046875, 0.29438209533691406, 0.4335174560546875, -0.3033905029296875, -0.00589752197265625, -0.48712921142578125, -1.0936660766601562], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000044.npy"}
|
||||
{"epoch": 0.06651549508692366, "step": 45, "batch_size": 64, "mean": 0.08534523844718933, "std": 0.41353312134742737, "min": -0.9718437194824219, "p10": -0.38171920776367185, "median": 0.15151405334472656, "p90": 0.502343559265137, "max": 1.179840087890625, "pos_frac": 0.609375, "sample": [0.18524169921875, -0.39337158203125, 0.19233131408691406, 0.2837982177734375, -0.07878875732421875, -0.47400665283203125, -0.1736125946044922, -0.1427154541015625, 0.1662750244140625, -0.907196044921875, 0.20272064208984375, 0.5452423095703125, 0.3789634704589844, 0.20774078369140625, -0.033130645751953125, 0.2687492370605469, 0.44104576110839844, -0.9286346435546875, -0.27081298828125, 0.344207763671875, 0.0364227294921875, -0.07605743408203125, 0.5286140441894531, 0.07995033264160156, 0.24971389770507812, 1.179840087890625, -0.1209869384765625, 0.08908843994140625, 0.22415924072265625, 0.40189361572265625, 0.13719558715820312, 0.5377197265625, -0.4241943359375, 0.242919921875, 0.2923736572265625, 1.0130615234375, 0.2324085235595703, -0.0117950439453125, -0.35453033447265625, 0.1240081787109375, -0.9718437194824219, 0.8626861572265625, -0.1629314422607422, -0.2901115417480469, -0.2049713134765625, 0.08688735961914062, -0.90869140625, 0.24271392822265625, 0.36553955078125, 0.38304901123046875, -0.15894317626953125, 0.06728935241699219, 0.36554718017578125, -0.0054473876953125, -0.006565093994140625, 0.32379913330078125, 0.8769073486328125, 0.22690582275390625, 0.22526168823242188, 0.20444488525390625, -0.19573211669921875, -0.07464599609375, 0.16583251953125, -0.15073776245117188], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000045.npy"}
|
||||
{"epoch": 0.06802721088435375, "step": 46, "batch_size": 64, "mean": 0.1240440309047699, "std": 0.38512948155403137, "min": -0.7017822265625, "p10": -0.3130775451660156, "median": 0.09075355529785156, "p90": 0.6772949218750004, "max": 1.0634613037109375, "pos_frac": 0.59375, "sample": [0.30007171630859375, -0.2681427001953125, -0.46868133544921875, 0.147369384765625, 1.0634613037109375, 0.37396240234375, 0.05022430419921875, -0.6006546020507812, 0.19775390625, -0.7017822265625, 0.44698143005371094, 0.09469985961914062, 0.2613677978515625, 0.4055519104003906, -0.08866119384765625, -0.29512786865234375, -0.5203781127929688, 0.17144393920898438, 0.1135101318359375, 0.2914142608642578, 0.37999725341796875, 0.3046875, -0.03743934631347656, 0.207122802734375, 0.5415687561035156, 0.059173583984375, 0.4892425537109375, 0.8283767700195312, -0.08032989501953125, 0.0764007568359375, -0.22803497314453125, 0.38611602783203125, -0.15969085693359375, -0.01947784423828125, -0.22327041625976562, 0.891845703125, -0.1598644256591797, -0.018627166748046875, -0.18922805786132812, 0.889495849609375, 0.5789642333984375, -0.08847808837890625, 0.451171875, 0.09688568115234375, 0.7275276184082031, -0.02948760986328125, -0.06841659545898438, 0.01314544677734375, -0.16241073608398438, 0.09105300903320312, 0.38120269775390625, 0.0904541015625, -0.19855690002441406, 0.442718505859375, 0.127593994140625, 0.9057464599609375, -0.5178985595703125, -0.320770263671875, -0.41997718811035156, 0.7194366455078125, -0.0125274658203125, 0.28873634338378906, -0.15821456909179688, 0.088470458984375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000046.npy"}
|
||||
{"epoch": 0.06953892668178382, "step": 47, "batch_size": 64, "mean": 0.16103604435920715, "std": 0.380586713552475, "min": -1.0150604248046875, "p10": -0.3187801361083984, "median": 0.22018814086914062, "p90": 0.5915145874023439, "max": 0.9197158813476562, "pos_frac": 0.671875, "sample": [0.30376243591308594, 0.26682281494140625, 0.5672874450683594, 0.34937286376953125, 0.20475006103515625, 0.1613178253173828, 0.040256500244140625, 0.6992568969726562, 0.07013702392578125, 0.5702590942382812, -0.24122047424316406, -0.2605171203613281, -0.07677459716796875, -0.4875640869140625, 0.267425537109375, 0.235626220703125, 0.43927001953125, -0.43392181396484375, 0.0188140869140625, -0.17438507080078125, -0.006683349609375, 0.1654815673828125, -0.13158416748046875, -0.015085220336914062, 0.1672515869140625, 0.7539825439453125, 0.5066375732421875, 0.3917236328125, -0.15935134887695312, 0.3224296569824219, 0.03242683410644531, 0.7286758422851562, -0.07332611083984375, 0.2739753723144531, -0.34375, 0.42995452880859375, 0.5974311828613281, 0.5229415893554688, 0.26678466796875, 0.7642974853515625, -0.2367095947265625, 0.028472900390625, 0.12790679931640625, 0.24108123779296875, 0.7848167419433594, -0.17961502075195312, 0.3647613525390625, -0.13934707641601562, 0.3917198181152344, 0.39211463928222656, -0.2320556640625, 0.5460739135742188, -0.473236083984375, -1.0150604248046875, -0.5287303924560547, 0.9197158813476562, 0.1334819793701172, 0.4387168884277344, 0.5777091979980469, 0.2935943603515625, 0.42787933349609375, 0.3530845642089844, -0.452178955078125, -0.17207908630371094], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000047.npy"}
|
||||
{"epoch": 0.0710506424792139, "step": 48, "batch_size": 64, "mean": 0.003746122121810913, "std": 0.4235166609287262, "min": -0.8472137451171875, "p10": -0.5027669906616211, "median": -0.011738777160644531, "p90": 0.5189979553222657, "max": 1.2762069702148438, "pos_frac": 0.5, "sample": [0.250396728515625, -0.2830352783203125, 0.3400249481201172, -0.0373687744140625, 0.07989501953125, 0.9060478210449219, -0.46685791015625, 0.5036544799804688, -0.5155963897705078, 0.4940223693847656, 0.4615936279296875, -0.023828506469726562, 0.12260055541992188, -0.359588623046875, 0.11605072021484375, 0.707916259765625, 0.4273414611816406, 0.09419059753417969, -0.624664306640625, 0.52557373046875, 0.16800689697265625, -0.20699691772460938, -0.19925689697265625, 0.4902801513671875, 0.16299819946289062, 0.12474822998046875, 0.01007080078125, 0.07970809936523438, -0.47283172607421875, 0.568572998046875, -0.105560302734375, -0.0650787353515625, -0.0692901611328125, -0.37551116943359375, 0.1797027587890625, -0.2625274658203125, 0.002948760986328125, -0.26508331298828125, 0.7348747253417969, -0.27512359619140625, -0.24769210815429688, 0.04099273681640625, 0.0003509521484375, 0.14669036865234375, -0.6901931762695312, -0.24898910522460938, -0.11307334899902344, 0.08547592163085938, -0.16022491455078125, -0.20672035217285156, -0.1356353759765625, -0.54620361328125, -0.8128910064697266, 0.044147491455078125, -0.23408889770507812, 0.3801136016845703, 1.2762069702148438, -0.023942947387695312, -0.4595184326171875, -0.2762107849121094, 0.08686637878417969, 0.90203857421875, -0.6635532379150391, -0.8472137451171875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000048.npy"}
|
||||
{"epoch": 0.07256235827664399, "step": 49, "batch_size": 64, "mean": 0.07798841595649719, "std": 0.481463760137558, "min": -1.392913818359375, "p10": -0.6372604370117188, "median": 0.1187295913696289, "p90": 0.6788421630859376, "max": 1.180938720703125, "pos_frac": 0.609375, "sample": [0.1632537841796875, -0.19861984252929688, 0.0006160736083984375, 0.8762664794921875, 0.2945709228515625, -0.09878730773925781, 0.08234596252441406, 0.33443450927734375, -0.0069866180419921875, -0.17504119873046875, 0.21864891052246094, -0.23082351684570312, 0.71502685546875, 0.5837860107421875, -0.3368797302246094, -0.3382415771484375, -0.5147705078125, -0.6420669555664062, 0.23714828491210938, -0.6518096923828125, 0.38422203063964844, 0.14426040649414062, 0.16065406799316406, 0.84796142578125, 0.17859649658203125, 0.207366943359375, -0.1872100830078125, 0.09319877624511719, 0.2442951202392578, -0.7701568603515625, 0.04915428161621094, 0.27864837646484375, -0.575836181640625, 0.8829021453857422, 0.6868896484375, 0.660064697265625, -0.19983673095703125, 0.17484283447265625, 0.5064697265625, 0.5479888916015625, 0.0465240478515625, -0.06700897216796875, 0.07256889343261719, -0.0088043212890625, -0.86871337890625, -0.00103759765625, 0.08640289306640625, 0.3580894470214844, 0.39046478271484375, -0.1001739501953125, 0.22310638427734375, 0.8142471313476562, -1.392913818359375, -0.6260452270507812, -0.1418914794921875, -0.7061691284179688, 0.35883331298828125, 1.180938720703125, 0.49332427978515625, 0.3402118682861328, -0.08524322509765625, 0.22850799560546875, 0.5163803100585938, -0.74688720703125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000049.npy"}
|
||||
{"epoch": 0.07407407407407407, "step": 50, "batch_size": 64, "mean": 0.12342384457588196, "std": 0.4061659276485443, "min": -0.6492919921875, "p10": -0.30152053833007814, "median": 0.11153602600097656, "p90": 0.5523838043212892, "max": 1.1259994506835938, "pos_frac": 0.59375, "sample": [0.2839508056640625, 0.28736114501953125, 0.0668182373046875, 0.15509033203125, -0.04300689697265625, 0.35115814208984375, 0.021938323974609375, -0.4309539794921875, 1.0646286010742188, 0.370025634765625, 0.38367271423339844, 0.3823089599609375, -0.271942138671875, -0.15383529663085938, 0.08523178100585938, -0.3048095703125, 0.517608642578125, 0.5672874450683594, 0.9033908843994141, 0.4552154541015625, -0.0582122802734375, 0.1208343505859375, -0.19382858276367188, 0.11905670166015625, 0.14535903930664062, 1.099853515625, 0.10401535034179688, 0.27782440185546875, -0.19115829467773438, 0.176422119140625, 0.26396751403808594, -0.018016815185546875, 0.30078697204589844, 0.50604248046875, -0.6492919921875, -0.2180023193359375, 0.1908721923828125, 0.9086532592773438, 0.08814430236816406, -0.1770477294921875, 1.1259994506835938, 1.067047119140625, -0.09056472778320312, -0.6089935302734375, -0.11839103698730469, 0.24710655212402344, -0.08315277099609375, -0.209228515625, -0.626312255859375, -0.11428070068359375, 0.3640003204345703, -0.2006378173828125, 0.21945762634277344, 0.17624664306640625, 0.5107631683349609, -0.2930793762207031, -0.4362297058105469, 0.17596817016601562, -0.29384613037109375, -0.3360557556152344, 0.011747360229492188, -0.12645721435546875, 0.197052001953125, -0.14644622802734375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000050.npy"}
|
||||
{"epoch": 0.07558578987150416, "step": 51, "batch_size": 64, "mean": 0.16471770405769348, "std": 0.5075812935829163, "min": -1.0662994384765625, "p10": -0.5401063919067383, "median": 0.14212512969970703, "p90": 0.803820037841797, "max": 1.5289840698242188, "pos_frac": 0.65625, "sample": [0.62762451171875, 0.03494071960449219, -0.029653549194335938, 0.22788238525390625, 0.14516448974609375, 0.4123687744140625, 0.21775054931640625, -0.08519172668457031, -0.5486660003662109, 0.00290679931640625, 0.058460235595703125, 0.8714561462402344, -0.2983741760253906, 1.0738601684570312, 0.0662078857421875, 0.769805908203125, -0.02150726318359375, 0.4562225341796875, 0.15338897705078125, 0.370697021484375, 0.550994873046875, 0.37027549743652344, 0.4244842529296875, 0.6402435302734375, 0.6661300659179688, -0.6275634765625, -0.4248199462890625, -0.7287979125976562, 0.14383506774902344, 0.27175331115722656, 0.32663726806640625, 1.084320068359375, 0.0044956207275390625, 0.74951171875, -1.0662994384765625, -0.5201339721679688, 0.6023674011230469, -0.3967399597167969, -0.46863555908203125, 0.181976318359375, -0.11260223388671875, -0.07733154296875, 0.22220993041992188, 0.916107177734375, -0.0020294189453125, -0.14773941040039062, -0.729034423828125, 1.2000732421875, -0.10992431640625, 0.8183975219726562, 0.46042823791503906, 0.5865249633789062, 0.5293655395507812, -0.55596923828125, 0.210052490234375, -0.09091949462890625, -0.10214614868164062, 1.5289840698242188, 0.050201416015625, 0.0072479248046875, -0.6313743591308594, 0.0792999267578125, 0.14041519165039062, 0.06231689453125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000051.npy"}
|
||||
{"epoch": 0.07709750566893424, "step": 52, "batch_size": 64, "mean": 0.18324412405490875, "std": 0.5819587707519531, "min": -1.05975341796875, "p10": -0.45673370361328125, "median": 0.0835428237915039, "p90": 0.9523941040039064, "max": 1.58306884765625, "pos_frac": 0.546875, "sample": [1.3942337036132812, -0.4048004150390625, 0.556243896484375, 0.894683837890625, 0.5645599365234375, -0.258453369140625, 0.34203529357910156, -0.1511993408203125, -0.21327781677246094, 1.4570541381835938, 1.383544921875, 0.5616531372070312, 0.02396392822265625, 1.2738571166992188, 0.9771270751953125, -0.6730728149414062, 0.21015167236328125, -0.24356842041015625, -0.3839073181152344, -0.18861961364746094, -0.46593475341796875, -0.13940811157226562, 0.46891021728515625, -0.041622161865234375, 0.706756591796875, 0.8457107543945312, -0.395172119140625, -0.03762054443359375, 0.3575286865234375, -0.13997268676757812, 0.23693275451660156, -1.05975341796875, 0.1645050048828125, -0.43526458740234375, 0.13958168029785156, 1.190521240234375, 0.1782989501953125, 0.3533458709716797, 0.4149017333984375, -0.8437347412109375, -0.13526344299316406, 0.5739517211914062, 0.5725498199462891, 0.24591827392578125, -0.08703041076660156, 0.41396331787109375, -0.1023111343383789, -0.168426513671875, 1.58306884765625, -0.559173583984375, 0.6780509948730469, -0.4915008544921875, -0.12998199462890625, 0.8079719543457031, 0.02750396728515625, -0.8209609985351562, 0.681915283203125, -0.19935226440429688, -0.1767578125, 0.2758331298828125, 0.26259613037109375, -0.08762931823730469, -0.08275032043457031, 0.02471923828125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000052.npy"}
|
||||
{"epoch": 0.07860922146636433, "step": 53, "batch_size": 64, "mean": 0.16743558645248413, "std": 0.4840546250343323, "min": -1.6668014526367188, "p10": -0.3371061325073242, "median": 0.13730716705322266, "p90": 0.7288795471191407, "max": 1.1843109130859375, "pos_frac": 0.640625, "sample": [0.044116973876953125, 0.47054290771484375, 0.8933982849121094, 0.25327301025390625, 0.44347572326660156, 0.25972938537597656, 0.30950927734375, 0.42929840087890625, 0.07390403747558594, -0.07137680053710938, -0.0623779296875, 0.4027862548828125, -0.31980133056640625, 0.16570663452148438, -0.82391357421875, -0.043483734130859375, 0.31278419494628906, 0.611419677734375, -0.13741493225097656, -0.5902252197265625, 0.09380912780761719, -0.117523193359375, 0.540069580078125, 0.6987419128417969, -1.6668014526367188, 0.48709869384765625, -0.32598304748535156, 0.739410400390625, 0.4816417694091797, 0.7043075561523438, 0.4316902160644531, -0.7390823364257812, -0.08789443969726562, -0.08896636962890625, 0.06821441650390625, 0.8953704833984375, 0.3025684356689453, 1.1843109130859375, -0.06368064880371094, -0.03846931457519531, -0.08029556274414062, -0.3418731689453125, 0.5203685760498047, -0.4431304931640625, 0.9057388305664062, 0.87841796875, 0.36481475830078125, 0.45090484619140625, 0.5071640014648438, 0.9434814453125, 0.01830291748046875, 0.5782241821289062, -0.2593555450439453, 0.4200859069824219, 0.23325347900390625, -0.24554824829101562, 0.6275787353515625, -0.5126609802246094, 0.0850830078125, 0.10890769958496094, 0.0446319580078125, -0.1938323974609375, -0.06184196472167969, 0.04727363586425781], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000053.npy"}
|
||||
{"epoch": 0.0801209372637944, "step": 54, "batch_size": 64, "mean": 0.13746550679206848, "std": 0.5660649538040161, "min": -1.711181640625, "p10": -0.48796691894531247, "median": 0.1209726333618164, "p90": 0.6741554260253909, "max": 2.3835906982421875, "pos_frac": 0.59375, "sample": [0.4616584777832031, 1.573944091796875, -0.44671630859375, 0.48430442810058594, -0.0689849853515625, -0.02301025390625, 0.5673255920410156, -0.08856964111328125, 0.8782196044921875, -0.5702438354492188, -0.1818084716796875, 0.0002727508544921875, 0.282928466796875, -1.711181640625, 0.2784099578857422, -0.07076072692871094, 0.277252197265625, 0.1934528350830078, 0.05753326416015625, 0.7652511596679688, 0.2615203857421875, 2.3835906982421875, 0.0282440185546875, 0.5591373443603516, -0.1881256103515625, 0.576995849609375, 0.4349250793457031, -0.35607147216796875, -0.11090850830078125, -0.07653045654296875, -0.61444091796875, -0.051425933837890625, -0.505645751953125, -0.3410625457763672, 0.3233203887939453, 0.15024566650390625, -0.212127685546875, 0.064971923828125, 1.00390625, 0.9396018981933594, 0.27088356018066406, -0.10385894775390625, 0.5955314636230469, 0.49469757080078125, -0.24810028076171875, 0.09559440612792969, 0.46562957763671875, 0.40076255798339844, 0.0011501312255859375, 0.4634590148925781, 0.22360992431640625, -0.5686111450195312, 0.14804840087890625, 0.14635086059570312, 0.43670082092285156, -0.29497528076171875, 0.20229148864746094, -0.21369361877441406, 0.7078514099121094, -0.30071258544921875, -0.5784835815429688, -0.07472991943359375, 0.35152435302734375, -0.7525253295898438], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000054.npy"}
|
||||
{"epoch": 0.08163265306122448, "step": 55, "batch_size": 64, "mean": 0.274186372756958, "std": 0.6704981327056885, "min": -1.0438461303710938, "p10": -0.4537628173828125, "median": 0.2230548858642578, "p90": 1.1565803527832037, "max": 2.353759765625, "pos_frac": 0.640625, "sample": [1.692230224609375, 0.5113906860351562, 0.3645896911621094, 0.35500335693359375, -0.19318389892578125, -0.19659423828125, 0.6916351318359375, 0.19373321533203125, -0.5179004669189453, -0.21424484252929688, 0.3274726867675781, 0.8170623779296875, -1.0438461303710938, -0.04257965087890625, 0.27910614013671875, 1.3638916015625, 0.7847824096679688, 0.2187652587890625, -0.057708740234375, -0.6115760803222656, 0.09534454345703125, -0.547149658203125, 1.0193328857421875, 0.44147682189941406, 1.88616943359375, 0.2643241882324219, 0.22734451293945312, 0.890960693359375, -0.157501220703125, 0.44281005859375, 0.5164070129394531, 0.358306884765625, 0.01177215576171875, -0.3645744323730469, -0.13014984130859375, -0.3801116943359375, 0.21584129333496094, 0.13939285278320312, 0.07140350341796875, -0.051227569580078125, 0.03356170654296875, 0.8116073608398438, -0.041522979736328125, 0.7939453125, 0.47071075439453125, 0.3065147399902344, 1.6416015625, -0.439300537109375, -0.7609939575195312, -0.4599609375, 0.44341278076171875, 0.4409160614013672, -0.35121917724609375, -0.12240219116210938, 0.2554969787597656, 1.000946044921875, 1.2154006958007812, 0.06607818603515625, 0.7440299987792969, 2.353759765625, -0.3748779296875, -0.43865966796875, 1.2871856689453125, -1.0005035400390625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000055.npy"}
|
||||
{"epoch": 0.08314436885865457, "step": 56, "batch_size": 64, "mean": 0.18233805894851685, "std": 0.6790233254432678, "min": -2.6298370361328125, "p10": -0.49550056457519526, "median": 0.153900146484375, "p90": 1.082723236083985, "max": 1.60369873046875, "pos_frac": 0.640625, "sample": [-0.3103790283203125, 0.3967628479003906, -0.09852790832519531, -0.2408905029296875, 0.6821136474609375, -0.32657623291015625, 1.1841659545898438, -1.2540435791015625, -0.3863105773925781, 0.29878997802734375, 0.4520072937011719, -0.517059326171875, -0.336639404296875, 0.2413330078125, 0.0450897216796875, 1.1862945556640625, 0.31036376953125, -0.7632846832275391, 0.1486053466796875, 1.1353340148925781, -0.7339992523193359, 1.60369873046875, 0.6577014923095703, 0.5000839233398438, 0.33744239807128906, 0.6743278503417969, -0.0355072021484375, 0.2957115173339844, 0.1591949462890625, 0.6454372406005859, -0.5166893005371094, 0.602264404296875, 1.169921875, 0.9340400695800781, 0.8537750244140625, 0.12033843994140625, -2.6298370361328125, 0.4485969543457031, 0.12435150146484375, 0.5234909057617188, 0.4498767852783203, 0.7543258666992188, -0.35760498046875, -0.7555809020996094, 0.044986724853515625, 0.053924560546875, -0.2801513671875, 0.069366455078125, -0.18175125122070312, 0.3077526092529297, 1.3055839538574219, 0.6849136352539062, 0.7497787475585938, 0.3362007141113281, 0.00848388671875, -0.2860584259033203, -0.3981208801269531, 0.0074615478515625, 0.9599647521972656, -0.1647796630859375, -0.025142669677734375, 1.284027099609375, -0.03325462341308594, -0.4460601806640625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000056.npy"}
|
||||
{"epoch": 0.08465608465608465, "step": 57, "batch_size": 64, "mean": 0.2848610281944275, "std": 0.5948304533958435, "min": -0.9525833129882812, "p10": -0.62215576171875, "median": 0.4009370803833008, "p90": 1.0433452606201172, "max": 2.2092971801757812, "pos_frac": 0.703125, "sample": [-0.634552001953125, -0.19287109375, 0.40581512451171875, 1.067657470703125, 0.035919189453125, 1.0915069580078125, 0.4054756164550781, -0.28577423095703125, 0.49124908447265625, 0.8793792724609375, 1.3525848388671875, 0.6352500915527344, -0.818145751953125, 0.09061431884765625, 0.6318435668945312, -0.7157306671142578, 0.2628631591796875, 0.6371116638183594, 2.2092971801757812, 0.1893444061279297, 0.7425765991210938, -0.40270233154296875, 0.6240081787109375, 1.1323089599609375, -0.3921775817871094, -0.4485149383544922, 0.3689079284667969, 0.494415283203125, 0.032238006591796875, -0.7209243774414062, 0.4745330810546875, 0.3486785888671875, 0.4604682922363281, 0.4675445556640625, -0.05496978759765625, 0.8145217895507812, -0.593231201171875, 0.5038700103759766, -0.7393798828125, 1.05743408203125, -0.30210113525390625, -0.12686920166015625, -0.9525833129882812, 0.5723781585693359, -0.16370391845703125, 0.4491729736328125, 0.6673049926757812, 0.5967941284179688, 0.5168609619140625, 0.39639854431152344, 0.47503662109375, 0.4706878662109375, 0.15959930419921875, 1.1194000244140625, 0.08887481689453125, 1.0104713439941406, 0.1810455322265625, 0.8088150024414062, 0.0964813232421875, 0.5161056518554688, -0.0011749267578125, -0.00965118408203125, 0.38941192626953125, -0.6360931396484375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000057.npy"}
|
||||
{"epoch": 0.08616780045351474, "step": 58, "batch_size": 64, "mean": 0.22464051842689514, "std": 1.0195177793502808, "min": -1.3321456909179688, "p10": -0.6923656463623047, "median": 0.10265922546386719, "p90": 1.198656463623047, "max": 4.45147705078125, "pos_frac": 0.546875, "sample": [-1.3321456909179688, 0.6585464477539062, 0.7165298461914062, -0.4423255920410156, 1.1097793579101562, -0.1446685791015625, 0.031158447265625, 0.10211944580078125, -0.0461883544921875, -0.4957122802734375, 0.114959716796875, -0.4123687744140625, 0.2618255615234375, -0.352203369140625, -0.6751651763916016, 0.7364387512207031, 1.5071640014648438, -0.6959304809570312, -0.552459716796875, 1.285308837890625, 0.22102928161621094, 0.0673370361328125, 0.44681549072265625, 0.6992263793945312, -0.8126220703125, -0.2030029296875, -1.182891845703125, 0.2097015380859375, -0.5233993530273438, 0.5224456787109375, -0.0010986328125, -1.12640380859375, 0.19535064697265625, 0.41542816162109375, 4.45147705078125, -0.6840476989746094, -0.824066162109375, 3.8968963623046875, 1.7631454467773438, 0.305511474609375, -0.324798583984375, 0.16401290893554688, -0.15279006958007812, -0.763336181640625, -0.2516326904296875, 2.7473297119140625, -0.171539306640625, -0.54168701171875, -0.3441314697265625, 0.8077888488769531, 0.4807586669921875, 1.2220230102539062, 0.6549453735351562, 0.10319900512695312, 0.12031936645507812, 0.9530029296875, 0.20513343811035156, -0.4644908905029297, -0.41661834716796875, 1.144134521484375, -0.644439697265625, -0.08657073974609375, 0.27220726013183594, 0.4526786804199219], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000058.npy"}
|
||||
{"epoch": 0.08767951625094482, "step": 59, "batch_size": 64, "mean": 0.2886250913143158, "std": 0.8572422862052917, "min": -1.3921966552734375, "p10": -0.6278026580810547, "median": 0.1256580352783203, "p90": 1.3224472045898439, "max": 3.1612091064453125, "pos_frac": 0.578125, "sample": [-0.07530593872070312, 0.5088424682617188, 0.083404541015625, -0.6263580322265625, -0.25044822692871094, -0.4680519104003906, 0.728790283203125, -1.3921966552734375, 0.7258224487304688, 0.6016845703125, 2.46795654296875, 0.08341598510742188, 1.4952239990234375, 1.5199089050292969, 0.906158447265625, -0.1293201446533203, 1.099822998046875, -0.9765739440917969, -0.4406585693359375, -0.34914398193359375, 0.20917892456054688, 0.21868896484375, -0.22711944580078125, 0.09296035766601562, -0.21952056884765625, 0.5122909545898438, -0.14728546142578125, 1.3297119140625, 2.453704833984375, 0.37932586669921875, -0.057140350341796875, 0.3764076232910156, 0.6530055999755859, -0.11326217651367188, -0.3475494384765625, 0.22400856018066406, -0.001903533935546875, 0.6928882598876953, -0.23301315307617188, -0.26964378356933594, 3.1612091064453125, 0.76611328125, 1.2763214111328125, 0.7968597412109375, -0.5126266479492188, 0.6070308685302734, -0.6284217834472656, 0.8365402221679688, -0.015157699584960938, 1.3054962158203125, -0.9752960205078125, 0.5550918579101562, -0.11910629272460938, -1.33941650390625, 0.4322357177734375, -0.06632232666015625, -1.123382568359375, 0.0523834228515625, 0.10533523559570312, 0.1459808349609375, 0.7048568725585938, 1.4087066650390625, 0.7384529113769531, -0.6795864105224609], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000059.npy"}
|
||||
{"epoch": 0.08919123204837491, "step": 60, "batch_size": 64, "mean": 0.0646200180053711, "std": 0.7359238862991333, "min": -1.7512359619140625, "p10": -0.9209453582763671, "median": 0.1814889907836914, "p90": 0.9641765594482423, "max": 1.6208724975585938, "pos_frac": 0.59375, "sample": [0.3346405029296875, 0.5439376831054688, 0.4710731506347656, -0.00606536865234375, -0.365997314453125, 1.2845115661621094, -1.329986572265625, -0.7112274169921875, -0.3059120178222656, -1.7512359619140625, 0.22339630126953125, 0.3779449462890625, -0.30158042907714844, -0.08486747741699219, -1.0459747314453125, 0.6313552856445312, -0.491302490234375, 0.9835548400878906, 1.2616939544677734, 0.408935546875, -0.9279022216796875, 0.210174560546875, 0.30987548828125, -0.6144332885742188, 0.3638877868652344, 0.2252044677734375, 0.4018135070800781, -0.09574127197265625, 1.6208724975585938, -0.8556632995605469, 0.18451881408691406, -0.04755401611328125, 0.7291641235351562, 0.17845916748046875, -0.5085868835449219, 0.3367462158203125, 0.7639675140380859, -0.9695968627929688, -0.30641937255859375, -0.7523593902587891, -0.6884307861328125, 0.8797073364257812, 0.5504035949707031, -0.9047126770019531, 0.1590290069580078, 1.1555633544921875, 0.16937637329101562, -0.8581695556640625, 0.9189605712890625, 0.11724662780761719, 0.5612411499023438, 0.3999519348144531, 0.6408424377441406, 1.0309524536132812, 1.50567626953125, -0.40718841552734375, 0.455841064453125, 0.29718780517578125, 0.036144256591796875, 0.053455352783203125, -1.6637420654296875, -1.041238784790039, 0.43800926208496094, -0.0437469482421875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000060.npy"}
|
||||
{"epoch": 0.09070294784580499, "step": 61, "batch_size": 64, "mean": 0.08620262145996094, "std": 0.9734062552452087, "min": -2.2948379516601562, "p10": -1.362565994262695, "median": 0.057804107666015625, "p90": 1.2258140563964846, "max": 2.2848052978515625, "pos_frac": 0.5625, "sample": [0.052886962890625, 0.162689208984375, 1.73248291015625, 2.0709609985351562, 0.43355560302734375, -0.008150100708007812, 0.3286895751953125, 1.7376708984375, 0.6522750854492188, -0.15048599243164062, 0.0062255859375, -0.10902786254882812, -0.147308349609375, 0.3729248046875, 0.5585861206054688, -0.6478996276855469, 0.5933151245117188, -0.43499183654785156, -0.3007354736328125, 0.38153076171875, -0.01607513427734375, 0.4738006591796875, 1.1246795654296875, 0.1306896209716797, -0.08462333679199219, -0.22241783142089844, -0.9168701171875, -1.7462005615234375, 0.34461212158203125, -0.22718429565429688, 0.9679431915283203, -0.24484825134277344, -1.2166213989257812, 0.954437255859375, -1.064727783203125, 0.3976325988769531, 0.06272125244140625, -0.5934505462646484, -0.4437103271484375, 1.244964599609375, 2.2848052978515625, -2.2948379516601562, 0.02321624755859375, -1.4251136779785156, 0.7929916381835938, 1.5088005065917969, 0.3004016876220703, 0.3124122619628906, -0.5857696533203125, -0.2613372802734375, 1.1811294555664062, -1.6916961669921875, 0.014986038208007812, -0.0762176513671875, 0.8082122802734375, -2.0794830322265625, 1.4003448486328125, 0.5353431701660156, -0.01224517822265625, 0.15203857421875, 1.0195159912109375, 0.96429443359375, -1.9927864074707031, -1.5719833374023438], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000061.npy"}
|
||||
{"epoch": 0.09221466364323508, "step": 62, "batch_size": 64, "mean": 0.3026942014694214, "std": 0.9148489832878113, "min": -2.58013916015625, "p10": -0.7731746673583983, "median": 0.3589897155761719, "p90": 1.3343570709228518, "max": 2.5889434814453125, "pos_frac": 0.703125, "sample": [0.9294891357421875, -1.4111557006835938, -0.537384033203125, 0.3240032196044922, 0.9361724853515625, 0.05991363525390625, 1.3668251037597656, -0.2434234619140625, -1.6511764526367188, -0.19940185546875, 1.7885589599609375, -0.6560211181640625, 1.709014892578125, -1.463531494140625, 1.2585983276367188, 0.609161376953125, 0.61492919921875, 0.4307403564453125, 0.07150077819824219, 0.5880661010742188, -0.42691802978515625, -0.8275527954101562, -0.6068115234375, 0.050235748291015625, 0.2657814025878906, 0.3598785400390625, -0.606842041015625, 0.6054840087890625, 0.8434333801269531, 1.2454910278320312, -0.48949432373046875, -0.13315200805664062, 0.06752777099609375, 0.3767242431640625, 1.5478096008300781, -2.58013916015625, 0.1019134521484375, 1.6223373413085938, 0.10558700561523438, 0.35810089111328125, 0.3294639587402344, 1.1489486694335938, 0.5573081970214844, 0.6909561157226562, 0.9816684722900391, 0.7621002197265625, -0.375152587890625, -0.8233833312988281, 0.39441490173339844, 1.2223129272460938, -0.06623077392578125, 0.9012546539306641, 0.9249591827392578, 0.29624176025390625, -0.30251312255859375, 0.40348052978515625, 2.5889434814453125, 2.2992706298828125, 0.04972076416015625, 0.3777008056640625, 0.5618858337402344, 0.15937042236328125, -0.964080810546875, 0.8495140075683594], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000062.npy"}
|
||||
{"epoch": 0.09372637944066516, "step": 63, "batch_size": 64, "mean": 0.5362226963043213, "std": 0.8066269755363464, "min": -2.1915283203125, "p10": -0.3180465698242187, "median": 0.4763031005859375, "p90": 1.4385505676269537, "max": 2.603912353515625, "pos_frac": 0.8125, "sample": [0.49404144287109375, 0.1797943115234375, -0.21112060546875, 0.020862579345703125, 2.58929443359375, -0.200408935546875, -0.0403900146484375, 2.603912353515625, 0.281158447265625, 0.41199493408203125, -2.1915283203125, -0.5075492858886719, 0.41326904296875, -0.8485565185546875, 1.1007537841796875, 0.36905670166015625, 1.2716064453125, 0.152679443359375, 0.5066452026367188, 0.9177360534667969, -0.6090240478515625, 0.3688621520996094, 1.3244438171386719, 2.0059356689453125, 0.7174530029296875, 0.7448978424072266, 0.5386810302734375, 0.25669097900390625, 0.41863059997558594, 0.3241424560546875, 0.55572509765625, 0.7156295776367188, 1.4874534606933594, 0.9334602355957031, 0.48252105712890625, 0.23664093017578125, 0.5045967102050781, -0.4090576171875, 0.3668479919433594, 0.2972087860107422, 0.47008514404296875, 2.5908203125, 0.12929725646972656, 1.2969322204589844, 0.6602191925048828, 1.1999931335449219, 1.009918212890625, 1.3234481811523438, 0.7913818359375, -0.26634979248046875, 0.30263328552246094, 0.189453125, 0.12282943725585938, -0.0320892333984375, 0.81011962890625, 0.5000457763671875, 0.5471572875976562, 0.6697254180908203, -0.34020233154296875, 0.09183883666992188, 0.7660064697265625, 1.8716278076171875, 1.6486968994140625, -0.6103286743164062], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000063.npy"}
|
||||
{"epoch": 0.09523809523809523, "step": 64, "batch_size": 64, "mean": 0.07834625244140625, "std": 0.8487309813499451, "min": -2.858184814453125, "p10": -0.6569753646850586, "median": 0.09377670288085938, "p90": 1.0929264068603517, "max": 2.53692626953125, "pos_frac": 0.5625, "sample": [1.738616943359375, 2.53692626953125, -0.4945716857910156, -0.5623703002929688, -0.11565780639648438, -0.0384674072265625, 1.2579669952392578, 0.002056121826171875, 0.33234214782714844, 1.830535888671875, 0.5147609710693359, 0.38986968994140625, 1.0415153503417969, -0.2906341552734375, 0.394378662109375, 0.016767501831054688, 0.118896484375, 0.5227890014648438, 0.3436241149902344, 0.3211517333984375, 0.2799835205078125, 0.224853515625, -0.03166961669921875, -0.11897087097167969, -0.9127044677734375, -0.40642738342285156, 0.0457305908203125, -2.858184814453125, -0.5401268005371094, 0.4321174621582031, -0.016048431396484375, -0.2442474365234375, 0.21533966064453125, -0.6956787109375, 0.24934005737304688, 0.06865692138671875, 0.35040283203125, -2.6680908203125, -0.6634693145751953, 0.5521087646484375, 1.114959716796875, 0.13384628295898438, 0.6634140014648438, -0.08897590637207031, 0.1315021514892578, -0.2056427001953125, -0.5126705169677734, -0.2978668212890625, -0.42815399169921875, 1.5236587524414062, -0.6418228149414062, -0.36266136169433594, 1.2985458374023438, 0.12311553955078125, 0.2851905822753906, 0.31407737731933594, 0.5993537902832031, -0.3247566223144531, -1.0106964111328125, -0.4340019226074219, 0.5374374389648438, -1.0468902587890625, -0.23295974731445312, 0.75274658203125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000064.npy"}
|
||||
{"epoch": 0.09674981103552532, "step": 65, "batch_size": 64, "mean": 0.49741989374160767, "std": 1.2052773237228394, "min": -2.5785789489746094, "p10": -1.0191402435302732, "median": 0.5033645629882812, "p90": 1.8595497131347658, "max": 4.309783935546875, "pos_frac": 0.6875, "sample": [0.3978605270385742, 0.9703903198242188, 1.7939300537109375, -0.18615341186523438, -0.09303474426269531, 0.8413352966308594, -1.716796875, -0.256195068359375, -0.32156944274902344, 1.8876724243164062, -0.70489501953125, 1.0275230407714844, 0.9157772064208984, 0.45766448974609375, 0.6842041015625, 0.6451740264892578, 0.7175521850585938, 0.4139213562011719, -0.674163818359375, 2.3239593505859375, -0.005157470703125, 0.0977325439453125, -1.2391204833984375, 2.0843124389648438, 1.6330909729003906, 0.11571884155273438, -2.5785789489746094, 0.5151138305664062, -1.3161849975585938, 0.560791015625, 0.025644302368164062, 0.5518112182617188, -0.8401641845703125, -1.0958442687988281, 0.6543350219726562, 2.3717727661132812, 0.11480331420898438, 2.706756591796875, 1.7841567993164062, 1.1811466217041016, 1.6247406005859375, 0.1591358184814453, 1.5184555053710938, 1.1121177673339844, -0.4639930725097656, 0.11678314208984375, 0.7139930725097656, -0.10157012939453125, -1.3404617309570312, 0.034332275390625, 0.49161529541015625, 0.8454093933105469, -0.139617919921875, 0.8903350830078125, 0.5713424682617188, 1.3106613159179688, 3.3956680297851562, -0.3159923553466797, 1.1480560302734375, 1.0634841918945312, 0.489959716796875, -1.96197509765625, 4.309783935546875, -0.08368110656738281], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000065.npy"}
|
||||
{"epoch": 0.0982615268329554, "step": 66, "batch_size": 64, "mean": 0.3557119369506836, "std": 1.0975146293640137, "min": -2.467864990234375, "p10": -0.9244514465332029, "median": 0.3196887969970703, "p90": 1.9300533294677744, "max": 3.449920654296875, "pos_frac": 0.640625, "sample": [0.2894706726074219, 3.449920654296875, 0.9178333282470703, 0.5900478363037109, 0.2623176574707031, 0.47953033447265625, 0.40953826904296875, 0.17198944091796875, -0.1511707305908203, 0.3892478942871094, -0.45613861083984375, 0.6534347534179688, 2.1088714599609375, 0.05524444580078125, -0.595855712890625, -0.6669235229492188, 0.2923126220703125, 0.3247337341308594, 2.0751495361328125, 2.039031982421875, -0.08648681640625, 1.6757698059082031, -1.6446762084960938, -0.04357147216796875, 0.6243743896484375, 0.9980945587158203, 0.3722381591796875, 1.5467376708984375, 0.94464111328125, 0.5709457397460938, 0.8877410888671875, 0.6715908050537109, -0.509490966796875, 0.1651153564453125, -0.020849227905273438, -0.2959938049316406, 0.30779266357421875, 0.355316162109375, -1.0859260559082031, 0.31464385986328125, -2.467864990234375, -2.2643890380859375, 0.0484771728515625, -0.03185272216796875, -0.17441558837890625, -0.64007568359375, 2.3118324279785156, -0.2600250244140625, 0.6980133056640625, 1.0294437408447266, -0.19816017150878906, 0.41680145263671875, -0.10094070434570312, 3.182891845703125, 2.1808547973632812, -1.2844200134277344, 1.080841064453125, -1.034820556640625, -1.1873931884765625, 1.652496337890625, 0.5213546752929688, -0.3910064697265625, 0.5680313110351562, 0.723297119140625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000066.npy"}
|
||||
{"epoch": 0.09977324263038549, "step": 67, "batch_size": 64, "mean": 0.2541486620903015, "std": 0.941967785358429, "min": -2.3326568603515625, "p10": -0.8259384155273437, "median": 0.183868408203125, "p90": 1.4058059692382814, "max": 3.01025390625, "pos_frac": 0.609375, "sample": [0.2365570068359375, -0.03411865234375, -0.7167510986328125, -0.4787921905517578, -1.2053680419921875, 0.3356056213378906, 1.9755363464355469, 1.0357017517089844, -0.46079254150390625, -0.544464111328125, 1.2477607727050781, 0.1354084014892578, -0.7399978637695312, -1.1851425170898438, 1.3886947631835938, 0.7268142700195312, -0.29856109619140625, 0.5559425354003906, -0.28723907470703125, 1.0573577880859375, -0.2214202880859375, 0.7882080078125, -2.3326568603515625, 3.01025390625, 0.17791748046875, -0.277191162109375, 0.45648193359375, 0.6405754089355469, 1.2857666015625, -0.9213047027587891, 0.43135833740234375, 0.5971508026123047, 0.23625946044921875, 1.8258056640625, 0.44434356689453125, 0.14654922485351562, 0.6552848815917969, 0.2862892150878906, -0.4256019592285156, 0.9494438171386719, 1.9955596923828125, -0.133209228515625, 0.6330032348632812, 1.4131393432617188, 0.1898193359375, -0.03646087646484375, -0.0888214111328125, 0.08753204345703125, -0.10430145263671875, -0.8627700805664062, 0.17367172241210938, 2.121429443359375, 0.4491386413574219, 0.09177017211914062, -0.2170257568359375, -0.2469329833984375, 1.0737075805664062, 0.0852203369140625, -0.9527778625488281, -2.116302490234375, 1.5451240539550781, -0.13897705078125, 0.467071533203125, 0.33924102783203125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000067.npy"}
|
||||
{"epoch": 0.10128495842781557, "step": 68, "batch_size": 64, "mean": 0.3424687385559082, "std": 1.3044476509094238, "min": -4.37603759765625, "p10": -0.930421447753906, "median": 0.33229732513427734, "p90": 1.6167503356933601, "max": 4.21551513671875, "pos_frac": 0.640625, "sample": [4.21551513671875, 1.330810546875, 0.41884803771972656, 2.0366363525390625, -0.48401641845703125, 2.5296173095703125, 0.8903274536132812, -0.22344970703125, 0.8889141082763672, -0.2130279541015625, 1.944000244140625, -4.37603759765625, 1.1851043701171875, 1.296142578125, 1.0557403564453125, 0.9759483337402344, -0.6397781372070312, -0.5316505432128906, 0.1139068603515625, 0.9729843139648438, 0.940277099609375, -0.04460716247558594, 1.0658721923828125, 0.8421974182128906, 0.48455047607421875, 0.3676280975341797, 0.29285430908203125, -0.3491992950439453, 1.6885757446289062, -0.3093109130859375, -0.08036231994628906, 0.5229759216308594, 0.4349937438964844, -0.029582977294921875, -1.0135078430175781, 0.8221244812011719, 3.48223876953125, 0.24675750732421875, -0.16632080078125, 0.07100677490234375, -2.0756378173828125, -1.473663330078125, 0.2660369873046875, 0.9412612915039062, -2.0603790283203125, -2.7234649658203125, 0.5328292846679688, 0.2255401611328125, -0.7365531921386719, -0.20507049560546875, 1.44915771484375, 0.296966552734375, -0.29560089111328125, 1.2543182373046875, 1.0454940795898438, 0.19220733642578125, -0.02721405029296875, 1.2237701416015625, 1.7029876708984375, 0.5247154235839844, 1.1523628234863281, -0.3983001708984375, 0.08902740478515625, -1.63848876953125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000068.npy"}
|
||||
{"epoch": 0.10279667422524566, "step": 69, "batch_size": 64, "mean": 0.6216727495193481, "std": 1.2031971216201782, "min": -1.9988937377929688, "p10": -0.5833641052246094, "median": 0.4077415466308594, "p90": 1.9651153564453128, "max": 5.079132080078125, "pos_frac": 0.6875, "sample": [-0.2654075622558594, -0.08231353759765625, 2.9215927124023438, 0.40508270263671875, 0.790557861328125, -0.0286712646484375, -0.2135448455810547, -0.05804252624511719, 1.0711822509765625, 0.5634307861328125, 0.4038810729980469, 0.6170845031738281, 0.684234619140625, 1.7399330139160156, 0.85400390625, 1.9910888671875, 0.6519088745117188, 1.7600021362304688, 2.3273773193359375, -1.3007583618164062, 3.473358154296875, -1.8811264038085938, 1.6528778076171875, -0.2610321044921875, 2.2645263671875, 0.11852645874023438, 0.64630126953125, 0.746490478515625, -0.02764892578125, -0.5294189453125, 1.2681598663330078, 0.391204833984375, 0.44382286071777344, 0.32940673828125, 0.6392631530761719, -0.7641448974609375, 1.8101692199707031, 0.410400390625, 0.000518798828125, 0.2022705078125, 0.62213134765625, 0.3492927551269531, 0.3870086669921875, 0.21956634521484375, 1.7663803100585938, -0.6064834594726562, 1.8327598571777344, 0.8287334442138672, -0.3997821807861328, -1.4014663696289062, -0.045131683349609375, 1.016866683959961, -1.9988937377929688, -0.43966102600097656, 1.24896240234375, -0.027523040771484375, 1.8804473876953125, 0.2521553039550781, 1.904510498046875, -0.7609329223632812, 5.079132080078125, -0.05664634704589844, 2.2322540283203125, 0.1368274688720703], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000069.npy"}
|
||||
{"epoch": 0.10430839002267574, "step": 70, "batch_size": 64, "mean": 0.395748108625412, "std": 1.2223933935165405, "min": -3.253692626953125, "p10": -0.9181308746337891, "median": 0.5020332336425781, "p90": 1.8784893035888677, "max": 3.501007080078125, "pos_frac": 0.640625, "sample": [0.24068450927734375, -0.9183540344238281, 1.4234466552734375, -1.819732666015625, 0.2475147247314453, -1.0839176177978516, -0.2650909423828125, 1.3979110717773438, 1.4221115112304688, -0.19228363037109375, 1.0892791748046875, 1.7762413024902344, -3.253692626953125, 0.8271713256835938, 0.27396583557128906, -0.6384735107421875, 2.181549072265625, -1.8928146362304688, -0.8476219177246094, -0.2550945281982422, 3.501007080078125, 0.7030868530273438, 1.0224361419677734, 0.7504425048828125, -1.5928993225097656, 0.8535385131835938, 0.3712882995605469, 2.4085006713867188, -0.9176101684570312, 3.193878173828125, -0.10222625732421875, 0.20347976684570312, -0.7857284545898438, -0.605010986328125, 0.2513580322265625, 2.0892715454101562, 1.9223098754882812, 0.7696533203125, 0.6560993194580078, 1.409627914428711, 0.6327781677246094, -0.8879871368408203, -0.08350944519042969, 0.8321075439453125, 1.4767913818359375, 1.0283851623535156, -0.41959381103515625, 2.390899658203125, 0.2569084167480469, 0.7097854614257812, 0.8906784057617188, 1.4267196655273438, 0.7079200744628906, -0.7499237060546875, 0.010833740234375, -0.13457870483398438, 0.3671875, -1.5638771057128906, 1.1588592529296875, 0.7037391662597656, -0.6380767822265625, 1.1056632995605469, -0.3592948913574219, 0.6501617431640625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000070.npy"}
|
||||
{"epoch": 0.10582010582010581, "step": 71, "batch_size": 64, "mean": 0.23435284197330475, "std": 1.4345401525497437, "min": -3.97637939453125, "p10": -1.1473716735839843, "median": 0.23393535614013672, "p90": 2.015304565429689, "max": 3.1530685424804688, "pos_frac": 0.578125, "sample": [-0.17204856872558594, 1.4134330749511719, 0.8765735626220703, 0.41420745849609375, 0.1708526611328125, -0.8601837158203125, 0.9709644317626953, -1.1573257446289062, -0.04004669189453125, 2.24261474609375, 0.44645118713378906, -2.313465118408203, -0.251373291015625, 1.1625900268554688, -3.01092529296875, -0.028474807739257812, 0.9588623046875, 2.7938308715820312, 0.9178695678710938, 1.452056884765625, -3.659942626953125, 2.4437255859375, 0.11681365966796875, -0.5581169128417969, 0.9286766052246094, 2.763885498046875, 0.7073974609375, -0.17939376831054688, 0.5690059661865234, 0.0362091064453125, -3.97637939453125, -0.8159332275390625, -1.35565185546875, -0.5591449737548828, 0.11175155639648438, 0.4640998840332031, 0.29701805114746094, 0.7245330810546875, 1.4002513885498047, -0.6318550109863281, -0.4911956787109375, -3.1631622314453125, 0.5197677612304688, 2.1586761474609375, -0.13532257080078125, -0.2457418441772461, -1.1241455078125, 1.6807708740234375, -0.3699989318847656, -0.07675933837890625, 1.2648773193359375, 2.3085060119628906, -0.7116546630859375, 1.6026535034179688, 1.1645660400390625, 0.9702987670898438, 0.5016365051269531, -0.02031707763671875, 0.349151611328125, 0.168792724609375, 1.6314620971679688, -0.06554985046386719, 3.1530685424804688, -0.8852119445800781], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000071.npy"}
|
||||
{"epoch": 0.1073318216175359, "step": 72, "batch_size": 64, "mean": 0.7390097379684448, "std": 1.3422960042953491, "min": -3.066314697265625, "p10": -0.6831859588623047, "median": 0.8583593368530273, "p90": 2.028182411193848, "max": 4.6888580322265625, "pos_frac": 0.734375, "sample": [1.4617500305175781, 1.8805618286132812, 0.40041351318359375, 1.8590660095214844, -3.066314697265625, 0.207763671875, 0.16265869140625, 4.6888580322265625, -1.5088844299316406, -0.5354633331298828, 1.3218498229980469, 1.1692886352539062, -0.320770263671875, 1.9808101654052734, 0.0629119873046875, -0.2763786315917969, -0.5510730743408203, 0.9046649932861328, 1.9281997680664062, -0.303466796875, 2.9856414794921875, 0.6391181945800781, 1.8804054260253906, 0.3499011993408203, 0.9697265625, 1.1658706665039062, 0.47918128967285156, -0.7164669036865234, -0.6929397583007812, -0.23084259033203125, 2.7269744873046875, 1.3195953369140625, 1.076263427734375, 2.396087646484375, 1.228240966796875, 0.5019378662109375, 0.4151897430419922, 1.2700443267822266, 0.8120536804199219, -0.05306243896484375, 1.3238258361816406, 1.3377838134765625, 1.6196212768554688, 0.38677024841308594, 1.2820281982421875, 0.04595184326171875, 1.5848541259765625, 2.0484848022460938, -1.5693931579589844, 1.0871009826660156, -1.3668670654296875, 1.30487060546875, 3.4845809936523438, -0.6604270935058594, 3.4153060913085938, -0.4001331329345703, 1.4563941955566406, 0.7266273498535156, -2.5320892333984375, 0.38364601135253906, -0.18474578857421875, 1.1772308349609375, 0.92755126953125, 0.42828369140625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000072.npy"}
|
||||
{"epoch": 0.10884353741496598, "step": 73, "batch_size": 64, "mean": 0.11978399753570557, "std": 1.6252601146697998, "min": -3.290496826171875, "p10": -1.4833786010742187, "median": -0.08919715881347656, "p90": 1.4404737472534184, "max": 7.7078857421875, "pos_frac": 0.4375, "sample": [1.2207717895507812, 0.9903526306152344, -0.3770904541015625, -1.158050537109375, -1.1359596252441406, -1.485992431640625, -1.6512527465820312, -0.7338294982910156, -3.290496826171875, 1.3359813690185547, -1.4989242553710938, 0.506378173828125, -0.00628662109375, -1.195068359375, -1.3474998474121094, 0.6367969512939453, -1.652587890625, -0.24682235717773438, -0.30670166015625, 0.6361846923828125, -1.094970703125, -0.07011032104492188, -0.8396282196044922, 0.03759002685546875, 7.7078857421875, 2.0572738647460938, -0.5593338012695312, -0.8728656768798828, -0.013109207153320312, 0.9243011474609375, 0.8454685211181641, -0.30948829650878906, 0.8883590698242188, -0.1508026123046875, -1.39013671875, 0.3324699401855469, 1.4852561950683594, 2.82470703125, -1.859426498413086, -1.291961669921875, -0.64410400390625, 2.8090362548828125, 0.200592041015625, -1.4772796630859375, -0.19138336181640625, 0.9251327514648438, 0.7013473510742188, -0.5592670440673828, -0.0020904541015625, -0.5430793762207031, -2.1825103759765625, 0.8914985656738281, 0.8087539672851562, 0.5761528015136719, 1.2832908630371094, 0.7561187744140625, 0.9321937561035156, -0.10828399658203125, -0.5750312805175781, -0.4586944580078125, 0.20079994201660156, 2.1646385192871094, 4.798431396484375, -0.5314674377441406], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000073.npy"}
|
||||
{"epoch": 0.11035525321239607, "step": 74, "batch_size": 64, "mean": 0.6968612670898438, "std": 1.3816484212875366, "min": -2.6110382080078125, "p10": -0.9498554229736328, "median": 0.7726783752441406, "p90": 2.0677261352539076, "max": 6.30047607421875, "pos_frac": 0.71875, "sample": [-0.7841567993164062, 2.1864471435546875, 0.2883453369140625, 0.6618804931640625, 1.5005340576171875, -0.4023323059082031, 2.39739990234375, 6.30047607421875, 0.9987335205078125, 2.2915782928466797, 1.3957138061523438, 0.2080078125, 1.79071044921875, 1.3316421508789062, -0.22960853576660156, 1.0972709655761719, 1.390838623046875, 0.7552032470703125, -0.18692588806152344, -2.6110382080078125, 0.5155220031738281, 1.744781494140625, 1.0956344604492188, 0.7491188049316406, 1.2084236145019531, 0.09977340698242188, -0.6905670166015625, 0.853057861328125, 2.4325637817382812, -1.451568603515625, 0.36174964904785156, 0.8104934692382812, 1.3389015197753906, -1.0930652618408203, -0.9571418762207031, 0.98297119140625, -0.7210826873779297, 1.0332183837890625, -1.6191177368164062, -0.6211109161376953, 0.9329757690429688, -1.1292095184326172, 1.2246551513671875, 0.7901535034179688, 0.7350006103515625, 2.92626953125, 1.5901641845703125, -0.4902763366699219, 1.5378456115722656, 1.627838134765625, -1.95587158203125, -0.060337066650390625, 0.6163902282714844, -0.16634368896484375, -0.9328536987304688, 0.4446125030517578, 0.35015869140625, 3.799713134765625, 1.7341995239257812, 1.5826034545898438, 1.4613399505615234, 0.8460731506347656, 0.3880043029785156, 0.2927398681640625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000074.npy"}
|
||||
{"epoch": 0.11186696900982615, "step": 75, "batch_size": 64, "mean": 0.7842553853988647, "std": 1.378349781036377, "min": -2.3259239196777344, "p10": -0.74901123046875, "median": 0.5506277084350586, "p90": 2.6332645416259775, "max": 4.418701171875, "pos_frac": 0.71875, "sample": [-0.44583702087402344, -0.7312698364257812, 0.07977867126464844, 1.1232757568359375, 3.4769821166992188, -0.19142913818359375, 0.24769210815429688, 0.9887657165527344, 1.8647613525390625, 3.3462295532226562, -0.3554840087890625, -0.08527374267578125, -1.3041229248046875, 0.314910888671875, 0.028177261352539062, 1.4917945861816406, 1.7646942138671875, -1.276906967163086, 0.40553855895996094, 0.4282264709472656, 3.69024658203125, 1.3247756958007812, 1.8664627075195312, 2.4303359985351562, 1.2427425384521484, 0.3316802978515625, 1.7569198608398438, -1.0461578369140625, 2.356689453125, 0.8942604064941406, 0.6352100372314453, 2.720233917236328, -0.016998291015625, 2.1790637969970703, 1.3957061767578125, 0.5577983856201172, 4.418701171875, 0.8787002563476562, 0.5091094970703125, 0.4820060729980469, -0.27698516845703125, 1.1570186614990234, -2.3259239196777344, 0.6507377624511719, 1.5765419006347656, -0.7566146850585938, 0.11701583862304688, 0.27588653564453125, 0.4422760009765625, -0.06143760681152344, 0.5810756683349609, 2.2246627807617188, -1.3465461730957031, -0.38100433349609375, 0.8452720642089844, -0.6853256225585938, 2.96533203125, -0.2329273223876953, -1.1320762634277344, 0.54345703125, 0.5589218139648438, 0.0333709716796875, 4.16827392578125, 1.4733543395996094], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000075.npy"}
|
||||
{"epoch": 0.11337868480725624, "step": 76, "batch_size": 64, "mean": 0.3494717478752136, "std": 1.566878318786621, "min": -4.253700256347656, "p10": -1.420491027832031, "median": 0.27880859375, "p90": 2.3544540405273446, "max": 4.23529052734375, "pos_frac": 0.578125, "sample": [-4.253700256347656, 1.5410079956054688, -1.95794677734375, -0.2218017578125, 0.8394241333007812, -0.6948318481445312, -0.4762744903564453, 1.4788818359375, 0.5612468719482422, 3.4018020629882812, 1.949850082397461, 4.23529052734375, -0.38707733154296875, -4.180999755859375, 1.070556640625, -1.5218505859375, 0.16248321533203125, 0.9306144714355469, 0.5305023193359375, 0.3176155090332031, 0.203582763671875, 2.4547653198242188, -1.4652099609375, 2.8859634399414062, -1.3161468505859375, 1.9148979187011719, 0.6092529296875, 1.2668685913085938, -0.2435016632080078, -0.059833526611328125, 0.015470504760742188, -0.5904178619384766, -1.5322151184082031, -0.46469879150390625, -0.5121250152587891, 0.3293304443359375, 2.4513778686523438, 1.16387939453125, -0.6877670288085938, 1.5426101684570312, -0.30114173889160156, 0.3676338195800781, 0.18571853637695312, -0.054088592529296875, 1.921905517578125, 0.6623649597167969, 0.7978057861328125, -1.2412796020507812, 0.2794952392578125, -0.33823394775390625, -0.0381011962890625, 0.7246475219726562, 0.2781219482421875, 2.1487884521484375, 3.86358642578125, 1.1121845245361328, -0.5834484100341797, -0.9976329803466797, -2.2277297973632812, 2.442596435546875, 0.9746284484863281, -0.2581596374511719, 1.4407958984375, -0.08514213562011719], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000076.npy"}
|
||||
{"epoch": 0.11489040060468632, "step": 77, "batch_size": 64, "mean": 0.7362152934074402, "std": 1.4746347665786743, "min": -3.1624908447265625, "p10": -1.2330257415771482, "median": 0.5685768127441406, "p90": 2.591776657104493, "max": 4.7983551025390625, "pos_frac": 0.734375, "sample": [-0.44120216369628906, 1.284912109375, -1.2958869934082031, -1.3179473876953125, 1.8930797576904297, 2.4154052734375, -1.5645179748535156, 0.06355094909667969, -3.1624908447265625, 0.32025146484375, 1.15679931640625, 1.1339912414550781, 0.004772186279296875, -1.5051078796386719, 0.5130348205566406, 2.838085174560547, 0.8387603759765625, 0.6531829833984375, 0.70367431640625, 2.94464111328125, -0.3647613525390625, 0.2581520080566406, 2.2643356323242188, 2.4268836975097656, 0.4955863952636719, 0.2689208984375, 0.3797435760498047, 2.3507156372070312, 0.22910499572753906, 0.9919509887695312, -0.4871997833251953, -1.5299606323242188, -0.2837982177734375, 0.566436767578125, 1.3741722106933594, 2.662445068359375, 1.00689697265625, 1.0706233978271484, 1.1956119537353516, 2.0785980224609375, 1.9832534790039062, 0.002292633056640625, 0.7945556640625, -0.1458415985107422, 0.18257904052734375, 4.7983551025390625, 0.1739368438720703, -0.26779937744140625, 2.7929611206054688, 1.0415763854980469, 4.346336364746094, -0.687255859375, 0.3514728546142578, 1.9885711669921875, 2.1655807495117188, 0.5707168579101562, -1.4943923950195312, -0.703887939453125, 1.4242477416992188, -0.2496929168701172, 0.40386962890625, 3.6527023315429688, 0.6485443115234375, -1.0863494873046875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000077.npy"}
|
||||
{"epoch": 0.1164021164021164, "step": 78, "batch_size": 64, "mean": 0.43755343556404114, "std": 1.4050379991531372, "min": -3.263946533203125, "p10": -1.2714649200439454, "median": 0.38922119140625, "p90": 2.138326644897461, "max": 3.6610260009765625, "pos_frac": 0.640625, "sample": [2.4099693298339844, -3.263946533203125, 0.8485679626464844, 2.1484031677246094, 0.590606689453125, 2.507692337036133, 0.3891639709472656, -0.4022674560546875, 0.22641944885253906, 0.14048004150390625, 1.5566635131835938, -1.4921417236328125, -0.29578399658203125, 1.713653564453125, 0.46435546875, -0.6889266967773438, -1.3217010498046875, 3.2189788818359375, -1.4838409423828125, 0.6327667236328125, -0.486175537109375, -0.7391128540039062, 0.1040496826171875, -0.6370887756347656, 1.3436126708984375, -0.137451171875, 0.054534912109375, -3.1264114379882812, 1.63665771484375, -2.48486328125, 1.774312973022461, 1.4026412963867188, 0.03827667236328125, 1.9197006225585938, -0.13190460205078125, 0.063140869140625, -1.278167724609375, -0.21234130859375, 1.7725086212158203, 2.0458831787109375, 0.4214630126953125, 0.40340423583984375, 0.4041748046875, 1.4042129516601562, -0.7359695434570312, 1.3839340209960938, -1.2558250427246094, 0.21011734008789062, 0.221893310546875, 1.5042266845703125, -0.5099296569824219, 2.6535568237304688, -0.2765960693359375, -0.6989059448242188, 0.8956451416015625, -0.638153076171875, 0.4327888488769531, -0.020263671875, 1.2416667938232422, 2.8873519897460938, 1.088592529296875, 0.3892784118652344, 3.6610260009765625, 2.1148147583007812], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000078.npy"}
|
||||
{"epoch": 0.11791383219954649, "step": 79, "batch_size": 64, "mean": 1.1198006868362427, "std": 2.0166492462158203, "min": -3.0485763549804688, "p10": -0.8741256713867186, "median": 0.5790319442749023, "p90": 3.202930450439454, "max": 7.750823974609375, "pos_frac": 0.734375, "sample": [0.15181922912597656, 0.35150909423828125, -1.1236209869384766, 1.6038665771484375, -0.9454841613769531, 0.9146385192871094, -3.0485763549804688, -0.07625961303710938, 0.01848602294921875, 0.4613800048828125, -0.3551673889160156, 0.48122406005859375, 1.7905616760253906, 7.750823974609375, -1.97296142578125, 0.40528106689453125, 2.3121795654296875, 0.44408226013183594, -0.5488624572753906, 0.2870597839355469, 0.9595355987548828, 6.8423919677734375, 2.3130035400390625, 3.28240966796875, 0.7622337341308594, -1.4600410461425781, 4.856456756591797, 2.3467025756835938, 1.8547821044921875, 1.8644943237304688, -1.55963134765625, -0.4248313903808594, 2.103179931640625, 2.108612060546875, 1.9195556640625, 0.4425067901611328, 1.2256584167480469, 1.7758598327636719, -0.9840927124023438, 2.5660934448242188, -0.7076225280761719, 3.0174789428710938, 0.5777626037597656, -0.3585071563720703, 0.5803012847900391, 0.9484405517578125, 2.691680908203125, 0.23064613342285156, 0.5355873107910156, 0.22663497924804688, 1.3895225524902344, -0.5561294555664062, 5.049652099609375, -0.0805511474609375, 0.9427528381347656, 7.3935546875, 0.31720733642578125, 1.1492843627929688, 0.4632415771484375, 2.223133087158203, -0.2450714111328125, 1.2167701721191406, 3.322784423828125, -0.3581695556640625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000079.npy"}
|
||||
{"epoch": 0.11942554799697656, "step": 80, "batch_size": 64, "mean": 0.6097861528396606, "std": 1.9898452758789062, "min": -4.4220123291015625, "p10": -1.612518501281738, "median": 0.2246694564819336, "p90": 3.4047161102294927, "max": 6.089080810546875, "pos_frac": 0.53125, "sample": [-0.61163330078125, 3.2365760803222656, 0.495269775390625, 2.3743896484375, 1.081634521484375, -0.529266357421875, -0.3646392822265625, -0.31426048278808594, -1.7935409545898438, -1.8541107177734375, 2.1392993927001953, 3.483123779296875, -0.81829833984375, -1.386993408203125, -0.10519218444824219, 4.412879943847656, -1.2818145751953125, -0.01453399658203125, 0.6738376617431641, -0.040271759033203125, 0.6035842895507812, -1.0439071655273438, -0.24837493896484375, 1.8236236572265625, 3.927093505859375, 3.223674774169922, 2.8133926391601562, 0.14886474609375, 3.200531005859375, 4.0554656982421875, -1.6833076477050781, 1.7779502868652344, 0.5741195678710938, 0.2457866668701172, 1.1109390258789062, 1.869720458984375, -0.6164226531982422, -0.38800811767578125, 4.372932434082031, -4.4220123291015625, 2.74493408203125, 0.20355224609375, -3.0299530029296875, 1.8755722045898438, -0.3807849884033203, 0.7967147827148438, 1.8206024169921875, 0.6362419128417969, -1.7154998779296875, -1.287454605102539, 3.476776123046875, -0.12983131408691406, -0.9321441650390625, -0.34503173828125, -0.24407386779785156, -0.16274070739746094, 0.9647064208984375, -1.4473438262939453, 1.7030811309814453, -2.869029998779297, 0.3298988342285156, -0.444671630859375, 6.089080810546875, 1.2456130981445312], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000080.npy"}
|
||||
{"epoch": 0.12093726379440665, "step": 81, "batch_size": 64, "mean": 0.8030184507369995, "std": 1.8265401124954224, "min": -3.0509872436523438, "p10": -1.2350631713867188, "median": 0.6536579132080078, "p90": 2.7849723815917975, "max": 8.004119873046875, "pos_frac": 0.640625, "sample": [1.021820068359375, 2.0500049591064453, 1.5415496826171875, -1.237823486328125, 1.2020606994628906, -2.2851104736328125, 0.428131103515625, 0.3524322509765625, -0.5989055633544922, -0.68377685546875, 5.2761993408203125, -0.20442962646484375, 0.32521820068359375, 2.3222274780273438, 0.666168212890625, -0.06681060791015625, 0.294525146484375, -0.10183906555175781, 3.2442703247070312, 2.1730194091796875, 0.9854469299316406, -0.6445446014404297, 1.5499267578125, 1.16180419921875, 2.6598052978515625, 0.8546905517578125, 0.20066070556640625, -0.19580078125, -0.2913684844970703, -1.6562957763671875, 1.1101646423339844, -2.061237335205078, 0.9596328735351562, 0.44112205505371094, -0.7994384765625, 3.472991943359375, -0.3106231689453125, 8.004119873046875, 1.224761962890625, 4.8842926025390625, -0.22186279296875, 1.2658634185791016, 0.41020965576171875, 0.3742828369140625, 0.8006210327148438, -0.03228569030761719, 1.8748531341552734, 1.0596389770507812, 2.5225448608398438, 1.4564666748046875, 3.5934104919433594, -1.2286224365234375, 2.5583877563476562, -0.66754150390625, -0.3445549011230469, 1.292327880859375, -1.3181934356689453, 0.6411476135253906, 0.9921722412109375, 2.8386154174804688, -0.7000331878662109, 1.4321632385253906, -3.0509872436523438, -1.4244842529296875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000081.npy"}
|
||||
{"epoch": 0.12244897959183673, "step": 82, "batch_size": 64, "mean": 1.0804365873336792, "std": 1.5326249599456787, "min": -3.9834747314453125, "p10": -0.7753906249999999, "median": 1.0893363952636719, "p90": 3.250669288635256, "max": 5.100517272949219, "pos_frac": 0.8125, "sample": [0.6738300323486328, 1.5478763580322266, 0.8662261962890625, -0.8310604095458984, 2.4291534423828125, 0.18378448486328125, 1.179718017578125, 2.032135009765625, 3.9333953857421875, 1.260061264038086, 1.2758769989013672, 0.6597347259521484, 1.18475341796875, 2.7520904541015625, 0.38309478759765625, 1.4795455932617188, 1.0558090209960938, 0.6155853271484375, 1.2644519805908203, 0.7062530517578125, 0.6065177917480469, -0.19775772094726562, 1.2107105255126953, 0.1430206298828125, -0.9229812622070312, 3.8770675659179688, 1.8380126953125, 1.12286376953125, 1.959066390991211, 1.2656517028808594, -0.06631851196289062, -0.1420001983642578, 5.100517272949219, 0.694183349609375, 0.9672088623046875, 3.949951171875, 2.6102218627929688, 2.0396575927734375, 1.606689453125, 3.464345932006836, 0.6737327575683594, -1.2983627319335938, 2.7164382934570312, -1.4487800598144531, 1.9259967803955078, 0.22226715087890625, 2.1298484802246094, 0.15372085571289062, 3.8880233764648438, 0.3302764892578125, 0.12758636474609375, 0.004730224609375, 1.4499969482421875, 3.7774734497070312, -0.221099853515625, -0.9188747406005859, 1.5024490356445312, 1.6127967834472656, 1.62884521484375, 0.5291252136230469, -1.3855361938476562, -3.9834747314453125, -0.6454944610595703, 0.5973167419433594], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000082.npy"}
|
||||
{"epoch": 0.12396069538926682, "step": 83, "batch_size": 64, "mean": 1.1440081596374512, "std": 1.8029627799987793, "min": -3.522672653198242, "p10": -0.8448387145996094, "median": 1.0162668228149414, "p90": 2.698078536987305, "max": 8.236030578613281, "pos_frac": 0.8125, "sample": [2.3870086669921875, 1.0294036865234375, -1.1702957153320312, 0.32513999938964844, 0.5885486602783203, 0.04524993896484375, 1.5209197998046875, 1.7823257446289062, 2.3690567016601562, 4.778717041015625, 0.45516204833984375, 1.1909008026123047, -1.8137474060058594, 1.2741355895996094, 1.8544273376464844, 1.0031299591064453, 1.1367263793945312, 2.661407470703125, 2.4957218170166016, 1.1253738403320312, 1.918182373046875, -0.954803466796875, 3.8969573974609375, 0.5085811614990234, -0.7541732788085938, 1.5878524780273438, 3.6158905029296875, 0.8006820678710938, 0.4341278076171875, -0.24477386474609375, -0.8267822265625, 0.5513210296630859, 2.580310821533203, -3.522672653198242, -1.0254173278808594, 2.1744022369384766, 1.2153644561767578, -0.08179473876953125, 1.2847766876220703, -2.04534912109375, 0.30716896057128906, 0.7504653930664062, 0.6801605224609375, 0.2736968994140625, 0.052921295166015625, -0.3749523162841797, 0.46477508544921875, 2.2088165283203125, 3.3654327392578125, 1.6937980651855469, 8.236030578613281, 1.7962970733642578, 2.713794708251953, 0.8811073303222656, 0.28734588623046875, 1.7020988464355469, 1.4072437286376953, 2.0234413146972656, 2.1153640747070312, 6.090087890625, 0.9901123046875, 0.095672607421875, 0.1562213897705078, -0.8525772094726562], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000083.npy"}
|
||||
{"epoch": 0.1254724111866969, "step": 84, "batch_size": 64, "mean": 0.8919090628623962, "std": 1.8700032234191895, "min": -3.273040771484375, "p10": -1.6226226806640625, "median": 0.6411609649658203, "p90": 3.6639083862304695, "max": 5.7155303955078125, "pos_frac": 0.75, "sample": [0.9159011840820312, 2.2198333740234375, 0.7674427032470703, 1.8299369812011719, 4.661140441894531, 1.7448062896728516, 3.5047378540039062, -0.493499755859375, -0.5207233428955078, -2.7018203735351562, 0.4670257568359375, 2.0556373596191406, 2.9594764709472656, 0.56744384765625, 0.20880889892578125, 0.05846405029296875, 0.22787857055664062, -1.1746444702148438, 3.7646560668945312, -1.2880058288574219, 1.005340576171875, 1.2341842651367188, 1.2411384582519531, 2.0227584838867188, 0.40248870849609375, 0.532623291015625, -1.9774932861328125, 3.8104248046875, 0.057826995849609375, 0.45147705078125, 0.9957733154296875, 0.51837158203125, -0.19507598876953125, 5.7155303955078125, -1.8030624389648438, 0.8358383178710938, 1.5660076141357422, 2.2665939331054688, 1.2650413513183594, -2.1905288696289062, -3.273040771484375, 5.00262451171875, 1.9485702514648438, 0.3379535675048828, 3.7321243286132812, 0.7734470367431641, 0.2468395233154297, 3.0653209686279297, 0.7148780822753906, -1.6319427490234375, 0.5605926513671875, -0.0740966796875, 2.436239242553711, 1.7045974731445312, 1.0186500549316406, -1.6008758544921875, 4.0854949951171875, 0.3571147918701172, 0.22304534912109375, -1.3155364990234375, 2.9519271850585938, 0.12392807006835938, -1.804422378540039, -0.031005859375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000084.npy"}
|
||||
{"epoch": 0.12698412698412698, "step": 85, "batch_size": 64, "mean": 1.253187894821167, "std": 2.0538854598999023, "min": -3.942546844482422, "p10": -1.4433849334716797, "median": 1.1862688064575195, "p90": 3.5431537628173833, "max": 6.9196319580078125, "pos_frac": 0.75, "sample": [2.284626007080078, -1.3861045837402344, 1.0863456726074219, 2.4972896575927734, 1.1871261596679688, 1.2047042846679688, 1.6307601928710938, 0.5593280792236328, -0.0454254150390625, 3.4452476501464844, 3.3468399047851562, 0.9933013916015625, 2.2563419342041016, -0.23065185546875, 1.1463489532470703, 1.9650382995605469, 1.6142616271972656, 3.185455322265625, 2.5977096557617188, 2.2156753540039062, -2.1726303100585938, 2.4440689086914062, -2.84173583984375, 2.701324462890625, 0.9072971343994141, 3.585113525390625, 0.3621559143066406, 1.2644882202148438, 5.178836822509766, -0.08058738708496094, -1.6389694213867188, 1.8285446166992188, 2.35638427734375, 0.544464111328125, 0.9928874969482422, 0.09124755859375, 4.382484436035156, 2.557842254638672, 0.5836448669433594, -0.13422584533691406, -3.942546844482422, 1.491668701171875, -0.5625991821289062, 1.8740425109863281, 6.9196319580078125, 3.0644454956054688, 5.5369720458984375, 0.45589256286621094, 1.2254104614257812, 1.1854114532470703, 1.1077804565429688, -2.4191436767578125, 0.9002799987792969, -0.02854156494140625, 1.1972732543945312, 0.7211456298828125, -2.1643943786621094, -1.4679336547851562, 4.367053985595703, 0.8521461486816406, 5.673439025878906, -0.9208831787109375, -0.5722427368164062, 1.2428646087646484], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000085.npy"}
|
||||
{"epoch": 0.12849584278155707, "step": 86, "batch_size": 64, "mean": 1.0026484727859497, "std": 2.1038193702697754, "min": -3.5746917724609375, "p10": -1.7377281188964842, "median": 0.8700618743896484, "p90": 4.053602600097657, "max": 6.555938720703125, "pos_frac": 0.640625, "sample": [6.555938720703125, -0.1947040557861328, -1.5925445556640625, 0.054241180419921875, 2.1969470977783203, -0.372039794921875, 0.34865570068359375, 1.1059951782226562, -3.5114059448242188, 1.4771480560302734, -0.28905677795410156, -2.1131820678710938, 4.0639801025390625, 5.325447082519531, -0.9599685668945312, 3.1574630737304688, 1.926309585571289, -0.8355426788330078, -3.5746917724609375, 0.45345306396484375, 4.13165283203125, -0.01586151123046875, -0.14026641845703125, 2.1505985260009766, 4.188789367675781, 0.2557659149169922, 1.540496826171875, 1.3249740600585938, 2.7681427001953125, 0.3465232849121094, 3.2441177368164062, 4.2450103759765625, 2.529266357421875, -2.3175811767578125, 3.66943359375, 2.0148391723632812, -1.7999496459960938, 0.9732437133789062, -0.7896537780761719, 0.42569732666015625, -1.8090972900390625, 2.4978408813476562, 1.8064117431640625, -0.3489418029785156, 1.4058990478515625, 0.5880851745605469, 3.474569320678711, -0.9823112487792969, 1.8290824890136719, -0.5481643676757812, 1.78985595703125, 4.1200408935546875, -0.19998931884765625, -0.32576942443847656, 2.4308815002441406, 4.029388427734375, -2.459390640258789, 2.1311492919921875, 0.6824417114257812, -0.83038330078125, 0.7782783508300781, -0.15359115600585938, 1.3336906433105469, 0.9618453979492188], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000086.npy"}
|
||||
{"epoch": 0.13000755857898716, "step": 87, "batch_size": 64, "mean": 1.0939970016479492, "std": 2.4889278411865234, "min": -5.92559814453125, "p10": -1.7737236022949219, "median": 0.9431571960449219, "p90": 4.34305877685547, "max": 7.08673095703125, "pos_frac": 0.71875, "sample": [1.188995361328125, 6.145111083984375, 7.08673095703125, 0.1627521514892578, 1.4616279602050781, 0.0127716064453125, 3.95916748046875, -1.6337051391601562, 3.3642807006835938, 2.817270278930664, -1.8174476623535156, -1.2699851989746094, 1.5367603302001953, -2.9574642181396484, -5.92559814453125, 5.827964782714844, -2.4202041625976562, 4.419681549072266, 1.2333641052246094, 1.3297576904296875, -1.8031005859375, 0.24971771240234375, 1.537668228149414, -0.7782020568847656, 3.6220169067382812, -4.339378356933594, -0.4508628845214844, -0.17182540893554688, 1.9642791748046875, -1.3323783874511719, -0.81243896484375, 0.7653961181640625, 4.164272308349609, -1.8945083618164062, 1.2371826171875, -1.6668930053710938, 0.45825958251953125, 1.817068099975586, 2.5859546661376953, 0.6028594970703125, 0.5564727783203125, 2.95562744140625, 0.8183975219726562, -1.7051773071289062, 0.7871322631835938, 0.98394775390625, 0.9023666381835938, 3.0677871704101562, -0.2150592803955078, 2.6954498291015625, 3.8670501708984375, 1.1906280517578125, 5.8419952392578125, 0.26430511474609375, 0.3449211120605469, 1.3794021606445312, 0.51666259765625, -0.3536834716796875, 6.00799560546875, 4.541954040527344, 1.50872802734375, 1.2462730407714844, 0.782867431640625, 1.7528457641601562], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000087.npy"}
|
||||
{"epoch": 0.13151927437641722, "step": 88, "batch_size": 64, "mean": 0.834026038646698, "std": 2.997704029083252, "min": -7.7045135498046875, "p10": -2.3865222930908203, "median": 0.6718616485595703, "p90": 3.9537261962890637, "max": 10.546676635742188, "pos_frac": 0.640625, "sample": [6.4336700439453125, 1.085123062133789, -2.363544464111328, -0.9577407836914062, 6.666893005371094, 3.6632843017578125, 6.364959716796875, -0.025859832763671875, 1.2685165405273438, -0.26035118103027344, 3.3005409240722656, 3.55560302734375, 1.6877250671386719, 1.8339405059814453, -0.7387542724609375, 1.7357940673828125, 4.386314392089844, -1.4277572631835938, -2.3321380615234375, 1.4645843505859375, 2.7596435546875, 0.7413120269775391, 0.6400718688964844, -0.43206024169921875, -2.3963699340820312, 1.6595001220703125, -0.056468963623046875, 10.546676635742188, -1.3535614013671875, -2.7107410430908203, -7.6034088134765625, -2.8618698120117188, 3.6314849853515625, 0.4015655517578125, 1.5607528686523438, 0.06905364990234375, -1.0603256225585938, 2.1800403594970703, 1.9665794372558594, 1.11358642578125, -0.0609283447265625, 2.8741226196289062, 2.039396286010742, 0.8704414367675781, -3.1110382080078125, 0.269927978515625, 0.5171241760253906, 0.2876873016357422, 0.1327667236328125, 0.20955848693847656, 2.5812854766845703, 0.7036514282226562, 6.03021240234375, 4.0782012939453125, 2.4810028076171875, -2.1895294189453125, 1.9497146606445312, 0.06975173950195312, -0.2791595458984375, 3.024953842163086, -1.2444190979003906, -7.7045135498046875, -1.756296157836914, -2.532512664794922], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000088.npy"}
|
||||
{"epoch": 0.1330309901738473, "step": 89, "batch_size": 64, "mean": 1.3890717029571533, "std": 2.5747203826904297, "min": -4.1431121826171875, "p10": -0.8015640258789061, "median": 1.1871299743652344, "p90": 4.298174285888672, "max": 11.370391845703125, "pos_frac": 0.71875, "sample": [0.6461963653564453, 1.8984794616699219, -0.09918975830078125, 1.4669342041015625, 3.53594970703125, 4.028522491455078, -0.5752029418945312, 0.6987762451171875, 1.5362014770507812, 4.316734313964844, 1.8749122619628906, 3.7192230224609375, 2.227996826171875, -0.4842987060546875, 0.4137229919433594, 0.40622901916503906, -0.3857154846191406, 1.959991455078125, -0.6177749633789062, -2.9527816772460938, 2.6140594482421875, -1.1910686492919922, 0.17816162109375, 3.643798828125, -0.6046772003173828, 11.370391845703125, 2.7375411987304688, 1.3057403564453125, -0.6628265380859375, 4.2548675537109375, -0.23931884765625, 2.139190673828125, 1.0685195922851562, 0.873321533203125, 4.3478546142578125, -0.08599090576171875, 3.2557296752929688, 1.3138809204101562, -0.86102294921875, 2.518646240234375, 2.7872676849365234, 2.5994224548339844, 6.222969055175781, 7.495916366577148, 0.9276752471923828, 0.051910400390625, 0.0795135498046875, 3.0905303955078125, 4.594821929931641, -3.2980575561523438, -0.6119842529296875, 2.7283706665039062, 0.18672561645507812, 0.44785308837890625, 0.4020347595214844, -3.2892684936523438, 1.7580223083496094, -0.6145172119140625, -3.1305694580078125, -4.1431121826171875, 4.847869873046875, 0.8230743408203125, 1.4293022155761719, 1.9231128692626953], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000089.npy"}
|
||||
{"epoch": 0.1345427059712774, "step": 90, "batch_size": 64, "mean": 1.2850478887557983, "std": 2.545086622238159, "min": -5.029571533203125, "p10": -1.848719787597656, "median": 1.0348339080810547, "p90": 4.232831573486329, "max": 7.193626403808594, "pos_frac": 0.734375, "sample": [-1.1077919006347656, 3.038177490234375, 0.9594249725341797, 2.2589492797851562, 3.5298805236816406, 0.4412384033203125, 3.2698936462402344, 7.193626403808594, -3.9201507568359375, -5.029571533203125, 3.865966796875, 3.278606414794922, 3.0470848083496094, 0.9183502197265625, 2.919464111328125, -1.0453681945800781, 1.2185478210449219, 5.592071533203125, -1.9839706420898438, 0.6106224060058594, 3.8359832763671875, -1.4997787475585938, 6.8394775390625, 4.474433898925781, 6.5270233154296875, 1.0362586975097656, 1.167449951171875, 4.298614501953125, 0.5376663208007812, 0.802978515625, 0.22528839111328125, -0.48728179931640625, 1.6598091125488281, 6.1188812255859375, 1.5173377990722656, 3.5973892211914062, -2.17230224609375, -1.4713363647460938, 4.073066711425781, 1.6947669982910156, -1.5331344604492188, 1.81732177734375, 2.2178955078125, -1.116485595703125, 4.079338073730469, 0.2395458221435547, 0.5558357238769531, 0.6745223999023438, 0.3250160217285156, -0.34032440185546875, -2.9023513793945312, 1.5552101135253906, 2.2840957641601562, -0.087066650390625, 0.7730617523193359, 1.0334091186523438, 0.3888092041015625, -3.0291080474853516, -2.2411117553710938, 2.37542724609375, 1.506256103515625, 2.205718994140625, -0.8104934692382812, 0.4409008026123047], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000090.npy"}
|
||||
{"epoch": 0.1360544217687075, "step": 91, "batch_size": 64, "mean": 1.142439603805542, "std": 2.777684211730957, "min": -6.023582458496094, "p10": -1.8689865112304687, "median": 0.9825344085693359, "p90": 4.425189208984375, "max": 11.21875, "pos_frac": 0.640625, "sample": [3.800933837890625, 3.8802261352539062, 1.4407424926757812, -1.8907852172851562, 0.456298828125, 6.8552398681640625, 0.24657058715820312, 1.1576690673828125, 2.8625030517578125, 7.247993469238281, 11.21875, -0.6366958618164062, 1.199575424194336, -1.4104270935058594, 1.6654834747314453, -2.0548839569091797, -0.74407958984375, -0.6700363159179688, 2.5822067260742188, -3.1614837646484375, 1.8491439819335938, 3.336963653564453, 0.8145980834960938, 0.3862953186035156, 1.1887130737304688, 1.9381389617919922, 3.257966995239258, -1.8922042846679688, -2.6242828369140625, 3.2972946166992188, 0.9878997802734375, -1.7809715270996094, -2.0187835693359375, 1.2595081329345703, 4.885839462280273, -0.6705780029296875, -0.88153076171875, 2.8354263305664062, -0.8914108276367188, -6.023582458496094, 2.6137771606445312, 1.137664794921875, -1.8181228637695312, -1.115386962890625, 0.23848724365234375, 0.9771690368652344, 0.36269378662109375, 5.022706985473633, 1.853759765625, 1.4518051147460938, -0.2532081604003906, -1.659372329711914, 0.09466171264648438, -0.1999197006225586, 4.380451202392578, 3.4571876525878906, 5.953369140625, -0.04618072509765625, 4.444362640380859, 2.4697418212890625, -1.6846084594726562, -0.01447296142578125, 1.7547988891601562, 0.3945178985595703], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000091.npy"}
|
||||
{"epoch": 0.13756613756613756, "step": 92, "batch_size": 64, "mean": 1.6008660793304443, "std": 3.14909029006958, "min": -5.020999908447266, "p10": -1.7174264907836911, "median": 1.1203575134277344, "p90": 5.716589546203614, "max": 11.663116455078125, "pos_frac": 0.71875, "sample": [2.5957794189453125, -0.5509452819824219, -3.3488922119140625, 0.07885360717773438, -1.0303878784179688, -0.7299308776855469, 0.09665298461914062, 0.7783870697021484, 1.5344390869140625, -0.8321819305419922, -1.5486259460449219, 1.522756576538086, 3.9230098724365234, -2.005950927734375, 4.925701141357422, 2.524768829345703, 1.9736862182617188, 0.06728935241699219, 5.517267227172852, 0.8034553527832031, 3.2867431640625, -0.2240753173828125, 1.4670276641845703, 1.1504669189453125, 4.6837310791015625, 0.022626876831054688, -3.2312469482421875, 1.7405319213867188, -3.9808731079101562, 2.7961807250976562, -1.787466049194336, 2.538341522216797, 5.909202575683594, 0.7291507720947266, 3.4552078247070312, 1.0902481079101562, 0.21573638916015625, 3.653900146484375, -3.2628936767578125, 2.6961135864257812, 0.2562408447265625, -0.120025634765625, 6.174766540527344, 3.06646728515625, 1.8231315612792969, -1.2159156799316406, 4.2817535400390625, 6.211174011230469, 3.3443145751953125, -1.5540008544921875, -0.5493621826171875, 10.21661376953125, 1.2208251953125, -0.3750743865966797, 4.059051513671875, -5.020999908447266, 11.663116455078125, 5.223167419433594, 0.4954490661621094, 7.34161376953125, 5.802013397216797, 0.6478519439697266, 0.1321258544921875, 0.08735275268554688], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000092.npy"}
|
||||
{"epoch": 0.13907785336356765, "step": 93, "batch_size": 64, "mean": 1.023630976676941, "std": 3.236750602722168, "min": -5.2625885009765625, "p10": -2.7371765136718746, "median": 0.4124584197998047, "p90": 5.1916038513183596, "max": 13.728691101074219, "pos_frac": 0.609375, "sample": [0.7051925659179688, 2.9988632202148438, 5.123687744140625, 3.895519256591797, -2.8317413330078125, 3.7897415161132812, -2.2893218994140625, -0.8767547607421875, -0.41837120056152344, 2.4171104431152344, 1.3105812072753906, 2.259113311767578, 13.728691101074219, 0.026546478271484375, -0.4589118957519531, 0.3779029846191406, 0.4121665954589844, 3.0673065185546875, -0.8573493957519531, -5.0326080322265625, 2.8784942626953125, 7.70294189453125, -2.2764663696289062, 0.20192718505859375, 2.5237388610839844, -0.33765411376953125, 4.454887390136719, -0.07495498657226562, -1.4105091094970703, 1.1883697509765625, -0.16259765625, 0.3587608337402344, -0.5765323638916016, 5.769866943359375, -1.0961532592773438, -1.1827545166015625, -0.28043365478515625, 0.3123321533203125, -5.2625885009765625, -0.1247406005859375, 0.9134140014648438, 0.412750244140625, 2.236804962158203, 1.7276535034179688, 1.6390151977539062, -3.2722320556640625, 0.8958148956298828, 0.11687850952148438, 5.3089447021484375, 1.4219894409179688, 4.264026641845703, 1.5958499908447266, 5.220710754394531, 2.140777587890625, -2.5165252685546875, -2.491727828979492, 5.541606903076172, -4.834465026855469, -3.0496559143066406, 5.833518981933594, 4.014854431152344, -3.4043350219726562, -1.1707229614257812, 3.014141082763672], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000093.npy"}
|
||||
{"epoch": 0.14058956916099774, "step": 94, "batch_size": 64, "mean": 1.843604326248169, "std": 3.967053174972534, "min": -9.891334533691406, "p10": -1.1066062927246092, "median": 1.8554325103759766, "p90": 5.794982910156251, "max": 14.032012939453125, "pos_frac": 0.75, "sample": [4.0237274169921875, 0.672637939453125, 2.1597366333007812, 3.3554458618164062, 1.1243553161621094, -8.755645751953125, -0.0078125, 0.6180610656738281, 1.1903533935546875, 2.9518775939941406, 0.7711219787597656, 5.5720672607421875, 1.689422607421875, -8.630172729492188, 2.3691749572753906, 5.076465606689453, -1.8450851440429688, -0.4766063690185547, 5.8905181884765625, 2.39251708984375, 2.4060497283935547, 2.5677261352539062, 3.2533531188964844, 2.021442413330078, 2.43780517578125, 1.0766830444335938, 0.2294139862060547, 0.33969879150390625, -3.33758544921875, 3.4415283203125, 3.476367950439453, -0.9278831481933594, 0.4388446807861328, -0.3728485107421875, 3.016357421875, 3.286052703857422, 3.6817245483398438, 2.6508941650390625, 3.8125648498535156, 0.4792022705078125, 6.282764434814453, 6.207679748535156, -0.4174232482910156, 12.48583984375, 4.414592742919922, 4.4564208984375, -0.93841552734375, 11.355133056640625, 4.1579742431640625, 2.434854507446289, 14.032012939453125, 0.9605178833007812, -0.21811676025390625, -0.7380161285400391, 2.092660903930664, 0.5109100341796875, -9.891334533691406, 1.5704631805419922, 7.3907012939453125, 1.6218719482421875, 0.04513359069824219, -1.1786880493164062, -2.1544857025146484, -0.6119308471679688], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000094.npy"}
|
||||
{"epoch": 0.1421012849584278, "step": 95, "batch_size": 64, "mean": 1.5847429037094116, "std": 3.448092460632324, "min": -6.334503173828125, "p10": -2.5594512939453122, "median": 1.0379228591918945, "p90": 6.555044555664063, "max": 8.811355590820312, "pos_frac": 0.640625, "sample": [6.462738037109375, -2.9766464233398438, 1.3567390441894531, -0.6595039367675781, -1.9305267333984375, -0.17559051513671875, 8.811355590820312, 1.049478530883789, 2.5716705322265625, 1.4301910400390625, -1.426544189453125, 1.0263671875, 0.0852508544921875, -0.4777565002441406, -0.6840362548828125, 0.8602676391601562, -3.5863800048828125, -2.04681396484375, -0.2902393341064453, 1.8964347839355469, 7.289764404296875, 6.5946044921875, -0.4763031005859375, 0.6609306335449219, -5.538421630859375, -0.42779541015625, 7.5867919921875, 6.9127044677734375, -3.3577041625976562, -6.334503173828125, 2.833293914794922, 7.427452087402344, 0.8288002014160156, 2.2562026977539062, -0.6557540893554688, 1.4020156860351562, 5.704505920410156, -1.0316848754882812, -3.5185546875, 2.4476242065429688, 3.293853759765625, 0.3298530578613281, 5.507469177246094, 2.75164794921875, -2.6763458251953125, 4.80255126953125, 5.2091064453125, 0.19771575927734375, 2.0259437561035156, 6.450843811035156, -0.1639404296875, 4.694297790527344, -0.31806373596191406, 3.441436767578125, 5.292510986328125, 8.2103271484375, -1.1753387451171875, 0.37781524658203125, 1.5235462188720703, 3.7540435791015625, 0.47206687927246094, 1.806924819946289, 6.001556396484375, -2.2866973876953125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000095.npy"}
|
||||
{"epoch": 0.1436130007558579, "step": 96, "batch_size": 64, "mean": 1.8668088912963867, "std": 4.589854717254639, "min": -7.9406585693359375, "p10": -2.3931632995605465, "median": 1.2683124542236328, "p90": 8.1637451171875, "max": 16.0352783203125, "pos_frac": 0.671875, "sample": [3.2548179626464844, -2.4854507446289062, 8.469558715820312, -0.37236785888671875, 11.736907958984375, -7.9406585693359375, 3.838043212890625, 0.4274768829345703, 2.3785018920898438, 0.2689666748046875, 4.8149261474609375, -0.6270656585693359, 1.7578277587890625, 10.346389770507812, 3.0320281982421875, 1.8505134582519531, -5.102668762207031, 0.01197052001953125, 1.2171859741210938, -0.9131240844726562, 2.4744911193847656, 2.8599624633789062, 8.1748046875, -5.474433898925781, 0.66119384765625, 0.8375492095947266, 5.15087890625, -0.6518325805664062, -2.177825927734375, -0.1045074462890625, 1.8984336853027344, 8.137939453125, 16.0352783203125, 5.257347106933594, -1.2741565704345703, -2.1050949096679688, -7.407218933105469, 0.5711803436279297, -3.0447845458984375, 1.8078231811523438, 3.4561920166015625, 1.2034072875976562, 1.7862491607666016, 0.476287841796875, 13.303558349609375, -1.864593505859375, 3.927886962890625, -0.470123291015625, 1.3194389343261719, -0.46030426025390625, 4.19964599609375, -1.621673583984375, 0.9714851379394531, 6.779487609863281, 10.684135437011719, 2.6953201293945312, -1.4063663482666016, 0.29704856872558594, 1.470998764038086, 3.7703323364257812, 7.271049499511719, 1.6938323974609375, -1.8390579223632812, -5.759281158447266], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000096.npy"}
|
||||
{"epoch": 0.14512471655328799, "step": 97, "batch_size": 64, "mean": 0.9901017546653748, "std": 5.087546348571777, "min": -18.111846923828125, "p10": -3.283107757568359, "median": 0.8274612426757812, "p90": 5.989920043945316, "max": 17.760238647460938, "pos_frac": 0.578125, "sample": [-10.258026123046875, -1.5750045776367188, -0.0817718505859375, 7.579233169555664, 0.8517074584960938, 12.774330139160156, -3.4220504760742188, -3.4624176025390625, -2.3344802856445312, -18.111846923828125, -2.6935272216796875, -0.8790817260742188, 0.8032150268554688, 2.7027721405029297, -4.1539306640625, -9.493385314941406, 0.3731346130371094, 1.2810993194580078, 2.866107940673828, 0.3661479949951172, 1.0083656311035156, 3.7941131591796875, 1.5563716888427734, -2.4292373657226562, 3.7322158813476562, -0.8988418579101562, -1.1378326416015625, 5.077033996582031, -1.37969970703125, 11.047035217285156, 1.5874404907226562, -2.8306121826171875, 3.0924911499023438, 0.5846405029296875, 1.1091728210449219, 4.030029296875, 7.901506423950195, 6.381156921386719, -1.0254268646240234, 3.93353271484375, 2.7793045043945312, -5.016033172607422, 17.760238647460938, 2.552143096923828, 3.7028274536132812, -1.5929145812988281, 3.150766372680664, 3.7189407348632812, 9.487457275390625, -2.9589080810546875, -0.157806396484375, 2.1389999389648438, -0.0103912353515625, -1.2569732666015625, -1.1821823120117188, 3.0743942260742188, 3.974344253540039, -2.3979568481445312, 2.484285354614258, 2.3845138549804688, 0.7649497985839844, -0.182708740234375, 2.6771888732910156, -0.7936534881591797], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000097.npy"}
|
||||
{"epoch": 0.14663643235071808, "step": 98, "batch_size": 64, "mean": 0.6617430448532104, "std": 3.23582124710083, "min": -8.46221923828125, "p10": -2.801913261413574, "median": 1.0666656494140625, "p90": 4.7046638488769545, "max": 7.24432373046875, "pos_frac": 0.625, "sample": [-0.15818405151367188, 1.4408550262451172, 5.265171051025391, 0.7311363220214844, 2.347564697265625, 1.8751983642578125, 2.2188644409179688, 1.8385848999023438, -1.2149734497070312, 6.434867858886719, -8.46221923828125, -2.789609909057617, -1.5108489990234375, -1.4628372192382812, -2.7854690551757812, -0.37453460693359375, -0.3595542907714844, 1.9596443176269531, 1.2150897979736328, -0.30267333984375, 3.392608642578125, 6.3499603271484375, -3.9136962890625, 4.81585693359375, 0.6777305603027344, -1.7508354187011719, 3.4614181518554688, 1.2845191955566406, 1.1921043395996094, 0.8051223754882812, 1.9008560180664062, 1.4040451049804688, 6.018516540527344, 1.8269195556640625, -2.8332786560058594, 1.0703353881835938, 1.4521369934082031, 2.6121139526367188, 0.9138240814208984, -0.1944580078125, 2.558807373046875, 0.15538787841796875, -7.9316864013671875, 1.5727005004882812, -1.52252197265625, -3.7766647338867188, 1.8827362060546875, 1.0629959106445312, -1.675445556640625, 5.588844299316406, -2.3980026245117188, 4.116279602050781, -1.967681884765625, -2.0524749755859375, -8.23138427734375, 3.9225616455078125, 2.580303192138672, 0.05398368835449219, 0.6477241516113281, 7.24432373046875, 2.8500328063964844, 4.445213317871094, -0.359161376953125, -2.8071861267089844], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000098.npy"}
|
||||
{"epoch": 0.14814814814814814, "step": 99, "batch_size": 64, "mean": 1.0287803411483765, "std": 4.698493480682373, "min": -12.668609619140625, "p10": -3.4788864135742186, "median": 0.8994102478027344, "p90": 6.355838394165039, "max": 15.670318603515625, "pos_frac": 0.625, "sample": [8.680274963378906, 7.39006233215332, 6.834381103515625, 11.191093444824219, 0.5919113159179688, -3.2373313903808594, -0.39887237548828125, -1.6414756774902344, 0.6923294067382812, 6.334403991699219, 2.751007080078125, 1.28167724609375, 2.1634902954101562, 4.667945861816406, -0.37072181701660156, 0.9932327270507812, 0.8326034545898438, 0.09092140197753906, -1.0267219543457031, 1.5983829498291016, 6.268470764160156, 0.3223915100097656, -5.380767822265625, -2.4276351928710938, -5.3133697509765625, -3.4914703369140625, 15.670318603515625, -9.367202758789062, -2.470947265625, 6.365024566650391, 1.5835342407226562, -3.44952392578125, -0.8283767700195312, 0.31435394287109375, 1.1673164367675781, 1.3310470581054688, 1.2257957458496094, -3.285991668701172, 0.19886016845703125, 5.569438934326172, 0.966217041015625, -12.668609619140625, 5.7564544677734375, 11.25408935546875, -0.4295654296875, 1.735321044921875, 6.216300964355469, 0.8160667419433594, -8.994049072265625, 1.2735309600830078, -0.12287521362304688, 3.7017860412597656, 2.0827102661132812, 1.0166053771972656, 2.709564208984375, -0.33643341064453125, -0.003017425537109375, 4.396949768066406, 1.822824478149414, -1.1942291259765625, 1.2275772094726562, -2.1018943786621094, -2.4165878295898438, -4.286655426025391], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000099.npy"}
|
||||
{"epoch": 0.14965986394557823, "step": 100, "batch_size": 64, "mean": 1.509624719619751, "std": 6.2507734298706055, "min": -17.116500854492188, "p10": -3.8107315063476555, "median": 1.177337646484375, "p90": 7.882118225097659, "max": 25.99053955078125, "pos_frac": 0.671875, "sample": [1.61273193359375, -1.6528549194335938, -4.5822601318359375, 2.8717269897460938, 8.183624267578125, -1.2096405029296875, 0.5309181213378906, 3.2584075927734375, 6.3900146484375, 4.4577178955078125, 1.1674957275390625, 1.216644287109375, 4.530387878417969, 12.498497009277344, 1.8285713195800781, 4.558036804199219, -1.0956878662109375, 1.6322822570800781, 0.6839008331298828, 1.1684112548828125, 0.6661834716796875, 0.4683990478515625, -5.4729461669921875, 0.3443603515625, -4.159149169921875, 5.2657928466796875, -1.9498023986816406, 9.004531860351562, 1.2180709838867188, -14.45184326171875, -0.7544898986816406, -0.4730224609375, -2.7897109985351562, 1.1862640380859375, 0.8910903930664062, -0.5585098266601562, 3.6345443725585938, 0.3198509216308594, -12.864860534667969, 5.7095184326171875, 1.43896484375, 18.240447998046875, -4.217334747314453, 0.6406707763671875, -2.050373077392578, -0.6356925964355469, 2.7842979431152344, 1.418935775756836, -1.1285934448242188, -1.094390869140625, 7.1786041259765625, 8.557449340820312, -1.192657470703125, 2.7274208068847656, 1.931121826171875, 1.322195053100586, -2.9977569580078125, 12.17437744140625, 5.015373229980469, 1.3915481567382812, 25.99053955078125, 0.2716064453125, -17.116500854492188, 2.682525634765625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000100.npy"}
|
||||
{"epoch": 0.15117157974300832, "step": 101, "batch_size": 64, "mean": 2.062567710876465, "std": 4.268919944763184, "min": -11.400726318359375, "p10": -2.0307256698608396, "median": 1.5417251586914062, "p90": 7.0913381576538095, "max": 13.12396240234375, "pos_frac": 0.734375, "sample": [-1.3225517272949219, -0.6190032958984375, 3.2587356567382812, 0.5762252807617188, 6.9497528076171875, 6.394866943359375, 4.152492523193359, 0.8463211059570312, -0.3087310791015625, -0.14368820190429688, 0.9704475402832031, 4.85711669921875, -11.400726318359375, 7.927459716796875, 4.651123046875, -0.028766632080078125, 4.412788391113281, 2.479297637939453, -1.2352371215820312, -3.4059219360351562, 4.501270294189453, 0.7533950805664062, 5.2895050048828125, -1.5022315979003906, 3.7172012329101562, -1.5861053466796875, 4.120330810546875, 4.197929382324219, 0.9562091827392578, 6.527032852172852, 0.744384765625, 11.33856201171875, 5.558713912963867, 0.5583000183105469, -0.8471450805664062, -2.4688587188720703, 1.1243133544921875, -6.471199035644531, -2.2212772369384766, 4.2452545166015625, -5.1437530517578125, 4.0562744140625, 1.3252544403076172, 1.5684585571289062, 7.264503479003906, 2.1785736083984375, 0.529815673828125, -9.423477172851562, 0.21878623962402344, 1.4731407165527344, 7.274528503417969, 0.5302524566650391, 7.152017593383789, 2.8722286224365234, 3.3953475952148438, 0.4525604248046875, 13.12396240234375, 2.8156356811523438, -0.5700302124023438, 2.414295196533203, 5.930328369140625, 3.1105804443359375, 10.392459869384766, 1.5149917602539062], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000101.npy"}
|
||||
{"epoch": 0.15268329554043839, "step": 102, "batch_size": 64, "mean": 0.8552302122116089, "std": 5.451023101806641, "min": -9.854393005371094, "p10": -4.598507690429687, "median": 0.37966156005859375, "p90": 6.744478225708008, "max": 20.998291015625, "pos_frac": 0.5625, "sample": [6.747779846191406, -3.326629638671875, -1.9843177795410156, 1.3813095092773438, -7.3533172607421875, -1.3182239532470703, -4.3406219482421875, 2.8350601196289062, -8.372764587402344, 0.12054443359375, 2.4259033203125, 3.8963241577148438, 0.11133575439453125, 1.7483978271484375, 7.0298614501953125, -3.284942626953125, -2.255260467529297, 16.619033813476562, -3.5359344482421875, 10.37945556640625, -3.010538101196289, 0.9310169219970703, -2.8227386474609375, -1.023284912109375, 2.8836612701416016, 2.756439208984375, 0.6746673583984375, 6.736774444580078, -8.450454711914062, 1.08660888671875, 6.430274963378906, 1.2549362182617188, -4.7090301513671875, 0.5752906799316406, -3.9934768676757812, -0.9213485717773438, 3.8316802978515625, 7.952239990234375, 11.725173950195312, -2.9485702514648438, -0.3534889221191406, 1.3913421630859375, -9.854393005371094, -0.685821533203125, -1.6634407043457031, 0.2577667236328125, 0.501556396484375, 5.624675750732422, 0.22592735290527344, -1.3248977661132812, 2.271270751953125, 4.631378173828125, 1.6748580932617188, -4.762401580810547, -1.4359016418457031, -3.8920631408691406, -1.559356689453125, 20.998291015625, -3.065673828125, -6.9946441650390625, 3.647045135498047, 5.5302581787109375, 5.9524688720703125, 1.1376628875732422], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000102.npy"}
|
||||
{"epoch": 0.15419501133786848, "step": 103, "batch_size": 64, "mean": 1.53092360496521, "std": 3.8010010719299316, "min": -6.654621124267578, "p10": -2.7128883361816407, "median": 0.9538240432739258, "p90": 6.855435180664064, "max": 10.389122009277344, "pos_frac": 0.640625, "sample": [-3.631256103515625, -0.7975692749023438, 7.685089111328125, -6.654621124267578, 0.18070411682128906, 3.902843475341797, -1.0051803588867188, -0.929107666015625, 1.0081253051757812, 7.612331390380859, -2.488636016845703, 1.692718505859375, -2.63037109375, 1.2521438598632812, -2.0126876831054688, 5.2924957275390625, 2.4673385620117188, -1.0128250122070312, -4.017791748046875, 6.029094696044922, 1.4425277709960938, 4.526645660400391, -0.029817581176757812, 2.746234893798828, 2.6407394409179688, -0.17102813720703125, -1.1604118347167969, 9.44561767578125, -2.7482528686523438, 2.8438796997070312, 4.239341735839844, 6.3509979248046875, 8.6817626953125, -1.6423110961914062, 0.4957427978515625, 4.315216064453125, -6.533821105957031, 3.90484619140625, 0.35315704345703125, 0.5660457611083984, 5.178092956542969, 0.7178173065185547, 0.8995227813720703, 2.256214141845703, 3.3334598541259766, 0.26629638671875, 6.562702178955078, -2.195270538330078, 6.980892181396484, 0.4464073181152344, 9.14520263671875, 2.9437255859375, -4.977638244628906, 0.8800315856933594, -3.4706573486328125, -0.2276611328125, -1.596893310546875, -0.9028358459472656, -1.3612079620361328, 1.3136558532714844, 3.5483627319335938, 3.3943328857421875, 10.389122009277344, 2.2454833984375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000103.npy"}
|
||||
{"epoch": 0.15570672713529857, "step": 104, "batch_size": 64, "mean": 2.2824254035949707, "std": 6.508499622344971, "min": -12.275619506835938, "p10": -4.2904533386230455, "median": 1.3072013854980469, "p90": 9.078355407714847, "max": 28.619842529296875, "pos_frac": 0.671875, "sample": [9.370330810546875, 4.916046142578125, 3.8218307495117188, -1.8988380432128906, -4.729148864746094, 2.3233642578125, -1.5362396240234375, 1.0889511108398438, 4.464454650878906, -0.69287109375, 0.8382415771484375, -12.275619506835938, 28.619842529296875, 7.4188079833984375, -5.951194763183594, -2.119518280029297, 13.78961181640625, 2.192605972290039, -0.5813026428222656, 7.915763854980469, -5.3451080322265625, -9.913436889648438, 7.32867431640625, 8.397079467773438, -0.007568359375, -0.6897621154785156, 0.3867149353027344, -7.642280578613281, 5.417287826538086, -0.4918346405029297, 0.4963207244873047, 0.6556892395019531, -0.8410587310791016, 2.386730194091797, 8.115276336669922, 2.7921199798583984, 1.3035964965820312, -1.0407047271728516, 3.1448516845703125, 2.2605743408203125, 1.3407821655273438, -3.0210018157958984, 18.24591064453125, 1.0094852447509766, 3.5413970947265625, 11.175926208496094, 3.3586196899414062, 1.3108062744140625, -9.850425720214844, 0.2589874267578125, 1.6825008392333984, 2.9909801483154297, 6.354602813720703, 4.509838104248047, 1.2726249694824219, -3.2668304443359375, 0.13514328002929688, -1.396728515625, 0.7778949737548828, 1.431060791015625, -2.7247467041015625, 5.993011474609375, 12.1519775390625, 15.105140686035156], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000104.npy"}
|
||||
{"epoch": 0.15721844293272866, "step": 105, "batch_size": 64, "mean": 1.8524706363677979, "std": 5.00480842590332, "min": -7.685211181640625, "p10": -3.887051391601562, "median": 1.3391494750976562, "p90": 8.027920532226565, "max": 19.049468994140625, "pos_frac": 0.65625, "sample": [1.7897720336914062, 5.4939422607421875, 2.8270721435546875, 3.5814132690429688, 1.5981121063232422, 5.158576965332031, -7.6159210205078125, 10.4974365234375, -3.3109207153320312, 2.361581802368164, 0.44475555419921875, 4.1281585693359375, -1.4274063110351562, 8.27947998046875, 0.09890365600585938, 8.506439208984375, -0.9761810302734375, -6.191459655761719, -0.9401626586914062, 3.9264984130859375, 3.0483055114746094, -2.1397247314453125, 1.4029464721679688, -4.133964538574219, 1.2753524780273438, 4.287353515625, 0.7357025146484375, -5.78350830078125, 3.2436294555664062, -6.3943939208984375, -2.1754150390625, 0.8651714324951172, -1.593505859375, 1.96337890625, -7.685211181640625, 19.049468994140625, 1.1856212615966797, -0.1219024658203125, 2.0156097412109375, 0.10604476928710938, -0.21743392944335938, -1.0845680236816406, 0.5727691650390625, 1.7173042297363281, -6.3340301513671875, 5.097766876220703, 5.1217498779296875, -0.7270584106445312, 5.7783203125, -2.2002182006835938, -2.8490753173828125, 2.2511520385742188, 1.0227890014648438, 11.303138732910156, 14.2530517578125, 0.6943359375, 7.213947296142578, 6.0264129638671875, -0.4693145751953125, 2.579488754272461, -0.960723876953125, 10.364479064941406, 4.5818328857421875, 7.440948486328125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000105.npy"}
|
||||
{"epoch": 0.15873015873015872, "step": 106, "batch_size": 64, "mean": 1.5819799900054932, "std": 5.403035640716553, "min": -16.891067504882812, "p10": -5.526586914062499, "median": 1.6561393737792969, "p90": 6.505997848510743, "max": 15.8011474609375, "pos_frac": 0.6875, "sample": [-6.4708404541015625, 5.051673889160156, 3.785022735595703, 5.083469390869141, -6.7851104736328125, -1.03485107421875, 9.009834289550781, 2.902507781982422, 1.3964920043945312, 1.0861968994140625, 4.987113952636719, 6.140171051025391, 10.403633117675781, 4.266853332519531, 0.5194187164306641, -3.6329193115234375, 1.7218780517578125, 5.952178955078125, 3.6485671997070312, 0.5667266845703125, -0.6015205383300781, 0.16177749633789062, -8.658645629882812, 3.131439208984375, -0.9702825546264648, -0.3713226318359375, -0.43732261657714844, -4.2858428955078125, 1.8170623779296875, 14.942840576171875, -0.4277915954589844, -0.4894123077392578, -1.7445220947265625, -2.7979583740234375, 5.593841552734375, 5.1974945068359375, 0.8575210571289062, -1.3200111389160156, 5.523563385009766, 15.8011474609375, 4.40325927734375, 2.8707618713378906, 4.753488540649414, 0.8082389831542969, -6.0583343505859375, 0.998260498046875, 1.6218585968017578, 1.7825355529785156, -16.891067504882812, 1.4826221466064453, -9.894866943359375, 1.7614669799804688, 5.36431884765625, 4.646434783935547, 6.66278076171875, 8.733406066894531, 0.5385093688964844, 2.4295482635498047, -0.7112884521484375, 10.379287719726562, -8.147445678710938, 1.690420150756836, 0.200103759765625, 2.3023433685302734], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000106.npy"}
|
||||
{"epoch": 0.1602418745275888, "step": 107, "batch_size": 64, "mean": 0.8538090586662292, "std": 6.318426609039307, "min": -18.36810302734375, "p10": -5.240845489501953, "median": 1.3153047561645508, "p90": 7.268230056762697, "max": 18.979095458984375, "pos_frac": 0.640625, "sample": [1.319711685180664, -5.158638000488281, 2.7776947021484375, 0.08567047119140625, 6.311614990234375, 4.137168884277344, -16.616058349609375, 6.985694885253906, -0.08223724365234375, 2.302764892578125, -18.36810302734375, -0.19005584716796875, 1.6451072692871094, 1.3108978271484375, 3.8259220123291016, -3.755340576171875, -3.166229248046875, 5.163909912109375, 4.00672721862793, 0.23175048828125, -1.5113906860351562, 10.94818115234375, 8.988525390625, 0.5230045318603516, 4.523876190185547, -9.899971008300781, 3.940906524658203, -2.8640823364257812, -1.2454605102539062, 1.2941360473632812, 3.0141525268554688, -7.174251556396484, 2.439727783203125, 5.932392120361328, 0.9122772216796875, 5.692474365234375, 6.228755950927734, 1.2755126953125, 2.7385330200195312, 9.789161682128906, 11.92431640625, 7.389316558837891, -4.768959045410156, -2.1490402221679688, 3.1131515502929688, 0.5433273315429688, -0.057926177978515625, 2.680999755859375, 2.6266002655029297, -11.224288940429688, 9.048042297363281, 3.2933311462402344, -2.2419986724853516, -3.856475830078125, -0.16662025451660156, 0.4906158447265625, 2.243560791015625, 18.979095458984375, 2.822755813598633, -2.6595230102539062, -5.2760772705078125, -14.810882568359375, 1.62847900390625, -3.2424545288085938], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000107.npy"}
|
||||
{"epoch": 0.1617535903250189, "step": 108, "batch_size": 64, "mean": 1.8839268684387207, "std": 5.535211086273193, "min": -12.017410278320312, "p10": -4.90072021484375, "median": 1.6515655517578125, "p90": 8.278745269775394, "max": 19.325149536132812, "pos_frac": 0.609375, "sample": [2.680419921875, 4.556793212890625, 10.114883422851562, -0.7854995727539062, 0.022584915161132812, 5.253044128417969, 4.685455322265625, 4.092353820800781, 6.2772369384765625, -4.96356201171875, 1.6212158203125, 6.0312652587890625, 3.572296142578125, -1.4002971649169922, 5.7278289794921875, 1.7396469116210938, -3.1232986450195312, -0.5204982757568359, 0.8241195678710938, 0.613250732421875, 7.29046630859375, -0.06966400146484375, -0.5931377410888672, 5.54779052734375, 0.7098236083984375, -6.0266265869140625, 3.4949417114257812, -12.017410278320312, 12.764846801757812, 0.5199203491210938, -9.929412841796875, -1.7859783172607422, 6.844825744628906, -0.7382850646972656, 8.970558166503906, 5.128824234008789, -0.014186859130859375, -0.5030422210693359, -6.418792724609375, 5.437042236328125, -4.3105621337890625, 8.702293395996094, 2.5479984283447266, -1.5875320434570312, -0.29833221435546875, 2.3084754943847656, -1.4210147857666016, 2.950540542602539, 1.681915283203125, -4.75408935546875, -6.982280731201172, 3.7023544311523438, -5.011505126953125, 2.5141754150390625, 11.598220825195312, 0.048892974853515625, 3.1478195190429688, 19.325149536132812, 1.6944808959960938, -0.505126953125, -0.5103759765625, 5.517662048339844, -1.693817138671875, 16.27423095703125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000108.npy"}
|
||||
{"epoch": 0.16326530612244897, "step": 109, "batch_size": 64, "mean": 3.055436372756958, "std": 5.453472137451172, "min": -4.931694030761719, "p10": -1.5741687774658204, "median": 1.5789213180541992, "p90": 8.33336410522461, "max": 24.705322265625, "pos_frac": 0.71875, "sample": [-0.13593292236328125, 13.178947448730469, 15.811798095703125, 0.5804557800292969, 4.373838424682617, 3.7545433044433594, 7.211887359619141, 0.712493896484375, 2.1513614654541016, 4.803073883056641, -0.3886566162109375, 1.37445068359375, 1.7920799255371094, 6.049167633056641, 2.724365234375, -3.809112548828125, 2.508058547973633, 0.6608753204345703, -1.099273681640625, 1.1254005432128906, 1.3731136322021484, 4.787200927734375, -3.8715362548828125, -0.4944295883178711, -1.285858154296875, 24.705322265625, 0.6893234252929688, -0.44415283203125, 4.111734390258789, 20.122589111328125, 3.652250289916992, 1.5700912475585938, 1.4234199523925781, 0.16031646728515625, 0.26305198669433594, 2.3743209838867188, -1.454559326171875, 5.6844940185546875, -0.60394287109375, 4.753364562988281, -1.5773124694824219, 4.500017166137695, 1.7205772399902344, 6.5007171630859375, -0.9844570159912109, -1.56683349609375, 0.384765625, 1.5877513885498047, -4.931694030761719, -3.483844757080078, 1.2156982421875, 3.4321441650390625, 5.125999450683594, 7.8894500732421875, 8.523612976074219, 14.680068969726562, 11.368717193603516, 1.1849517822265625, 5.658210754394531, 3.5459442138671875, -3.325714111328125, -1.0173931121826172, -2.260955810546875, 6.481575012207031], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000109.npy"}
|
||||
{"epoch": 0.16477702191987906, "step": 110, "batch_size": 64, "mean": 0.33765092492103577, "std": 5.487300395965576, "min": -20.207717895507812, "p10": -5.546797943115234, "median": 0.8136634826660156, "p90": 6.793875122070315, "max": 13.5562744140625, "pos_frac": 0.609375, "sample": [1.8258895874023438, 0.33935546875, -3.6363677978515625, 1.1346931457519531, -0.7120285034179688, 6.9836273193359375, 3.8105316162109375, -0.8101959228515625, -10.106475830078125, 1.5807037353515625, 5.364910125732422, 3.3220176696777344, 0.7981185913085938, 7.474647521972656, 0.6317520141601562, -3.1853408813476562, 0.5346336364746094, 10.776351928710938, -0.8282871246337891, -3.935028076171875, 2.3966407775878906, -2.3400955200195312, 2.0825557708740234, 6.3511199951171875, 0.4451942443847656, 2.6468353271484375, 0.9475994110107422, 1.0874061584472656, 0.8292083740234375, -5.853569030761719, -3.138153076171875, 2.1877899169921875, 2.309101104736328, -1.4847297668457031, -5.685447692871094, 0.846588134765625, -5.2232818603515625, 5.572265625, -1.3651809692382812, 1.6668357849121094, 8.67071533203125, 4.62225341796875, -0.9709930419921875, -2.6112594604492188, -0.9132423400878906, 2.5093154907226562, 4.548583984375, -6.90606689453125, -8.22442626953125, -16.7418212890625, -0.6368408203125, 1.9023666381835938, -0.7946395874023438, 0.15057373046875, -0.67364501953125, 13.5562744140625, 8.828208923339844, 3.589143753051758, 1.8148307800292969, 7.0410003662109375, -4.372026443481445, 1.14886474609375, 0.6380157470703125, -20.207717895507812], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000110.npy"}
|
||||
{"epoch": 0.16628873771730915, "step": 111, "batch_size": 64, "mean": 2.008547782897949, "std": 5.216946125030518, "min": -7.504539489746094, "p10": -3.18231201171875, "median": 0.983057975769043, "p90": 9.379944992065429, "max": 19.31585693359375, "pos_frac": 0.5625, "sample": [-1.027151107788086, -2.7044200897216797, -0.40847015380859375, 12.576705932617188, 2.353801727294922, 7.7093963623046875, -1.1262168884277344, 3.6986541748046875, 2.379375457763672, 0.2701549530029297, 5.755542755126953, -0.949066162109375, -3.8332061767578125, 2.387786865234375, -6.891242980957031, -0.013555526733398438, -0.31505584716796875, 0.9443607330322266, -2.0053863525390625, 2.4114322662353516, 10.101631164550781, 5.73541259765625, 11.263404846191406, 9.372257232666016, 12.325927734375, 4.089471817016602, 4.294189453125, 6.047142028808594, 2.074369430541992, -7.504539489746094, -0.18294906616210938, -2.7485904693603516, -3.7227783203125, -6.223262786865234, -0.143280029296875, -2.9692230224609375, -0.14275169372558594, 4.8196868896484375, -2.188983917236328, 5.806098937988281, 9.38323974609375, 8.591758728027344, 10.101882934570312, -1.7869625091552734, -1.8179512023925781, -0.5932292938232422, 3.138113021850586, 8.683658599853516, -2.579071044921875, 1.0217552185058594, 5.655387878417969, 3.4475746154785156, 3.0484561920166016, -2.649993896484375, -5.647880554199219, 0.8803749084472656, -3.2736358642578125, 3.1108169555664062, -2.9015541076660156, 1.4793510437011719, 19.31585693359375, -2.1078872680664062, 2.205585479736328, 0.5247344970703125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000111.npy"}
|
||||
{"epoch": 0.16780045351473924, "step": 112, "batch_size": 64, "mean": 2.420254707336426, "std": 4.793520927429199, "min": -10.24365234375, "p10": -2.69803466796875, "median": 2.487314224243164, "p90": 8.375328063964846, "max": 13.14056396484375, "pos_frac": 0.734375, "sample": [11.830413818359375, 7.0517425537109375, 9.31182861328125, 4.038745880126953, 0.023153305053710938, 1.4239501953125, 12.49560546875, 3.5460243225097656, 5.633026123046875, 3.1303672790527344, 0.264923095703125, 6.728790283203125, -4.808967590332031, 2.5187110900878906, 1.0961475372314453, -7.36322021484375, -2.5050048828125, 4.0440216064453125, -5.0415496826171875, -0.9437179565429688, 2.9424972534179688, -2.78076171875, 3.2238197326660156, 6.034172058105469, 2.2702713012695312, 2.87628173828125, 5.531494140625, 1.8648796081542969, 13.14056396484375, 3.575845718383789, 7.9276123046875, 1.2845001220703125, -2.3802242279052734, 0.9122657775878906, 6.194957733154297, -0.309722900390625, 2.4559173583984375, -6.032020568847656, 8.560104370117188, 2.9257125854492188, -1.6385841369628906, 3.444610595703125, 5.142917633056641, 2.5446109771728516, 2.6339492797851562, 11.261215209960938, 5.8174896240234375, 5.964836120605469, -0.8461952209472656, 0.7415046691894531, -8.123451232910156, 2.1589279174804688, -0.5609207153320312, 1.8889274597167969, 7.944183349609375, 1.032257080078125, 2.014423370361328, -2.4009227752685547, -2.1017837524414062, 4.25177001953125, -0.592559814453125, -10.24365234375, 1.8870162963867188, 9.982582092285156], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000112.npy"}
|
||||
{"epoch": 0.1693121693121693, "step": 113, "batch_size": 64, "mean": 1.3413527011871338, "std": 4.687411308288574, "min": -15.402099609375, "p10": -3.5617118835449215, "median": 0.9732570648193359, "p90": 6.617084121704104, "max": 14.320526123046875, "pos_frac": 0.625, "sample": [-2.384246826171875, -6.025238037109375, -2.4411849975585938, 0.9578056335449219, -0.3943061828613281, 6.007175445556641, 3.039337158203125, -8.628654479980469, 0.2050018310546875, 3.559316635131836, 1.6154518127441406, -1.4178237915039062, -3.2497177124023438, 2.469696044921875, -3.7075538635253906, -0.044219970703125, 1.7306365966796875, 8.33443832397461, -0.22967910766601562, -3.6775970458984375, -2.89471435546875, 3.3616256713867188, 6.832218170166016, -5.1036224365234375, -0.4964275360107422, 3.045684814453125, 4.394840240478516, 2.98651123046875, 7.235752105712891, -0.5521831512451172, 3.691356658935547, 14.320526123046875, 0.7632846832275391, -0.6335868835449219, 3.5292396545410156, 7.291107177734375, -15.402099609375, 3.379629135131836, 5.0555419921875, -1.3878097534179688, 1.4879131317138672, 4.921117782592773, -4.705322265625, 5.962638854980469, -3.2442169189453125, 9.116744995117188, 0.98870849609375, 0.15048599243164062, -0.32871055603027344, 4.618936538696289, 3.1115264892578125, -2.91925048828125, 1.8213310241699219, 11.9510498046875, 5.2587127685546875, 0.036457061767578125, 3.0794219970703125, 0.4714317321777344, 4.842376708984375, 0.411865234375, 0.8693084716796875, -0.0152587890625, -3.2913131713867188, 6.115104675292969], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000113.npy"}
|
||||
{"epoch": 0.1708238851095994, "step": 114, "batch_size": 64, "mean": 2.3785853385925293, "std": 4.761023998260498, "min": -5.7782745361328125, "p10": -3.4326423645019526, "median": 1.5142135620117188, "p90": 8.358612442016605, "max": 16.51336669921875, "pos_frac": 0.6875, "sample": [-5.7782745361328125, -3.5693740844726562, 0.9382896423339844, 6.638599395751953, 4.406198501586914, 1.6624221801757812, 3.4047698974609375, 6.903709411621094, 4.3674774169921875, -2.4846134185791016, -3.817981719970703, 1.3071479797363281, 3.5886878967285156, -2.908905029296875, -0.3964385986328125, 1.4618911743164062, 3.17657470703125, -0.2249755859375, 0.4740753173828125, -0.9196662902832031, 3.1568374633789062, -2.8366012573242188, 16.51336669921875, -2.8240928649902344, 0.6758537292480469, 6.940284729003906, 4.291515350341797, 5.334386825561523, 0.7759246826171875, 6.770570755004883, -1.9338417053222656, 1.5665359497070312, 8.959259033203125, 3.66851806640625, 2.377197265625, 0.837188720703125, 6.8628997802734375, 2.2465286254882812, 5.357368469238281, -5.002346038818359, 16.44488525390625, 7.3846435546875, 7.1483154296875, 0.2828693389892578, 4.991794586181641, 1.2730159759521484, 11.487091064453125, -3.1136016845703125, -1.8394088745117188, -4.224460601806641, 0.436859130859375, -1.095184326171875, 3.787874221801758, 9.130706787109375, 8.77602767944336, 9.78192138671875, 1.9128189086914062, -0.23479652404785156, -4.412055969238281, -3.958101272583008, 5.524738311767578, -1.4117202758789062, 1.0047492980957031, 1.183511734008789], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000114.npy"}
|
||||
{"epoch": 0.17233560090702948, "step": 115, "batch_size": 64, "mean": 1.9374088048934937, "std": 4.34823751449585, "min": -7.755645751953125, "p10": -2.6400077819824217, "median": 1.7272415161132812, "p90": 7.375688552856446, "max": 15.480697631835938, "pos_frac": 0.703125, "sample": [-1.7205352783203125, -2.6042442321777344, 4.0245208740234375, -5.261322021484375, 15.480697631835938, 2.5388927459716797, 3.1622085571289062, 4.509891510009766, 0.9665374755859375, 12.986587524414062, 2.0879764556884766, -2.2035903930664062, 9.516708374023438, -5.230865478515625, -3.0966949462890625, -2.6538619995117188, 0.04642295837402344, 7.13623046875, 1.751220703125, 3.2045936584472656, 4.349346160888672, 3.4258575439453125, 2.406982421875, 0.441009521484375, -3.4001522064208984, 1.7032623291015625, 4.50946044921875, 2.040008544921875, -7.755645751953125, 7.478313446044922, -2.6076812744140625, -0.7205104827880859, 0.4119071960449219, 7.964752197265625, 1.4774246215820312, 4.5345611572265625, -0.13907623291015625, 3.613382339477539, -2.3755722045898438, 2.1819992065429688, 2.8877410888671875, -6.1665496826171875, 6.165924072265625, 2.977081298828125, 10.60430908203125, 2.7652969360351562, 1.4430160522460938, 1.800760269165039, -1.18621826171875, 6.19610595703125, 0.9244098663330078, 0.20945167541503906, 6.447513580322266, -0.4059600830078125, 0.4414081573486328, -0.9466304779052734, -2.3294143676757812, -0.8462104797363281, 0.8123931884765625, 1.1422157287597656, 4.2185516357421875, 0.18550682067871094, 2.503805160522461, 9.968658447265625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000115.npy"}
|
||||
{"epoch": 0.17384731670445955, "step": 116, "batch_size": 64, "mean": 2.828784227371216, "std": 4.170994758605957, "min": -6.001220703125, "p10": -2.0699453353881836, "median": 3.0767688751220703, "p90": 7.018170166015626, "max": 13.1409912109375, "pos_frac": 0.703125, "sample": [6.334892272949219, -2.1152267456054688, 0.4528846740722656, 8.659439086914062, 5.62359619140625, 5.0550994873046875, 3.6420745849609375, 1.1294403076171875, 5.153127670288086, 7.131095886230469, 5.9729156494140625, -2.5288333892822266, -5.3168792724609375, 11.223129272460938, 0.22097396850585938, -1.240304946899414, 0.5001144409179688, -0.07084846496582031, 3.9973793029785156, 5.7391815185546875, -5.111492156982422, 1.9513473510742188, 5.837184906005859, 3.2848854064941406, -0.24013137817382812, 6.092689514160156, 5.838417053222656, 6.210414886474609, 5.367828369140625, 1.1170501708984375, 1.2631416320800781, -0.9037723541259766, -2.3997020721435547, 0.9151954650878906, 10.8504638671875, -0.16411590576171875, 4.552970886230469, 5.560661315917969, -0.44599151611328125, -1.6118850708007812, 5.3647613525390625, 4.0478057861328125, 3.878223419189453, 1.3351287841796875, 13.1409912109375, 4.377952575683594, 10.914993286132812, -3.9782257080078125, 0.43060302734375, 2.86865234375, -0.03739738464355469, 6.754676818847656, 0.876007080078125, -6.001220703125, -0.8259544372558594, 6.2613677978515625, -1.4171180725097656, 3.7023086547851562, 1.6351890563964844, 9.613784790039062, 6.221809387207031, -0.07033157348632812, 6.384063720703125, -1.9642887115478516], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000116.npy"}
|
||||
{"epoch": 0.17535903250188964, "step": 117, "batch_size": 64, "mean": 1.9091267585754395, "std": 4.646184921264648, "min": -12.27386474609375, "p10": -3.2306442260742183, "median": 1.881509780883789, "p90": 7.138000106811525, "max": 16.22991943359375, "pos_frac": 0.71875, "sample": [0.652984619140625, 4.893583297729492, 3.4718475341796875, 0.7121734619140625, 9.707077026367188, 3.1267929077148438, 5.328641891479492, 1.8778152465820312, -2.2728652954101562, 4.209329605102539, 0.18027114868164062, -1.6294078826904297, 6.77618408203125, -3.6679000854492188, 0.9274978637695312, 3.240234375, -3.4034271240234375, 2.385772705078125, 2.5751113891601562, 5.1814422607421875, 2.4372482299804688, 3.689075469970703, 4.925235748291016, 2.7023239135742188, -1.5163955688476562, -5.370258331298828, 1.878753662109375, -0.26905059814453125, 3.5586605072021484, 10.758716583251953, -2.167194366455078, 9.838226318359375, 7.293064117431641, 0.087677001953125, 8.715179443359375, 2.1199874877929688, 4.29411506652832, 8.757827758789062, 0.9702072143554688, 6.3024749755859375, 2.5343551635742188, -0.6161575317382812, 1.7815399169921875, 3.3642578125, -1.9248123168945312, -10.024894714355469, 1.5977363586425781, 1.2277412414550781, 2.7940216064453125, 4.401100158691406, 1.7898139953613281, -0.39354705810546875, 6.06219482421875, -2.827484130859375, 16.22991943359375, -6.442222595214844, -3.7938575744628906, 1.8842658996582031, -12.27386474609375, -1.8757095336914062, 5.197994232177734, 0.5749130249023438, 0.29113197326660156, -0.6533603668212891], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000117.npy"}
|
||||
{"epoch": 0.17687074829931973, "step": 118, "batch_size": 64, "mean": 2.616124153137207, "std": 3.9079034328460693, "min": -7.402050018310547, "p10": -2.0804800033569335, "median": 2.589292526245117, "p90": 8.050180816650391, "max": 10.960548400878906, "pos_frac": 0.71875, "sample": [2.39910888671875, 1.8510665893554688, 4.8084869384765625, -3.27325439453125, -7.402050018310547, -2.167633056640625, 5.47930908203125, 2.2742691040039062, 4.967498779296875, -1.3974800109863281, 3.0765762329101562, 2.9432830810546875, 1.0944442749023438, 7.8393707275390625, 0.8378772735595703, -0.5701751708984375, -1.9966583251953125, 5.456504821777344, 0.049896240234375, 0.9370174407958984, -1.9351997375488281, 8.103805541992188, 3.6063404083251953, 4.40673828125, 2.9654769897460938, -2.3788909912109375, 0.1014251708984375, -2.1932907104492188, 4.54460334777832, -0.04055595397949219, 4.192718505859375, 2.978628158569336, 8.497085571289062, -1.7116146087646484, 0.8697586059570312, 4.507049560546875, 2.5967941284179688, 10.960548400878906, 9.3995361328125, -2.8945388793945312, -2.116403579711914, 7.5442962646484375, 0.32263946533203125, 5.055412292480469, -1.069183349609375, 7.925056457519531, -1.3536300659179688, 9.221389770507812, 4.576641082763672, 8.883033752441406, 4.7318267822265625, 0.8336601257324219, 3.4684677124023438, 9.882892608642578, 6.746425628662109, 1.6781806945800781, -1.228424072265625, 4.250370025634766, 6.1792755126953125, 7.555521011352539, 2.5817909240722656, -1.8099956512451172, 1.109048843383789, -1.320220947265625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000118.npy"}
|
||||
{"epoch": 0.17838246409674982, "step": 119, "batch_size": 64, "mean": 1.8326729536056519, "std": 5.519773006439209, "min": -13.576675415039062, "p10": -4.163282203674316, "median": 1.705047607421875, "p90": 8.905706787109377, "max": 21.64910888671875, "pos_frac": 0.640625, "sample": [-2.1917190551757812, 4.450798034667969, -3.881010055541992, 0.284210205078125, -4.2842559814453125, -0.9063491821289062, 5.029195785522461, 3.2376861572265625, -1.402902603149414, 0.8780746459960938, -4.434490203857422, -0.6065521240234375, 2.1202621459960938, 9.869354248046875, 7.394580841064453, -13.576675415039062, -0.24484634399414062, -3.6173095703125, 0.7956256866455078, 4.3638458251953125, -1.2975749969482422, 5.998510360717773, -2.1318817138671875, 2.5731945037841797, 9.20220947265625, 4.328987121582031, -5.042346954345703, 2.376110076904297, 9.379440307617188, 11.108444213867188, 6.608917236328125, -2.2923736572265625, 3.6621131896972656, 4.295719146728516, 1.4890594482421875, 21.64910888671875, 2.3191757202148438, -0.75494384765625, 0.9405975341796875, 8.2138671875, 2.760772705078125, 7.7746124267578125, -2.569326400756836, 0.24071502685546875, -6.5350494384765625, -5.232969284057617, 11.552322387695312, 5.0741729736328125, 2.2403793334960938, 3.8627471923828125, 1.240753173828125, -11.414932250976562, -3.5749969482421875, -1.4837532043457031, 1.2445812225341797, 1.7232208251953125, 4.704656600952148, 10.471092224121094, 5.435325622558594, 1.6868743896484375, 2.3106040954589844, 2.1946182250976562, -1.7465553283691406, -0.5726470947265625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000119.npy"}
|
||||
{"epoch": 0.17989417989417988, "step": 120, "batch_size": 64, "mean": 2.842111110687256, "std": 3.8490335941314697, "min": -4.178249359130859, "p10": -1.5109916687011717, "median": 2.1895017623901367, "p90": 6.851264190673829, "max": 13.999153137207031, "pos_frac": 0.78125, "sample": [1.3523101806640625, 0.76031494140625, -2.0245285034179688, 6.475078582763672, 4.8944091796875, 0.6649818420410156, 6.131473541259766, 1.1746597290039062, 1.5070343017578125, -3.7712783813476562, 7.3845977783203125, 1.8199501037597656, 2.9452037811279297, 4.244110107421875, 4.367362976074219, 13.096282958984375, 4.26495361328125, -0.30098915100097656, 2.2721023559570312, 4.952720642089844, 5.5803070068359375, -1.2380752563476562, 4.9158172607421875, 2.870512008666992, 1.385498046875, 1.7893829345703125, 6.597602844238281, 0.8983001708984375, 5.822906494140625, 4.979461669921875, 13.999153137207031, -2.436309814453125, 9.781837463378906, -1.0385456085205078, 8.481842041015625, 0.3644752502441406, 2.0804901123046875, 0.5947227478027344, 5.846443176269531, -1.6068267822265625, 0.20267677307128906, 2.3506011962890625, 1.0333251953125, 12.548431396484375, -1.3553466796875, -1.56365966796875, -4.178249359130859, 6.9599761962890625, 4.882102966308594, 1.0494461059570312, 0.1276836395263672, 0.3741645812988281, 6.343379974365234, -0.106292724609375, 3.720102310180664, -0.38800048828125, 2.8426666259765625, 3.6631011962890625, -1.3880996704101562, -3.5971145629882812, 5.56781005859375, 2.106901168823242, 5.824554443359375, 2.9951934814453125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000120.npy"}
|
||||
{"epoch": 0.18140589569160998, "step": 121, "batch_size": 64, "mean": 3.003298759460449, "std": 5.04355001449585, "min": -4.479663848876953, "p10": -1.7826656341552731, "median": 1.7814931869506836, "p90": 9.403696060180668, "max": 22.2235107421875, "pos_frac": 0.765625, "sample": [0.9976043701171875, 14.759063720703125, 0.688079833984375, -2.11407470703125, 5.806818008422852, 13.440643310546875, 8.404193878173828, -1.8985519409179688, 2.319408416748047, 5.409828186035156, -1.1678543090820312, -2.2774200439453125, 12.404800415039062, 3.5661544799804688, 0.3503265380859375, 0.47722816467285156, 4.542871475219727, 0.03237152099609375, 5.523017883300781, -4.479663848876953, 9.832054138183594, -0.3859519958496094, 6.774791717529297, 3.2985687255859375, 4.5801239013671875, 2.7856521606445312, 1.9655227661132812, -2.4803199768066406, 2.970001220703125, 1.440155029296875, -0.7465038299560547, 22.2235107421875, 2.497051239013672, 3.12213134765625, 0.6427383422851562, -0.90875244140625, -0.6951828002929688, 1.2426071166992188, 2.517578125, 0.456298828125, -0.2665729522705078, 0.1469898223876953, 3.3943405151367188, 5.420555114746094, 0.6585845947265625, 2.334564208984375, -3.7379913330078125, 15.954032897949219, -0.419586181640625, 2.8725109100341797, 2.233806610107422, 1.3133201599121094, 2.4662399291992188, 4.39324951171875, 1.0781269073486328, 15.698822021484375, 1.597463607788086, -1.5122642517089844, 0.9964580535888672, 0.3773040771484375, 0.190460205078125, 3.4791107177734375, 7.940460205078125, -2.3157825469970703], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000121.npy"}
|
||||
{"epoch": 0.18291761148904007, "step": 122, "batch_size": 64, "mean": 2.676017999649048, "std": 5.746242523193359, "min": -7.8445587158203125, "p10": -3.7898223876953123, "median": 1.706131935119629, "p90": 9.362384414672858, "max": 22.588577270507812, "pos_frac": 0.671875, "sample": [22.588577270507812, -3.7852401733398438, 0.1177978515625, 6.886077880859375, -4.4619903564453125, -2.41754150390625, 14.365692138671875, 10.020210266113281, -0.8306331634521484, 6.0749664306640625, 4.475818634033203, 1.153472900390625, 0.42220115661621094, 2.3025970458984375, 0.9439029693603516, 16.365234375, -2.2792205810546875, 2.8696746826171875, -4.712673187255859, 6.3038787841796875, 0.9703598022460938, 11.437772750854492, 4.616359710693359, 5.851095199584961, 7.827457427978516, 6.1212310791015625, -2.513162612915039, -4.130168914794922, 4.6829071044921875, 4.852210998535156, -7.8445587158203125, 4.722892761230469, -0.39288902282714844, -1.97991943359375, -4.330230712890625, 4.140651702880859, 2.1904659271240234, -0.8356781005859375, 2.911865234375, -0.7766265869140625, -0.7808265686035156, 0.501373291015625, -3.871845245361328, 11.47198486328125, -2.8174209594726562, 5.567338943481445, 1.8274707794189453, 2.195110321044922, -0.48465728759765625, 1.9751358032226562, 1.5847930908203125, 1.5391845703125, 0.2789154052734375, 0.7524032592773438, -3.7917861938476562, 2.9829235076904297, 0.27825164794921875, -0.7761306762695312, -0.3678741455078125, 6.5170135498046875, 3.935993194580078, 4.838798522949219, 2.8298492431640625, 21.154312133789062], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000122.npy"}
|
||||
{"epoch": 0.18442932728647016, "step": 123, "batch_size": 64, "mean": 2.6469385623931885, "std": 4.561816215515137, "min": -8.047637939453125, "p10": -1.8377769470214844, "median": 1.8993043899536133, "p90": 8.385489273071292, "max": 20.222976684570312, "pos_frac": 0.6875, "sample": [7.183738708496094, 6.752845764160156, 1.2963790893554688, 20.222976684570312, -4.714847564697266, -1.331085205078125, 2.4247398376464844, 1.10333251953125, 11.220024108886719, 5.39961051940918, 1.5319976806640625, 9.321640014648438, -0.298858642578125, -1.3443450927734375, 6.195426940917969, 5.9105377197265625, 2.9597911834716797, 3.8794212341308594, -1.1549301147460938, 8.691963195800781, 1.8450164794921875, 7.083885192871094, 1.5629234313964844, 7.670383453369141, 3.479461669921875, 10.960716247558594, 2.966440200805664, -1.8649978637695312, 6.2386322021484375, 1.9645538330078125, 1.2243080139160156, 11.541755676269531, -1.774261474609375, 6.515569686889648, -0.6932411193847656, -4.945701599121094, 1.5325393676757812, -2.3694629669189453, -0.14881515502929688, 1.6416969299316406, 1.712158203125, -1.5808601379394531, 2.2171459197998047, -0.6400985717773438, 1.953592300415039, 6.0697479248046875, -0.7061080932617188, -1.554849624633789, 1.4072284698486328, 5.42780876159668, -1.6742439270019531, 2.6674041748046875, 1.7198562622070312, 3.4884796142578125, -8.047637939453125, 9.010406494140625, -0.7860565185546875, 2.8013343811035156, -2.1260147094726562, 2.30657958984375, 1.39532470703125, 2.498889923095703, 4.03094482421875, -1.8687286376953125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000123.npy"}
|
||||
{"epoch": 0.18594104308390022, "step": 124, "batch_size": 64, "mean": 3.6488072872161865, "std": 4.846010684967041, "min": -4.806632995605469, "p10": -2.2775075912475584, "median": 3.3275394439697266, "p90": 10.268969726562501, "max": 17.210174560546875, "pos_frac": 0.765625, "sample": [-1.0451889038085938, 17.210174560546875, 5.180763244628906, -4.348560333251953, 3.8942012786865234, 4.353660583496094, 16.153518676757812, 5.295581817626953, 12.40924072265625, 11.148834228515625, 6.869514465332031, 6.955242156982422, 1.5056438446044922, 13.075027465820312, 1.1077728271484375, 2.2338008880615234, 1.9899368286132812, 6.046808242797852, 1.4599151611328125, -2.2960681915283203, -4.629974365234375, 3.22259521484375, 3.2659950256347656, 1.952728271484375, 2.2486419677734375, 4.926578521728516, 1.2166881561279297, -0.3811187744140625, -3.9885711669921875, -3.182952880859375, 4.573991775512695, 4.90618896484375, -2.2341995239257812, 9.948577880859375, 5.520145416259766, 8.640213012695312, 11.726036071777344, -3.470916748046875, 7.081123352050781, 2.903167724609375, 0.6507110595703125, 8.063583374023438, -1.0034255981445312, 4.847511291503906, 3.0029449462890625, 4.470130920410156, 1.9925537109375, 6.610237121582031, -2.0454845428466797, 4.431737899780273, 10.406280517578125, 1.3382377624511719, 0.18105697631835938, 3.231609344482422, 4.091888427734375, 3.8531341552734375, 3.3890838623046875, -1.0645675659179688, -4.806632995605469, 4.921112060546875, -0.5594444274902344, 5.098995208740234, -0.6619911193847656, 9.6396484375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000124.npy"}
|
||||
{"epoch": 0.1874527588813303, "step": 125, "batch_size": 64, "mean": 3.1023077964782715, "std": 5.185013294219971, "min": -7.0477294921875, "p10": -3.2720901489257814, "median": 2.953947067260742, "p90": 9.105371284484864, "max": 21.247894287109375, "pos_frac": 0.734375, "sample": [-0.906463623046875, 7.5770416259765625, 6.564537048339844, 3.07489013671875, 0.8168411254882812, -1.4555034637451172, 11.683242797851562, 0.26720428466796875, 6.60211181640625, 3.3988685607910156, 21.247894287109375, -4.064842224121094, 0.643341064453125, 3.4940567016601562, 2.187021255493164, 9.554630279541016, 9.259063720703125, 3.4835052490234375, 2.8330039978027344, -1.3493423461914062, 8.746755599975586, 0.5812568664550781, 3.2347030639648438, 3.580821990966797, -1.5010299682617188, 1.7048473358154297, 8.134803771972656, -3.313457489013672, -1.3026580810546875, -0.021087646484375, 8.444137573242188, 4.440193176269531, 1.7161483764648438, 12.94537353515625, 6.095417022705078, -3.3629913330078125, 5.349910736083984, 1.7235488891601562, 11.243598937988281, 16.634521484375, 6.7585906982421875, 3.1118392944335938, 6.710777282714844, 5.514102935791016, 3.1809425354003906, -7.0477294921875, -0.35762786865234375, 8.118240356445312, 0.5640335083007812, 0.7329998016357422, 3.075347900390625, 3.4174461364746094, 4.087890625, 2.4005355834960938, -2.78448486328125, 4.030853271484375, 0.7747230529785156, -3.2680740356445312, 1.9899368286132812, -0.6948680877685547, 2.177583694458008, -3.2738113403320312, -6.900520324707031, -3.7569427490234375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000125.npy"}
|
||||
{"epoch": 0.1889644746787604, "step": 126, "batch_size": 64, "mean": 1.7743775844573975, "std": 4.689510822296143, "min": -10.142547607421875, "p10": -3.1097206115722655, "median": 2.0333309173583984, "p90": 6.403403472900391, "max": 16.674545288085938, "pos_frac": 0.625, "sample": [3.366964340209961, 0.5901756286621094, 6.205802917480469, 2.5972347259521484, 0.025600433349609375, 6.355724334716797, 6.422691345214844, 0.15694236755371094, 3.4442291259765625, -2.1724472045898438, 6.06089973449707, 7.0408477783203125, -9.091865539550781, 0.746490478515625, 5.74957275390625, 1.9911155700683594, -2.3287124633789062, -1.8723182678222656, -0.42249488830566406, -3.4706192016601562, 2.23193359375, 6.3583984375, 1.5628089904785156, 3.056124687194824, -1.29620361328125, -0.21243667602539062, 5.416130065917969, -5.574302673339844, -2.9423675537109375, -1.652669906616211, -0.1070556640625, -3.0216598510742188, -2.614248275756836, 3.115507125854492, -4.254152297973633, -1.0006256103515625, -2.3794994354248047, 16.674545288085938, 4.658720016479492, -10.142547607421875, 4.3789520263671875, 4.0760498046875, 3.262022018432617, 4.977077484130859, 4.733283996582031, 5.724311828613281, 7.333686828613281, 13.527618408203125, -0.1657867431640625, 2.0755462646484375, 0.8533515930175781, 7.805809020996094, 3.351898193359375, 3.449329376220703, 1.381011962890625, -1.2409133911132812, -3.1474609375, 5.456840515136719, -1.954833984375, -0.7643890380859375, 3.5786285400390625, 9.113471984863281, -6.254289627075195, 2.7367095947265625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000126.npy"}
|
||||
{"epoch": 0.19047619047619047, "step": 127, "batch_size": 64, "mean": 3.324540615081787, "std": 4.377881050109863, "min": -6.8220367431640625, "p10": -1.8232597351074218, "median": 2.8284740447998047, "p90": 8.766565513610841, "max": 14.495162963867188, "pos_frac": 0.765625, "sample": [-0.598907470703125, 14.495162963867188, 7.701683044433594, 3.9195022583007812, -2.1430511474609375, 9.055023193359375, 3.5170135498046875, 8.207695007324219, 1.5812225341796875, 11.1488037109375, 7.113245010375977, 3.5838470458984375, 4.495872497558594, 8.549600601196289, 1.4756011962890625, 1.8070487976074219, 5.872932434082031, 3.8128890991210938, 1.1710128784179688, 1.5368690490722656, 6.8041229248046875, -6.8220367431640625, 4.4353485107421875, 7.10528564453125, 3.9321441650390625, -5.3729095458984375, 3.8164138793945312, 1.5573043823242188, 2.6210784912109375, 0.9187698364257812, 0.2617034912109375, 3.5132675170898438, -0.10743522644042969, 8.524314880371094, -0.017377853393554688, 8.236801147460938, 1.6118297576904297, 8.859550476074219, 9.83254623413086, 14.397621154785156, -2.904987335205078, 6.4182281494140625, 2.6276321411132812, -1.9962844848632812, 2.2385177612304688, 2.8663864135742188, -1.8910064697265625, -1.6651840209960938, 6.1926116943359375, -4.2178802490234375, -0.28521728515625, 8.099441528320312, 2.7905616760253906, 1.6695632934570312, 1.1783866882324219, -0.8561019897460938, -1.66168212890625, 2.098724365234375, 4.035835266113281, 3.460765838623047, 4.139249801635742, 0.6352939605712891, 10.129264831542969, -0.7429313659667969], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000127.npy"}
|
||||
{"epoch": 0.19198790627362056, "step": 128, "batch_size": 64, "mean": 3.6484885215759277, "std": 6.603226661682129, "min": -13.16619873046875, "p10": -4.000250434875487, "median": 3.099306106567383, "p90": 11.809138488769532, "max": 20.604339599609375, "pos_frac": 0.75, "sample": [3.44696044921875, 11.715133666992188, 3.1153831481933594, 3.6651382446289062, 3.3505821228027344, 1.8972854614257812, 5.1970062255859375, 11.84942626953125, 1.4030990600585938, 16.801544189453125, 2.042633056640625, -13.16619873046875, 11.608497619628906, -1.47015380859375, 3.6178531646728516, -7.206153869628906, 0.8520259857177734, -1.0251846313476562, -7.2013702392578125, 12.797988891601562, 16.667236328125, -5.021123886108398, -2.3073768615722656, 20.604339599609375, 19.76800537109375, 7.45672607421875, 0.5417327880859375, 11.105537414550781, -1.824075698852539, 7.00244140625, 1.125091552734375, 5.0949249267578125, 3.274738311767578, 8.915809631347656, 7.754436492919922, 6.5910797119140625, -0.06056976318359375, 9.970306396484375, -6.65507698059082, 2.1486663818359375, -2.7895965576171875, 2.5510597229003906, 2.6865921020507812, 0.17124176025390625, -4.519102096557617, 1.8668670654296875, 2.077608108520508, 3.2831192016601562, 3.0832290649414062, -2.074777603149414, 4.7761077880859375, 4.79638671875, 9.170425415039062, -1.087371826171875, 5.328037261962891, 0.6900787353515625, 1.7036285400390625, 6.952083587646484, 16.539505004882812, -1.8285961151123047, -7.6922607421875, 1.5922164916992188, 5.356052398681641, 5.4263763427734375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000128.npy"}
|
||||
{"epoch": 0.19349962207105065, "step": 129, "batch_size": 64, "mean": 2.0779051780700684, "std": 5.948974609375, "min": -15.106765747070312, "p10": -3.85045051574707, "median": 2.3824081420898438, "p90": 8.872796440124514, "max": 19.5223388671875, "pos_frac": 0.609375, "sample": [-0.8100738525390625, -1.0127639770507812, 5.237850189208984, -5.580108642578125, -0.7951850891113281, 7.1081085205078125, 0.4122161865234375, -1.7284412384033203, 2.58984375, 9.787174224853516, 7.287628173828125, 2.604522705078125, 6.4982452392578125, -2.7033157348632812, -2.5309066772460938, -1.4789276123046875, 7.788612365722656, 0.4729461669921875, 9.445808410644531, 7.593727111816406, 2.598968505859375, -12.904190063476562, 1.67010498046875, 4.680854797363281, -3.147411346435547, 19.5223388671875, 5.11176872253418, -1.3344497680664062, 3.7611465454101562, 3.8112945556640625, 5.806816101074219, -6.9851531982421875, -5.9417724609375, 8.392396926879883, 0.5768814086914062, 7.2643890380859375, -7.773151397705078, -1.1585006713867188, -3.3509082794189453, 9.708518981933594, 3.9925384521484375, -1.5509090423583984, -2.109416961669922, 1.3436965942382812, 4.842388153076172, 3.7469120025634766, -1.71099853515625, 16.15182876586914, 6.778175354003906, 8.082122802734375, 2.859081268310547, 9.078681945800781, 9.420242309570312, -1.2565135955810547, -1.2818870544433594, -3.395793914794922, 3.7678985595703125, -15.106765747070312, 5.6727294921875, 1.065408706665039, 2.1749725341796875, -4.0453033447265625, -0.053562164306640625, 4.023509979248047], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000129.npy"}
|
||||
{"epoch": 0.19501133786848074, "step": 130, "batch_size": 64, "mean": 3.3693928718566895, "std": 5.019582748413086, "min": -5.4618682861328125, "p10": -4.291590118408203, "median": 3.6653404235839844, "p90": 9.589524841308595, "max": 17.788681030273438, "pos_frac": 0.78125, "sample": [1.5151519775390625, 2.246082305908203, 9.834236145019531, 9.018531799316406, 5.810455322265625, 0.397308349609375, -3.00213623046875, 8.164701461791992, 3.8327274322509766, 0.7150974273681641, -4.6056060791015625, 7.336479187011719, 8.048133850097656, 1.763101577758789, -5.0991668701171875, 7.137626647949219, 2.4690380096435547, -4.410858154296875, 1.700103759765625, 11.543220520019531, 3.2215309143066406, 3.6552047729492188, 6.382225036621094, 8.173042297363281, -3.1433944702148438, -4.013298034667969, -4.742393493652344, 0.7367095947265625, -2.733011245727539, 0.28917694091796875, 8.655288696289062, 9.8895263671875, 7.5779266357421875, 4.8292236328125, 1.7209014892578125, 14.264972686767578, 8.203880310058594, 4.27067756652832, 17.788681030273438, 3.043865203857422, -1.9032249450683594, 4.500194549560547, -0.249786376953125, -5.250988006591797, 2.1940670013427734, 4.077201843261719, 5.908847808837891, 0.8850021362304688, 4.5508270263671875, -1.3589668273925781, -5.195102691650391, 2.1619319915771484, 4.4721527099609375, 5.385805130004883, -5.4618682861328125, 5.370830535888672, 4.987335205078125, 2.3159637451171875, 3.67547607421875, 11.21234130859375, 10.302001953125, 3.9776153564453125, 1.7718868255615234, 4.8266448974609375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000130.npy"}
|
||||
{"epoch": 0.1965230536659108, "step": 131, "batch_size": 64, "mean": 4.330511093139648, "std": 6.478960037231445, "min": -7.681549072265625, "p10": -3.672314834594726, "median": 2.8215179443359375, "p90": 13.867646026611329, "max": 18.382461547851562, "pos_frac": 0.734375, "sample": [2.9786300659179688, 7.836648941040039, -0.5986213684082031, -1.5464801788330078, -1.8315849304199219, 18.382461547851562, 3.7006797790527344, 4.732227325439453, -1.9614849090576172, 13.343238830566406, 2.6644058227539062, 1.903879165649414, -3.990245819091797, -3.9843711853027344, 2.114002227783203, 6.20854377746582, -2.8640899658203125, 3.0770797729492188, 9.829940795898438, -7.433403015136719, -4.612556457519531, -2.944183349609375, 16.934799194335938, 5.8924102783203125, 2.03033447265625, 0.12487030029296875, 2.1024551391601562, -4.4093780517578125, -1.5978050231933594, 0.8416366577148438, 6.138759613037109, 13.998435974121094, 12.019271850585938, 6.265600204467773, 13.082618713378906, 16.840072631835938, 2.195484161376953, 14.875213623046875, 4.6316680908203125, 13.562469482421875, 2.164369583129883, 15.376945495605469, -2.0356178283691406, 8.245475769042969, 2.25811767578125, 0.6605911254882812, 0.7273197174072266, 5.503837585449219, -5.583709716796875, 9.690643310546875, 0.34920310974121094, -0.2202911376953125, 8.450973510742188, -0.13628005981445312, -7.681549072265625, 11.33660888671875, 2.5441818237304688, 2.26806640625, 13.050735473632812, 3.7589187622070312, 3.20721435546875, 9.262386322021484, 14.24030876159668, 9.180606842041016], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000131.npy"}
|
||||
{"epoch": 0.1980347694633409, "step": 132, "batch_size": 64, "mean": 3.6230602264404297, "std": 5.41845178604126, "min": -14.03741455078125, "p10": -2.6686227798461912, "median": 4.3552961349487305, "p90": 9.422170639038086, "max": 17.14055633544922, "pos_frac": 0.75, "sample": [9.160758972167969, 2.8161087036132812, 6.6721649169921875, 5.97509765625, -2.341829299926758, -1.7010555267333984, 7.8666229248046875, 0.22246932983398438, -2.3922119140625, 5.927886962890625, 5.970630645751953, 8.930303573608398, 4.555278778076172, 4.309486389160156, -0.20986557006835938, 1.63836669921875, -1.8292808532714844, 15.56689453125, 5.4572906494140625, -0.2899169921875, -2.9805450439453125, -14.03741455078125, 8.468128204345703, -1.0827255249023438, -3.1572532653808594, -6.9409027099609375, 5.8037261962890625, 10.091228485107422, 10.241132736206055, -1.5661392211914062, 6.216972351074219, 0.9459800720214844, -10.911947250366211, 4.575706481933594, 2.1754913330078125, 3.2457733154296875, 17.14055633544922, 6.82099723815918, 9.34469223022461, 5.075141906738281, -0.4456138610839844, 9.576919555664062, 4.0049285888671875, 9.455375671386719, 8.11912727355957, 4.183319091796875, 4.746517181396484, 2.91558837890625, 2.3067779541015625, 4.401105880737305, 6.143398284912109, 3.50811767578125, 3.6530303955078125, 5.805889129638672, 9.034479141235352, 5.3248748779296875, 10.546157836914062, -2.7870845794677734, 1.857635498046875, 3.1318817138671875, -4.513999938964844, 5.67230224609375, 3.80548095703125, 5.655853271484375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000132.npy"}
|
||||
{"epoch": 0.19954648526077098, "step": 133, "batch_size": 64, "mean": 2.8133130073547363, "std": 5.958312034606934, "min": -15.16632080078125, "p10": -4.130863571166992, "median": 2.423055648803711, "p90": 10.606833076477054, "max": 16.98603057861328, "pos_frac": 0.6875, "sample": [5.208740234375, -5.502471923828125, 3.416942596435547, 10.869171142578125, 0.04435920715332031, -1.6605377197265625, 4.0956268310546875, 12.018150329589844, 0.7821121215820312, 1.6533126831054688, 2.508502960205078, 15.312362670898438, 8.39654541015625, 1.1228370666503906, -6.574687957763672, 2.5084457397460938, 2.5136260986328125, -15.16632080078125, 8.010799407958984, 1.4304733276367188, -0.519744873046875, 7.220493316650391, -1.3955039978027344, 7.368762969970703, 4.750988006591797, -0.31111907958984375, 7.3138275146484375, 0.2262115478515625, 3.69500732421875, -0.8526077270507812, 2.337665557861328, 1.029703140258789, -9.335281372070312, 9.994710922241211, -3.729938507080078, 1.78082275390625, -0.7798500061035156, -4.3026885986328125, 9.10379409790039, 9.367889404296875, -1.5990982055664062, 2.8431015014648438, 5.5467071533203125, -6.805288314819336, 3.1465301513671875, 14.818679809570312, 6.7556610107421875, 8.396812438964844, 1.3724822998046875, 16.98603057861328, 1.300140380859375, -0.5989494323730469, -2.111530303955078, 13.62054443359375, 6.218467712402344, -0.5442123413085938, 4.9954986572265625, 11.038698196411133, 4.906730651855469, -1.4596118927001953, 2.7415237426757812, -2.1190261840820312, -4.363697052001953, 1.0146942138671875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000133.npy"}
|
||||
{"epoch": 0.20105820105820105, "step": 134, "batch_size": 64, "mean": 4.936406135559082, "std": 6.427099227905273, "min": -12.345230102539062, "p10": -1.466022491455078, "median": 4.129251480102539, "p90": 13.663032531738281, "max": 22.136932373046875, "pos_frac": 0.8125, "sample": [-1.2225723266601562, 0.4900970458984375, 3.5551910400390625, 4.061431884765625, 4.197071075439453, 3.883678436279297, 3.6840667724609375, 6.192756652832031, 13.508743286132812, 13.729156494140625, 4.300224304199219, 9.4764404296875, 5.312660217285156, 3.683727264404297, 4.45307731628418, -0.11569976806640625, 10.233440399169922, 8.165191650390625, -3.5801658630371094, 7.044635772705078, 7.995220184326172, 5.170207977294922, 9.24454116821289, -0.17718124389648438, 4.447235107421875, 9.116439819335938, 0.257049560546875, 1.372406005859375, -1.5703582763671875, 7.1414642333984375, -3.376249313354492, -0.23686981201171875, 12.736099243164062, 0.54559326171875, 0.643402099609375, -7.588384628295898, 4.6641387939453125, 3.027637481689453, 2.2435760498046875, 4.736814498901367, 22.136932373046875, 10.055946350097656, -12.345230102539062, 2.5529632568359375, 6.533611297607422, 1.59857177734375, 14.324310302734375, -3.936248779296875, 16.957046508789062, 2.9028778076171875, 2.199047088623047, 6.838581085205078, 5.520988464355469, 8.635696411132812, 19.103012084960938, 1.3321914672851562, -3.7383499145507812, 2.160186767578125, 3.5033340454101562, 10.394031524658203, 14.919677734375, 22.054351806640625, 1.910512924194336, -1.1299934387207031], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000134.npy"}
|
||||
{"epoch": 0.20256991685563114, "step": 135, "batch_size": 64, "mean": 3.1580426692962646, "std": 5.944058418273926, "min": -13.04241943359375, "p10": -2.8255405426025386, "median": 2.7736663818359375, "p90": 10.255488586425784, "max": 19.853500366210938, "pos_frac": 0.734375, "sample": [4.912605285644531, -2.0237808227539062, 7.3580322265625, 2.314075469970703, 2.2501564025878906, -2.5496864318847656, -0.5868148803710938, -5.24212646484375, -0.23345184326171875, 11.637687683105469, 8.038406372070312, -2.9437637329101562, 6.419685363769531, -0.8158130645751953, 2.0491409301757812, -6.251956939697266, 11.529022216796875, 4.046104431152344, -0.9028739929199219, 2.4930801391601562, 2.5647125244140625, 2.92962646484375, -12.500999450683594, 2.0670852661132812, 4.141044616699219, 8.42697525024414, 2.550933837890625, 5.351310729980469, 1.41650390625, 4.602153778076172, 4.01153564453125, 0.3469276428222656, 10.569061279296875, 2.2051162719726562, 5.619903564453125, 1.4881172180175781, 2.617706298828125, 3.4857177734375, 7.8836517333984375, -6.394554138183594, 0.08781051635742188, 19.853500366210938, 0.24498367309570312, 4.29376220703125, 3.2159595489501953, 3.0017013549804688, -0.5833892822265625, -6.063093185424805, 7.841808319091797, 7.9761962890625, 6.264156341552734, -13.04241943359375, 0.72979736328125, 7.321897506713867, 13.382251739501953, 14.964420318603516, -2.4764041900634766, 6.538473129272461, -2.081960678100586, 8.201152801513672, 3.6147918701171875, 9.523818969726562, 15.032508850097656, -0.60723876953125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000135.npy"}
|
||||
{"epoch": 0.20408163265306123, "step": 136, "batch_size": 64, "mean": 4.134727478027344, "std": 8.12443733215332, "min": -13.749267578125, "p10": -4.965328979492186, "median": 3.596538543701172, "p90": 15.949847793579101, "max": 23.818756103515625, "pos_frac": 0.6875, "sample": [15.392410278320312, 11.158132553100586, -9.879537582397461, -0.7828445434570312, -5.291229248046875, 9.944976806640625, -3.7024917602539062, 0.7371177673339844, -6.858787536621094, 7.428201675415039, 3.7837142944335938, -13.749267578125, -1.0335102081298828, 5.2992401123046875, -2.5545120239257812, 7.7678985595703125, 16.307096481323242, 3.922027587890625, 1.7336177825927734, -4.069664001464844, 10.11358642578125, 5.212333679199219, 10.570457458496094, 19.64490509033203, -1.8904495239257812, -0.09021759033203125, 15.974357604980469, 3.858654022216797, 2.1646728515625, 23.818756103515625, 10.2222900390625, -0.7287521362304688, 16.34454345703125, 13.310104370117188, 3.40936279296875, 3.144287109375, -1.1507644653320312, 4.740816116333008, 6.710533142089844, 1.0413894653320312, -12.552501678466797, 4.4516754150390625, 1.84295654296875, 11.057441711425781, 5.634853363037109, 7.803123474121094, -3.6574325561523438, 7.2226409912109375, -4.20489501953125, 18.319435119628906, 3.0953521728515625, -8.602317810058594, -1.0474090576171875, 9.9742431640625, 2.473064422607422, 16.513702392578125, -2.5552291870117188, 0.0110015869140625, 12.8048095703125, 15.892658233642578, 7.474334716796875, 0.6274871826171875, 1.9499015808105469, -11.879798889160156], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000136.npy"}
|
||||
{"epoch": 0.20559334845049132, "step": 137, "batch_size": 64, "mean": 4.194239139556885, "std": 6.575453281402588, "min": -10.662384033203125, "p10": -2.492035675048828, "median": 2.5168895721435547, "p90": 11.891746139526369, "max": 24.12017822265625, "pos_frac": 0.703125, "sample": [-1.609588623046875, -10.662384033203125, 6.622142791748047, 11.374893188476562, 6.634695053100586, 11.246109008789062, 12.527942657470703, 6.7247161865234375, -0.3506813049316406, 24.12017822265625, 9.858051300048828, 1.940826416015625, 12.11325454711914, 2.9297256469726562, 11.134376525878906, -0.727935791015625, 2.670654296875, 12.197189331054688, -1.42828369140625, 1.9452838897705078, -1.0324935913085938, 5.940238952636719, 9.996505737304688, -6.411884307861328, 1.8533573150634766, 0.1987457275390625, 7.937225341796875, 21.749961853027344, -1.2688121795654297, 1.4289093017578125, -2.6300125122070312, -2.1700897216796875, -1.2235565185546875, 5.02288818359375, 15.367218017578125, 1.5851058959960938, 0.10404586791992188, -3.939544677734375, 5.990501403808594, -2.163015365600586, 2.3631248474121094, 2.8863162994384766, 5.725587844848633, 13.225257873535156, 6.086051940917969, -3.5516605377197266, 1.7927837371826172, 9.486745834350586, 9.541152954101562, 11.164627075195312, 2.3421192169189453, -4.4149627685546875, -0.2060394287109375, -7.339225769042969, 10.387535095214844, 3.606351852416992, 1.265249252319336, -1.7130661010742188, -0.003387451171875, 7.721771240234375, 2.1037673950195312, 11.338485717773438, 0.33587646484375, 8.690399169921875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000137.npy"}
|
||||
{"epoch": 0.20710506424792138, "step": 138, "batch_size": 64, "mean": 3.6779236793518066, "std": 7.380885601043701, "min": -16.5150146484375, "p10": -3.3492538452148435, "median": 4.3624420166015625, "p90": 11.452651977539062, "max": 23.069732666015625, "pos_frac": 0.71875, "sample": [1.2928085327148438, 8.457221984863281, 4.507240295410156, 8.645587921142578, 5.887851715087891, 3.8650341033935547, 0.337310791015625, 3.9243087768554688, -7.920654296875, 4.913532257080078, 3.7857418060302734, -16.5150146484375, 13.556427001953125, -0.0970916748046875, 23.069732666015625, -8.925140380859375, -2.9681930541992188, 8.306692123413086, 10.234405517578125, 5.178554534912109, 7.0783233642578125, 1.76837158203125, 19.542205810546875, 9.52737045288086, 2.5314788818359375, 7.798460006713867, 10.514190673828125, 7.483802795410156, 1.1680488586425781, 1.3700485229492188, 7.858253479003906, -0.325286865234375, 11.647207260131836, -1.7736053466796875, -3.59832763671875, -2.9381179809570312, 0.1483745574951172, -14.008460998535156, -0.7236042022705078, 7.656993865966797, 6.44989013671875, 11.448005676269531, 0.6247444152832031, 5.034149169921875, -2.556385040283203, -15.276290893554688, 16.559263229370117, 4.217643737792969, 5.382621765136719, -0.7505035400390625, 5.3864898681640625, 7.595741271972656, 9.771003723144531, -2.1550979614257812, 0.1037445068359375, -0.8820209503173828, -3.5125656127929688, 5.929418563842773, -1.4560585021972656, 6.6868438720703125, 17.448890686035156, 4.559505462646484, 11.454643249511719, 1.06134033203125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000138.npy"}
|
||||
{"epoch": 0.20861678004535147, "step": 139, "batch_size": 64, "mean": 4.252243995666504, "std": 6.423991680145264, "min": -15.16510009765625, "p10": -3.1947303771972653, "median": 3.9697065353393555, "p90": 12.010040664672852, "max": 20.582977294921875, "pos_frac": 0.75, "sample": [3.318155288696289, 12.002449035644531, 4.811614990234375, 3.7306289672851562, -7.36566162109375, 8.655521392822266, 7.744773864746094, 11.781230926513672, 7.222141265869141, -1.7027912139892578, -1.4409980773925781, 17.327939987182617, 9.314437866210938, -2.6783695220947266, 5.1388702392578125, 12.527885437011719, 19.535202026367188, 4.091733932495117, 1.9847183227539062, 12.013294219970703, -3.326019287109375, -15.16510009765625, -3.46173095703125, 3.499011993408203, 7.945610046386719, 3.4994277954101562, 7.448516845703125, 5.611764907836914, 5.819766998291016, -0.8013458251953125, 6.774665832519531, 7.7005157470703125, 3.8476791381835938, 0.54522705078125, -0.21596527099609375, 7.69989013671875, 7.4061279296875, 1.7865409851074219, 7.6252899169921875, 1.7106475830078125, 9.805374145507812, -1.7606201171875, 12.320602416992188, -1.0070648193359375, -2.8883895874023438, -5.483375549316406, 5.482513427734375, 4.697601318359375, 20.582977294921875, 3.712818145751953, 8.535308837890625, 3.445270538330078, 2.6749420166015625, -7.52166748046875, -2.4687576293945312, 6.720233917236328, 0.5306396484375, 2.8800125122070312, -4.10809326171875, 7.8145904541015625, 5.343467712402344, 15.081565856933594, 3.7154369354248047, 0.074951171875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000139.npy"}
|
||||
{"epoch": 0.21012849584278157, "step": 140, "batch_size": 64, "mean": 5.969539165496826, "std": 9.354920387268066, "min": -23.541839599609375, "p10": -6.077751350402831, "median": 6.020343780517578, "p90": 17.930242156982423, "max": 22.588165283203125, "pos_frac": 0.734375, "sample": [18.890335083007812, 11.46148681640625, 0.2681159973144531, -0.5394363403320312, -0.176483154296875, 10.715339660644531, -8.639732360839844, -9.00762939453125, 10.374370574951172, 13.542381286621094, -23.541839599609375, 9.592445373535156, 5.2987213134765625, 5.1474609375, 17.560104370117188, 7.4787750244140625, -6.42335319519043, 1.902984619140625, 1.043548583984375, 15.283554077148438, 4.210931777954102, 20.440780639648438, 12.426895141601562, 8.976242065429688, 1.0761833190917969, 9.72650146484375, 3.450777053833008, 19.7480411529541, 2.26983642578125, -6.8388824462890625, -5.2713470458984375, -4.267112731933594, 11.469568252563477, -4.820808410644531, -3.6894989013671875, 6.5687255859375, 13.065132141113281, 17.945831298828125, 16.16033172607422, 22.588165283203125, 7.8406524658203125, 17.52106475830078, 17.89386749267578, 17.86899185180664, 5.421108245849609, -7.529348373413086, 18.93195343017578, 0.3549537658691406, 21.309715270996094, -3.2953739166259766, -8.556194305419922, 10.732137680053711, 3.9494056701660156, 8.296915054321289, 9.960601806640625, 2.205049514770508, 9.38873291015625, -3.5265274047851562, 6.540641784667969, 5.5000457763671875, -0.2663612365722656, 15.448200225830078, -4.0594940185546875, 4.652320861816406], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000140.npy"}
|
||||
{"epoch": 0.21164021164021163, "step": 141, "batch_size": 64, "mean": 5.320494651794434, "std": 9.001922607421875, "min": -14.0150146484375, "p10": -6.936542892456052, "median": 5.2833404541015625, "p90": 16.349287414550783, "max": 28.091644287109375, "pos_frac": 0.75, "sample": [-14.0150146484375, 0.5431976318359375, 16.6749267578125, 4.204105377197266, 9.076431274414062, -0.7188491821289062, 2.9601669311523438, 13.556941986083984, 3.3489837646484375, -8.828947067260742, 5.8251495361328125, 1.2260284423828125, 8.709260940551758, -7.812259674072266, 10.36642074584961, 6.593296051025391, 6.644294738769531, 1.207925796508789, -1.5773468017578125, -13.698028564453125, 3.7420730590820312, 0.7652492523193359, 0.419342041015625, 6.867298126220703, 7.165061950683594, 3.0528564453125, 2.1529388427734375, 8.738052368164062, -2.594879150390625, -8.816436767578125, -1.8780517578125, -8.223365783691406, 8.936683654785156, 10.383140563964844, 11.737197875976562, -2.369112014770508, 10.823909759521484, 5.124458312988281, 23.72037124633789, 20.1256103515625, 15.589462280273438, 12.331958770751953, 12.758514404296875, 9.633674621582031, 22.34357452392578, 1.1681671142578125, -0.7174396514892578, 5.442222595214844, 18.316547393798828, 14.798431396484375, -9.880340576171875, 2.9051437377929688, 5.9195404052734375, 24.284271240234375, 9.506240844726562, 28.091644287109375, -4.265188217163086, 13.899654388427734, 1.4417800903320312, 7.985954284667969, 10.304153442382812, 3.3482322692871094, -4.8932037353515625, -3.960399627685547], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000141.npy"}
|
||||
{"epoch": 0.21315192743764172, "step": 142, "batch_size": 64, "mean": 4.430446624755859, "std": 6.9725189208984375, "min": -13.99013900756836, "p10": -2.796828460693359, "median": 3.4437808990478516, "p90": 13.711533355712891, "max": 23.069381713867188, "pos_frac": 0.75, "sample": [5.011161804199219, -11.075508117675781, 2.155824661254883, 11.445880889892578, 12.675350189208984, 19.404464721679688, 10.649925231933594, -2.299365997314453, 1.9717597961425781, 6.626129150390625, 1.873779296875, -0.6186561584472656, 6.412717819213867, -13.99013900756836, -0.3529052734375, 9.432144165039062, 6.470500946044922, -6.535879135131836, 1.9640865325927734, 3.616849899291992, 2.5824050903320312, 10.759490966796875, 7.26597785949707, 3.685436248779297, 3.183216094970703, 14.892471313476562, -1.146240234375, -0.7601165771484375, 11.748985290527344, -2.7154006958007812, 1.8809356689453125, 1.9358367919921875, 15.981048583984375, 23.069381713867188, 2.2961502075195312, -2.6994705200195312, 3.0107955932617188, -7.028709411621094, 0.445220947265625, 4.562652587890625, 16.62737274169922, 13.551300048828125, 3.37322998046875, 1.8239822387695312, -0.8814544677734375, 12.238845825195312, -2.9926834106445312, 4.0076751708984375, 0.15758514404296875, 9.773979187011719, 3.514331817626953, -2.83172607421875, 0.74188232421875, 14.729843139648438, 4.615810394287109, 5.4696807861328125, 13.780204772949219, 10.29197883605957, 6.0879669189453125, -0.07614898681640625, 0.9751968383789062, -3.039949417114258, 10.117759704589844, 3.703704833984375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000142.npy"}
|
||||
{"epoch": 0.2146636432350718, "step": 143, "batch_size": 64, "mean": 3.8062262535095215, "std": 9.525252342224121, "min": -27.91302490234375, "p10": -5.551110839843749, "median": 3.830801010131836, "p90": 14.200115203857425, "max": 21.720977783203125, "pos_frac": 0.703125, "sample": [3.8753509521484375, 10.8411865234375, 13.488349914550781, 10.728775024414062, 4.6066436767578125, -1.99591064453125, 2.457000732421875, -1.146890640258789, 18.99884605407715, -10.50677490234375, 12.45335578918457, -2.1349639892578125, 21.720977783203125, 3.7862510681152344, 8.065208435058594, 6.820789337158203, -14.3106689453125, 0.4185791015625, 20.151351928710938, 1.207763671875, -2.63262939453125, -0.2693061828613281, 20.638965606689453, 13.438400268554688, 9.933547973632812, 3.6277713775634766, 5.30645751953125, -15.553314208984375, 2.5591373443603516, -4.451576232910156, 0.041961669921875, 4.375947952270508, -13.194366455078125, 7.7208709716796875, 17.213394165039062, -2.7219696044921875, -4.458984375, 0.6191234588623047, 0.8330154418945312, 7.098236083984375, 2.9106407165527344, -27.91302490234375, 13.32040023803711, 2.6359710693359375, -2.53778076171875, 10.1195068359375, 13.159515380859375, 10.434188842773438, -0.22521209716796875, -1.8517074584960938, -6.0191650390625, 5.045385360717773, 14.505157470703125, 7.152885437011719, -19.149818420410156, 17.369384765625, 11.04962158203125, 4.6748199462890625, 11.787788391113281, 10.0555419921875, 3.092489242553711, -2.3056278228759766, 4.295040130615234, 2.342571258544922], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000143.npy"}
|
||||
{"epoch": 0.2161753590325019, "step": 144, "batch_size": 64, "mean": 4.972504138946533, "std": 9.397297859191895, "min": -20.980506896972656, "p10": -6.047933959960937, "median": 3.4612321853637695, "p90": 17.665209007263186, "max": 27.213973999023438, "pos_frac": 0.65625, "sample": [17.890853881835938, 6.847850799560547, -6.5634918212890625, 1.3774089813232422, -2.3768672943115234, 3.6509647369384766, 0.7823848724365234, 12.872077941894531, 1.3757171630859375, -0.06659507751464844, -6.2425384521484375, 8.514732360839844, -0.47650909423828125, 3.483644485473633, 8.520011901855469, 3.0256271362304688, 17.04974365234375, 0.15752410888671875, -4.943756103515625, 10.935356140136719, 11.93572998046875, 4.4811248779296875, -1.3730316162109375, -2.7582778930664062, -2.7802467346191406, 1.6886138916015625, 6.4536895751953125, 15.094253540039062, -0.6396617889404297, -1.6998939514160156, 10.338882446289062, 3.4388198852539062, -5.5938568115234375, -2.3079986572265625, 14.71657943725586, -0.8456344604492188, -1.1981372833251953, 20.478525161743164, 27.042068481445312, 19.321121215820312, 11.932525634765625, 6.9943084716796875, -8.511711120605469, 16.167007446289062, -8.527374267578125, 8.442337036132812, 4.873268127441406, -6.527944564819336, 13.41903305053711, 2.770355224609375, 0.8539257049560547, -5.530609130859375, 4.3164215087890625, 21.137805938720703, -7.3650054931640625, 17.138704299926758, 21.558425903320312, 27.213973999023438, -0.4261016845703125, -20.980506896972656, 5.7850494384765625, 2.5144805908203125, 9.633119583129883, 9.751968383789062], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000144.npy"}
|
||||
{"epoch": 0.21768707482993196, "step": 145, "batch_size": 64, "mean": 5.278238296508789, "std": 9.991899490356445, "min": -24.55748748779297, "p10": -4.690245056152343, "median": 5.560205459594727, "p90": 18.315501403808597, "max": 28.49228286743164, "pos_frac": 0.71875, "sample": [6.891670227050781, 8.189014434814453, -4.001100540161133, 6.2446136474609375, -0.0798492431640625, -24.55748748779297, 11.324935913085938, -1.8775482177734375, 28.49228286743164, 2.6999969482421875, 8.596284866333008, 8.484996795654297, 3.4022388458251953, 10.104278564453125, 3.56060791015625, 9.701675415039062, -0.6870269775390625, 20.206268310546875, -6.9730987548828125, 7.533424377441406, 2.799907684326172, 2.6895179748535156, -4.7576141357421875, 6.12408447265625, 19.297237396240234, -1.6293563842773438, 1.9719161987304688, 8.282440185546875, 7.4373321533203125, -1.4968681335449219, 20.828773498535156, 14.860221862792969, 5.565887451171875, 24.322288513183594, -4.533050537109375, 18.790367126464844, 3.647195816040039, 10.386343002319336, 8.935447692871094, 14.112491607666016, -17.308250427246094, 7.769458770751953, 17.207481384277344, 5.705955505371094, 0.5636653900146484, -0.6353607177734375, 3.0442657470703125, 5.554523468017578, -3.2190017700195312, 9.595359802246094, 13.830581665039062, 10.829627990722656, 13.885547637939453, -2.486797332763672, 1.16168212890625, 27.9783935546875, 5.108245849609375, -5.704200744628906, -19.271408081054688, 3.9574050903320312, 16.863494873046875, -8.525604248046875, -3.303466796875, 0.31491851806640625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000145.npy"}
|
||||
{"epoch": 0.21919879062736206, "step": 146, "batch_size": 64, "mean": 5.276614189147949, "std": 8.944928169250488, "min": -13.505035400390625, "p10": -7.214482116699219, "median": 4.7659759521484375, "p90": 17.846662902832033, "max": 26.755386352539062, "pos_frac": 0.75, "sample": [14.658683776855469, 9.285625457763672, -6.75274658203125, 5.800590515136719, 2.2895278930664062, -7.4123687744140625, -3.1527862548828125, 8.474613189697266, 6.098960876464844, 5.209014892578125, -10.455368041992188, -7.4337310791015625, 13.652267456054688, 7.571184158325195, 2.4101638793945312, -6.6114654541015625, 10.773550033569336, -1.500396728515625, 15.11324691772461, -13.505035400390625, 7.555999755859375, 17.674087524414062, 2.0238037109375, -1.1115226745605469, -9.53592300415039, 0.6407508850097656, 4.047121047973633, 11.196815490722656, -9.430625915527344, 2.5222549438476562, 7.749872207641602, -8.871818542480469, 22.602035522460938, 6.475502014160156, 10.568260192871094, -1.0941085815429688, 3.8046035766601562, 0.652679443359375, 5.943023681640625, 0.3880786895751953, 17.920623779296875, 6.512264251708984, 4.16688346862793, 13.256546020507812, 12.983070373535156, 21.684654235839844, 14.421314239501953, 7.024074554443359, 16.118118286132812, 3.8854026794433594, 7.358287811279297, 1.11279296875, 18.9038028717041, 1.6978836059570312, -5.450916290283203, -3.5376129150390625, 26.755386352539062, -4.3694000244140625, 3.297739028930664, 4.32293701171875, 18.671310424804688, 9.314533233642578, 4.139373779296875, 19.1998348236084], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000146.npy"}
|
||||
{"epoch": 0.22071050642479215, "step": 147, "batch_size": 64, "mean": 4.230469703674316, "std": 9.380505561828613, "min": -25.8953857421875, "p10": -4.45965461730957, "median": 3.959989547729492, "p90": 16.22545127868653, "max": 24.325889587402344, "pos_frac": 0.625, "sample": [4.843481063842773, 8.136629104614258, 13.605018615722656, -4.2674713134765625, 2.156219482421875, 7.885993957519531, -11.014991760253906, 3.516693115234375, -1.2471923828125, -4.462436676025391, 8.3873291015625, 7.142169952392578, 12.757354736328125, 8.875518798828125, -3.2721405029296875, 8.377723693847656, 21.129623413085938, -25.8953857421875, 10.03217887878418, -1.6970062255859375, 6.944189071655273, 3.4449234008789062, -0.4836158752441406, 2.4673614501953125, -0.6837615966796875, 16.869474411010742, -4.453163146972656, 9.980712890625, 4.453399658203125, 4.403285980224609, -1.304697036743164, 5.584892272949219, 7.452239990234375, -9.527896881103516, -0.7477264404296875, 10.201499938964844, 14.72273063659668, 24.325889587402344, 21.353233337402344, 0.9168338775634766, 6.782073974609375, 2.8650360107421875, -2.5358734130859375, 12.964580535888672, 13.860401153564453, -5.925529479980469, 23.860801696777344, -2.9842147827148438, 14.111618041992188, -1.6954002380371094, -1.207916259765625, 5.584526062011719, 9.786041259765625, -0.5001049041748047, -4.10382080078125, 0.9917755126953125, 2.390207290649414, -0.9130706787109375, 17.595306396484375, -15.403228759765625, -12.669443130493164, 7.510986328125, 19.6313419342041, -0.15516090393066406], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000147.npy"}
|
||||
{"epoch": 0.2222222222222222, "step": 148, "batch_size": 64, "mean": 5.640023708343506, "std": 8.995809555053711, "min": -18.192298889160156, "p10": -7.295730590820312, "median": 5.879828453063965, "p90": 16.692081451416016, "max": 23.76251220703125, "pos_frac": 0.734375, "sample": [23.676132202148438, -9.818756103515625, 5.481403350830078, -8.792192459106445, 15.7706298828125, 18.588623046875, 10.150459289550781, -10.603036880493164, 17.895294189453125, 16.6004638671875, -2.0920448303222656, 11.615901947021484, 3.6912384033203125, 4.892951965332031, 10.021045684814453, -6.0597076416015625, -2.711200714111328, 14.588653564453125, 12.907539367675781, -2.979808807373047, 9.579471588134766, -8.314163208007812, 14.064201354980469, 2.6818809509277344, 11.453607559204102, 20.704391479492188, 12.275806427001953, -0.7048454284667969, 21.44805908203125, 1.632232666015625, 6.278253555297852, -1.4449005126953125, 11.285429000854492, -7.8254547119140625, 7.024471282958984, 15.900421142578125, 1.7366256713867188, 13.643402099609375, -18.192298889160156, 1.0654850006103516, 8.862785339355469, 1.3304100036621094, 9.377555847167969, 23.76251220703125, -11.632352828979492, 8.237581253051758, 6.642391204833984, 5.169116973876953, 16.731346130371094, -1.0739593505859375, 10.690067291259766, 3.399566650390625, 4.677888870239258, 1.7942161560058594, 0.5376472473144531, 6.432350158691406, 9.181625366210938, 8.418411254882812, 8.439838409423828, -1.3023300170898438, 2.86322021484375, 4.503021240234375, -1.7516250610351562, -1.4454345703125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000148.npy"}
|
||||
{"epoch": 0.2237339380196523, "step": 149, "batch_size": 64, "mean": 6.275809288024902, "std": 8.159056663513184, "min": -11.943305969238281, "p10": -3.2065525054931636, "median": 5.9670515060424805, "p90": 17.59967041015625, "max": 25.36528778076172, "pos_frac": 0.78125, "sample": [12.027557373046875, 18.4415283203125, 12.966426849365234, -2.6062850952148438, 1.566497802734375, 4.529857635498047, -3.4638099670410156, 12.825571060180664, -1.9087390899658203, 14.798768997192383, -11.943305969238281, 4.304437637329102, 2.1889991760253906, 4.06407356262207, 8.36700439453125, 0.17438507080078125, 3.5542144775390625, 2.391357421875, 1.6856880187988281, 10.800460815429688, 2.526691436767578, 11.461380004882812, 19.307329177856445, -3.578533172607422, 25.36528778076172, 6.014997482299805, 4.832576751708984, -2.5826778411865234, 8.248823165893555, -11.168464660644531, -2.5617618560791016, 17.666893005371094, -9.732742309570312, 21.546348571777344, 5.406089782714844, 18.708574295043945, 11.092096328735352, -6.718170166015625, 12.170074462890625, -4.5849761962890625, 2.68310546875, 8.9609375, 1.5785694122314453, 16.969696044921875, 9.931278228759766, 14.998472213745117, 5.788734436035156, 7.297885894775391, 4.666807174682617, 6.936546325683594, 10.191322326660156, 6.533660888671875, -0.9009552001953125, -1.240692138671875, 0.3447608947753906, 10.871978759765625, 6.756439208984375, 22.536376953125, 13.462615966796875, 5.919105529785156, 6.641258239746094, 17.44281768798828, -1.485992431640625, 6.582527160644531], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000149.npy"}
|
||||
{"epoch": 0.2252456538170824, "step": 150, "batch_size": 64, "mean": 8.277616500854492, "std": 10.141423225402832, "min": -12.9857177734375, "p10": -2.73778839111328, "median": 6.542186737060547, "p90": 23.611547851562506, "max": 31.617523193359375, "pos_frac": 0.796875, "sample": [8.297969818115234, 3.973543167114258, 25.2967529296875, 7.311676025390625, -10.974601745605469, -3.27777099609375, 6.292940139770508, 6.5238037109375, 0.7137451171875, 8.590028762817383, 2.7601470947265625, 14.061820983886719, -1.285919189453125, 6.30352783203125, 0.5271282196044922, 5.746368408203125, 4.789482116699219, 31.617523193359375, -0.927032470703125, 6.560569763183594, 10.022697448730469, 2.1425933837890625, 26.376449584960938, -10.470352172851562, 1.342437744140625, 12.016782760620117, 4.483182907104492, 7.940235137939453, 6.3386383056640625, 21.323455810546875, 10.22459602355957, 15.909011840820312, 21.125022888183594, 6.7644805908203125, 25.363426208496094, -5.3229827880859375, 1.3018169403076172, 9.968528747558594, 16.87171173095703, 24.844009399414062, 11.31353759765625, 19.01134490966797, -1.36346435546875, 11.692962646484375, 21.889671325683594, 1.944122314453125, 15.154077529907227, 17.316038131713867, -12.9857177734375, 31.335098266601562, 5.5819549560546875, -0.67474365234375, 8.221412658691406, -5.787109375, 3.1247215270996094, 24.34949493408203, -7.594516754150391, -1.4778289794921875, 15.946151733398438, -0.8129806518554688, 5.312808990478516, 18.605987548828125, 6.3288116455078125, 11.8681640625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000150.npy"}
|
||||
{"epoch": 0.22675736961451248, "step": 151, "batch_size": 64, "mean": 9.464058876037598, "std": 10.23266887664795, "min": -9.359123229980469, "p10": -2.5319158554077146, "median": 9.544729232788086, "p90": 19.377587890625, "max": 41.461334228515625, "pos_frac": 0.765625, "sample": [8.166633605957031, 16.0101318359375, 17.204788208007812, -4.102787017822266, 2.9870948791503906, 3.9351634979248047, 41.461334228515625, 0.7058982849121094, 19.145828247070312, -0.7486457824707031, 10.932075500488281, 15.119129180908203, 8.651596069335938, -2.186920166015625, 10.19830322265625, 13.381145477294922, 16.92601203918457, -1.1912975311279297, 7.717123031616211, -1.3817634582519531, -4.535697937011719, -2.6675243377685547, 13.497886657714844, -3.101085662841797, 33.57539367675781, -9.359123229980469, 13.620611190795898, 10.362213134765625, -1.8622722625732422, -1.5037612915039062, 6.262931823730469, -3.5620956420898438, 14.285285949707031, 15.79901123046875, 4.211723327636719, -1.3766021728515625, 12.99102783203125, 8.441024780273438, 8.577499389648438, 33.3897705078125, 0.55828857421875, 10.738677978515625, 17.895954132080078, 6.542057037353516, 25.125198364257812, 22.123313903808594, 19.476913452148438, 18.289886474609375, 17.887474060058594, 9.706974029541016, 9.382484436035156, 16.92668914794922, 3.3073368072509766, 21.215911865234375, -7.973834991455078, 14.757213592529297, 13.347711563110352, 6.962673187255859, -2.215496063232422, 0.59844970703125, 13.04754638671875, 18.747241973876953, 0.8030338287353516, 18.47100830078125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000151.npy"}
|
||||
{"epoch": 0.22826908541194255, "step": 152, "batch_size": 64, "mean": 6.127760887145996, "std": 11.022237777709961, "min": -18.4324951171875, "p10": -7.215486526489258, "median": 4.609134674072266, "p90": 20.717618751525883, "max": 29.141067504882812, "pos_frac": 0.734375, "sample": [-1.1487503051757812, 3.3143081665039062, 17.214733123779297, 2.8374481201171875, -18.4324951171875, 4.589622497558594, 13.567474365234375, 8.671516418457031, 6.4593353271484375, -7.4922943115234375, 6.189197540283203, -9.113868713378906, 11.356464385986328, -3.6151275634765625, 19.772918701171875, 18.00635528564453, 2.476675033569336, 9.193462371826172, 1.0071392059326172, -4.891090393066406, -17.500396728515625, 23.084754943847656, 11.792213439941406, 9.644630432128906, -5.5701446533203125, 0.8289909362792969, 16.376800537109375, 4.6286468505859375, 1.9639053344726562, 11.442550659179688, 2.9912281036376953, 20.990507125854492, 1.8763790130615234, 2.8674545288085938, -2.7955093383789062, -4.7938232421875, 2.1290111541748047, -5.768341064453125, 18.137542724609375, 2.002716064453125, -7.402019500732422, 21.464677810668945, 13.830841064453125, -6.780242919921875, 7.4821319580078125, 29.141067504882812, 14.816673278808594, 0.8151283264160156, 1.7556228637695312, 10.03369140625, -14.65936279296875, 14.669197082519531, 0.8244953155517578, 27.228622436523438, 16.963041305541992, 7.854625701904297, 24.89789581298828, 20.08087921142578, -2.8695526123046875, 23.6533203125, -0.9861831665039062, -12.213668823242188, 9.022262573242188, 18.261402130126953], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000152.npy"}
|
||||
{"epoch": 0.22978080120937264, "step": 153, "batch_size": 64, "mean": 7.216427803039551, "std": 9.342192649841309, "min": -14.141014099121094, "p10": -4.8027599334716795, "median": 6.3994951248168945, "p90": 20.354340744018558, "max": 25.71875, "pos_frac": 0.765625, "sample": [12.082212448120117, 4.107826232910156, 7.544471740722656, 15.230682373046875, 17.470741271972656, 22.484031677246094, 11.426666259765625, 3.059734344482422, 3.9973983764648438, 6.4222564697265625, 6.376733779907227, 18.807525634765625, 3.9745101928710938, 17.045425415039062, 6.2285003662109375, -4.996337890625, 6.908714294433594, 0.4377593994140625, 2.417022705078125, 22.133514404296875, 21.441184997558594, 14.490188598632812, -6.679847717285156, 9.824325561523438, 23.273990631103516, -11.400320053100586, 14.72027587890625, 2.1717529296875, 0.560821533203125, 17.879146575927734, -5.145973205566406, 5.251325607299805, -14.141014099121094, -1.0831489562988281, 12.9478759765625, 4.788440704345703, -2.6209793090820312, 19.99126434326172, 9.682861328125, 5.591392517089844, -3.71014404296875, -1.4549560546875, 23.90093994140625, 10.668651580810547, 11.92465591430664, -1.2118854522705078, 11.17083740234375, 8.83868408203125, 10.2584228515625, -2.8047122955322266, 0.277801513671875, 15.064949035644531, 5.982404708862305, -0.361480712890625, 6.4456634521484375, -8.047378540039062, -5.9719085693359375, 18.01106834411621, 2.507537841796875, 25.71875, 20.509944915771484, 1.5341205596923828, -4.351078033447266, 12.247566223144531], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000153.npy"}
|
||||
{"epoch": 0.23129251700680273, "step": 154, "batch_size": 64, "mean": 5.260444641113281, "std": 10.793181419372559, "min": -21.60418701171875, "p10": -10.654447174072263, "median": 6.260149002075195, "p90": 17.890965080261235, "max": 27.86284065246582, "pos_frac": 0.703125, "sample": [-12.699745178222656, 1.0956649780273438, -12.999618530273438, -21.60418701171875, 23.218978881835938, 14.831785202026367, -11.568500518798828, 6.227642059326172, 5.08746337890625, -15.066612243652344, 5.994890213012695, -0.36236000061035156, 14.651662826538086, 7.516021728515625, -7.921775817871094, -7.292182922363281, -2.3490867614746094, 6.457195281982422, 0.5979194641113281, -1.0672454833984375, 15.267707824707031, 27.86284065246582, 25.461402893066406, 6.292655944824219, -12.844589233398438, 9.401121139526367, 12.332054138183594, -2.5466690063476562, 13.83367919921875, -19.563919067382812, 8.890340805053711, 3.1375961303710938, 15.838088989257812, 16.789566040039062, 12.66061019897461, -0.9921684265136719, 12.003164291381836, 16.82598876953125, 9.625457763671875, 6.633354187011719, -1.6618804931640625, 0.4340362548828125, 12.882806777954102, -1.8785400390625, 3.363037109375, 21.748109817504883, 9.373394012451172, 5.818599700927734, 15.31951904296875, -8.521656036376953, 1.3327484130859375, -4.0629730224609375, 2.2093658447265625, 4.677825927734375, 9.968170166015625, 6.61407470703125, 19.819984436035156, -0.9229030609130859, 18.347383499145508, 21.34532928466797, 7.219383239746094, 10.093877792358398, 2.451478958129883, 11.041116714477539], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000154.npy"}
|
||||
{"epoch": 0.2328042328042328, "step": 155, "batch_size": 64, "mean": 7.4498138427734375, "std": 13.016178131103516, "min": -26.56668472290039, "p10": -6.0113174438476555, "median": 7.356359481811523, "p90": 25.511578750610354, "max": 34.39320373535156, "pos_frac": 0.703125, "sample": [7.530609130859375, -1.266937255859375, 33.23516845703125, 6.686920166015625, 5.355926513671875, 9.835399627685547, 1.0135078430175781, 22.964908599853516, -2.5500221252441406, 14.754386901855469, 10.969573974609375, 15.92228889465332, 1.7290191650390625, -5.663848876953125, -0.04401397705078125, -3.9379501342773438, 12.547653198242188, 29.348983764648438, 10.32672119140625, 27.86296844482422, 25.904956817626953, -1.7727489471435547, 13.167648315429688, -14.91242790222168, 13.797927856445312, 27.852359771728516, -0.30748748779296875, 20.667022705078125, -4.1027984619140625, 14.765205383300781, -17.440032958984375, 11.297758102416992, 23.386198043823242, 9.652053833007812, 34.39320373535156, 11.73480224609375, -1.1293373107910156, 2.3701248168945312, -19.75482177734375, 20.024261474609375, -3.6777496337890625, 0.6476173400878906, 13.63946533203125, 1.4552383422851562, 30.04277801513672, 8.983318328857422, 4.562461853027344, 21.23967742919922, -13.035789489746094, 7.182109832763672, 7.8894500732421875, 3.720123291015625, 3.0127410888671875, 3.095550537109375, 3.5383529663085938, -0.3725738525390625, 8.183074951171875, -12.537353515625, -26.56668472290039, -0.142791748046875, 10.839881896972656, 24.59369659423828, 20.440616607666016, -6.1602325439453125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000155.npy"}
|
||||
{"epoch": 0.23431594860166288, "step": 156, "batch_size": 64, "mean": 8.85677719116211, "std": 10.885432243347168, "min": -14.47921371459961, "p10": -4.1278850555419915, "median": 6.781105041503906, "p90": 24.478142547607423, "max": 31.918813705444336, "pos_frac": 0.78125, "sample": [5.9579010009765625, -4.467292785644531, -4.387882232666016, -14.47921371459961, 10.9573974609375, 29.697738647460938, 31.918813705444336, 12.104606628417969, 13.474853515625, 23.732772827148438, 3.0753021240234375, 22.07171630859375, -6.590766906738281, 17.322052001953125, 19.607666015625, 24.12628936767578, 24.628936767578125, 20.969924926757812, -0.13781166076660156, 15.063804626464844, 3.140573501586914, 2.4997081756591797, 7.234928131103516, 18.44467544555664, -3.3687820434570312, 5.626914978027344, 0.3068275451660156, 18.771644592285156, 4.115631103515625, 14.836875915527344, 9.549797058105469, 6.40081787109375, -3.067901611328125, 10.9501953125, 11.547805786132812, 30.83911895751953, -10.259414672851562, 6.7804412841796875, -5.786590576171875, 26.695274353027344, 12.859619140625, 22.78680419921875, 3.966297149658203, 3.19281005859375, 0.8618755340576172, 4.871740341186523, -0.650238037109375, 0.249664306640625, 9.53466796875, -3.5212249755859375, 8.308238983154297, 3.390625, 12.907623291015625, -0.06108856201171875, 6.781768798828125, 4.0292816162109375, -2.566314697265625, 18.49798583984375, 25.675323486328125, 2.462850570678711, 7.831085205078125, 27.065383911132812, 5.772449493408203, -7.318817138671875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000156.npy"}
|
||||
{"epoch": 0.23582766439909297, "step": 157, "batch_size": 64, "mean": 7.025417804718018, "std": 11.397170066833496, "min": -22.282562255859375, "p10": -5.700795745849609, "median": 6.629253387451172, "p90": 21.56104850769043, "max": 34.96319580078125, "pos_frac": 0.734375, "sample": [28.16558837890625, 26.98200798034668, -4.649681091308594, 14.312793731689453, 19.06915283203125, -22.282562255859375, 5.2861785888671875, -5.13629150390625, 15.652671813964844, -5.942726135253906, 0.7804183959960938, 7.233222961425781, 17.2950439453125, 6.036825180053711, -17.374588012695312, 0.2222747802734375, -0.319244384765625, 7.610740661621094, 15.96025276184082, 21.479537963867188, 7.4280853271484375, 21.59598159790039, 34.96319580078125, 10.266853332519531, 8.872480392456055, 16.304847717285156, -7.182670593261719, -1.2158679962158203, 8.944927215576172, -2.5848922729492188, 17.67875862121582, -0.8434295654296875, 11.971553802490234, 6.750007629394531, 14.274726867675781, 24.32727813720703, -8.836593627929688, 24.794572830200195, 27.81920623779297, 2.9020919799804688, -2.1745986938476562, 3.7080307006835938, -0.8516082763671875, 17.431678771972656, 8.119598388671875, 4.419075012207031, 2.4495925903320312, 11.077781677246094, 9.952438354492188, 3.108766555786133, 1.7443008422851562, -3.2773590087890625, 11.3955078125, 0.9848442077636719, 4.938499450683594, 15.687360763549805, 3.9595794677734375, 16.00539779663086, -8.606735229492188, 14.562721252441406, 0.6547698974609375, -4.046577453613281, -16.737548828125, 6.5084991455078125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000157.npy"}
|
||||
{"epoch": 0.23733938019652306, "step": 158, "batch_size": 64, "mean": 7.629086971282959, "std": 12.363077163696289, "min": -19.61758041381836, "p10": -6.6406616210937495, "median": 4.950023651123047, "p90": 25.78134059906006, "max": 36.04296875, "pos_frac": 0.71875, "sample": [-2.4545059204101562, 15.156906127929688, 1.6003570556640625, 8.304206848144531, -19.374710083007812, 17.325300216674805, 28.214553833007812, 12.1092529296875, -1.142730712890625, 1.0093841552734375, 22.785743713378906, 12.192802429199219, 10.313064575195312, 18.91945457458496, -19.61758041381836, 14.85807991027832, 36.04296875, 4.742090225219727, 4.747615814208984, 6.3963470458984375, -2.3965835571289062, 5.35968017578125, 14.697996139526367, 1.5088577270507812, 25.874916076660156, 14.677871704101562, -1.93548583984375, 25.485061645507812, 20.591079711914062, 13.778202056884766, -6.7846832275390625, 12.052696228027344, 26.07311248779297, -1.9263534545898438, 25.562997817993164, 27.255884170532227, 15.211753845214844, -6.3046112060546875, 2.477109909057617, -8.5235595703125, 22.289352416992188, 10.289318084716797, 1.8174591064453125, 27.402252197265625, 14.256599426269531, 16.95983123779297, -4.244119644165039, 4.082618713378906, 30.903045654296875, -5.306539535522461, 4.2957000732421875, -4.95458984375, -8.967926025390625, 0.7904129028320312, 0.8219509124755859, 0.1632976531982422, -4.632848739624023, 5.152431488037109, 13.728546142578125, 4.506473541259766, -9.963672637939453, 2.9287109375, -0.23089218139648438, -8.690399169921875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000158.npy"}
|
||||
{"epoch": 0.23885109599395313, "step": 159, "batch_size": 64, "mean": 8.67977237701416, "std": 12.95463752746582, "min": -35.84356689453125, "p10": -5.348406791687012, "median": 7.293201446533203, "p90": 26.735820007324218, "max": 37.1463623046875, "pos_frac": 0.75, "sample": [19.865726470947266, 14.866397857666016, 1.7578659057617188, 22.99384307861328, 11.441741943359375, -14.399864196777344, -6.6292724609375, 26.66205596923828, 2.9528350830078125, 37.1463623046875, 28.217697143554688, 4.741477966308594, 21.886537551879883, 7.4760284423828125, -3.2531776428222656, 1.5561561584472656, 9.128646850585938, 33.603477478027344, 25.057899475097656, -0.30733299255371094, 2.0379638671875, -8.001884460449219, 6.339630126953125, 0.464111328125, 6.876502990722656, 3.3343143463134766, 1.880868911743164, -8.979766845703125, 20.658035278320312, 26.767433166503906, -5.416015625, 18.616165161132812, 16.093259811401367, -0.5872573852539062, 5.676048278808594, 10.843439102172852, 23.308483123779297, 9.703567504882812, 4.349700927734375, -0.71539306640625, 7.110374450683594, -10.8380126953125, 29.058914184570312, 0.7900123596191406, 12.651554107666016, 27.956140518188477, -5.190652847290039, 16.253103256225586, -0.40758514404296875, 8.873863220214844, -0.034000396728515625, 29.06540298461914, 9.737876892089844, -2.3212890625, -0.1780567169189453, -35.84356689453125, 20.428701400756836, 15.122840881347656, 0.41939544677734375, 7.829851150512695, 18.228492736816406, 6.2352752685546875, 12.380867004394531, 10.161613464355469], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000159.npy"}
|
||||
{"epoch": 0.24036281179138322, "step": 160, "batch_size": 64, "mean": 7.533838272094727, "std": 13.229598999023438, "min": -19.20226287841797, "p10": -7.068058013916016, "median": 4.786008834838867, "p90": 25.357495880126958, "max": 40.16166687011719, "pos_frac": 0.734375, "sample": [1.947235107421875, -3.6857833862304688, 16.6385498046875, 33.550933837890625, 0.8596305847167969, 25.800453186035156, 4.791595458984375, -6.325986862182617, 13.50533676147461, 12.813400268554688, 40.16166687011719, -7.220405578613281, 23.804494857788086, 21.080764770507812, 8.931842803955078, 18.078125, 1.2026023864746094, -19.20226287841797, 15.318614959716797, 11.069944381713867, -4.109321594238281, -9.902267456054688, 2.807270050048828, 2.743377685546875, 14.098199844360352, 1.6601066589355469, 12.496925354003906, 2.0206775665283203, 5.4078216552734375, 1.6200599670410156, 17.23610496520996, 9.572450637817383, 38.961700439453125, 32.5778694152832, 2.995401382446289, -6.372804641723633, 14.770530700683594, 2.0234107971191406, -0.912689208984375, 25.866249084472656, 4.780422210693359, 15.567447662353516, 2.3181610107421875, -14.084381103515625, -4.803108215332031, 10.75408935546875, -14.209846496582031, -6.38908576965332, 15.439552307128906, 14.104660034179688, -7.154106140136719, -4.891401290893555, -3.5940399169921875, 14.51611328125, 12.44744873046875, 3.8967132568359375, 1.3077239990234375, 29.1534423828125, -16.796836853027344, 3.4050674438476562, 24.323928833007812, 22.045974731445312, 8.213134765625, -6.867279052734375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000160.npy"}
|
||||
{"epoch": 0.2418745275888133, "step": 161, "batch_size": 64, "mean": 8.29751205444336, "std": 12.909443855285645, "min": -32.006004333496094, "p10": -6.962907886505125, "median": 8.681161880493164, "p90": 25.38074054718018, "max": 40.22715759277344, "pos_frac": 0.765625, "sample": [10.4447021484375, -16.693408966064453, 3.4618072509765625, 11.885574340820312, 3.057525634765625, -0.7690811157226562, 18.96246337890625, 31.92486572265625, -7.7765655517578125, -5.095206260681152, -10.334823608398438, 7.964569091796875, 12.781585693359375, 13.572402954101562, 28.780033111572266, 13.79658317565918, 8.933427810668945, -32.006004333496094, 18.433181762695312, -11.226898193359375, -4.526969909667969, 2.7511444091796875, 8.255767822265625, 2.06488037109375, 21.10881805419922, -3.2223854064941406, 6.2454071044921875, 16.325807571411133, 15.58306884765625, 4.1291351318359375, 23.648683547973633, 16.60419464111328, 11.11430549621582, 17.001808166503906, 20.046859741210938, -3.9084625244140625, -12.396669387817383, 40.22715759277344, 28.78264617919922, 26.123050689697266, 4.636051177978516, 15.282737731933594, -1.0604209899902344, 7.657173156738281, 11.974357604980469, 2.7853851318359375, 9.744300842285156, 9.003129959106445, 6.8192596435546875, 21.474258422851562, -3.6900558471679688, 8.570701599121094, 30.202409744262695, 8.900978088378906, 0.9144439697265625, 0.4041118621826172, -7.7633514404296875, 2.885000228881836, 12.160140991210938, 12.147148132324219, 34.739967346191406, -1.7072525024414062, 0.11371421813964844, 8.791622161865234], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000161.npy"}
|
||||
{"epoch": 0.24338624338624337, "step": 162, "batch_size": 64, "mean": 10.862085342407227, "std": 13.414870262145996, "min": -24.706375122070312, "p10": -6.269386291503905, "median": 10.002204895019531, "p90": 27.20596694946289, "max": 48.52995300292969, "pos_frac": 0.828125, "sample": [15.52484130859375, 6.01068115234375, -24.706375122070312, 5.239662170410156, 5.187431335449219, 23.735309600830078, 16.069366455078125, 7.6378173828125, 21.696929931640625, 29.89177894592285, 18.305763244628906, 20.13918685913086, 5.469022750854492, 48.52995300292969, 25.596298217773438, 8.129972457885742, 11.72772216796875, 30.052169799804688, 17.5704345703125, 29.57870864868164, 28.937538146972656, 24.437580108642578, 5.861686706542969, -5.0367279052734375, 11.923507690429688, -1.0497970581054688, 9.623687744140625, 3.3154144287109375, -16.872879028320312, 16.374969482421875, 6.948692321777344, 5.5608367919921875, 2.12432861328125, 20.31972885131836, 10.380722045898438, 12.348358154296875, 18.20769500732422, 18.946767807006836, -0.5722827911376953, -11.269388198852539, -12.346824645996094, 0.023681640625, 27.24610137939453, 21.764062881469727, 22.416231155395508, 8.761184692382812, -12.716106414794922, 14.943023681640625, -6.79766845703125, 1.9533958435058594, 8.12472152709961, -0.4729156494140625, 12.977378845214844, 22.338361740112305, 27.112319946289062, 16.63970184326172, 2.633869171142578, -15.072288513183594, 4.38311767578125, 5.70623779296875, 3.3405895233154297, 25.054073333740234, 29.194822311401367, 6.069267272949219], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000162.npy"}
|
||||
{"epoch": 0.24489795918367346, "step": 163, "batch_size": 64, "mean": 11.196203231811523, "std": 13.904367446899414, "min": -20.32703399658203, "p10": -8.86513214111328, "median": 9.823606491088867, "p90": 27.655586433410644, "max": 36.57542037963867, "pos_frac": 0.765625, "sample": [-2.024155616760254, 10.475624084472656, -7.838897705078125, 19.654651641845703, 7.438346862792969, 3.4168128967285156, -20.32703399658203, 6.6572265625, 31.36810302734375, -9.304946899414062, 7.7253875732421875, 23.633262634277344, 33.09430694580078, 9.403579711914062, 7.003211975097656, 19.51757049560547, 23.507099151611328, -2.4279708862304688, 34.947052001953125, 16.558746337890625, 4.521903991699219, 27.243026733398438, 24.60883903503418, -10.232765197753906, -6.8987274169921875, 29.209884643554688, 16.218257904052734, 27.56444549560547, 7.823617935180664, 7.8537445068359375, 17.029062271118164, 5.5595855712890625, 21.24938201904297, -4.105556488037109, 19.4986572265625, 22.425613403320312, -9.956626892089844, 21.96539306640625, -11.062782287597656, 3.2588329315185547, 16.058467864990234, 36.57542037963867, 17.57392120361328, 27.529186248779297, 13.241607666015625, -2.787982940673828, -0.7224502563476562, 8.558208465576172, 5.902275085449219, -1.3148117065429688, -16.823875427246094, 27.69464683532715, 19.01299285888672, 1.3822822570800781, 1.7920722961425781, -12.347747802734375, 26.36803436279297, 26.048126220703125, 31.4381103515625, 25.273269653320312, 18.762863159179688, 3.7781505584716797, 7.068864822387695, 10.243633270263672], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000163.npy"}
|
||||
{"epoch": 0.24640967498110355, "step": 164, "batch_size": 64, "mean": 10.824628829956055, "std": 14.99836540222168, "min": -30.855003356933594, "p10": -6.818921089172363, "median": 9.817508697509766, "p90": 31.37398185729981, "max": 40.8642578125, "pos_frac": 0.796875, "sample": [-0.0110626220703125, 0.7252388000488281, -2.0817794799804688, 7.256023406982422, 4.902626037597656, 13.400154113769531, 36.926795959472656, 22.620704650878906, 0.9014968872070312, 5.630519866943359, 12.3671875, -4.746246337890625, 19.87689971923828, -13.941661834716797, 10.107105255126953, 14.591171264648438, 21.024314880371094, 12.320653915405273, 40.8642578125, -6.917182922363281, 6.051868438720703, -14.665390014648438, 8.157878875732422, 9.619766235351562, -0.35352134704589844, 11.338447570800781, 30.094196319580078, 2.253814697265625, 19.493499755859375, 4.688129425048828, 29.819852828979492, 31.807830810546875, -6.589643478393555, 15.48807144165039, 2.4141006469726562, -15.210472106933594, 33.507972717285156, 38.3974609375, 33.526695251464844, -13.294090270996094, 10.26844596862793, 3.7975082397460938, 4.193500518798828, 22.170312881469727, 18.04977798461914, 16.126598358154297, 27.368026733398438, 2.392120361328125, 24.314361572265625, 5.870166778564453, 7.315185546875, -0.6017570495605469, 30.36166763305664, -13.81218147277832, 0.3678569793701172, -30.855003356933594, 9.808708190917969, 27.809385299682617, 29.976760864257812, 1.45068359375, 33.16194152832031, 16.349685668945312, 14.702510833740234, 9.826309204101562], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000164.npy"}
|
||||
{"epoch": 0.24792139077853365, "step": 165, "batch_size": 64, "mean": 9.748390197753906, "std": 15.279340744018555, "min": -28.01287078857422, "p10": -8.124484252929687, "median": 9.571918487548828, "p90": 28.8927116394043, "max": 37.42925262451172, "pos_frac": 0.75, "sample": [37.42925262451172, 7.755132675170898, 11.421470642089844, 24.04935073852539, 12.00830078125, 2.9501495361328125, 6.203922271728516, -2.9218292236328125, 12.001373291015625, 3.8771896362304688, 4.171775817871094, 12.404277801513672, 33.128631591796875, 0.11832618713378906, -2.3854827880859375, 9.139495849609375, 10.282150268554688, 13.3070068359375, -25.07683563232422, -6.406745910644531, -20.672531127929688, -2.6094284057617188, 23.45917510986328, -3.5216903686523438, 20.513717651367188, 24.31134605407715, 10.972808837890625, 19.560327529907227, 32.669761657714844, 32.96611022949219, 7.750328063964844, 1.1541728973388672, 34.555389404296875, 30.857574462890625, 19.18310546875, -28.01287078857422, 3.0110702514648438, 20.168540954589844, 24.86568832397461, 23.6708984375, -8.675962448120117, -4.531551361083984, -2.4951629638671875, 5.570043563842773, 20.465518951416016, -8.429931640625, 3.5629844665527344, 1.528594970703125, 21.073951721191406, -2.5936355590820312, -7.411773681640625, 26.6181640625, -26.968849182128906, 27.64582061767578, 5.38142204284668, -9.559146881103516, 10.004341125488281, 8.403266906738281, 22.67959976196289, 8.269973754882812, 27.064857482910156, 29.427093505859375, 20.61420440673828, 17.942710876464844], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000165.npy"}
|
||||
{"epoch": 0.2494331065759637, "step": 166, "batch_size": 64, "mean": 7.752241134643555, "std": 13.15926742553711, "min": -13.500301361083984, "p10": -7.343096923828124, "median": 6.513950347900391, "p90": 26.521157836914064, "max": 36.53727340698242, "pos_frac": 0.65625, "sample": [18.55335235595703, -0.37248802185058594, -3.1282196044921875, 13.395027160644531, 25.952735900878906, 3.284759521484375, -5.6968994140625, 28.170089721679688, 8.253280639648438, 7.160308837890625, -2.801483154296875, 5.712947845458984, 30.871013641357422, -5.011970520019531, 25.894622802734375, 7.9652862548828125, -0.3129119873046875, 17.55586814880371, -11.198776245117188, 4.89007568359375, -2.894794464111328, 22.946514129638672, -7.5880279541015625, 5.867591857910156, 30.694602966308594, 7.979337692260742, -11.082687377929688, 20.43792724609375, 10.643783569335938, 10.034942626953125, -11.43267822265625, -13.500301361083984, 4.212968826293945, 26.22856903076172, 24.29334259033203, -1.2165107727050781, 26.64655303955078, 2.4792098999023438, 12.55670166015625, 15.809255599975586, 7.173065185546875, 34.283538818359375, -6.635833740234375, -6.7715911865234375, 11.4251708984375, -5.7851409912109375, -4.629158020019531, 1.3293418884277344, 12.034917831420898, 36.53727340698242, 15.157218933105469, -11.693428039550781, 2.177215576171875, 13.679336547851562, -11.985599517822266, -3.0186996459960938, 13.835746765136719, -5.403957366943359, 33.0445556640625, -1.8935737609863281, 10.535629272460938, 17.53319549560547, 1.1813831329345703, 1.7799263000488281], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000166.npy"}
|
||||
{"epoch": 0.2509448223733938, "step": 167, "batch_size": 64, "mean": 11.829839706420898, "std": 13.746853828430176, "min": -17.261077880859375, "p10": -3.3814781188964838, "median": 8.39249324798584, "p90": 30.313464355468753, "max": 38.71377944946289, "pos_frac": 0.796875, "sample": [7.267333984375, 5.358173370361328, 4.430414199829102, 11.36513900756836, -8.412986755371094, 25.99859619140625, 12.239738464355469, 6.514862060546875, 9.51765251159668, 20.761436462402344, 24.595510482788086, 3.7940826416015625, 12.40671157836914, -3.6155624389648438, -1.6032676696777344, -0.21459579467773438, 29.186317443847656, 29.928131103515625, 19.935623168945312, -0.6372146606445312, -17.261077880859375, -2.252887725830078, 4.165651321411133, 1.0866928100585938, 14.316041946411133, -1.2446136474609375, 3.5783309936523438, 0.3211536407470703, 4.226020812988281, 15.14190673828125, 5.4941558837890625, 24.000072479248047, 4.447484970092773, 18.472919464111328, 15.549415588378906, 11.085067749023438, 36.30937957763672, 37.618568420410156, 25.0587158203125, 24.079002380371094, 2.934009552001953, 29.66266632080078, 2.0263137817382812, 6.6566009521484375, 21.822946548461914, -13.051559448242188, -10.590972900390625, 37.055152893066406, -2.8352813720703125, 23.713706970214844, 26.36068344116211, 1.0805015563964844, 7.082040786743164, 38.71377944946289, 15.227622985839844, 30.478607177734375, -4.514179229736328, 35.0792236328125, 25.382488250732422, 33.681922912597656, 15.778831481933594, 1.8099727630615234, 4.6040191650390625, -4.057453155517578], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000167.npy"}
|
||||
{"epoch": 0.25245653817082386, "step": 168, "batch_size": 64, "mean": 9.57107162475586, "std": 14.204455375671387, "min": -17.8770751953125, "p10": -5.942430114746093, "median": 5.430274963378906, "p90": 31.396245193481448, "max": 37.84230041503906, "pos_frac": 0.671875, "sample": [8.335433959960938, -6.515525817871094, 33.25972366333008, -4.230224609375, 6.367483139038086, 30.380390167236328, 16.232742309570312, -3.2340774536132812, -17.8770751953125, -11.933868408203125, 7.427848815917969, -1.5511703491210938, 5.311859130859375, 33.510887145996094, 14.379180908203125, 5.036369323730469, 21.285167694091797, 5.5486907958984375, 1.2098236083984375, 18.618133544921875, 13.820167541503906, -1.9785175323486328, 1.470428466796875, 6.080453872680664, -0.6311359405517578, 14.775005340576172, 32.7205924987793, 2.111339569091797, -6.039558410644531, 23.38098907470703, 1.776031494140625, -2.0144271850585938, 15.303695678710938, 31.83161163330078, 29.038196563720703, -4.639373779296875, 24.9591064453125, -1.5254745483398438, -0.8178329467773438, 0.91259765625, 4.338399887084961, 36.51836395263672, -4.306877136230469, 27.686126708984375, 17.476089477539062, -1.6541213989257812, 17.69305419921875, -7.776458740234375, 4.555076599121094, -3.6120758056640625, 15.2406005859375, 15.677505493164062, 28.174846649169922, 3.0710372924804688, -5.891929626464844, 28.334577560424805, -5.964073181152344, -7.9654083251953125, -2.2442703247070312, 1.2258377075195312, 29.489166259765625, 35.87030792236328, 37.84230041503906, 6.674797058105469], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000168.npy"}
|
||||
{"epoch": 0.25396825396825395, "step": 169, "batch_size": 64, "mean": 10.600841522216797, "std": 17.53333282470703, "min": -37.11260223388672, "p10": -6.775985336303709, "median": 8.362903594970703, "p90": 32.03947296142578, "max": 60.286895751953125, "pos_frac": 0.71875, "sample": [7.158256530761719, 32.18315124511719, 30.385353088378906, -11.663711547851562, 8.870134353637695, 2.725433349609375, -7.626163482666016, -0.6132164001464844, 10.295787811279297, 15.904579162597656, 10.463462829589844, -0.2293548583984375, 24.457366943359375, 29.8973388671875, 7.897918701171875, -3.4273605346679688, 7.349826812744141, 18.8236083984375, 21.15595245361328, -0.3285980224609375, -4.792236328125, 27.535202026367188, 31.7042236328125, 12.899154663085938, -2.3172149658203125, 52.917030334472656, 2.5146026611328125, 7.85443115234375, -4.031772613525391, 13.735332489013672, 6.873693466186523, 6.924018859863281, 23.565868377685547, -2.005552291870117, 27.76068115234375, 6.132091522216797, 24.314790725708008, -37.11260223388672, 14.34242057800293, 10.872291564941406, -0.4365043640136719, 8.776924133300781, 1.9297676086425781, -0.12777328491210938, 7.294889450073242, 17.225975036621094, 12.529586791992188, 7.948883056640625, 33.668304443359375, 15.467041015625, 0.79840087890625, 14.980504989624023, 12.767013549804688, -26.166412353515625, -16.897232055664062, 60.286895751953125, 36.958251953125, 35.76774597167969, -17.235246658325195, 42.078739166259766, 2.1459808349609375, 26.931562423706055, -20.950347900390625, -0.6553268432617188], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000169.npy"}
|
||||
{"epoch": 0.25547996976568405, "step": 170, "batch_size": 64, "mean": 10.891260147094727, "std": 18.39212989807129, "min": -38.15709686279297, "p10": -15.74198474884033, "median": 11.036672592163086, "p90": 32.99863891601563, "max": 49.36555480957031, "pos_frac": 0.75, "sample": [10.246421813964844, -0.14064788818359375, 15.518135070800781, 13.440963745117188, 2.1929378509521484, 40.54425811767578, 8.73609733581543, 11.378715515136719, 6.0981597900390625, 37.46208953857422, -25.20850372314453, 4.284694671630859, 19.01685333251953, 41.86956787109375, 34.835693359375, -16.53092384338379, 2.228761672973633, 11.87643051147461, 26.505386352539062, -4.5332794189453125, 6.107509613037109, 30.73650360107422, 22.935319900512695, 3.126821517944336, -20.69355010986328, 15.426155090332031, -28.169815063476562, -3.08282470703125, -38.15709686279297, 23.960693359375, 31.09151840209961, 25.639755249023438, 25.480430603027344, 13.478464126586914, -13.901126861572266, 14.98175048828125, -8.000244140625, 18.635787963867188, 25.634796142578125, -6.259468078613281, 15.15057373046875, -0.3518257141113281, -0.7030868530273438, 25.969955444335938, 8.854469299316406, 49.36555480957031, 16.22802734375, 33.46539306640625, 8.768627166748047, 2.3002700805664062, 9.421577453613281, -6.622560501098633, 3.0340347290039062, 39.058448791503906, -16.689239501953125, 7.228435516357422, 26.12643051147461, 31.9095458984375, 1.611419677734375, 10.694629669189453, 29.012863159179688, 30.149551391601562, 17.491249084472656, -23.226882934570312], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000170.npy"}
|
||||
{"epoch": 0.25699168556311414, "step": 171, "batch_size": 64, "mean": 11.86996841430664, "std": 19.277650833129883, "min": -44.299171447753906, "p10": -7.195816040039062, "median": 8.256553649902344, "p90": 37.71534538269044, "max": 43.750709533691406, "pos_frac": 0.703125, "sample": [12.145431518554688, 5.915798187255859, 34.22987365722656, 5.218631744384766, 29.287940979003906, 38.111454010009766, 28.515682220458984, 33.261505126953125, 0.9127902984619141, 3.7470741271972656, -1.7264690399169922, 2.1256561279296875, 36.79109191894531, 16.527420043945312, 43.537784576416016, -6.525093078613281, 27.182720184326172, 8.415580749511719, 25.36758804321289, -1.5297775268554688, -0.7961215972900391, 10.562807083129883, 23.957988739013672, 2.3443603515625, 31.631973266601562, 1.2491722106933594, 17.384857177734375, -0.0645599365234375, 43.750709533691406, -32.89940643310547, 24.268260955810547, 23.211105346679688, 43.14923095703125, -13.735851287841797, 18.232593536376953, 34.52151870727539, -7.483268737792969, -14.936210632324219, 41.09379196166992, 32.71185302734375, -44.299171447753906, 4.435371398925781, 25.52374267578125, 3.90069580078125, 6.172401428222656, 38.207786560058594, 24.945236206054688, -0.02816009521484375, -3.4384689331054688, 17.0908203125, 8.097526550292969, 38.37693786621094, 33.74665832519531, 2.10693359375, -2.595306396484375, -8.299896240234375, 11.962020874023438, -0.38696861267089844, -2.995136260986328, -32.766876220703125, 18.793121337890625, -0.0837554931640625, 2.246307373046875, -0.7013320922851562], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000171.npy"}
|
||||
{"epoch": 0.2585034013605442, "step": 172, "batch_size": 64, "mean": 11.191009521484375, "std": 17.726346969604492, "min": -32.759498596191406, "p10": -8.120796966552733, "median": 9.610294342041016, "p90": 36.946923446655276, "max": 55.0196533203125, "pos_frac": 0.78125, "sample": [-13.865623474121094, 8.015037536621094, 1.5670013427734375, 9.367996215820312, 0.541900634765625, 0.9917945861816406, -0.27857398986816406, 18.172557830810547, 26.57292938232422, 41.26560974121094, 55.0196533203125, 11.761993408203125, 12.378829956054688, -0.4704322814941406, 22.629852294921875, 46.164268493652344, 15.93341064453125, -15.650115966796875, 1.721771240234375, 9.876106262207031, 10.891830444335938, 2.69317626953125, -23.31340217590332, 2.4903182983398438, 12.936531066894531, 7.991691589355469, 2.8187408447265625, 22.670562744140625, 24.442039489746094, -4.50933837890625, 19.163349151611328, 26.67748260498047, 10.826461791992188, 22.60455322265625, 1.2767868041992188, 3.2045116424560547, 37.13383865356445, 23.506690979003906, 38.26551818847656, 2.5097808837890625, -1.9333629608154297, 12.042259216308594, 4.476104736328125, 25.35291862487793, 26.859996795654297, -24.377822875976562, -0.43064117431640625, -10.757362365722656, 1.828887939453125, 9.86025619506836, -7.459510803222656, 36.20758056640625, 40.84935760498047, -32.759498596191406, 1.4850807189941406, 23.852432250976562, 1.894378662109375, 9.852592468261719, 1.9358272552490234, 36.51078796386719, -8.404205322265625, -0.9737148284912109, 39.93901824951172, 34.376129150390625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000172.npy"}
|
||||
{"epoch": 0.2600151171579743, "step": 173, "batch_size": 64, "mean": 12.120534896850586, "std": 21.68239402770996, "min": -35.61724853515625, "p10": -12.091279983520506, "median": 12.316081047058105, "p90": 42.384620666503906, "max": 55.609832763671875, "pos_frac": 0.671875, "sample": [-15.703420639038086, -1.9434890747070312, -2.0673675537109375, 17.682567596435547, 29.40258026123047, 33.26289367675781, 9.77314567565918, -4.0742034912109375, -4.520984649658203, 24.64482879638672, 3.2695999145507812, 19.1650390625, 40.34864044189453, -10.131446838378906, -6.018102645874023, 36.86817932128906, 49.12077331542969, 5.448829650878906, 22.622285842895508, 1.3757133483886719, 18.341094970703125, 43.348594665527344, -6.240375518798828, 33.649444580078125, -35.61724853515625, -33.43281555175781, 16.14550018310547, 29.053375244140625, 5.155055999755859, 0.1904296875, -13.430068969726562, 44.18562316894531, -12.717180252075195, 41.82225799560547, 15.139877319335938, 3.034036636352539, 32.93545913696289, -5.679912567138672, 20.489105224609375, 22.999496459960938, 14.859016418457031, 1.2623825073242188, 4.407068252563477, 42.625633239746094, 20.21207046508789, -34.16582489013672, -0.8013267517089844, -8.868980407714844, 55.609832763671875, 28.66907501220703, 44.30253601074219, -26.77335548400879, 15.829925537109375, 33.53968048095703, 29.702693939208984, -3.2257308959960938, -0.9500274658203125, -10.63084602355957, 18.877004623413086, 51.328155517578125, 3.5276336669921875, 28.321128845214844, -1.354217529296875, 1.512908935546875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000173.npy"}
|
||||
{"epoch": 0.2615268329554044, "step": 174, "batch_size": 64, "mean": 13.36252212524414, "std": 18.862363815307617, "min": -36.43104553222656, "p10": -4.370594787597656, "median": 11.59916877746582, "p90": 38.33910446166992, "max": 47.956695556640625, "pos_frac": 0.78125, "sample": [14.290786743164062, 16.044265747070312, 23.916610717773438, 5.47291374206543, 13.603799819946289, -4.455596923828125, 4.406257629394531, 27.654693603515625, 7.484931945800781, 13.889713287353516, 39.993385314941406, 37.98710632324219, 5.265235900878906, -4.1722564697265625, 21.8675537109375, -1.7414512634277344, 34.57145690917969, 31.324939727783203, 28.286293029785156, -36.43104553222656, 40.68351745605469, 9.690696716308594, -2.999401092529297, 28.29610824584961, 46.89780044555664, -1.2024974822998047, 14.076751708984375, -3.870880126953125, 45.020294189453125, 11.3638916015625, 3.374725341796875, 47.956695556640625, 44.23454284667969, -4.132097244262695, 25.264202117919922, 16.550048828125, 11.155059814453125, 7.8979949951171875, 2.870311737060547, 37.24800109863281, -17.75334930419922, 38.248779296875, 11.83444595336914, 1.5451126098632812, 0.2765350341796875, 35.57078552246094, 28.84516143798828, -18.239845275878906, 38.37781524658203, 15.49183464050293, -29.168975830078125, 24.714279174804688, 5.1667938232421875, 1.0049400329589844, 26.71269989013672, -2.8463516235351562, 17.15929412841797, -13.27691650390625, 28.46933364868164, -19.552322387695312, 7.752195358276367, 5.9064178466796875, 9.078758239746094, 0.2486572265625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000174.npy"}
|
||||
{"epoch": 0.26303854875283444, "step": 175, "batch_size": 64, "mean": 14.801218032836914, "std": 20.94384002685547, "min": -26.361610412597656, "p10": -9.957416534423828, "median": 10.310089111328125, "p90": 41.738701629638676, "max": 62.020263671875, "pos_frac": 0.71875, "sample": [-24.623327255249023, 25.250776290893555, -3.8733978271484375, 4.825700759887695, -13.883598327636719, 27.856647491455078, 0.1476898193359375, 23.961395263671875, 10.088150024414062, -4.040771484375, 3.4473495483398438, 42.85505676269531, 22.936542510986328, 32.323760986328125, -12.554458618164062, 39.74859619140625, 14.98309326171875, 35.08898162841797, 40.29443359375, 3.8049850463867188, 6.1902008056640625, 5.574550628662109, 5.8340911865234375, 42.21830749511719, 32.15870666503906, 15.594863891601562, 0.07505416870117188, -3.102386474609375, 40.30155944824219, -20.734283447265625, 39.29429626464844, -3.1158905029296875, 17.265098571777344, -0.24069976806640625, 12.220596313476562, -1.3808937072753906, 22.31420135498047, 44.46790313720703, -2.581960678100586, -26.361610412597656, 13.247726440429688, -10.282989501953125, 1.2410354614257812, 7.388179779052734, -7.4343414306640625, -1.1784210205078125, 40.15593719482422, -9.197746276855469, 37.75129699707031, 40.257164001464844, 9.996706008911133, 37.81901550292969, 10.532028198242188, -11.502281188964844, 5.622611999511719, 40.61962127685547, -8.002323150634766, 48.04559326171875, 33.103782653808594, 62.020263671875, 2.052642822265625, 46.68775939941406, 18.191665649414062, 45.513729095458984], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000175.npy"}
|
||||
{"epoch": 0.26455026455026454, "step": 176, "batch_size": 64, "mean": 9.531242370605469, "std": 21.78780174255371, "min": -45.85345458984375, "p10": -14.1731575012207, "median": 6.279598236083984, "p90": 38.87358131408693, "max": 58.50146484375, "pos_frac": 0.65625, "sample": [47.34808349609375, -35.95580291748047, 22.551429748535156, 22.14105224609375, 0.6700592041015625, 6.6847381591796875, -6.031349182128906, 21.171234130859375, 6.157861709594727, 2.4272613525390625, -31.145950317382812, -15.464668273925781, -8.410308837890625, 8.39913558959961, -28.63459014892578, 41.601783752441406, 0.6657524108886719, 27.121829986572266, -3.3354644775390625, -2.4572906494140625, -2.879852294921875, -2.04620361328125, 6.412525177001953, 48.336055755615234, 28.298126220703125, 23.80303955078125, -11.868629455566406, 33.12415313720703, 4.425407409667969, 33.21356201171875, 9.079605102539062, -0.14538002014160156, -2.398284912109375, 35.30438232421875, 24.757795333862305, 45.078250885009766, 2.10675048828125, 23.186614990234375, -2.4895477294921875, -15.160812377929688, 1.4347152709960938, 15.486572265625, 21.708070755004883, 40.23032760620117, 12.99346923828125, 10.304882049560547, -6.88470458984375, -4.591560363769531, 25.535446166992188, 35.70783996582031, -2.11236572265625, 30.380462646484375, 1.984954833984375, 20.668846130371094, 2.8024215698242188, 6.401334762573242, 58.50146484375, -1.0754623413085938, -4.43731689453125, 53.52195739746094, 5.790435791015625, -45.85345458984375, 8.093879699707031, -32.235076904296875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000176.npy"}
|
||||
{"epoch": 0.2660619803476946, "step": 177, "batch_size": 64, "mean": 18.600608825683594, "std": 19.54615592956543, "min": -34.127288818359375, "p10": -3.3713745117187495, "median": 16.33118438720703, "p90": 46.79258117675782, "max": 54.26659393310547, "pos_frac": 0.84375, "sample": [5.1728668212890625, 7.309490203857422, 30.266427993774414, 54.26659393310547, 24.913063049316406, -1.11224365234375, 16.301841735839844, 12.802902221679688, 53.36131286621094, 22.867401123046875, 51.22138977050781, 43.19282150268555, 25.056182861328125, -1.5047683715820312, 20.51443099975586, -13.577474594116211, 45.49847412109375, 6.306262969970703, -4.661958694458008, 12.023412704467773, 17.157859802246094, 16.36052703857422, 4.755012512207031, 41.74109649658203, 18.55075454711914, 1.07647705078125, 9.391141891479492, 38.19971466064453, -6.073759078979492, 2.296846389770508, 16.855865478515625, 11.033462524414062, 41.7574462890625, 12.7406005859375, 9.500043869018555, 42.888851165771484, 40.977394104003906, 1.8372726440429688, 12.482025146484375, 34.46052551269531, 2.9920578002929688, 17.992902755737305, 34.27482223510742, 12.546401977539062, 32.0921630859375, -11.611373901367188, -3.64727783203125, -6.7210845947265625, 47.95174026489258, 1.0970306396484375, 47.347198486328125, 1.3124752044677734, -34.127288818359375, 50.63872146606445, 13.9935302734375, 18.06609344482422, 25.593050003051758, 53.32289123535156, 33.18170166015625, 19.50651741027832, 6.962333679199219, 43.704795837402344, -2.72760009765625, 8.489585876464844], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000177.npy"}
|
||||
{"epoch": 0.2675736961451247, "step": 178, "batch_size": 64, "mean": 11.341386795043945, "std": 18.58299446105957, "min": -27.256553649902344, "p10": -11.359383392333983, "median": 9.64388656616211, "p90": 37.365438842773436, "max": 49.93864440917969, "pos_frac": 0.75, "sample": [13.99833869934082, 29.89508056640625, 7.908664703369141, -2.0814743041992188, 30.083457946777344, 0.010023117065429688, -1.1864013671875, 22.234134674072266, 33.59336853027344, 10.529537200927734, -11.625190734863281, 6.2200927734375, 18.873260498046875, -22.04665184020996, 11.731651306152344, 5.128669738769531, 44.98827362060547, -0.2131214141845703, -7.980625152587891, -3.9410476684570312, 40.64504623413086, 9.102876663208008, 12.822273254394531, -26.307586669921875, 2.2797393798828125, 8.399154663085938, 2.9969024658203125, 11.48101806640625, 3.07275390625, 7.723123550415039, 27.054214477539062, 0.018093109130859375, 9.711662292480469, 49.93864440917969, 10.749313354492188, 9.57611083984375, 14.489959716796875, 5.778850555419922, 43.567161560058594, 10.967266082763672, -16.569656372070312, 29.467262268066406, 36.883056640625, 3.6976661682128906, -24.386756896972656, 42.890602111816406, -3.4817352294921875, -20.709083557128906, 40.46277618408203, 2.8854293823242188, 37.572174072265625, 34.790740966796875, 14.363916397094727, 30.930755615234375, 12.938863754272461, 3.9489974975585938, 25.702789306640625, -10.739166259765625, -27.256553649902344, -1.2839202880859375, 20.966716766357422, 29.596214294433594, 27.000404357910156, -4.009376525878906], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000178.npy"}
|
||||
{"epoch": 0.2690854119425548, "step": 179, "batch_size": 64, "mean": 14.473957061767578, "std": 22.50478172302246, "min": -43.13660430908203, "p10": -15.978480720520013, "median": 15.216552734375, "p90": 43.0000228881836, "max": 66.67460632324219, "pos_frac": 0.765625, "sample": [23.230560302734375, -5.111785888671875, -3.59967041015625, 15.227493286132812, 0.7064552307128906, 21.281158447265625, 14.681352615356445, 26.754356384277344, 49.062774658203125, 21.213947296142578, 43.422157287597656, 37.04649353027344, 39.38406753540039, -28.01512908935547, 8.431894302368164, -22.406143188476562, -43.13660430908203, -18.722333908081055, 3.7527236938476562, 16.865936279296875, -2.8995933532714844, 31.59311294555664, 15.842941284179688, 35.61376190185547, 42.082244873046875, -33.67315673828125, -3.3380889892578125, 37.493675231933594, 11.27816390991211, 19.225107192993164, 19.047447204589844, 29.48259735107422, 7.031970977783203, -4.9563751220703125, 1.4054069519042969, -9.576156616210938, 15.205612182617188, 64.35884094238281, 43.39335632324219, 5.573829650878906, 22.64640235900879, 32.852516174316406, 28.220888137817383, 14.693470001220703, -20.736717224121094, 22.8206787109375, 53.72198486328125, 5.9743499755859375, 46.842628479003906, 66.67460632324219, 31.710758209228516, 25.763599395751953, 30.476333618164062, 4.2969207763671875, 3.820911407470703, 18.788253784179688, 13.047187805175781, 0.4774322509765625, 7.171062469482422, -1.3765106201171875, -19.882482528686523, 0.5567646026611328, 17.916015625, -4.39820671081543], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000179.npy"}
|
||||
{"epoch": 0.2705971277399849, "step": 180, "batch_size": 64, "mean": 11.48359489440918, "std": 21.292068481445312, "min": -42.96349334716797, "p10": -10.08348159790039, "median": 8.128623008728027, "p90": 44.252764892578135, "max": 53.32037353515625, "pos_frac": 0.71875, "sample": [18.05877685546875, 2.0972061157226562, 7.227790832519531, 6.246574401855469, 2.795257568359375, 12.794815063476562, -10.683799743652344, 15.662425994873047, 1.1782207489013672, -2.346487045288086, 45.805816650390625, 7.546112060546875, -8.6827392578125, -0.6786212921142578, 8.067108154296875, 1.395111083984375, -16.362380981445312, 42.42707824707031, -1.4591865539550781, -37.86940002441406, 49.361572265625, 19.286405563354492, 5.006132125854492, 41.48846435546875, 9.829164505004883, 28.24784278869629, 0.3324775695800781, -0.29059600830078125, 8.19013786315918, 21.4267578125, 13.654563903808594, 37.245094299316406, -20.484161376953125, 5.058013916015625, 11.43198013305664, 15.247390747070312, 8.82767105102539, 52.931129455566406, -2.172943115234375, -2.55780029296875, 4.930456161499023, 37.97871398925781, 45.03520202636719, -0.6788005828857422, -3.220602035522461, 1.5822563171386719, 50.80694580078125, -5.40509033203125, 53.32037353515625, 9.735071182250977, 36.95726776123047, -19.06536102294922, 8.476089477539062, -6.457183837890625, 19.252479553222656, 10.172954559326172, 48.91419219970703, -42.96349334716797, 30.010631561279297, -21.33934783935547, 22.509307861328125, 14.807170867919922, 41.75543975830078, 2.556396484375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000180.npy"}
|
||||
{"epoch": 0.272108843537415, "step": 181, "batch_size": 64, "mean": 16.700328826904297, "std": 19.598102569580078, "min": -21.17767333984375, "p10": -1.803005218505859, "median": 10.684929847717285, "p90": 46.56635475158693, "max": 60.40382385253906, "pos_frac": 0.84375, "sample": [24.051834106445312, 15.241737365722656, -0.2474212646484375, -0.0391082763671875, 6.036746978759766, 0.5844097137451172, 13.038520812988281, -9.615936279296875, 3.5889434814453125, 1.2729301452636719, 35.95661926269531, 6.339023590087891, 20.469802856445312, -15.722793579101562, 19.826045989990234, 5.8033599853515625, 36.46686553955078, 3.596038818359375, 38.054771423339844, -11.573150634765625, 28.93505859375, 47.92259979248047, 28.118907928466797, 23.797958374023438, 53.504966735839844, 26.08062744140625, 8.091384887695312, 60.40382385253906, 0.06732940673828125, -21.17767333984375, 30.930389404296875, 3.4543304443359375, 57.60015869140625, -6.290012359619141, 28.958465576171875, 0.8261566162109375, -1.4102325439453125, 54.48567199707031, 4.397945404052734, 17.80200958251953, 39.12773513793945, 4.805702209472656, 40.017913818359375, 47.76945495605469, 38.158382415771484, 6.005830764770508, 8.494588851928711, 11.471918106079102, 11.506523132324219, 1.7831249237060547, 3.0707855224609375, 43.75912094116211, 57.64363098144531, 1.4004135131835938, 9.761749267578125, 31.60964584350586, 4.149406433105469, 10.568603515625, 10.80125617980957, -1.9713363647460938, 20.412643432617188, 0.03722572326660156, 31.631248474121094, -2.8236236572265625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000181.npy"}
|
||||
{"epoch": 0.273620559334845, "step": 182, "batch_size": 64, "mean": 17.048091888427734, "std": 24.17027473449707, "min": -48.76819610595703, "p10": -11.569139862060544, "median": 12.045967102050781, "p90": 49.429543304443364, "max": 58.19331359863281, "pos_frac": 0.765625, "sample": [38.82392883300781, -19.09068489074707, -15.188331604003906, 22.14574432373047, 24.531593322753906, 25.036964416503906, 24.217487335205078, -17.989700317382812, -8.2645263671875, 37.93479919433594, 21.779216766357422, 3.561126708984375, 58.19331359863281, 7.1956939697265625, 47.96680450439453, 25.706085205078125, 8.990058898925781, 10.54193115234375, 44.26474380493164, 18.381141662597656, 32.87981414794922, 0.5267906188964844, 1.2310638427734375, -12.594528198242188, 32.97643280029297, 0.8867950439453125, 40.216453552246094, 7.263580322265625, 52.608673095703125, 47.754417419433594, 57.3817138671875, 16.811233520507812, 7.5247344970703125, 8.076580047607422, 53.59833526611328, 1.3219451904296875, -1.03265380859375, 5.581672668457031, 49.751312255859375, -0.8100757598876953, 15.726936340332031, -29.737085342407227, 56.324729919433594, -2.4573516845703125, 44.11700439453125, 37.89015197753906, 4.832069396972656, -9.088829040527344, 40.00458526611328, 4.875587463378906, 36.18141555786133, 48.678749084472656, -9.176567077636719, 35.72496795654297, 51.95856475830078, 5.934661865234375, -5.2873077392578125, 9.129278182983398, 45.31305694580078, 2.748992919921875, 13.550003051757812, -17.78954315185547, -0.29970741271972656, -48.76819610595703], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000182.npy"}
|
||||
{"epoch": 0.2751322751322751, "step": 183, "batch_size": 64, "mean": 7.181671619415283, "std": 20.981002807617188, "min": -50.84894561767578, "p10": -13.38967056274414, "median": 5.014842987060547, "p90": 36.94911384582522, "max": 49.964012145996094, "pos_frac": 0.640625, "sample": [9.556167602539062, 46.93336486816406, -2.7213668823242188, -1.1101646423339844, 9.520149230957031, -5.8689422607421875, -32.18800354003906, 43.464744567871094, -12.218070983886719, -5.272430419921875, -50.84894561767578, -2.030059814453125, 29.18684196472168, -7.0120086669921875, 15.527551651000977, 12.308097839355469, -1.1483173370361328, 23.422279357910156, 32.16856384277344, 17.292943954467773, 24.608688354492188, -5.1902923583984375, 26.17723846435547, -6.090707778930664, 3.8887710571289062, -5.033393859863281, 14.330108642578125, 0.76629638671875, -33.353485107421875, -13.89178466796875, 49.964012145996094, 12.044448852539062, 5.737373352050781, 4.232330322265625, -11.979393005371094, 9.48675537109375, 12.419143676757812, 46.10590362548828, -49.67588806152344, 38.997920989990234, -5.476413726806641, -1.219482421875, 45.306793212890625, -8.024589538574219, 2.29681396484375, 7.522037506103516, 14.223077774047852, 3.1927051544189453, -18.98339080810547, -1.015493392944336, 19.53050994873047, 11.202499389648438, 2.0425243377685547, 2.2699966430664062, 31.5216064453125, 0.3061504364013672, 25.99871063232422, -19.110107421875, 5.864418029785156, 4.2923126220703125, 12.290611267089844, 39.91481018066406, 23.93813133239746, 19.236297607421875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000183.npy"}
|
||||
{"epoch": 0.2766439909297052, "step": 184, "batch_size": 64, "mean": 12.999320983886719, "std": 19.21928596496582, "min": -19.878353118896484, "p10": -3.506578826904297, "median": 5.010625839233398, "p90": 44.7578338623047, "max": 54.87085723876953, "pos_frac": 0.65625, "sample": [-0.9006500244140625, 42.31145477294922, 45.80628204345703, 0.14155006408691406, 17.099891662597656, 1.55877685546875, -0.7907752990722656, -3.5264434814453125, -1.4858474731445312, -0.8517913818359375, -0.12882423400878906, 46.15544128417969, -0.6939926147460938, 12.416120529174805, 39.28410339355469, -5.23797607421875, 22.630783081054688, 13.723442077636719, 11.924663543701172, 46.87840270996094, -19.02161407470703, 54.87085723876953, 37.22557830810547, 48.727149963378906, -19.23468780517578, 45.94596862792969, 14.80534553527832, 15.845108032226562, 37.381404876708984, 2.8726425170898438, 35.52703094482422, -0.7847671508789062, 5.2908477783203125, -19.878353118896484, -0.5899581909179688, 16.76666259765625, 52.36503601074219, 16.95430564880371, -1.0086822509765625, 36.28472137451172, 19.54421043395996, 1.0388565063476562, 22.75078582763672, -0.237518310546875, 17.525840759277344, 4.041097640991211, 29.955322265625, 13.790931701660156, 0.2594470977783203, 41.26261901855469, -18.323713302612305, -3.305633544921875, 18.10692596435547, 4.730403900146484, 23.661544799804688, 13.87353515625, -5.313730239868164, 1.8464126586914062, -1.5204524993896484, 4.092063903808594, -3.4602279663085938, 4.430027008056641, -0.4723491668701172, -2.9790573120117188], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000184.npy"}
|
||||
{"epoch": 0.2781557067271353, "step": 185, "batch_size": 64, "mean": 15.423196792602539, "std": 26.841257095336914, "min": -49.40876770019531, "p10": -16.945000839233398, "median": 15.792984962463379, "p90": 49.72411575317383, "max": 64.79251098632812, "pos_frac": 0.671875, "sample": [4.575706481933594, 55.523162841796875, -1.7936248779296875, 30.282861709594727, 59.29437255859375, -26.87057876586914, 13.720592498779297, 14.345897674560547, -9.415130615234375, 3.9731216430664062, 51.31214904785156, 49.69928741455078, -1.33685302734375, -23.814208984375, 59.85176086425781, 4.566904067993164, 64.79251098632812, 16.1390380859375, 45.92215347290039, 50.045650482177734, -5.243106842041016, 45.09138488769531, 35.356529235839844, 11.258888244628906, 1.0667362213134766, 40.180477142333984, 24.675811767578125, 17.459636688232422, 10.012367248535156, 41.976806640625, -11.009927749633789, -3.0125885009765625, 22.45951271057129, 27.668121337890625, -44.2197265625, 34.444252014160156, 42.17083740234375, 34.01286315917969, 33.51321029663086, -3.2504959106445312, -49.40876770019531, 17.258522033691406, -5.911327362060547, 26.41278076171875, 45.87275695800781, -31.15240478515625, 45.76811981201172, -5.220819473266602, 49.73475646972656, -5.603157043457031, 24.588272094726562, -17.044832229614258, 27.7254638671875, -9.445999145507812, 0.2141265869140625, 19.82270622253418, -23.37702178955078, -16.712060928344727, 15.446931838989258, -4.842247009277344, -16.29184341430664, 5.454189300537109, 36.715232849121094, 41.654884338378906], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000185.npy"}
|
||||
{"epoch": 0.2796674225245654, "step": 186, "batch_size": 64, "mean": 16.857139587402344, "std": 26.33567237854004, "min": -57.804039001464844, "p10": -8.797193717956542, "median": 14.51089096069336, "p90": 50.908354187011724, "max": 75.94732666015625, "pos_frac": 0.734375, "sample": [20.421875, 60.99720764160156, -26.875030517578125, 16.164134979248047, 36.43096160888672, -18.108200073242188, 8.664596557617188, -25.344573974609375, -6.094749450683594, 43.27759552001953, 20.329071044921875, -4.485141754150391, 26.168045043945312, 21.39046859741211, 68.35940551757812, 35.60942459106445, 4.970773696899414, 6.294406890869141, 13.032402038574219, -38.52458190917969, 8.783710479736328, 44.311729431152344, 41.76042175292969, 58.505210876464844, -57.804039001464844, -4.458118438720703, -5.598701477050781, -5.846855163574219, -4.6535491943359375, 0.4080467224121094, 5.459056854248047, 22.848379135131836, 15.9893798828125, -5.0381927490234375, -8.780206680297852, -3.037677764892578, 22.67548370361328, 44.555747985839844, 36.404598236083984, 0.3592376708984375, -6.098659515380859, -12.19390869140625, 11.043350219726562, 9.81097412109375, 28.931381225585938, 6.916515350341797, 51.7611083984375, 30.790374755859375, -8.804473876953125, 26.876243591308594, 67.60050964355469, 7.2662353515625, 19.35521697998047, 75.94732666015625, 37.78766632080078, 23.683792114257812, 59.46122741699219, 48.91859436035156, 41.48823547363281, 32.258460998535156, 11.134719848632812, 33.312835693359375, 0.2000274658203125, 11.887453079223633], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000186.npy"}
|
||||
{"epoch": 0.2811791383219955, "step": 187, "batch_size": 64, "mean": 14.735928535461426, "std": 27.35327911376953, "min": -47.198081970214844, "p10": -18.19419250488281, "median": 11.188514709472656, "p90": 52.636834716796876, "max": 71.61447143554688, "pos_frac": 0.734375, "sample": [14.210960388183594, -17.0791015625, 47.46461486816406, 61.475425720214844, 71.61447143554688, 20.66828155517578, 26.231353759765625, 15.587387084960938, 9.486801147460938, 52.46814727783203, -47.198081970214844, 15.182113647460938, -0.9286613464355469, 15.915725708007812, 17.446517944335938, 1.4040355682373047, 1.227609634399414, 32.217315673828125, 39.75407409667969, 4.203193664550781, -18.672088623046875, -37.04204559326172, 52.19438934326172, -2.7620277404785156, 3.6034164428710938, -12.218826293945312, 1.964935302734375, 2.629486083984375, 16.940216064453125, 54.22106170654297, 45.716575622558594, 51.27326202392578, -13.262611389160156, -2.054462432861328, 22.485198974609375, 16.776885986328125, 5.773223876953125, 2.6486740112304688, -23.744056701660156, 27.969215393066406, -4.1254425048828125, -20.020496368408203, 41.66050720214844, 0.2306041717529297, 5.01556396484375, 23.881103515625, 60.54847717285156, -11.851993560791016, 12.890228271484375, 31.9559326171875, 46.681251525878906, -30.64417266845703, 35.058860778808594, 68.24044799804688, 3.658018112182617, 5.9289703369140625, 65.8682861328125, 52.709129333496094, 1.1530876159667969, 7.961677551269531, -8.607124328613281, -27.499629974365234, -12.75927734375, 19.372772216796875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000187.npy"}
|
||||
{"epoch": 0.28269085411942557, "step": 188, "batch_size": 64, "mean": 17.42372703552246, "std": 25.96239471435547, "min": -44.98365783691406, "p10": -6.755057334899901, "median": 8.132278442382812, "p90": 55.101623153686525, "max": 72.98715209960938, "pos_frac": 0.796875, "sample": [6.19659423828125, -4.229623794555664, 58.64883041381836, -12.928693771362305, -1.8752517700195312, 55.4208984375, 44.64582824707031, 5.209318161010742, 48.23028564453125, 26.271408081054688, 3.138561248779297, 72.98715209960938, 54.35664749145508, 1.8553466796875, -0.5046653747558594, 2.7481136322021484, 35.14567565917969, 57.57311248779297, 0.8624591827392578, -5.123010635375977, 20.872478485107422, 5.8924407958984375, 0.9192733764648438, 3.2839584350585938, 12.551078796386719, -3.7149276733398438, -14.361560821533203, 41.7705078125, 17.63492202758789, 46.3564338684082, 14.822593688964844, 13.569419860839844, 5.713653564453125, 72.18828582763672, 46.05101776123047, 1.6190776824951172, 53.52232360839844, 9.098648071289062, 17.262836456298828, 12.237815856933594, 58.49973678588867, 1.085968017578125, 4.396617889404297, 24.358238220214844, 1.1728591918945312, -0.420166015625, 1.4357738494873047, 49.924705505371094, 5.412605285644531, 1.0354423522949219, 50.17256164550781, -31.33819580078125, -7.454505920410156, 7.913604736328125, 22.268096923828125, 8.3509521484375, 7.138288497924805, -27.17848777770996, 39.952354431152344, 52.62186813354492, 59.061912536621094, 14.164390563964844, -8.391687393188477, -44.98365783691406], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000188.npy"}
|
||||
{"epoch": 0.2842025699168556, "step": 189, "batch_size": 64, "mean": 15.455196380615234, "std": 25.427148818969727, "min": -47.87300491333008, "p10": -16.07439384460449, "median": 10.903085708618164, "p90": 52.078271102905276, "max": 74.59102630615234, "pos_frac": 0.796875, "sample": [-3.80352783203125, 19.289447784423828, 10.657051086425781, 32.968048095703125, 18.119049072265625, 59.425689697265625, 58.5777587890625, 16.503629684448242, 2.997762680053711, 17.190261840820312, -29.741958618164062, 58.74900436401367, -20.55872344970703, 5.668739318847656, -1.7827301025390625, -2.785146713256836, 58.213966369628906, 36.78097915649414, -12.601398468017578, 22.01642608642578, 10.814960479736328, 24.340600967407227, 7.5553741455078125, 14.4853515625, 0.7254791259765625, 2.0947513580322266, 33.60535430908203, 31.551921844482422, 4.739410400390625, 7.105033874511719, 51.129852294921875, -17.562820434570312, 5.045356750488281, 51.231231689453125, 50.94915008544922, 16.096397399902344, 0.63916015625, 7.248992919921875, 10.460039138793945, 17.828411102294922, 52.441287994384766, 20.24298095703125, 38.211524963378906, 20.370502471923828, 21.76287841796875, 5.340869903564453, -7.80303955078125, -32.31798553466797, 10.9912109375, 5.6188201904296875, -2.24517822265625, 21.61134910583496, -29.943695068359375, 74.59102630615234, 45.70951843261719, 7.0053558349609375, -26.289169311523438, -47.87300491333008, 11.91107177734375, 7.436737060546875, 9.018943786621094, 2.2265472412109375, 64.33878326416016, 40.806922912597656], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000189.npy"}
|
||||
{"epoch": 0.2857142857142857, "step": 190, "batch_size": 64, "mean": 21.888038635253906, "std": 27.24253273010254, "min": -35.79725646972656, "p10": -10.736612319946289, "median": 18.11964511871338, "p90": 61.98346786499024, "max": 74.69822692871094, "pos_frac": 0.796875, "sample": [8.675613403320312, 45.258514404296875, 73.20828247070312, -7.6094970703125, 29.198909759521484, 50.81907653808594, 3.0612030029296875, 2.7779998779296875, -13.247634887695312, 62.798248291015625, 20.112892150878906, 48.80375671386719, 12.653636932373047, 2.019359588623047, 20.580657958984375, 4.079193115234375, -10.949352264404297, 63.0145263671875, 16.5609188079834, 7.603506088256836, 48.66987991333008, 64.43067932128906, -1.9617919921875, 5.43231201171875, -3.5766735076904297, 24.47956085205078, -35.79725646972656, 10.203529357910156, 6.134681701660156, 31.95944595336914, 19.67837142944336, -10.240219116210938, 11.822372436523438, 30.626678466796875, 43.122337341308594, 2.8046493530273438, 50.62037658691406, -16.00401496887207, -2.832744598388672, 32.279842376708984, -22.835777282714844, 30.176315307617188, 51.85015106201172, 74.69822692871094, 9.852781295776367, 3.5174407958984375, -1.148977279663086, 36.36968994140625, -25.55413818359375, -12.575210571289062, 28.62529182434082, 15.079277038574219, 4.70060920715332, 59.85081481933594, 19.909881591796875, 66.0406723022461, 73.77938842773438, 1.7006607055664062, 42.23143005371094, 60.082313537597656, 2.1551895141601562, 56.60028076171875, 45.718170166015625, 28.738079071044922], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000190.npy"}
|
||||
{"epoch": 0.2872260015117158, "step": 191, "batch_size": 64, "mean": 17.132522583007812, "std": 28.50922393798828, "min": -57.44776916503906, "p10": -9.159723663330077, "median": 16.908313751220703, "p90": 57.379807281494145, "max": 71.63683319091797, "pos_frac": 0.75, "sample": [52.514259338378906, -3.7137889862060547, 21.091506958007812, 39.35699462890625, 67.06759643554688, -4.1635589599609375, 17.250350952148438, 65.21224212646484, 19.57855796813965, 7.200019836425781, -0.0727386474609375, 71.63683319091797, 27.624961853027344, -9.486988067626953, 13.290573120117188, 2.899850845336914, -57.44776916503906, 8.732208251953125, 55.58685302734375, 28.928739547729492, 16.56627655029297, 31.563140869140625, 19.388256072998047, 58.148216247558594, -36.57910919189453, 49.04588317871094, 31.86151885986328, 22.846508026123047, -5.550537109375, -0.3487262725830078, 20.346195220947266, -2.4767608642578125, 0.09193992614746094, 3.3143882751464844, 59.587738037109375, -42.10887145996094, -0.4848899841308594, -33.55583190917969, 9.36099624633789, 48.6885986328125, 28.766963958740234, 18.38437271118164, 12.099502563476562, 21.825546264648438, 44.96569061279297, 2.7654972076416016, 48.73029327392578, 60.20827102661133, -8.396106719970703, 4.4685516357421875, -4.788545608520508, 11.539617538452148, 60.12176513671875, 17.764076232910156, -9.538410186767578, 28.468109130859375, 51.843177795410156, -54.21327209472656, 35.16050720214844, 6.702430725097656, 2.9851531982421875, 39.51908874511719, 1.3292465209960938, 2.9783248901367188], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000191.npy"}
|
||||
{"epoch": 0.2887377173091459, "step": 192, "batch_size": 64, "mean": 10.817346572875977, "std": 24.290788650512695, "min": -59.62793731689453, "p10": -11.518802642822266, "median": 5.867954254150391, "p90": 45.8226848602295, "max": 64.60953521728516, "pos_frac": 0.6875, "sample": [-59.62793731689453, -24.500404357910156, 3.6256675720214844, 30.16341781616211, 29.437068939208984, 4.750494003295898, -4.686031341552734, 19.061172485351562, 60.757667541503906, -30.211700439453125, -11.869747161865234, -9.565574645996094, 0.35517120361328125, 1.9830093383789062, 6.3116912841796875, 9.916427612304688, 43.40061569213867, 2.5151824951171875, -8.03110122680664, 5.424217224121094, 47.54802703857422, 35.617897033691406, -3.5900421142578125, -4.8735198974609375, 52.137855529785156, -0.7592334747314453, 22.783721923828125, 44.559139251708984, 2.4964447021484375, 4.32159423828125, 13.19512939453125, 46.36420440673828, 42.91072082519531, 40.137535095214844, -5.019464492797852, -10.680187225341797, 8.198188781738281, -2.3120784759521484, -14.223396301269531, 9.018653869628906, 9.804798126220703, -10.699932098388672, -35.39695358276367, 64.60953521728516, 24.241119384765625, 7.055809020996094, -2.6846542358398438, 1.7173347473144531, 16.62206268310547, -29.17066192626953, 11.84454345703125, 51.249366760253906, 3.7304153442382812, 19.400062561035156, 0.26925086975097656, 33.228660583496094, 17.14822006225586, -3.467071533203125, 36.052772521972656, 49.296470642089844, 31.148120880126953, -10.563882827758789, 2.738861083984375, 7.095403671264648], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000192.npy"}
|
||||
{"epoch": 0.29024943310657597, "step": 193, "batch_size": 64, "mean": 20.507247924804688, "std": 26.68012237548828, "min": -53.168949127197266, "p10": -5.464558029174803, "median": 14.476173400878906, "p90": 57.10362129211426, "max": 71.68263244628906, "pos_frac": 0.796875, "sample": [0.7108001708984375, 11.18034553527832, 50.9559440612793, 12.837181091308594, 6.75114631652832, -18.840072631835938, 9.283531188964844, 48.26110076904297, -11.396697998046875, -6.260847091674805, 56.723236083984375, 3.250507354736328, 25.138710021972656, 45.40087890625, 53.32293701171875, -0.54376220703125, 30.952789306640625, -15.573318481445312, 43.189537048339844, 44.44209289550781, -1.7706031799316406, 52.14777374267578, 71.68263244628906, -9.293865203857422, 36.26592254638672, 60.46726989746094, 1.5088748931884766, 2.0871429443359375, -0.9909133911132812, 22.999544143676758, 48.41333770751953, 37.282325744628906, 4.116306304931641, 16.168365478515625, 15.881973266601562, 23.202667236328125, -2.60211181640625, 8.583015441894531, 23.821022033691406, 15.6824951171875, -17.75836181640625, 6.142589569091797, 52.89388656616211, 69.81591796875, -3.6065502166748047, 4.4479217529296875, 27.292015075683594, -3.1115379333496094, 16.406005859375, 64.29480743408203, 44.76294708251953, 1.1553573608398438, 63.26158905029297, 65.3909683227539, 57.26664352416992, 0.5637283325195312, 25.822677612304688, -53.168949127197266, 13.269851684570312, 56.379783630371094, 0.17038726806640625, 3.6748485565185547, 1.4130363464355469, 0.24508094787597656], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000193.npy"}
|
||||
{"epoch": 0.29176114890400606, "step": 194, "batch_size": 64, "mean": 16.74282455444336, "std": 32.4864387512207, "min": -61.509376525878906, "p10": -15.465280914306641, "median": 14.296509742736816, "p90": 62.186615753173825, "max": 77.56195831298828, "pos_frac": 0.734375, "sample": [72.7546157836914, 2.073089599609375, 22.06499481201172, -26.544639587402344, 56.273475646972656, 12.489421844482422, -47.88566589355469, 2.0711708068847656, 8.855384826660156, 18.1484375, 8.482078552246094, 28.423568725585938, 6.609031677246094, 56.87782287597656, -15.318153381347656, -9.414108276367188, 28.21532440185547, -13.085700988769531, 25.285648345947266, -2.9091567993164062, 38.2984619140625, 4.709316253662109, 47.074974060058594, 49.54662322998047, -9.226505279541016, 23.60791778564453, -15.528335571289062, 42.81208801269531, -2.0024490356445312, 23.36766815185547, -7.2977142333984375, 62.069908142089844, -27.027618408203125, -9.61709976196289, -55.220741271972656, -2.435161590576172, 72.72317504882812, 28.338653564453125, 17.303417205810547, 64.30642700195312, -61.41352081298828, 15.901924133300781, 5.826093673706055, 3.319164276123047, 25.833763122558594, 69.37115478515625, 11.024581909179688, -61.509376525878906, 77.56195831298828, 39.2396240234375, 34.529293060302734, 2.84820556640625, 41.145042419433594, 61.8779296875, 62.23663330078125, 69.68694305419922, 13.082246780395508, 0.6971549987792969, 15.510772705078125, 43.254241943359375, 7.536836624145508, 1.9344596862792969, -4.712093353271484, 17.48807716369629], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000194.npy"}
|
||||
{"epoch": 0.29327286470143615, "step": 195, "batch_size": 64, "mean": 21.851619720458984, "std": 31.09200096130371, "min": -44.491729736328125, "p10": -14.75364303588867, "median": 20.22272491455078, "p90": 64.84676399230958, "max": 75.96662902832031, "pos_frac": 0.71875, "sample": [8.53082275390625, 17.504425048828125, 0.875030517578125, 15.247222900390625, 71.92848205566406, 54.942405700683594, -5.753517150878906, 20.712921142578125, 26.021835327148438, 54.79850769042969, 30.12676239013672, 53.98118591308594, -16.289390563964844, -34.72230529785156, -27.843170166015625, -43.08882141113281, -1.7223377227783203, 18.222530364990234, 27.556482315063477, 58.001426696777344, -13.714981079101562, 4.861114501953125, 27.948837280273438, 42.368255615234375, 35.000831604003906, 37.51702880859375, 67.25875091552734, 65.51467895507812, 17.81877899169922, 35.71217346191406, 7.15764045715332, -3.4263916015625, 63.53273391723633, 7.059761047363281, 65.54139709472656, 5.484718322753906, -11.572935104370117, 65.40991973876953, 42.78648376464844, 53.103668212890625, 20.715375900268555, -13.075538635253906, -1.391448974609375, -6.7927703857421875, -1.9336605072021484, 2.8415584564208984, -4.113555908203125, 63.0927734375, -44.491729736328125, 1.9056987762451172, -29.74677276611328, -2.5116729736328125, 75.96662902832031, 55.27029800415039, 45.60212707519531, 43.05410385131836, 50.98297119140625, 54.15852355957031, -15.198783874511719, 72.12333679199219, 31.741058349609375, 3.5218582153320312, 19.732528686523438, 32.657859802246094], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000195.npy"}
|
||||
{"epoch": 0.2947845804988662, "step": 196, "batch_size": 64, "mean": 17.221647262573242, "std": 26.857433319091797, "min": -71.29605102539062, "p10": -9.450908470153806, "median": 14.364524841308594, "p90": 56.412210845947264, "max": 78.52377319335938, "pos_frac": 0.75, "sample": [13.926376342773438, -71.29605102539062, -11.173322677612305, -2.2585296630859375, 22.78769874572754, -13.394819259643555, 2.9835739135742188, 8.168342590332031, 16.06800079345703, -6.645484924316406, 7.2655029296875, 58.35870361328125, 58.590789794921875, 23.147979736328125, 14.30194091796875, 16.439186096191406, -44.79057693481445, 56.598609924316406, 3.4537506103515625, 26.918182373046875, -4.04588508605957, 7.636707305908203, 62.73707962036133, -0.014972686767578125, -10.580020904541016, 8.889129638671875, 26.141387939453125, 14.995964050292969, 15.954025268554688, 78.52377319335938, 19.469608306884766, 37.08865737915039, 5.3042449951171875, -6.816312789916992, -5.561408996582031, 16.715118408203125, 8.32628059387207, 40.892364501953125, 15.974323272705078, 52.16175842285156, 26.83776092529297, 18.082992553710938, 55.97727966308594, 0.4587974548339844, 39.91050720214844, -13.451637268066406, -0.7357807159423828, 24.390594482421875, 41.24152374267578, 50.44158935546875, 57.935089111328125, 51.40870666503906, -2.6814308166503906, -13.291488647460938, 12.23751449584961, 47.170745849609375, 13.303613662719727, 30.10582733154297, 4.895904541015625, 6.8854827880859375, 4.481422424316406, -6.262813568115234, 14.427108764648438, 75.17442321777344], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000196.npy"}
|
||||
{"epoch": 0.2962962962962963, "step": 197, "batch_size": 64, "mean": 21.445934295654297, "std": 36.58386993408203, "min": -64.1834716796875, "p10": -29.92660827636718, "median": 16.443222045898438, "p90": 67.78229751586915, "max": 79.11541748046875, "pos_frac": 0.75, "sample": [-3.340404510498047, 72.93914031982422, 15.863079071044922, 5.717628479003906, 32.821258544921875, 51.443016052246094, -23.48434066772461, -58.78533935546875, 1.6753387451171875, -0.5269088745117188, 42.360511779785156, 69.37980651855469, -37.93938446044922, -40.052268981933594, 13.308086395263672, 72.91658020019531, 43.35686492919922, 4.482746124267578, 76.62935638427734, 67.27452087402344, 15.972923278808594, 19.556671142578125, 42.17387008666992, 49.68653869628906, 16.91352081298828, 1.3762359619140625, 12.007171630859375, -32.68758010864258, 62.56413269042969, 28.161041259765625, 67.99991607666016, 79.11541748046875, -53.111785888671875, 53.294822692871094, -4.925323486328125, -64.1834716796875, 5.912239074707031, 29.452667236328125, 35.94996643066406, 22.943450927734375, -0.48552703857421875, 64.9088363647461, 65.14059448242188, -3.981231689453125, -1.550262451171875, 64.60734558105469, 6.59283447265625, 1.9381752014160156, 11.565105438232422, 50.816986083984375, -21.210603713989258, -11.243797302246094, 51.552101135253906, 40.68490982055664, 8.190892219543457, -51.96794891357422, 76.17903900146484, 48.909446716308594, 4.784852981567383, 60.514644622802734, 5.6379852294921875, 42.91656494140625, 10.276899337768555, 53.55012512207031], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000197.npy"}
|
||||
{"epoch": 0.29780801209372637, "step": 198, "batch_size": 64, "mean": 22.841447830200195, "std": 29.94135093688965, "min": -47.74845886230469, "p10": -13.209060478210448, "median": 21.691421508789062, "p90": 61.65577964782715, "max": 78.21117401123047, "pos_frac": 0.78125, "sample": [-8.790283203125, 59.049110412597656, 0.10204505920410156, 48.82965850830078, 26.290111541748047, 16.447654724121094, -34.12023162841797, 4.207679748535156, 27.583402633666992, 10.822784423828125, 61.156219482421875, 16.367721557617188, 3.1873130798339844, -35.56549835205078, 78.21117401123047, 66.69310760498047, 40.071502685546875, 45.32159423828125, 30.78167724609375, -13.455781936645508, 46.598228454589844, 45.86802673339844, 66.8096923828125, -13.722471237182617, 63.28953552246094, -30.914077758789062, -40.584320068359375, -0.5380058288574219, 51.14891052246094, 55.70403289794922, 61.45539855957031, 32.42054748535156, 24.669723510742188, -47.74845886230469, 19.951858520507812, -2.923828125, 4.346782684326172, -1.1183547973632812, -12.633377075195312, 29.744888305664062, 7.8445587158203125, 57.670677185058594, 56.720035552978516, 20.12596893310547, 41.85236358642578, 14.897872924804688, -2.02728271484375, 66.86090850830078, 72.16869354248047, 17.324295043945312, 6.07037353515625, 0.8930816650390625, 51.78587341308594, 61.74165725708008, 0.4141845703125, 20.219589233398438, 23.163253784179688, 26.442012786865234, -1.5877552032470703, 20.21865463256836, 26.961883544921875, 11.70315933227539, 23.665817260742188, 41.70707702636719], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000198.npy"}
|
||||
{"epoch": 0.29931972789115646, "step": 199, "batch_size": 64, "mean": 18.034997940063477, "std": 35.01445007324219, "min": -68.80093383789062, "p10": -21.47590923309326, "median": 20.93012809753418, "p90": 59.47180404663086, "max": 85.0091552734375, "pos_frac": 0.6875, "sample": [23.659465789794922, 51.95927429199219, 7.617271423339844, 21.957172393798828, -20.430978775024414, -53.30393981933594, 10.934806823730469, 61.61131286621094, 41.114906311035156, 14.787933349609375, 37.02569580078125, -7.955600738525391, 85.0091552734375, 49.751243591308594, 76.69830322265625, 54.91456604003906, 29.235767364501953, 58.253013610839844, 44.478851318359375, 58.934696197509766, -12.035125732421875, -34.054527282714844, 31.387561798095703, 63.152435302734375, -6.741846084594727, -0.07158660888671875, 44.05603790283203, -16.670333862304688, 9.367790222167969, 59.07372283935547, 54.60577392578125, 21.049821853637695, 20.899349212646484, -3.2279586791992188, -1.971816062927246, 20.960906982421875, 27.785736083984375, -7.205802917480469, -46.31293487548828, 32.779502868652344, -68.80093383789062, 36.182220458984375, 10.015518188476562, 53.807579040527344, -49.532920837402344, 9.997505187988281, 13.811807632446289, 6.475566864013672, -17.37120819091797, 12.627151489257812, 0.9925270080566406, -55.377471923828125, -13.905952453613281, 25.625896453857422, -14.870697021484375, -5.7549285888671875, 59.64241027832031, -21.923736572265625, 58.081085205078125, 28.615795135498047, 71.95008087158203, 13.780609130859375, 75.17387390136719, 21.918472290039062], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000199.npy"}
|
||||
{"epoch": 0.30083144368858655, "step": 200, "batch_size": 64, "mean": 25.72006607055664, "std": 37.68013000488281, "min": -80.99081420898438, "p10": -15.750474739074706, "median": 22.021873474121094, "p90": 77.49344329833986, "max": 89.17887878417969, "pos_frac": 0.75, "sample": [9.90658950805664, 27.787647247314453, 72.95730590820312, 18.50707244873047, 28.948013305664062, 2.5276050567626953, 73.1158447265625, 13.605537414550781, 10.429122924804688, -3.8937835693359375, 11.186691284179688, 63.181365966796875, -21.804141998291016, 79.1107177734375, 3.675933837890625, -8.083599090576172, 18.380104064941406, -2.6100921630859375, 86.95746612548828, -1.183441162109375, 1.6209335327148438, -6.485343933105469, 46.72643280029297, 62.241336822509766, 2.1583690643310547, 49.73561096191406, 61.73029327392578, -41.419166564941406, 51.45645523071289, -3.904327392578125, 63.63838195800781, 45.39250183105469, 21.406234741210938, 89.17887878417969, 85.66867065429688, -15.509725570678711, -80.99081420898438, 52.00019073486328, 80.28421020507812, 73.71980285644531, 63.39984893798828, 52.714752197265625, 79.97340393066406, 28.595741271972656, 13.032943725585938, -65.05351257324219, 4.804878234863281, 22.63751220703125, 8.89002799987793, 82.85134887695312, -15.853652954101562, 30.97797393798828, 19.662940979003906, -16.099807739257812, 24.029937744140625, 29.811492919921875, -8.531820297241211, -29.261951446533203, 44.98234558105469, 26.710765838623047, 5.238990783691406, -7.495146751403809, 58.499732971191406, 70.2146987915039], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000200.npy"}
|
||||
{"epoch": 0.30234315948601664, "step": 201, "batch_size": 64, "mean": 20.047910690307617, "std": 35.06766891479492, "min": -61.48139953613281, "p10": -27.131152153015133, "median": 16.761378288269043, "p90": 71.38794708251953, "max": 97.55956268310547, "pos_frac": 0.765625, "sample": [12.997451782226562, 28.084136962890625, 27.07278823852539, 27.283981323242188, 16.759695053100586, 22.974281311035156, 39.74257278442383, 10.826444625854492, 48.007110595703125, 2.1388626098632812, 9.745353698730469, 40.94780731201172, 57.22782897949219, 25.9207763671875, 2.659536361694336, -47.300048828125, 97.55956268310547, 71.6616439819336, -1.44464111328125, 76.71601867675781, -28.909515380859375, 4.03509521484375, 12.365257263183594, -32.645076751708984, 4.4500732421875, 74.53289794921875, 39.96758270263672, 24.207778930664062, -28.749086380004883, 70.74932098388672, 4.45695686340332, 37.1689453125, 1.738973617553711, -14.346931457519531, 51.365291595458984, 43.82060623168945, 0.061412811279296875, -39.048065185546875, 22.217437744140625, 10.780532836914062, 59.67185974121094, 6.896610260009766, -3.2190189361572266, 17.914474487304688, -11.06903076171875, 31.667068481445312, 19.22515106201172, -49.821807861328125, 15.788991928100586, 94.41539001464844, -8.091146469116211, 16.7630615234375, 44.99411392211914, 51.2831916809082, 51.942604064941406, 79.56060028076172, 55.72517395019531, -21.79938507080078, 3.8379974365234375, -23.355972290039062, 74.67259216308594, -0.3792724609375, 10.123779296875, -61.48139953613281], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000201.npy"}
|
||||
{"epoch": 0.30385487528344673, "step": 202, "batch_size": 64, "mean": 25.918418884277344, "std": 35.135704040527344, "min": -76.89508819580078, "p10": -6.994546508789062, "median": 29.180343627929688, "p90": 63.373181152343754, "max": 113.11215209960938, "pos_frac": 0.78125, "sample": [-11.444574356079102, 30.958404541015625, 80.44215393066406, 113.11215209960938, 64.16842651367188, -7.325386047363281, 40.74988555908203, 56.042755126953125, 51.845298767089844, 26.16899871826172, 7.143444061279297, 39.043270111083984, 16.765445709228516, 14.595148086547852, -3.2279739379882812, 55.508506774902344, 52.47764587402344, 2.1539459228515625, 57.351646423339844, -6.222587585449219, 28.230804443359375, 53.07977294921875, 30.490097045898438, 1.2383651733398438, -4.979204177856445, 24.706283569335938, 8.265802383422852, 82.39459228515625, 2.3081283569335938, 34.814796447753906, -3.4132957458496094, 16.304452896118164, 39.83214569091797, 37.48539733886719, 10.143096923828125, -76.89508819580078, 41.49400329589844, 27.731569290161133, 53.96892547607422, -57.80726623535156, 30.1298828125, 61.20735168457031, 38.37957000732422, 18.728897094726562, 69.86491394042969, 41.79154968261719, 60.769508361816406, -0.8439540863037109, 78.16246032714844, -4.480171203613281, -0.33933448791503906, -35.32744216918945, 0.07515144348144531, -12.57237434387207, 6.037811279296875, 61.517608642578125, 71.29661560058594, 15.265129089355469, 5.954578399658203, 49.088504791259766, -65.3732681274414, 46.18559265136719, 40.93601989746094, 52.62409210205078], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000202.npy"}
|
||||
{"epoch": 0.30536659108087677, "step": 203, "batch_size": 64, "mean": 23.00586700439453, "std": 36.0931396484375, "min": -65.9398193359375, "p10": -10.811884307861327, "median": 13.107344627380371, "p90": 74.0936477661133, "max": 100.84646606445312, "pos_frac": 0.765625, "sample": [19.34130096435547, 4.300331115722656, -36.806304931640625, 1.4560546875, 6.34423828125, 53.056182861328125, 8.023551940917969, 55.466827392578125, 26.8304443359375, -7.3366851806640625, 82.43592834472656, 11.629669189453125, 28.140260696411133, 89.53326416015625, 12.750768661499023, 100.84646606445312, 63.77661895751953, 13.463920593261719, 55.86346435546875, 35.1009521484375, -65.9398193359375, 1.3397750854492188, 17.066612243652344, -5.320278167724609, -1.853597640991211, -17.47796630859375, 15.509033203125, 87.17678833007812, 6.169364929199219, 40.94725799560547, 17.727088928222656, 4.230323791503906, 49.982177734375, 11.562175750732422, 55.520904541015625, -0.078948974609375, 76.49715423583984, 54.59669494628906, 67.56454467773438, 39.84020233154297, -1.2124099731445312, 9.102035522460938, 10.409156799316406, 42.726539611816406, 58.98988342285156, -9.6326904296875, 68.48546600341797, 80.32421875, -11.317253112792969, 85.29953002929688, 53.94113540649414, 4.267108917236328, 7.914764404296875, 8.769195556640625, -3.3429107666015625, -47.056739807128906, 2.9566421508789062, 11.346855163574219, 53.041748046875, 41.773338317871094, 16.900650024414062, -49.57012939453125, -40.253597259521484, -0.7637596130371094], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000203.npy"}
|
||||
{"epoch": 0.30687830687830686, "step": 204, "batch_size": 64, "mean": 24.542932510375977, "std": 35.77454376220703, "min": -68.38969421386719, "p10": -14.839581108093261, "median": 16.51257038116455, "p90": 77.75672683715821, "max": 88.86788940429688, "pos_frac": 0.734375, "sample": [84.3974609375, 73.0372543334961, -20.62896728515625, 5.1308746337890625, 11.631904602050781, 85.47601318359375, 6.2777557373046875, 59.017181396484375, -22.06645965576172, 44.19915771484375, 88.86788940429688, -40.730186462402344, -5.462953567504883, 4.703456878662109, 36.3201904296875, 15.115447998046875, 49.012657165527344, 5.16900634765625, -33.576751708984375, 4.246955871582031, 45.2589111328125, 23.910621643066406, 64.77445983886719, 71.26143646240234, -15.005823135375977, -5.7000579833984375, 34.40179443359375, 16.748271942138672, 5.232036590576172, 10.962936401367188, 79.22441101074219, -7.76104736328125, 4.122646331787109, 19.853317260742188, 53.03976821899414, -10.750553131103516, 33.78291320800781, 82.17707824707031, -1.402109146118164, 14.417999267578125, 13.707927703857422, 20.489482879638672, 66.25084686279297, 68.8843002319336, -13.109611511230469, 29.55986785888672, 25.967147827148438, -1.861276626586914, 61.13490295410156, -0.3026618957519531, 74.3321304321289, 81.04945373535156, -0.5415306091308594, 18.589035034179688, 38.44862365722656, 85.04696655273438, 59.132957458496094, 1.0031356811523438, 4.9341278076171875, 51.112274169921875, -68.38969421386719, -14.451683044433594, 16.27686882019043, -15.202865600585938], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000204.npy"}
|
||||
{"epoch": 0.30839002267573695, "step": 205, "batch_size": 64, "mean": 23.332550048828125, "std": 40.25199890136719, "min": -81.85629272460938, "p10": -19.337232971191405, "median": 23.35891819000244, "p90": 74.66689987182619, "max": 124.79412841796875, "pos_frac": 0.734375, "sample": [124.79412841796875, 22.08750343322754, -1.7872791290283203, 27.329345703125, 39.2267951965332, 92.16853332519531, 14.587186813354492, 51.011314392089844, -63.077392578125, 75.78436279296875, 7.582485198974609, 9.549617767333984, 104.63423919677734, 27.34564971923828, 34.885292053222656, 24.630332946777344, 10.967086791992188, -9.569854736328125, -18.40576171875, 17.82463836669922, 25.485538482666016, 28.94464874267578, 7.469633102416992, 13.266494750976562, 50.55584716796875, -5.490142822265625, -21.288619995117188, -6.540973663330078, 26.31319808959961, 99.97534942626953, -9.411933898925781, -18.072050094604492, 41.90380859375, 19.064346313476562, 1.6046905517578125, 52.414520263671875, 31.421743392944336, 60.32567596435547, -2.0582504272460938, 8.457206726074219, -29.65045166015625, 72.05948638916016, 55.807884216308594, 43.10185241699219, 87.59088134765625, 67.18057250976562, 1.259307861328125, 32.123680114746094, 62.635345458984375, -9.074024200439453, 32.573577880859375, 2.04901123046875, 38.27023696899414, 10.057334899902344, -68.96382141113281, 64.54235076904297, 49.49168014526367, -26.118242263793945, 82.67981719970703, -16.056053161621094, -19.736434936523438, -81.85629272460938, 2.7044830322265625, 44.70210266113281], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000205.npy"}
|
||||
{"epoch": 0.30990173847316704, "step": 206, "batch_size": 64, "mean": 16.571247100830078, "std": 41.919105529785156, "min": -69.09011840820312, "p10": -32.63568935394287, "median": 9.88003158569336, "p90": 78.58210372924806, "max": 92.64964294433594, "pos_frac": 0.671875, "sample": [-31.179155349731445, 9.282485961914062, 2.8217010498046875, -61.250953674316406, -25.611284255981445, -0.718109130859375, 10.477577209472656, 11.57010269165039, 67.6436996459961, 25.16954803466797, -24.2783145904541, 85.50403594970703, 29.17923355102539, -11.164510726928711, -16.55638885498047, 49.224334716796875, -1.8667526245117188, 90.37637329101562, -33.259918212890625, 4.171682357788086, 90.48410034179688, 59.25822067260742, 11.764511108398438, 4.10089111328125, -23.59661865234375, 71.30964660644531, 13.39486312866211, -54.00480651855469, 52.727088928222656, 57.32440948486328, 64.64077758789062, 80.0671157836914, 1.2419815063476562, 7.349124908447266, 72.29252624511719, 8.478973388671875, -69.09011840820312, -43.449195861816406, -10.045976638793945, -7.07275390625, 41.308815002441406, 0.4014167785644531, 29.536056518554688, -24.44062042236328, -7.851432800292969, -50.563655853271484, 75.32464599609375, 65.01181030273438, 6.065910339355469, 16.66180419921875, 45.751068115234375, 86.17564392089844, 18.235443115234375, -24.202308654785156, 49.736602783203125, 79.97815704345703, 6.196514129638672, 12.431533813476562, 24.54949188232422, 92.64964294433594, -62.08015441894531, 0.6051788330078125, -3.8063583374023438, 16.174522399902344], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000206.npy"}
|
||||
{"epoch": 0.31141345427059713, "step": 207, "batch_size": 64, "mean": 22.76787757873535, "std": 37.94688034057617, "min": -59.24778747558594, "p10": -20.749091339111324, "median": 10.937920570373535, "p90": 79.34127044677736, "max": 104.82565307617188, "pos_frac": 0.734375, "sample": [97.95997619628906, 11.43376350402832, -24.503883361816406, 7.037269592285156, 6.2228851318359375, 34.771934509277344, -17.026992797851562, 7.496437072753906, -1.0926742553710938, 88.81660461425781, 9.339691162109375, 16.323638916015625, 1.7383079528808594, 80.36013793945312, -4.029573440551758, -13.223556518554688, 26.179779052734375, 55.074371337890625, 39.492835998535156, 76.10882568359375, 68.21913146972656, 19.34368896484375, 10.419525146484375, -9.68670654296875, -25.863115310668945, -9.676559448242188, -4.330991744995117, 75.4997329711914, -38.01978302001953, 9.711357116699219, 73.04944610595703, 15.706737518310547, 104.82565307617188, -22.344276428222656, 6.3852691650390625, -11.49521255493164, -33.115562438964844, 15.740203857421875, -8.71917724609375, 2.364351272583008, 38.930999755859375, 68.52169799804688, 8.017837524414062, 54.472145080566406, 3.9137344360351562, 49.90290832519531, 59.87550354003906, -59.24778747558594, 81.44792175292969, 25.70880126953125, -36.48223114013672, 38.26866912841797, 27.711318969726562, 13.427932739257812, 5.1082305908203125, 94.32868194580078, 76.96391296386719, 5.243812561035156, 33.52040100097656, 47.43244934082031, -13.390060424804688, 10.44207763671875, 2.5621490478515625, 83.96952819824219], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000207.npy"}
|
||||
{"epoch": 0.3129251700680272, "step": 208, "batch_size": 64, "mean": 11.888145446777344, "std": 39.587745666503906, "min": -75.53514862060547, "p10": -39.47302742004394, "median": 9.328887939453125, "p90": 64.33688888549804, "max": 88.64706420898438, "pos_frac": 0.59375, "sample": [0.7588024139404297, 2.7992725372314453, 59.34607696533203, -16.268840789794922, 62.75628662109375, 73.04647064208984, 43.6683349609375, -8.415237426757812, -30.268890380859375, -60.810340881347656, -15.8759765625, -18.014251708984375, 57.70814514160156, 59.197967529296875, 37.800838470458984, 8.768775939941406, 12.735210418701172, 29.43951416015625, -7.913965225219727, -27.4078369140625, 57.061279296875, -13.383934020996094, 7.260524749755859, -44.473045349121094, 48.860084533691406, 24.413145065307617, 64.34265899658203, -38.255619049072266, 15.757041931152344, 28.14937400817871, -35.27851486206055, -10.468803405761719, -19.403564453125, 20.635604858398438, -53.678123474121094, 46.99333190917969, 81.73910522460938, 64.32342529296875, 39.52903747558594, -0.29187583923339844, 88.64706420898438, 24.536712646484375, 10.614288330078125, -39.994773864746094, 72.07846069335938, -23.164817810058594, 87.96078491210938, -7.481609344482422, 12.159250259399414, 21.199005126953125, -24.821617126464844, 25.177528381347656, -2.1490478515625, 43.26726531982422, -43.13142776489258, 9.888999938964844, 28.570449829101562, -7.090488433837891, -59.314937591552734, -6.119609832763672, -75.53514862060547, 5.739173889160156, 4.8054351806640625, 68.11888885498047], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000208.npy"}
|
||||
{"epoch": 0.3144368858654573, "step": 209, "batch_size": 64, "mean": 29.188037872314453, "std": 45.934329986572266, "min": -100.00890350341797, "p10": -10.556300354003906, "median": 25.13037109375, "p90": 85.51901702880859, "max": 117.06533813476562, "pos_frac": 0.78125, "sample": [87.87339782714844, 78.3802261352539, 8.242713928222656, 30.395092010498047, -9.359375, 117.06533813476562, 17.121824264526367, 78.19439697265625, 22.800949096679688, 77.37933349609375, 5.880455017089844, 2.8763198852539062, -77.05694580078125, 77.31998443603516, -3.221160888671875, 11.864517211914062, -5.8238067626953125, 2.2508468627929688, -1.6547813415527344, 110.16044616699219, 7.781333923339844, 5.10699462890625, -3.1367645263671875, 29.658935546875, 7.9862518310546875, 5.288482666015625, 40.20014190673828, 54.6004638671875, 92.18275451660156, 26.663360595703125, 40.121185302734375, 50.754669189453125, 17.132469177246094, 85.86954498291016, 85.77293395996094, -1.8764572143554688, 84.74111938476562, 56.187103271484375, -100.00890350341797, 6.88629150390625, -23.33570098876953, 5.327863693237305, -10.883514404296875, -9.792800903320312, 78.209228515625, 18.650768280029297, 114.32320404052734, 68.6064224243164, 48.975555419921875, 69.12489318847656, 30.26469612121582, -65.77861785888672, 16.213655471801758, 41.56831359863281, 23.597381591796875, 5.237064361572266, 27.53032684326172, -84.14077758789062, 26.955795288085938, 44.644866943359375, 84.92654418945312, 80.39645385742188, 69.10508728027344, -14.294029235839844], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000209.npy"}
|
||||
{"epoch": 0.31594860166288735, "step": 210, "batch_size": 64, "mean": 28.413719177246094, "std": 40.82936096191406, "min": -60.558204650878906, "p10": -22.13609275817871, "median": 19.37389373779297, "p90": 83.63666915893556, "max": 118.85359191894531, "pos_frac": 0.78125, "sample": [87.81453704833984, 6.962150573730469, 15.78387451171875, 18.607223510742188, -2.4865760803222656, 32.515167236328125, 100.76226806640625, 62.65913391113281, 93.02396392822266, 81.32144165039062, -0.5161476135253906, 118.85359191894531, 93.03335571289062, 12.205307006835938, 81.74746704101562, 4.5211334228515625, -60.558204650878906, -34.40977096557617, 81.0416259765625, -36.764747619628906, 65.68417358398438, 19.54003143310547, 11.911050796508789, -21.1273193359375, 78.9520492553711, 10.902084350585938, 0.41179466247558594, 14.284172058105469, 68.23009490966797, -0.5259552001953125, 84.44632720947266, 73.29216766357422, 24.79846954345703, 22.246246337890625, 22.55263900756836, -3.761260986328125, 9.044025421142578, -22.568424224853516, 3.7640724182128906, 81.6812515258789, -14.136495590209961, 43.30145263671875, 26.157211303710938, 7.760505676269531, 60.84454345703125, -0.6304149627685547, 13.14497184753418, 16.86191177368164, 21.93132781982422, 6.1492462158203125, 32.8109016418457, 67.88616180419922, 22.692455291748047, -48.92655944824219, 5.769950866699219, 97.87621307373047, 63.341651916503906, 19.20775604248047, 64.64419555664062, -35.41449737548828, 5.291810989379883, -29.35871124267578, 22.9625244140625, 48.435508728027344], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000210.npy"}
|
||||
{"epoch": 0.31746031746031744, "step": 211, "batch_size": 64, "mean": 28.919544219970703, "std": 38.09408950805664, "min": -49.82965850830078, "p10": -9.70748996734619, "median": 16.72985076904297, "p90": 90.7733741760254, "max": 134.44271850585938, "pos_frac": 0.75, "sample": [0.4574928283691406, -49.82965850830078, 47.498558044433594, 9.582290649414062, 1.2465972900390625, -3.539796829223633, -12.328033447265625, 6.2796630859375, 47.929779052734375, 6.818769454956055, 23.056427001953125, -2.214611053466797, 40.43876647949219, 92.36044311523438, 8.333362579345703, 53.64936447143555, -0.9581279754638672, -1.8693428039550781, 82.39340209960938, -19.574581146240234, 19.334491729736328, -8.423675537109375, 35.72590255737305, -17.86761474609375, 10.323040008544922, -2.013662338256836, -0.19954681396484375, 17.334991455078125, 93.27511596679688, 6.8692169189453125, 29.397048950195312, 11.446853637695312, 15.029973983764648, 4.291297912597656, 92.45620727539062, 25.843402862548828, 9.973068237304688, 16.124710083007812, -18.69051742553711, 41.84040832519531, 9.135047912597656, 134.44271850585938, 113.8812255859375, 7.914907455444336, -2.8911972045898438, 42.43734359741211, 87.0702133178711, 60.205047607421875, 31.2374267578125, 92.64311218261719, -13.105422973632812, 30.961917877197266, 82.05653381347656, 73.0468521118164, 68.42408752441406, 79.51505279541016, 24.83203125, 63.93336486816406, -10.257696151733398, 97.57966613769531, 14.987136840820312, -2.7165298461914062, 34.139015197753906, 19.577552795410156], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000211.npy"}
|
||||
{"epoch": 0.31897203325774753, "step": 212, "batch_size": 64, "mean": 28.71660614013672, "std": 42.67084503173828, "min": -112.18833923339844, "p10": -16.528522872924803, "median": 31.560553550720215, "p90": 89.09589614868165, "max": 102.1485824584961, "pos_frac": 0.765625, "sample": [89.14814758300781, 45.34487533569336, 58.833648681640625, 75.13056945800781, -10.646421432495117, 9.263771057128906, 98.35507202148438, 48.57439422607422, -17.16564178466797, 78.11724090576172, -12.431896209716797, -112.18833923339844, 71.34366607666016, 89.77944946289062, 9.879920959472656, 73.82231140136719, 64.46636962890625, 54.60515594482422, -32.96306610107422, 28.903242111206055, 36.3101806640625, 7.999185562133789, 2.1998348236083984, 41.025352478027344, -62.50103759765625, 0.5248870849609375, 45.49078369140625, 2.1123409271240234, 82.06543731689453, 94.90513610839844, 38.65143966674805, 66.06719970703125, 45.29261016845703, 34.217864990234375, 15.702911376953125, 12.673271179199219, -14.588565826416016, 49.763343811035156, 2.0728721618652344, 9.84591293334961, 5.59576416015625, 24.858108520507812, -4.912296295166016, -2.3908843994140625, 35.117340087890625, 48.616615295410156, 43.41645050048828, 88.9739761352539, -15.041912078857422, -22.177902221679688, -4.442150115966797, -0.4609527587890625, 53.04872131347656, -20.5040283203125, 53.5595703125, 92.89643096923828, 89.48685455322266, 102.1485824584961, 59.7562255859375, 2.147430419921875, 1.667572021484375, 27.272979736328125, -41.081336975097656, 0.30829620361328125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000212.npy"}
|
||||
{"epoch": 0.3204837490551776, "step": 213, "batch_size": 64, "mean": 32.623939514160156, "std": 44.5143928527832, "min": -88.1232681274414, "p10": -21.5225959777832, "median": 29.34423828125, "p90": 87.42460098266602, "max": 145.97427368164062, "pos_frac": 0.78125, "sample": [66.9876708984375, 59.89909362792969, 9.311996459960938, 83.61892700195312, 53.27484893798828, 25.92908477783203, -28.950393676757812, -24.428390502929688, -13.254974365234375, 65.30607604980469, 103.47845458984375, 3.53546142578125, 29.8529052734375, -12.612783432006836, -88.1232681274414, -22.569686889648438, 60.924339294433594, 145.97427368164062, -19.079383850097656, 40.53424072265625, 75.80264282226562, -49.32748031616211, 94.71076202392578, 114.59188079833984, 65.41038513183594, 22.21870994567871, -7.771465301513672, -51.75714874267578, 7.655803680419922, 22.92629051208496, -25.40686798095703, 28.8355712890625, 31.779090881347656, 27.60809898376465, 49.771453857421875, 46.138946533203125, 6.19035530090332, 57.882102966308594, 86.56439971923828, 9.212104797363281, 20.96817398071289, 32.454566955566406, 87.79325866699219, 2.6469078063964844, 78.80599975585938, 27.879531860351562, 2.1299285888671875, 55.702510833740234, 97.38221740722656, 9.928916931152344, -1.7142791748046875, 63.12567901611328, -18.61834716796875, 5.808868408203125, 8.586524963378906, 31.102645874023438, 20.34210205078125, 79.4808349609375, 33.34236145019531, -5.0082855224609375, 78.12312316894531, 84.73492431640625, 49.677978515625, 90.61198425292969], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000213.npy"}
|
||||
{"epoch": 0.3219954648526077, "step": 214, "batch_size": 64, "mean": 31.853899002075195, "std": 47.2297477722168, "min": -58.41252136230469, "p10": -19.71971569061279, "median": 25.54559898376465, "p90": 89.70892944335938, "max": 124.68098449707031, "pos_frac": 0.671875, "sample": [69.55626678466797, 53.756690979003906, -13.78485107421875, 26.50326156616211, 24.587936401367188, 87.43234252929688, 7.110103607177734, 32.29338455200195, 60.48272705078125, 77.94548797607422, 86.64824676513672, 72.6724853515625, 36.56713104248047, -15.873321533203125, 81.25650024414062, 70.62037658691406, 63.86050033569336, -53.719451904296875, 55.89684295654297, 37.59580612182617, -5.208778381347656, 52.74507522583008, 22.054824829101562, -18.7374267578125, -1.879098892211914, 55.62825012207031, -38.19059753417969, 24.4993953704834, 97.54606628417969, 59.69239807128906, 23.91889190673828, -15.803321838378906, -19.212522506713867, 1.630767822265625, 46.59252166748047, 70.14533996582031, -56.1864013671875, -11.052215576171875, 87.9395751953125, 124.68098449707031, -11.290206909179688, -19.937084197998047, 84.52764129638672, 47.337730407714844, 101.00904846191406, 121.84358215332031, 9.55184555053711, -0.700775146484375, 23.164169311523438, -7.929744720458984, -43.0528564453125, 104.77531433105469, 15.642560958862305, -12.737117767333984, 7.239622116088867, -0.0323944091796875, 118.89224243164062, 22.50432586669922, -58.41252136230469, 61.05841064453125, 71.02555084228516, 90.46722412109375, -5.805866241455078, -42.70336151123047], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000214.npy"}
|
||||
{"epoch": 0.3235071806500378, "step": 215, "batch_size": 64, "mean": 32.916534423828125, "std": 49.40408706665039, "min": -80.34130096435547, "p10": -32.50116806030273, "median": 26.812471389770508, "p90": 91.21856689453125, "max": 129.60775756835938, "pos_frac": 0.734375, "sample": [31.389057159423828, -62.87644958496094, -80.34130096435547, 43.08319091796875, 31.44153594970703, -52.57237243652344, 11.957382202148438, -4.829795837402344, 89.34336853027344, 91.32884216308594, 21.268386840820312, -3.821514129638672, -40.020423889160156, 33.24803924560547, 40.34674072265625, 95.56373596191406, 9.126617431640625, 88.12932586669922, 6.0257415771484375, -7.5217742919921875, 22.235885620117188, 0.9797821044921875, -7.706005096435547, 44.776771545410156, 83.67024993896484, 5.049465179443359, -0.07375335693359375, 66.65003967285156, 81.85494995117188, -27.688257217407227, 8.330623626708984, 84.35883331298828, 4.555393218994141, -12.904373168945312, -31.615203857421875, -46.443511962890625, 74.1485366821289, 61.76231384277344, 60.93414306640625, 124.47183227539062, 16.725337982177734, 21.50524139404297, 16.568389892578125, -4.181556701660156, 85.82046508789062, 90.96125793457031, 104.57408905029297, 98.14938354492188, 84.9206314086914, 129.60775756835938, 49.064144134521484, 85.1751937866211, 61.08350372314453, -32.88086700439453, 20.16722297668457, 3.674959182739258, 45.8828125, 122.20985412597656, 74.8623046875, -1.2586517333984375, 67.18907165527344, 2.357311248779297, 68.05176544189453, -41.187564849853516], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000215.npy"}
|
||||
{"epoch": 0.3250188964474679, "step": 216, "batch_size": 64, "mean": 35.08480453491211, "std": 40.6789665222168, "min": -86.80792999267578, "p10": -4.242982864379882, "median": 34.481801986694336, "p90": 92.77354049682619, "max": 105.341796875, "pos_frac": 0.8125, "sample": [11.056991577148438, 80.64736938476562, 78.716552734375, 73.33950805664062, 49.08716583251953, 16.03636932373047, 28.767189025878906, 15.346038818359375, 55.68269348144531, 80.40504455566406, 63.71051788330078, 37.23517608642578, 1.8254165649414062, 14.11578369140625, 33.06560516357422, -17.955032348632812, 13.331104278564453, 38.634254455566406, -2.90509033203125, 102.11294555664062, 41.25933837890625, 102.42398071289062, 36.10761260986328, 82.37897491455078, 51.29719543457031, 34.308162689208984, 95.75738525390625, 1.6486015319824219, 0.9572715759277344, 95.39859008789062, 89.35538482666016, -86.80792999267578, 34.65544128417969, 49.09135818481445, 46.99432373046875, 56.29547119140625, -38.48426818847656, 1.4066333770751953, 31.88265609741211, -4.7384490966796875, 1.4216842651367188, 94.38070678710938, 81.01048278808594, -31.23876190185547, 27.427528381347656, -19.96862030029297, 29.80227279663086, 4.291072845458984, 11.80426025390625, 44.51190185546875, 105.341796875, -32.828487396240234, -2.080291748046875, -1.154348373413086, 90.04923248291016, -0.4848041534423828, 25.54750633239746, 70.00200653076172, 4.089775085449219, 93.94110107421875, -3.086894989013672, 76.60137939453125, 37.804931640625, 44.798583984375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000216.npy"}
|
||||
{"epoch": 0.32653061224489793, "step": 217, "batch_size": 64, "mean": 36.457122802734375, "std": 44.0004997253418, "min": -61.81396484375, "p10": -15.499212265014643, "median": 30.52617835998535, "p90": 96.68674468994142, "max": 129.05551147460938, "pos_frac": 0.765625, "sample": [74.77279663085938, 75.50181579589844, -18.480789184570312, -4.7724456787109375, -2.2688961029052734, -9.73470687866211, 5.128627777099609, 129.05551147460938, -10.692840576171875, 25.261444091796875, 125.39454650878906, 114.48149108886719, 9.152275085449219, -17.559085845947266, 22.344467163085938, 12.678733825683594, 98.05882263183594, 58.61442947387695, 19.467864990234375, 76.84745025634766, 28.68729591369629, 38.804222106933594, 17.80621337890625, 89.70236206054688, 15.579353332519531, 79.35540771484375, -61.81396484375, -5.769523620605469, 8.365264892578125, 19.923282623291016, 88.95182800292969, 43.705848693847656, 47.736968994140625, -18.37543487548828, 60.64482879638672, -22.902523040771484, 15.349990844726562, 48.233436584472656, 1.9294815063476562, -36.43464660644531, -49.31085205078125, 70.0948257446289, 45.55424499511719, -0.0286712646484375, -9.747238159179688, 10.443305969238281, 17.929031372070312, 28.891983032226562, 67.70880889892578, 118.70518493652344, 47.07028579711914, 78.37267303466797, 33.30394744873047, 3.4463729858398438, 100.40657806396484, 67.21052551269531, 102.68830108642578, 43.216033935546875, 55.17527770996094, 80.34519958496094, 56.94804382324219, 93.4852294921875, 32.16037368774414, -3.5449066162109375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000217.npy"}
|
||||
{"epoch": 0.328042328042328, "step": 218, "batch_size": 64, "mean": 21.409412384033203, "std": 50.066829681396484, "min": -117.71417999267578, "p10": -38.99569244384766, "median": 24.417236328125, "p90": 88.20860366821289, "max": 116.63768005371094, "pos_frac": 0.703125, "sample": [9.32861328125, 0.7525863647460938, 44.95259094238281, 53.17182922363281, 0.656951904296875, -12.327102661132812, 24.215286254882812, 0.273529052734375, -7.556488037109375, 88.6486587524414, 28.050308227539062, 18.413169860839844, 91.9840316772461, -76.93048095703125, -39.13304901123047, -63.76612854003906, 3.3402862548828125, 99.12054443359375, -56.998504638671875, -8.37789535522461, 45.57954406738281, 37.02531433105469, 0.9838485717773438, 35.52400588989258, 60.250545501708984, 109.9178466796875, 84.8156967163086, 116.63768005371094, 28.668867111206055, 56.96836853027344, -1.8225898742675781, 44.719696044921875, 50.05836486816406, -19.689050674438477, -17.755901336669922, -37.70690155029297, 7.969512939453125, 38.18629455566406, 4.776634216308594, 74.97859191894531, 76.85016632080078, 46.74090576171875, -6.640491485595703, 14.823348999023438, 29.734439849853516, -38.675193786621094, 67.04535675048828, -5.258298873901367, -66.48350524902344, -19.365081787109375, 87.18180847167969, -82.54102325439453, 26.592201232910156, 0.6942424774169922, 28.729759216308594, 75.6807861328125, 6.760875701904297, 96.39289093017578, -117.71417999267578, -7.455802917480469, 41.99709701538086, 24.619186401367188, 69.15617370605469, 103.43167114257812], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000218.npy"}
|
||||
{"epoch": 0.3295540438397581, "step": 219, "batch_size": 64, "mean": 23.374771118164062, "std": 40.488975524902344, "min": -73.62139129638672, "p10": -17.67337875366211, "median": 14.985527038574219, "p90": 88.09762344360352, "max": 104.7090072631836, "pos_frac": 0.734375, "sample": [11.181838989257812, 90.33204650878906, 66.36223602294922, 10.385217666625977, 39.484825134277344, -4.047889709472656, 4.560054779052734, 2.4526729583740234, 17.153480529785156, -17.415145874023438, -8.589338302612305, 104.7090072631836, 60.8563232421875, -47.08953857421875, -46.4866943359375, 42.409515380859375, 45.686622619628906, 1.1354618072509766, 33.14606475830078, 5.799945831298828, 3.0437393188476562, -17.090133666992188, -15.482048034667969, 30.080799102783203, -33.5950927734375, 1.5906829833984375, 7.896492004394531, -2.174358367919922, 45.61888122558594, 15.619728088378906, 60.16670227050781, 89.34358215332031, 88.52200317382812, 94.14448547363281, 41.722389221191406, 87.66134643554688, -17.78404998779297, 4.060462951660156, -3.9721908569335938, 39.02195358276367, 51.07798767089844, 23.139450073242188, -73.62139129638672, 37.73206329345703, -7.195747375488281, -3.7042617797851562, 49.207000732421875, 8.347244262695312, 45.227867126464844, 89.60828399658203, 68.64830017089844, 71.62342834472656, 15.695194244384766, -11.134353637695312, -64.08242797851562, 9.748859405517578, 14.351325988769531, 4.286369323730469, 69.66804504394531, 0.82550048828125, 88.28459930419922, -23.64508056640625, 67.61048889160156, 33.86444091796875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000219.npy"}
|
||||
{"epoch": 0.3310657596371882, "step": 220, "batch_size": 64, "mean": 28.318567276000977, "std": 49.690757751464844, "min": -92.93110656738281, "p10": -29.17484512329101, "median": 20.936965942382812, "p90": 89.58534164428713, "max": 138.55120849609375, "pos_frac": 0.65625, "sample": [51.34624481201172, -14.288856506347656, 54.20616149902344, 17.62627410888672, 11.213008880615234, 41.14573669433594, 138.55120849609375, 35.34492492675781, 6.466381072998047, 13.482643127441406, -14.829498291015625, 103.95433044433594, 5.87158203125, -62.08910369873047, -31.584327697753906, -31.93848419189453, 91.20751190185547, -9.192222595214844, -6.393962860107422, 3.7178916931152344, 73.20819091796875, 32.949180603027344, -6.814994812011719, -63.87383270263672, -54.899375915527344, -21.247636795043945, -2.0955848693847656, -23.552719116210938, -34.1475830078125, 65.98544311523438, 53.955078125, 55.39250183105469, 47.44087219238281, 74.53533935546875, 85.41795349121094, 7.303382873535156, 83.14268493652344, 60.889320373535156, 58.013999938964844, -9.330718994140625, 10.722000122070312, 53.60306930541992, -20.926307678222656, 62.273193359375, 132.04754638671875, -6.1878662109375, 105.29401397705078, -0.7272224426269531, 35.95398712158203, 94.88682556152344, 74.80169677734375, -4.750818252563477, 85.80027770996094, -2.7182579040527344, 25.91351318359375, 24.247657775878906, -92.93110656738281, 14.90255355834961, 68.2713623046875, 3.2999725341796875, 119.80609130859375, -9.369394302368164, 83.53791809082031, 68.54866027832031], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000220.npy"}
|
||||
{"epoch": 0.3325774754346183, "step": 221, "batch_size": 64, "mean": 31.438785552978516, "std": 47.2042121887207, "min": -76.48954772949219, "p10": -22.31003036499022, "median": 17.542898178100586, "p90": 98.29193649291996, "max": 157.2386474609375, "pos_frac": 0.765625, "sample": [15.457754135131836, -36.226741790771484, 36.463600158691406, 40.064697265625, -2.7141971588134766, 78.85008239746094, -7.99284553527832, 34.9901123046875, 50.943931579589844, -46.570640563964844, -4.458461761474609, 23.86871337890625, 16.799335479736328, 15.624099731445312, -34.014732360839844, 21.64889144897461, 5.3058013916015625, -2.5762195587158203, 59.17310333251953, 54.26947021484375, 83.53266906738281, 88.48604583740234, 15.042125701904297, 76.85054016113281, 157.2386474609375, 3.7403488159179688, 82.22660827636719, 103.09193420410156, -8.379302978515625, 18.971323013305664, 49.30943298339844, 105.46836853027344, 5.578540802001953, 47.10185623168945, 114.80741882324219, 13.653800964355469, 67.09915161132812, -28.28034210205078, 43.28791427612305, 83.728271484375, 67.201904296875, -3.4727630615234375, 4.715324401855469, -1.2360153198242188, 1.4458560943603516, 1.8634490966796875, 0.4681835174560547, 102.49446105957031, 6.853919982910156, 11.511585235595703, -40.168212890625, -76.48954772949219, 4.7093658447265625, 118.29733276367188, 69.12742614746094, -46.72930908203125, 18.286460876464844, 51.015159606933594, 131.80531311035156, 66.34703063964844, 5.120124816894531, -2.1169891357421875, 66.52665710449219, 13.044395446777344], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000221.npy"}
|
||||
{"epoch": 0.3340891912320484, "step": 222, "batch_size": 64, "mean": 43.04325866699219, "std": 51.234256744384766, "min": -69.10041809082031, "p10": -13.572195434570308, "median": 42.61950874328613, "p90": 106.90595626831056, "max": 170.79931640625, "pos_frac": 0.78125, "sample": [-3.314687728881836, -49.990013122558594, 45.29819869995117, 2.1253662109375, 84.32907104492188, 85.67076873779297, 58.6579475402832, -15.485668182373047, 89.76441955566406, 105.11309814453125, 61.843017578125, 6.299873352050781, -48.05077362060547, 6.4580841064453125, 74.64936065673828, 7.333099365234375, 20.966293334960938, 10.754634857177734, 16.94192123413086, 73.60171508789062, 54.859230041503906, -0.6749687194824219, 82.81451416015625, 108.54216766357422, -7.331794738769531, 102.59394073486328, -30.638389587402344, 168.29281616210938, -16.224565505981445, 66.31781005859375, 38.21922302246094, 18.986053466796875, 110.19058227539062, 85.38320922851562, 28.087509155273438, 61.455963134765625, 84.6844253540039, -3.7992095947265625, -9.107425689697266, -0.0051727294921875, -69.10041809082031, 39.940818786621094, 48.293373107910156, 10.942977905273438, 4.542949676513672, 170.79931640625, 24.800804138183594, -37.94444274902344, 6.6897735595703125, 64.04370880126953, 80.80872344970703, 1.7963180541992188, 85.92225646972656, 107.67432403564453, 63.18560791015625, 14.624137878417969, 91.01237487792969, 93.480712890625, 3.16094970703125, 109.65705871582031, 80.64060974121094, 72.58735656738281, 112.53909301757812, -0.9413299560546875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000222.npy"}
|
||||
{"epoch": 0.3356009070294785, "step": 223, "batch_size": 64, "mean": 32.60646057128906, "std": 46.262229919433594, "min": -49.64455795288086, "p10": -22.90240364074707, "median": 26.090608596801758, "p90": 89.46865539550781, "max": 130.59983825683594, "pos_frac": 0.6875, "sample": [74.55581665039062, 9.421258926391602, 74.6732177734375, 7.834205627441406, -10.824501037597656, 55.56868362426758, -12.045913696289062, 28.185623168945312, 0.107208251953125, 56.06472396850586, 2.459012985229492, 121.71507263183594, 70.55458068847656, 89.8692398071289, 17.06173324584961, -23.166500091552734, -2.3858509063720703, 69.75313568115234, 13.656494140625, -11.967910766601562, 81.79512023925781, 46.95390319824219, -32.69972610473633, 67.56419372558594, -24.674522399902344, -9.96539306640625, -24.07053565979004, 85.62727355957031, -5.4681396484375, -32.92822265625, 63.45672607421875, 18.317106246948242, -46.42967224121094, 6.236194610595703, 9.620346069335938, 104.03337860107422, 58.76012420654297, 43.41695022583008, -22.286178588867188, -49.64455795288086, -14.190780639648438, -21.15376853942871, 40.556121826171875, 73.4500732421875, 124.32904052734375, 23.995594024658203, 54.092193603515625, 68.7261962890625, 33.71971130371094, 99.95767211914062, 119.93134307861328, -13.925666809082031, 130.59983825683594, 88.5339584350586, 44.82000732421875, 55.381317138671875, 81.24757385253906, 83.94139862060547, 41.0164794921875, -4.531795501708984, -3.899444580078125, 9.27276611328125, 2.4268646240234375, -0.187103271484375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000223.npy"}
|
||||
{"epoch": 0.3371126228269085, "step": 224, "batch_size": 64, "mean": 17.776432037353516, "std": 48.1540412902832, "min": -94.97715759277344, "p10": -47.78180541992187, "median": 13.470844268798828, "p90": 83.84325027465825, "max": 129.80416870117188, "pos_frac": 0.703125, "sample": [44.5850944519043, 129.80416870117188, 24.88581657409668, 47.419952392578125, 14.2528076171875, 101.39503479003906, 45.05039978027344, 64.07042694091797, -2.9656219482421875, 19.54522705078125, 71.15283203125, 8.87443733215332, 71.75393676757812, 94.01792907714844, -59.62907791137695, -2.5076656341552734, 15.806951522827148, 21.18700408935547, 21.2579345703125, 11.110504150390625, 13.41265869140625, 32.95044708251953, -8.839065551757812, -70.22137451171875, -7.992485046386719, 2.421314239501953, -21.21302032470703, -15.233062744140625, 16.521575927734375, 4.2799072265625, 65.81941986083984, -94.97715759277344, -56.80419158935547, -38.00714111328125, 4.081268310546875, 0.0065765380859375, -6.654777526855469, -87.99345397949219, 8.032821655273438, 40.01191711425781, -35.601219177246094, 44.182106018066406, 74.82699584960938, 13.529029846191406, -61.01042175292969, 12.27783203125, 16.316650390625, 28.35733413696289, -51.970947265625, -20.244171142578125, -4.076562881469727, 41.5850830078125, 65.49944305419922, 87.70735931396484, 113.56877899169922, 42.509490966796875, 8.635005950927734, 14.075836181640625, 0.1626605987548828, 102.00675964355469, -4.827728271484375, 122.13650512695312, 2.7956924438476562, 4.579929351806641], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000224.npy"}
|
||||
{"epoch": 0.3386243386243386, "step": 225, "batch_size": 64, "mean": 37.4577751159668, "std": 46.15288162231445, "min": -60.5494384765625, "p10": -7.2829460144042955, "median": 25.06774139404297, "p90": 104.02540893554688, "max": 137.74530029296875, "pos_frac": 0.78125, "sample": [7.684902191162109, 42.342994689941406, 8.437156677246094, -7.73004150390625, 22.04437255859375, 76.2962646484375, -3.2050514221191406, 0.18994140625, 47.003501892089844, 19.95317840576172, 14.34112548828125, 89.65975952148438, 16.65444564819336, 112.64955139160156, -15.655033111572266, 13.068073272705078, 26.737831115722656, 23.39765167236328, 85.86381530761719, 95.04129028320312, 105.51643371582031, 36.74089813232422, 128.53671264648438, 74.76771545410156, 137.74530029296875, 20.47913360595703, 5.3372802734375, 68.31094360351562, 104.0660400390625, -3.8076534271240234, -3.0718307495117188, -36.428497314453125, 65.78612518310547, -5.26708984375, 113.17854309082031, 71.04148864746094, 2.1638031005859375, 11.610931396484375, 80.78131103515625, 33.37006759643555, 13.379570007324219, 20.78485107421875, 100.01688385009766, -5.626466751098633, 60.21674346923828, -60.5494384765625, -52.72589111328125, -6.239723205566406, 32.81610107421875, 34.97119140625, 88.37308502197266, 2.605093002319336, 8.307647705078125, 112.34431457519531, 84.11346435546875, -2.4620895385742188, 19.205902099609375, 26.78618621826172, 103.93060302734375, 50.39734649658203, -34.01551818847656, 48.20109558105469, 79.66761016845703, -12.83453369140625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000225.npy"}
|
||||
{"epoch": 0.3401360544217687, "step": 226, "batch_size": 64, "mean": 28.876598358154297, "std": 54.78819274902344, "min": -101.50779724121094, "p10": -47.306521606445294, "median": 32.51651954650879, "p90": 98.42935333251954, "max": 122.92127990722656, "pos_frac": 0.6875, "sample": [14.151176452636719, -59.06438446044922, 50.88592529296875, 32.499977111816406, 34.89039611816406, 52.3095703125, 59.93128204345703, 3.4972801208496094, 71.25141906738281, -12.375652313232422, 82.6631851196289, 34.99879455566406, 70.29682159423828, 88.73289489746094, -69.71739196777344, -6.839569091796875, 3.731740951538086, -84.10618591308594, 95.50840759277344, 46.14740753173828, 99.08969116210938, -13.46099853515625, 69.44824981689453, 6.7763824462890625, 51.48329162597656, -0.25765228271484375, 4.587188720703125, 85.6341781616211, 80.2265853881836, 56.919403076171875, -18.138273239135742, 95.52340698242188, 32.53306198120117, 43.81177520751953, 110.83876037597656, 26.664031982421875, 62.73657989501953, -101.50779724121094, -8.993703842163086, 5.949436187744141, -4.7949676513671875, 53.82136535644531, -27.30353546142578, -85.02888488769531, 102.80511474609375, 122.92127990722656, -54.280120849609375, 38.241912841796875, -80.9288558959961, 51.634849548339844, 20.52768325805664, 6.355682373046875, 31.238361358642578, 25.31516456604004, -31.0347900390625, 115.02883911132812, 111.88616943359375, 101.1429672241211, -17.71228790283203, -8.290887832641602, 88.65377807617188, -0.953521728515625, -7.288301467895508, 96.88856506347656], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000226.npy"}
|
||||
{"epoch": 0.3416477702191988, "step": 227, "batch_size": 64, "mean": 30.060535430908203, "std": 54.87797164916992, "min": -112.66329956054688, "p10": -33.756503295898426, "median": 18.667282104492188, "p90": 106.30960540771486, "max": 123.78355407714844, "pos_frac": 0.703125, "sample": [66.36558532714844, 46.13365173339844, 5.869224548339844, 102.76187133789062, -11.128204345703125, 3.1791648864746094, 82.15087127685547, 108.54464721679688, -79.83746337890625, -2.3124217987060547, 123.78355407714844, -41.86334991455078, -4.86781120300293, 102.12126922607422, 22.75830078125, 30.517333984375, 4.435625076293945, -3.479461669921875, 102.47482299804688, -24.189437866210938, 71.51432800292969, 98.99237823486328, -1.6970062255859375, 5.829200744628906, 110.44050598144531, 42.33716583251953, -1.6475257873535156, 44.837158203125, -9.571924209594727, -0.8116302490234375, 74.73411560058594, 107.83006286621094, 42.54010009765625, 21.481124877929688, 2.5253753662109375, 83.97297668457031, 121.77587890625, 11.172340393066406, 28.246658325195312, 111.80272674560547, 10.895118713378906, 36.22021484375, -37.85667419433594, 29.158218383789062, 92.904052734375, 15.853439331054688, 11.77374267578125, 4.75422477722168, -10.544097900390625, -2.0747032165527344, -81.37301635742188, 71.96585845947266, 14.81451416015625, 70.3284912109375, 10.184865951538086, -112.66329956054688, 25.383499145507812, -3.375926971435547, 100.9539566040039, 1.2500152587890625, -81.650146484375, -43.83314514160156, 91.06920623779297, 110.0140609741211], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000227.npy"}
|
||||
{"epoch": 0.3431594860166289, "step": 228, "batch_size": 64, "mean": 35.28078079223633, "std": 49.59514617919922, "min": -78.73158264160156, "p10": -29.357705688476557, "median": 35.136796951293945, "p90": 102.77861862182618, "max": 127.56890869140625, "pos_frac": 0.75, "sample": [-0.025106430053710938, 91.66241455078125, 6.947601318359375, 107.29279327392578, 24.40911865234375, 35.502593994140625, 49.48711013793945, 111.34175872802734, 8.095855712890625, -62.984169006347656, 43.71228790283203, 15.162887573242188, 40.18255615234375, -78.73158264160156, 36.634246826171875, 4.743583679199219, 79.32780456542969, -24.417266845703125, 13.387531280517578, 62.71454620361328, 4.903013229370117, 126.86715698242188, -57.79502868652344, 26.07892608642578, -34.81330871582031, 88.21278381347656, 72.51860046386719, 127.56890869140625, 57.63690185546875, 82.3149185180664, -16.865680694580078, -16.894561767578125, 76.59152221679688, 8.560623168945312, 63.551422119140625, -2.147798538208008, 34.770999908447266, -1.9565505981445312, -31.47503662109375, -11.89272689819336, 30.299880981445312, 72.51248168945312, 79.3828125, -42.02910614013672, 63.76768493652344, 9.088508605957031, 112.6120376586914, 113.65667724609375, -46.18567657470703, 13.8232421875, -17.72638702392578, 17.885936737060547, 77.25321960449219, 62.353309631347656, -8.101188659667969, 30.177566528320312, 95.9124755859375, 104.51840209960938, 98.71912384033203, 47.371986389160156, 8.043006896972656, 44.13880920410156, 66.64924621582031, 63.66429901123047], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000228.npy"}
|
||||
{"epoch": 0.34467120181405897, "step": 229, "batch_size": 64, "mean": 29.813419342041016, "std": 56.2115478515625, "min": -106.61679077148438, "p10": -30.956717681884754, "median": 15.72529125213623, "p90": 98.9943946838379, "max": 163.62051391601562, "pos_frac": 0.6875, "sample": [59.5645751953125, 93.78292846679688, 64.7088851928711, 55.5057373046875, 5.450038909912109, 20.80169677734375, 93.753173828125, 33.0250244140625, 36.06501388549805, -43.01068878173828, 78.87065887451172, 68.83984375, 98.89373016357422, 9.159019470214844, 53.052764892578125, 10.648885726928711, -68.1556396484375, -3.8154945373535156, 99.57537078857422, 97.69839477539062, 25.169857025146484, 2.4306201934814453, 49.187095642089844, 35.802452087402344, -72.92964172363281, 5.670558929443359, -35.34471893310547, -3.463113784790039, 10.391569137573242, -8.559089660644531, -11.34613037109375, 163.62051391601562, -106.61679077148438, 87.48443603515625, -8.300704956054688, 9.209165573120117, 99.03753662109375, -73.30726623535156, -2.9551315307617188, 35.398895263671875, 89.6478500366211, 54.41994094848633, -0.27960777282714844, -4.874454498291016, 60.45907974243164, 4.960044860839844, 107.7837142944336, 44.843017578125, 96.20423889160156, 9.553672790527344, -7.615394592285156, -2.793041229248047, 58.69322967529297, -20.718048095703125, 109.3918228149414, 6.5435638427734375, 1.6526718139648438, -7.714818954467773, 100.99374389648438, 148.39901733398438, 93.72173309326172, -6.294425964355469, 4.6988983154296875, -98.61172485351562], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000229.npy"}
|
||||
{"epoch": 0.34618291761148906, "step": 230, "batch_size": 64, "mean": 33.29789733886719, "std": 52.66197204589844, "min": -116.53839111328125, "p10": -4.8437274932861305, "median": 18.306468963623047, "p90": 101.62281494140625, "max": 193.0374755859375, "pos_frac": 0.859375, "sample": [-6.078060150146484, 33.18882751464844, 91.91343688964844, 4.4533843994140625, 17.04492950439453, -116.53839111328125, 6.66673469543457, 5.29364013671875, 13.192214965820312, 113.22230529785156, 23.697402954101562, 106.11087799072266, 106.2119369506836, 101.63084411621094, 19.568008422851562, 42.4395751953125, 45.212425231933594, 99.50791931152344, -39.238990783691406, 4.555938720703125, -114.50138854980469, -2.939056396484375, 54.01470947265625, 60.09628677368164, 70.8501968383789, 6.790252685546875, 101.60408020019531, 10.498489379882812, 93.29800415039062, 80.88288879394531, 62.310020446777344, 7.395397186279297, -2.744932174682617, 7.681772232055664, 3.690673828125, 13.579566955566406, 14.52032470703125, 23.029930114746094, 193.0374755859375, 26.232467651367188, 12.943572998046875, 7.947120666503906, 3.8008155822753906, 41.382652282714844, 141.2454833984375, 73.58634185791016, -36.941322326660156, 101.09632110595703, 22.20404052734375, 2.2838172912597656, 80.10380554199219, -27.81982421875, 90.41822052001953, 42.34392547607422, 30.43265151977539, 7.65740966796875, 5.723442077636719, 5.648496627807617, 26.857603073120117, 1.4879913330078125, 9.097734451293945, 104.68495178222656, -5.660015106201172, 9.158069610595703], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000230.npy"}
|
||||
{"epoch": 0.3476946334089191, "step": 231, "batch_size": 64, "mean": 51.4948844909668, "std": 53.57829666137695, "min": -112.90084838867188, "p10": -7.53708152770996, "median": 59.8993034362793, "p90": 112.04375228881835, "max": 157.07376098632812, "pos_frac": 0.828125, "sample": [112.06996154785156, 21.058181762695312, 80.39251708984375, 108.53832244873047, 9.080604553222656, 23.443374633789062, 16.231731414794922, 105.68280029296875, 28.625244140625, 6.717018127441406, 89.16219329833984, 8.988754272460938, 81.51217651367188, -5.420383453369141, -28.556884765625, 77.55625915527344, 31.282424926757812, 95.36563110351562, 33.61778259277344, -5.122917175292969, 61.19343566894531, -7.860149383544922, 76.5279312133789, 96.81986999511719, 8.98333740234375, 136.9925079345703, 14.664993286132812, -53.8802604675293, 114.0481185913086, 75.71058654785156, 110.9058609008789, -79.42343139648438, 42.67976379394531, 102.82423400878906, 97.80635070800781, -9.207763671875, 9.313119888305664, 65.30744934082031, 19.495208740234375, -6.783256530761719, 129.10252380371094, 5.133430480957031, 63.5594482421875, 78.73492431640625, 58.60517120361328, 37.763038635253906, 87.86959075927734, -17.882495880126953, 22.400012969970703, -3.9345703125, 101.89146423339844, 157.07376098632812, 58.1881103515625, 64.83660125732422, 89.7590560913086, 111.98259735107422, 55.385498046875, 80.87137603759766, 58.34527587890625, 115.48973083496094, -112.90084838867188, 68.66603088378906, 103.398193359375, 114.991943359375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000231.npy"}
|
||||
{"epoch": 0.3492063492063492, "step": 232, "batch_size": 64, "mean": 34.0102424621582, "std": 63.383724212646484, "min": -169.42715454101562, "p10": -25.772142410278317, "median": 19.979416847229004, "p90": 112.27208251953125, "max": 209.36856079101562, "pos_frac": 0.703125, "sample": [3.102508544921875, 102.95352172851562, 68.23987579345703, 126.87035369873047, 9.049724578857422, 57.4306640625, 29.07054901123047, 107.1575698852539, -4.6405029296875, -30.318374633789062, -27.635303497314453, 8.1063232421875, 3.4175262451171875, 25.74734115600586, 49.45106887817383, 109.98497009277344, -0.08470916748046875, -11.957344055175781, 67.49827575683594, 13.529964447021484, -11.596542358398438, 61.34056854248047, 82.88385009765625, 37.925201416015625, 15.816268920898438, 76.40966033935547, -12.686464309692383, 113.25227355957031, -17.771305084228516, 22.26280975341797, 62.44488525390625, 119.44987487792969, -21.424766540527344, -3.673440933227539, 40.897125244140625, 6.6588592529296875, 2.134279251098633, -74.6156997680664, 128.1095733642578, 41.866058349609375, -1.3214683532714844, 136.8770751953125, -21.21623992919922, 206.56634521484375, 104.89448547363281, 209.36856079101562, -13.817588806152344, -36.98987579345703, -13.900863647460938, 17.69602394104004, 24.767822265625, 104.25646209716797, 77.99960327148438, 9.035652160644531, 5.121593475341797, -51.48179626464844, 1.2823715209960938, 109.97525024414062, 12.744155883789062, 55.76931381225586, -30.14678192138672, -169.42715454101562, 26.728973388671875, 35.216552734375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000232.npy"}
|
||||
{"epoch": 0.3507180650037793, "step": 233, "batch_size": 64, "mean": 25.215673446655273, "std": 54.04281234741211, "min": -99.10405731201172, "p10": -34.15526866912841, "median": 7.195522308349609, "p90": 99.52672882080078, "max": 144.60435485839844, "pos_frac": 0.640625, "sample": [131.94253540039062, -7.4260711669921875, 97.80683898925781, -52.56887435913086, 13.125965118408203, 50.41387176513672, 39.651947021484375, 10.838191986083984, -16.82834243774414, -21.861305236816406, 11.79510498046875, 144.60435485839844, -4.829265594482422, -1.2565193176269531, -1.5059890747070312, 32.56446838378906, 83.42140197753906, 6.935632705688477, 1.2381057739257812, 7.005790710449219, 117.56367492675781, 7.159736633300781, 93.73294067382812, 100.26382446289062, -37.600555419921875, 33.36968231201172, -12.307022094726562, 62.743797302246094, -52.30253601074219, -43.85493469238281, 88.14318084716797, 2.1087188720703125, 94.6884765625, 57.09335708618164, 74.76127624511719, -99.10405731201172, 28.052635192871094, 5.399438858032227, -4.244415283203125, -1.5092277526855469, 16.878341674804688, -73.57646942138672, 41.49931335449219, -3.0279998779296875, 61.569740295410156, -14.213630676269531, 1.1927413940429688, 37.928192138671875, 86.18098449707031, 78.6470947265625, -21.09952163696289, -26.11626625061035, 2.148653030395508, 121.5098648071289, 71.27102661132812, 127.34342956542969, 5.7900543212890625, -20.00790786743164, 112.74305725097656, 7.2313079833984375, 44.83965301513672, -7.716804504394531, -64.40662384033203, -12.030967712402344], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000233.npy"}
|
||||
{"epoch": 0.35222978080120937, "step": 234, "batch_size": 64, "mean": 24.559268951416016, "std": 56.72664260864258, "min": -83.68434143066406, "p10": -53.613772201538076, "median": 11.492578506469727, "p90": 101.64917755126955, "max": 126.83247375488281, "pos_frac": 0.671875, "sample": [-9.296058654785156, -64.88665771484375, -4.946441650390625, -72.93704223632812, 61.286109924316406, -4.248729705810547, 4.378591537475586, -28.89129638671875, 88.80525207519531, -24.777862548828125, 38.489349365234375, 53.23810958862305, 90.29370880126953, 25.85362434387207, 117.54507446289062, 89.19058227539062, -43.48130798339844, -68.41099548339844, 49.75732421875, -13.853437423706055, 8.486289978027344, 0.5962295532226562, 98.45854949951172, 126.83247375488281, -0.9664516448974609, -7.641937255859375, 99.18807983398438, 4.645109176635742, -72.8047103881836, 11.209602355957031, 94.10958862304688, 119.458740234375, 108.02518463134766, 94.91767883300781, -4.551979064941406, -83.68434143066406, 0.76715087890625, 4.121337890625, 6.346357345581055, -56.884586334228516, 102.70393371582031, 38.54015350341797, 21.589946746826172, 66.30854797363281, 107.2132339477539, 14.064229965209961, 10.693367004394531, 55.84812927246094, 98.25140380859375, -30.987436294555664, -10.9854736328125, 60.620208740234375, -42.24046325683594, 32.67011642456055, -77.26644897460938, 4.300304412841797, 62.072364807128906, 11.194229125976562, 26.979209899902344, 11.775554656982422, 26.205307006835938, 115.68511199951172, 78.80329895019531, -45.98187255859375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000234.npy"}
|
||||
{"epoch": 0.35374149659863946, "step": 235, "batch_size": 64, "mean": 43.68671417236328, "std": 50.77742004394531, "min": -77.20268249511719, "p10": -11.920769500732419, "median": 34.543495178222656, "p90": 109.47724151611328, "max": 136.6314697265625, "pos_frac": 0.828125, "sample": [3.4006805419921875, 33.17779541015625, 22.579635620117188, 36.70085144042969, 80.34013366699219, 32.372718811035156, 17.039779663085938, 136.6314697265625, 131.55575561523438, 72.70027160644531, 6.7178192138671875, 95.97188568115234, 23.788619995117188, 35.90919494628906, 95.97416687011719, 94.61444091796875, 55.24061584472656, 17.38619613647461, 96.68141174316406, 110.15669250488281, 8.54388427734375, 17.657119750976562, -7.940948486328125, 108.01162719726562, 107.5618667602539, 94.20895385742188, -10.028587341308594, -16.24237823486328, 110.10536193847656, -77.20268249511719, 48.882293701171875, 3.4666500091552734, 11.41966438293457, 14.99749755859375, 112.88662719726562, 40.63092041015625, 29.401004791259766, 12.965347290039062, 76.06071472167969, -40.6441650390625, -36.181488037109375, 63.615753173828125, 87.72738647460938, 17.123584747314453, 73.68882751464844, 8.773067474365234, 1.0613327026367188, 102.09078979492188, 133.93431091308594, 17.03543472290039, 42.40253448486328, -12.731704711914062, 116.73857116699219, 59.498069763183594, 0.1820526123046875, -65.67147827148438, 85.42684173583984, 101.54817199707031, 103.66566467285156, 6.363304138183594, 68.54888916015625, -18.628307342529297, -3.4133148193359375, -0.5294151306152344], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000235.npy"}
|
||||
{"epoch": 0.35525321239606955, "step": 236, "batch_size": 64, "mean": 32.58956527709961, "std": 59.88448715209961, "min": -145.08352661132812, "p10": -15.19380645751953, "median": 20.633307456970215, "p90": 110.04649353027344, "max": 165.1830596923828, "pos_frac": 0.6875, "sample": [96.48428344726562, -12.655418395996094, 3.2690887451171875, 87.54042053222656, 11.724164962768555, 18.025894165039062, 100.43338775634766, 36.53990173339844, -19.266220092773438, 26.976852416992188, 52.819732666015625, 37.21875, -6.811788558959961, -0.2239837646484375, -7.70362663269043, -2.8443527221679688, 79.67317199707031, 80.58401489257812, 60.48371887207031, 109.84910583496094, -2.5215606689453125, -15.770492553710938, 21.59210205078125, -0.3459053039550781, -0.21466827392578125, 24.35064697265625, 74.76596069335938, 95.64739990234375, 110.13108825683594, 66.71504211425781, 107.48910522460938, -13.84820556640625, 8.541336059570312, 11.970027923583984, 115.37489318847656, -36.51319885253906, 0.8431549072265625, 165.1830596923828, 32.20918655395508, 3.830137252807617, -83.42059326171875, 42.1778450012207, 95.81820678710938, 120.66918182373047, 54.133209228515625, 62.0999755859375, 3.7985992431640625, -38.988609313964844, 115.26536560058594, 0.8422431945800781, 6.611209869384766, -145.08352661132812, -139.52256774902344, 115.60963439941406, 108.85394287109375, 116.63140106201172, 27.6632080078125, -7.855756759643555, -13.289375305175781, -9.228496551513672, 109.19451141357422, -4.171195983886719, 19.67451286315918, 6.703125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000236.npy"}
|
||||
{"epoch": 0.35676492819349964, "step": 237, "batch_size": 64, "mean": 40.178955078125, "std": 60.73076248168945, "min": -138.94985961914062, "p10": -18.503186035156247, "median": 40.74916458129883, "p90": 113.37568817138673, "max": 172.69607543945312, "pos_frac": 0.765625, "sample": [114.0918960571289, 90.3868179321289, 44.866851806640625, 116.84276580810547, -3.6516265869140625, -28.795120239257812, 18.520263671875, 31.17058563232422, 7.917736053466797, 101.02086639404297, 68.99772644042969, 51.509002685546875, -13.96514892578125, -13.7486572265625, 105.16984558105469, 11.338920593261719, -4.726139068603516, 54.74779510498047, -20.44805908203125, 83.58518981933594, -100.82463073730469, 84.65338134765625, 52.45954895019531, 71.185791015625, 153.26319885253906, -7.740367889404297, 77.58856201171875, 78.11894989013672, -31.39697265625, 26.939022064208984, 110.36182403564453, 28.566505432128906, 5.003730773925781, -138.94985961914062, 82.56375885009766, 1.2593803405761719, 6.746116638183594, -106.77698516845703, 172.69607543945312, 31.128679275512695, 43.25409698486328, 43.799461364746094, 15.003768920898438, 60.35835266113281, 4.506263732910156, 0.9980831146240234, 77.85760498046875, 117.16017150878906, 38.244232177734375, 98.1079330444336, 111.70453643798828, 2.8155059814453125, 105.06982421875, 150.76756286621094, -12.828338623046875, 12.149700164794922, -4.555803298950195, 101.76082611083984, 8.465415954589844, 56.804725646972656, -12.552310943603516, 49.16533660888672, -28.89533233642578, 120.61418151855469], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000237.npy"}
|
||||
{"epoch": 0.35827664399092973, "step": 238, "batch_size": 64, "mean": 38.09858322143555, "std": 56.91424560546875, "min": -87.19467163085938, "p10": -35.447867202758786, "median": 36.90169334411621, "p90": 113.80192108154299, "max": 151.01499938964844, "pos_frac": 0.734375, "sample": [-36.799198150634766, 15.737556457519531, 86.06294250488281, -32.294761657714844, 96.51956176757812, 100.27621459960938, 7.327373504638672, 16.988815307617188, 124.66492462158203, -60.64586639404297, 28.45885467529297, -20.191268920898438, 110.05992889404297, 11.866386413574219, 76.62593078613281, 45.34453201293945, -1.8261985778808594, -17.039670944213867, 55.644561767578125, 115.64969635009766, 73.71356964111328, 115.90876770019531, 92.6967544555664, 84.34404754638672, 5.258174896240234, 104.49241638183594, 23.47551727294922, -74.38179779052734, -50.880699157714844, 85.24394226074219, 62.194610595703125, 81.822021484375, 5.087224960327148, 108.40177154541016, -12.21822738647461, 19.789310455322266, 90.59494018554688, -0.5188369750976562, 121.55860137939453, 78.21629333496094, 5.408451080322266, 6.668357849121094, -3.442859649658203, 70.88740539550781, 80.1692886352539, 56.05384826660156, 151.01499938964844, 60.298667907714844, 48.298988342285156, 75.32980346679688, 18.96868896484375, 62.38048553466797, 115.40563201904297, -30.700428009033203, -11.082189559936523, 2.3834495544433594, -87.19467163085938, 17.584747314453125, 120.25652313232422, -56.19972229003906, 60.69580078125, -43.85652160644531, -25.551883697509766, 7.303796768188477], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000238.npy"}
|
||||
{"epoch": 0.35978835978835977, "step": 239, "batch_size": 64, "mean": 39.907203674316406, "std": 63.09526443481445, "min": -102.06304931640625, "p10": -31.01906661987304, "median": 36.944915771484375, "p90": 116.72872695922852, "max": 157.34774780273438, "pos_frac": 0.765625, "sample": [26.227081298828125, -2.7554779052734375, 87.73365783691406, -4.026325225830078, 142.35951232910156, 0.93536376953125, -11.00115966796875, 6.062599182128906, -26.176589965820312, 41.595191955566406, -33.09441375732422, 42.397369384765625, 13.245922088623047, 122.37727355957031, 93.63038635253906, 27.396270751953125, -65.29904174804688, -82.64920806884766, 45.89813232421875, 140.96563720703125, 2.8569717407226562, 23.297889709472656, 111.09617614746094, 114.40877532958984, 106.29808044433594, 45.13499450683594, 72.56452941894531, 70.15275573730469, 60.770843505859375, 2.5657958984375, 109.18125915527344, 69.68376159667969, -36.191444396972656, -21.71590232849121, 40.888206481933594, 108.50811767578125, -5.012504577636719, 18.815832138061523, 11.689178466796875, 71.90266418457031, -3.0638179779052734, 34.95372009277344, 26.164283752441406, 46.53749084472656, -98.86238861083984, 51.26117706298828, 157.34774780273438, 154.1011962890625, 154.14564514160156, -99.44140625, 0.1864299774169922, 8.728866577148438, -102.06304931640625, 105.85147094726562, 109.0244140625, 111.68440246582031, 117.72299194335938, -7.177135467529297, 38.93611145019531, 92.2098388671875, 1.6777076721191406, 64.53710174560547, 18.93793487548828, 27.94226837158203], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000239.npy"}
|
||||
{"epoch": 0.36130007558578986, "step": 240, "batch_size": 64, "mean": 49.933414459228516, "std": 62.648502349853516, "min": -115.35759735107422, "p10": -17.899212265014643, "median": 41.572021484375, "p90": 128.30041656494143, "max": 176.13723754882812, "pos_frac": 0.765625, "sample": [79.40615844726562, -21.093170166015625, 176.13723754882812, -3.918701171875, 92.85638427734375, 167.30023193359375, 59.24114990234375, 59.39366912841797, -49.221107482910156, -50.85863494873047, -13.513923645019531, -5.206571578979492, 101.92779541015625, 20.56427574157715, 72.43840026855469, 105.12029266357422, 105.20083618164062, -32.655426025390625, 122.9035873413086, 42.94920349121094, -12.890445709228516, 118.38143920898438, 36.19839859008789, -58.42455291748047, 49.530181884765625, 39.971412658691406, 59.15678405761719, 124.21089172363281, 151.4362030029297, 16.457740783691406, 116.68248748779297, 161.76858520507812, 11.421646118164062, 34.461090087890625, 6.75567626953125, 91.30892944335938, 49.94446563720703, 132.45516967773438, 116.1005859375, 6.57493782043457, 102.2608871459961, 14.444583892822266, -19.778621673583984, 40.47313690185547, 130.05307006835938, -115.35759735107422, -1.1673755645751953, -1.7286605834960938, 68.2091293334961, -2.79205322265625, 20.32455825805664, 121.3599853515625, -13.099937438964844, 15.686447143554688, 5.556835174560547, 11.819206237792969, 123.29949951171875, 142.85043334960938, 89.80475616455078, 42.67090606689453, 101.22296905517578, 17.021209716796875, 1.2298202514648438, 20.902008056640625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000240.npy"}
|
||||
{"epoch": 0.36281179138321995, "step": 241, "batch_size": 64, "mean": 33.55549621582031, "std": 62.29868698120117, "min": -115.36434173583984, "p10": -34.657321166992176, "median": 18.39470672607422, "p90": 116.22791137695313, "max": 165.73709106445312, "pos_frac": 0.6875, "sample": [0.41851806640625, -38.38361358642578, 59.79480743408203, -51.06947326660156, 94.48200225830078, -15.720611572265625, 16.83513641357422, 80.86121368408203, 108.25308990478516, 116.73146057128906, -2.4564971923828125, 80.41720581054688, -11.645084381103516, 36.2674560546875, 33.50981903076172, 84.78816223144531, 2.3585777282714844, 6.413610458374023, 17.20482635498047, 109.09291076660156, 63.09516906738281, 71.51992797851562, -9.754898071289062, -8.591041564941406, 60.24890899658203, -6.143548965454102, 75.70155334472656, -15.059106826782227, -43.865623474121094, -25.96263885498047, 120.508544921875, 158.71942138671875, 139.6068878173828, 38.609336853027344, -5.740997314453125, -80.05810546875, -15.876083374023438, 42.37211608886719, 55.15997314453125, 1.9034919738769531, -115.36434173583984, 7.279731750488281, 0.1970195770263672, 136.22906494140625, 9.311151504516602, 19.58458709716797, -103.22811889648438, 154.2354736328125, 27.190017700195312, -62.83233642578125, 165.73709106445312, 7.8236846923828125, 67.23092651367188, 36.5863037109375, 115.05296325683594, -13.265758514404297, 14.256278991699219, 112.64939880371094, 94.59925079345703, 67.66624450683594, 8.40245246887207, 66.00057983398438, -8.340463638305664, -3.9961299896240234], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000241.npy"}
|
||||
{"epoch": 0.36432350718065004, "step": 242, "batch_size": 64, "mean": 41.28602600097656, "std": 61.97880554199219, "min": -116.31026458740234, "p10": -13.30428009033203, "median": 25.347713470458984, "p90": 122.36584243774413, "max": 210.38897705078125, "pos_frac": 0.734375, "sample": [-89.08509063720703, 24.83324432373047, 54.20470428466797, 8.362712860107422, 32.212242126464844, -29.119495391845703, 134.63986206054688, 85.60722351074219, 25.456954956054688, 58.10773468017578, 91.4134521484375, 88.55963134765625, 1.5510807037353516, 147.42733764648438, 101.52777099609375, -13.474380493164062, 119.88916015625, -12.214752197265625, -0.46630096435546875, -16.501819610595703, 42.31116485595703, 25.23847198486328, 122.51358795166016, 7.963521957397461, -1.7473373413085938, 31.28607177734375, 24.6207275390625, -12.907379150390625, 106.33663940429688, 41.44841003417969, 107.14984130859375, 0.4262828826904297, 145.78273010253906, 14.285726547241211, 104.93180847167969, 9.278861999511719, -2.1928348541259766, 1.3409156799316406, 210.38897705078125, 40.58005905151367, -31.245010375976562, 70.54679107666016, 85.60821533203125, 2.976856231689453, 71.62650299072266, 3.0517120361328125, 90.02288055419922, -4.836601257324219, 125.16624450683594, 106.09672546386719, 122.02110290527344, 58.996116638183594, 158.5291748046875, -3.955249786376953, -1.373666763305664, 61.012779235839844, 80.28294372558594, -11.70829963684082, 5.515968322753906, -116.31026458740234, 5.062416076660156, -10.204051971435547, -72.80326080322266, 16.25823402404785], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000242.npy"}
|
||||
{"epoch": 0.36583522297808013, "step": 243, "batch_size": 64, "mean": 38.238338470458984, "std": 58.26858139038086, "min": -118.71485900878906, "p10": -26.780060577392565, "median": 29.001953125, "p90": 117.2769401550293, "max": 139.62246704101562, "pos_frac": 0.75, "sample": [7.784889221191406, 87.1876220703125, 78.03290557861328, -36.25323486328125, 79.55068969726562, 4.124361038208008, -118.71485900878906, -1.9570846557617188, 117.00092315673828, -5.261772155761719, 69.92619323730469, 49.478233337402344, 16.67919921875, 48.361541748046875, 20.015029907226562, 120.55074310302734, 1.2103691101074219, 7.083900451660156, 117.91134643554688, -6.7698974609375, -4.6019744873046875, 131.8483123779297, 68.2191390991211, -6.192588806152344, 93.41824340820312, 67.570068359375, 23.10645294189453, 88.32994079589844, 112.91728210449219, 46.16283416748047, -94.30633544921875, -56.21806716918945, 84.85922241210938, -8.293367385864258, 112.1689224243164, 51.12889099121094, 111.13743591308594, 109.15103912353516, -71.10696411132812, 28.36700439453125, 139.62246704101562, 71.41838836669922, 32.68086242675781, 124.40752410888672, 13.599197387695312, -1.4774971008300781, 0.17973899841308594, 1.807058334350586, 118.65571594238281, 70.05545806884766, -13.132949829101562, 52.24809265136719, 14.318893432617188, 106.13207244873047, -55.61708450317383, 29.63690185546875, 67.98831176757812, 15.052310943603516, 0.4538002014160156, -13.054214477539062, -32.628822326660156, 117.39523315429688, 19.143585205078125, 24.76202392578125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000243.npy"}
|
||||
{"epoch": 0.3673469387755102, "step": 244, "batch_size": 64, "mean": 49.40470504760742, "std": 57.875938415527344, "min": -94.4581069946289, "p10": -6.505328369140623, "median": 34.98782730102539, "p90": 125.34260787963868, "max": 148.17630004882812, "pos_frac": 0.8125, "sample": [21.87609100341797, 82.00049591064453, 44.79371643066406, 6.881797790527344, 5.865966796875, 39.938514709472656, -3.170886993408203, 112.13353729248047, 53.30833435058594, -36.839019775390625, 32.75093460083008, 103.34687805175781, -3.34625244140625, 125.61177062988281, 74.39014434814453, -22.064971923828125, 106.65483856201172, 119.78075408935547, -94.4581069946289, 119.50135803222656, 44.341766357421875, 12.096916198730469, -75.0867919921875, 86.66100311279297, 15.0888671875, 120.25496673583984, 87.19451141357422, 98.21759033203125, 128.0032958984375, 14.239786148071289, 134.18136596679688, 70.14342498779297, -2.158121109008789, -4.860527038574219, -18.4532470703125, 20.96341896057129, -0.7374420166015625, 114.23993682861328, 92.74459838867188, 141.11618041992188, 1.7495384216308594, 141.04281616210938, 1.2269611358642578, 4.8685760498046875, 3.8776702880859375, 115.2806396484375, 35.45043182373047, 97.5029296875, 5.303077697753906, 23.52492904663086, 4.850629806518555, -7.210243225097656, 116.41658020019531, 46.39549255371094, 140.46910095214844, 34.52522277832031, 0.5070114135742188, 13.1038818359375, 16.34566879272461, -8.313224792480469, 105.84913635253906, 29.095949172973633, 124.71456146240234, 148.17630004882812], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000244.npy"}
|
||||
{"epoch": 0.3688586545729403, "step": 245, "batch_size": 64, "mean": 57.16092300415039, "std": 59.047847747802734, "min": -103.31878662109375, "p10": -7.566907501220703, "median": 62.14887809753418, "p90": 128.8403923034668, "max": 149.798095703125, "pos_frac": 0.859375, "sample": [82.16228485107422, 92.23373413085938, 102.09707641601562, 117.76991271972656, 13.803571701049805, 11.628948211669922, 1.9788398742675781, 125.12786102294922, 95.58863067626953, 61.911170959472656, 1.2263946533203125, 54.5989990234375, 113.32247161865234, 28.687030792236328, 35.93748474121094, 115.80821990966797, 6.310630798339844, 118.03963470458984, -51.23834228515625, 139.55374145507812, 135.85787963867188, 5.129600524902344, 2.414337158203125, 75.44863891601562, 121.66439819335938, 44.176727294921875, 82.31239318847656, -12.470855712890625, 103.57902526855469, 4.18168830871582, 137.209228515625, -95.89762878417969, 2.8133163452148438, 83.87815856933594, -7.865715026855469, 90.69929504394531, 11.273056030273438, 63.05620574951172, 99.41567993164062, 146.64230346679688, 51.745574951171875, 25.535629272460938, 102.77242279052734, 2.6887130737304688, -8.863252639770508, 81.19007873535156, 126.91240692138672, 115.41790771484375, 0.1135101318359375, -33.501007080078125, 59.27130126953125, 104.50163269042969, -6.86968994140625, 80.97044372558594, 140.82977294921875, 75.3382568359375, -103.31878662109375, 129.6666717529297, 53.33529281616211, 26.684600830078125, -0.2170257568359375, 149.798095703125, 61.84366989135742, 62.3865852355957], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000245.npy"}
|
||||
{"epoch": 0.37037037037037035, "step": 246, "batch_size": 64, "mean": 41.97944641113281, "std": 73.75279235839844, "min": -136.96847534179688, "p10": -30.063172721862784, "median": 32.57831573486328, "p90": 126.404150390625, "max": 227.46676635742188, "pos_frac": 0.703125, "sample": [109.35357666015625, 116.25569152832031, 124.02265930175781, 118.51481628417969, 9.286407470703125, 12.902080535888672, -112.19126892089844, 135.49253845214844, 12.764808654785156, -20.484888076782227, 146.4949951171875, 120.37541198730469, 62.43544006347656, -10.126005172729492, 53.70738220214844, 111.26898193359375, 10.659355163574219, -9.128080368041992, -12.734565734863281, -0.0470428466796875, 47.514312744140625, 72.05106353759766, -0.9442825317382812, 127.42478942871094, 142.49050903320312, -84.00456237792969, 29.623519897460938, 227.46676635742188, -34.16815185546875, 15.104537963867188, 3.783395767211914, 121.3966064453125, 16.12261962890625, -2.9308109283447266, -81.33477783203125, 89.66326904296875, -104.59026336669922, 12.288793563842773, 114.98751068115234, 3.701751708984375, 72.13475036621094, -109.548095703125, -5.751274108886719, 133.08187866210938, 40.608333587646484, -9.65966796875, 102.54963684082031, 102.50534057617188, 118.7205810546875, -136.96847534179688, -7.5197906494140625, 44.15459442138672, 104.14494323730469, 143.32040405273438, 8.381832122802734, -16.9329833984375, 106.13970184326172, 106.26742553710938, -12.57501220703125, 35.533111572265625, 37.70026397705078, 22.132644653320312, 95.0282211303711, 18.767120361328125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000246.npy"}
|
||||
{"epoch": 0.37188208616780044, "step": 247, "batch_size": 64, "mean": 28.213390350341797, "std": 68.90077209472656, "min": -131.64981079101562, "p10": -56.60730209350585, "median": 20.762123107910156, "p90": 110.96814041137698, "max": 221.2738037109375, "pos_frac": 0.703125, "sample": [-102.33248901367188, 113.45133209228516, 77.94564819335938, 62.52044677734375, 1.8690147399902344, -19.369873046875, 15.631011962890625, 129.17007446289062, 119.31509399414062, -7.1771087646484375, 101.35755920410156, 20.9571533203125, 104.95226287841797, 27.57517433166504, 38.40986633300781, 161.33670043945312, 105.17402648925781, 25.560626983642578, -8.412837982177734, -131.64981079101562, 221.2738037109375, 132.5880126953125, -114.4759521484375, -14.111173629760742, 22.241668701171875, 21.46649169921875, 76.96522521972656, -4.950340270996094, -72.96910095214844, 92.8155288696289, -3.251739501953125, 8.605056762695312, 85.11459350585938, 12.511444091796875, 44.64051818847656, 13.634811401367188, 57.44209289550781, -116.53321838378906, -48.0611572265625, 5.942842483520508, -3.85235595703125, 14.82701301574707, 2.4423370361328125, 117.40129852294922, 65.21322631835938, 14.564167022705078, 5.916469573974609, -60.269935607910156, 33.17802810668945, 73.75326538085938, 20.567092895507812, 64.57100677490234, -39.40806579589844, 17.087495803833008, -93.14856719970703, 86.0036392211914, 91.70860290527344, 95.87957000732422, -27.164993286132812, -40.93913269042969, -24.214508056640625, 91.60469055175781, 26.11395263671875, 16.649282455444336], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000247.npy"}
|
||||
{"epoch": 0.37339380196523053, "step": 248, "batch_size": 64, "mean": 45.89332580566406, "std": 70.25704193115234, "min": -94.70526885986328, "p10": -45.51676406860351, "median": 30.216567993164062, "p90": 127.48515014648439, "max": 172.51397705078125, "pos_frac": 0.75, "sample": [7.45751953125, 117.64073181152344, -48.16618347167969, -83.81912231445312, 135.12301635742188, 102.7288818359375, 120.37712097167969, 108.13008880615234, -0.2631244659423828, -31.622861862182617, 2.1220130920410156, 121.05630493164062, 49.39884948730469, -39.33478546142578, 117.07681274414062, 76.11856079101562, 172.51397705078125, -36.40929412841797, -73.72584533691406, 28.80982208251953, -25.70299530029297, 14.987960815429688, 30.423568725585938, 15.602670669555664, 114.5185546875, 108.3890380859375, 116.60749816894531, 18.470069885253906, 83.94209289550781, 135.4901885986328, -94.70526885986328, 126.39862060546875, -8.526145935058594, -8.404298782348633, 127.9508056640625, 121.29808044433594, 30.009567260742188, 36.707359313964844, 47.340370178222656, 41.28326416015625, 108.14422607421875, 105.1836166381836, 16.29673194885254, 147.4580535888672, 143.9495391845703, 87.76766967773438, 11.916130065917969, -1.5726356506347656, 7.54094123840332, 106.65581512451172, 167.1800537109375, 9.24588394165039, 120.53202819824219, 4.362663269042969, -69.70304107666016, 8.607011795043945, -63.55841064453125, 126.11351013183594, 90.6162338256836, 7.754524230957031, 10.445650100708008, 14.339103698730469, -77.88168334960938, -21.51446533203125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000248.npy"}
|
||||
{"epoch": 0.3749055177626606, "step": 249, "batch_size": 64, "mean": 16.519784927368164, "std": 71.1880874633789, "min": -154.36094665527344, "p10": -79.58114852905273, "median": 12.778121948242188, "p90": 109.0370834350586, "max": 139.00668334960938, "pos_frac": 0.65625, "sample": [5.014595031738281, -64.06571960449219, 11.477279663085938, 56.3843994140625, 49.680110931396484, -59.19288635253906, -106.44204711914062, 96.8710708618164, 107.75588989257812, 29.515060424804688, 67.64468383789062, 92.1212158203125, 14.078964233398438, 100.53120422363281, -124.22869873046875, -10.575546264648438, 85.04296875, -44.73272705078125, -62.59953308105469, -73.99640655517578, 8.119508743286133, -5.3115386962890625, -88.40656280517578, 3.891773223876953, 63.72611999511719, 6.453269958496094, 2.071084976196289, -12.670730590820312, 120.29367065429688, -31.026763916015625, -19.329910278320312, 115.27484893798828, 0.5020885467529297, 38.589271545410156, 73.63433837890625, -148.27899169921875, 31.18305206298828, -5.0899810791015625, 20.059221267700195, -81.974609375, 7.58745002746582, 128.3860321044922, 83.21963500976562, 129.24319458007812, 48.08345031738281, 5.372203826904297, -112.04685974121094, 109.58616638183594, 24.52690887451172, 15.778167724609375, 35.59114074707031, 121.49143981933594, 72.02423095703125, -0.5564498901367188, 15.006929397583008, 46.02687072753906, -26.00541114807129, 84.01362609863281, 139.00668334960938, -154.36094665527344, 107.2427978515625, -71.6689453125, -17.68426513671875, 5.4091949462890625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000249.npy"}
|
||||
{"epoch": 0.3764172335600907, "step": 250, "batch_size": 64, "mean": 46.814453125, "std": 67.64754486083984, "min": -148.81861877441406, "p10": -22.408257293701162, "median": 36.4783992767334, "p90": 127.80384216308595, "max": 214.54962158203125, "pos_frac": 0.796875, "sample": [33.153038024902344, 9.397298812866211, 2.297403335571289, 6.207330703735352, 24.470703125, 50.65269470214844, 81.05043029785156, 147.2405242919922, 113.14683532714844, -11.306045532226562, 119.83505249023438, -11.359657287597656, 136.68502807617188, 30.05865478515625, 107.01182556152344, 21.032686233520508, -27.14337158203125, -148.81861877441406, 108.96755981445312, 37.29247283935547, 43.68690490722656, -7.876386642456055, 15.518169403076172, -90.51177215576172, 110.45427703857422, 66.17855834960938, 128.8855743408203, 51.578453063964844, -3.933624267578125, 69.42950439453125, -63.7111701965332, 79.31756591796875, 125.27980041503906, 5.700168609619141, 43.78999328613281, 28.58629608154297, 2.1743545532226562, 110.87307739257812, 149.52047729492188, 24.760894775390625, -76.37653350830078, 0.11974716186523438, 40.8939323425293, 2.5087051391601562, 179.83506774902344, 107.93698120117188, 115.24124145507812, 108.88423156738281, 59.913875579833984, 114.40296936035156, -40.02357482910156, -46.626548767089844, 136.2117156982422, 9.071807861328125, 18.133132934570312, 18.04187774658203, 214.54962158203125, 97.46260833740234, 35.66432571411133, -0.4811134338378906, -1.5106925964355469, 116.81383514404297, 58.143821716308594, 7.7409515380859375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000250.npy"}
|
||||
{"epoch": 0.3779289493575208, "step": 251, "batch_size": 64, "mean": 31.505043029785156, "std": 64.85224151611328, "min": -117.41549682617188, "p10": -57.736050033569335, "median": 32.11330795288086, "p90": 118.26225051879884, "max": 162.27890014648438, "pos_frac": 0.71875, "sample": [2.5136165618896484, 121.81224060058594, 123.0542984008789, -0.09465789794921875, 33.36265563964844, 86.34657287597656, -117.41549682617188, 32.99251174926758, 96.49927520751953, 160.11654663085938, 7.801765441894531, -108.88610076904297, 25.404220581054688, -17.923377990722656, 52.46759796142578, -60.36781692504883, 76.98777770996094, 162.27890014648438, 13.477773666381836, -102.01142883300781, -45.191627502441406, 137.14019775390625, 6.25920295715332, -38.86073303222656, 19.480484008789062, -59.30840301513672, 24.39020538330078, 51.75621032714844, -40.87065887451172, -4.9897613525390625, 100.97734069824219, 113.98326873779297, 31.23410415649414, 50.84864807128906, 75.8466567993164, 44.885536193847656, 100.13631439208984, 73.64765930175781, -54.06722640991211, -0.3005790710449219, 56.09434127807617, 0.536407470703125, 119.17748260498047, 17.414569854736328, 28.60308837890625, 49.40220642089844, 51.449310302734375, 11.364898681640625, 78.10298156738281, 79.14376831054688, 121.19979858398438, 4.6624603271484375, 92.20101928710938, -40.83185577392578, 3.6139144897460938, -3.6641101837158203, 60.09469985961914, -60.699371337890625, 35.45903778076172, -29.445533752441406, -77.19539642333984, 63.06201171875, 116.126708984375, 65.03657531738281], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000251.npy"}
|
||||
{"epoch": 0.3794406651549509, "step": 252, "batch_size": 64, "mean": 24.62472152709961, "std": 71.62081146240234, "min": -150.31369018554688, "p10": -87.57611007690429, "median": 20.321937561035156, "p90": 112.73354339599611, "max": 169.19419860839844, "pos_frac": 0.71875, "sample": [53.61450958251953, -106.24760437011719, -15.069360733032227, -150.31369018554688, 52.74800109863281, 24.815765380859375, 31.19546890258789, 108.29609680175781, 0.0492095947265625, 125.7077407836914, 108.30374908447266, 99.60401916503906, 32.07640838623047, -0.45325469970703125, 137.7843017578125, -10.191642761230469, 137.53549194335938, -59.49036407470703, 39.12944793701172, -0.9569911956787109, -101.48446655273438, 15.102853775024414, 58.46089172363281, 17.514663696289062, 12.3206787109375, 14.820602416992188, 20.889419555664062, -11.395675659179688, 107.17488098144531, 1.1144638061523438, 86.52603912353516, 169.19419860839844, 105.14128112792969, 19.75445556640625, 32.980873107910156, 3.0853309631347656, 72.87335205078125, 6.378660202026367, 62.06146240234375, -71.88294982910156, 47.14094161987305, -56.84685516357422, 0.2459869384765625, 103.02769470214844, -19.267547607421875, 159.55227661132812, -94.30175018310547, 30.22919464111328, -136.74591064453125, -106.64822387695312, -19.90315818786621, 31.91766357421875, 66.39796447753906, -100.94635009765625, 9.903846740722656, 127.2936019897461, 15.914840698242188, 114.63202667236328, 17.683639526367188, 37.394012451171875, 101.605224609375, 37.60455322265625, -38.02013397216797, 19.35012435913086], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000252.npy"}
|
||||
{"epoch": 0.38095238095238093, "step": 253, "batch_size": 64, "mean": 34.85858917236328, "std": 60.91413497924805, "min": -109.66841888427734, "p10": -33.25462341308592, "median": 24.445133209228516, "p90": 114.52665100097657, "max": 160.17401123046875, "pos_frac": 0.71875, "sample": [106.5207748413086, 5.0862274169921875, 130.63221740722656, 9.809089660644531, -1.3095016479492188, 16.40969467163086, 35.92115783691406, 120.55673217773438, 104.23431396484375, 46.30609130859375, 77.35154724121094, 5.775215148925781, 110.03160095214844, 6.684288024902344, -1.653341293334961, -95.30001831054688, -16.21380615234375, 64.49822998046875, -12.099800109863281, 75.09123229980469, 92.07310485839844, 81.27651977539062, -99.55718994140625, 88.02189636230469, 11.582536697387695, -12.393951416015625, 113.87448120117188, 25.719497680664062, 23.17076873779297, 60.15379333496094, 36.22618103027344, 26.41779327392578, 114.80615234375, -2.9313011169433594, 31.783905029296875, 160.17401123046875, 31.274459838867188, -109.66841888427734, 17.957763671875, 16.110971450805664, 17.061721801757812, -13.284797668457031, 55.45703125, -7.211973190307617, 141.47018432617188, -63.67889404296875, -12.460485458374023, 30.415145874023438, 127.92373657226562, 62.81044006347656, 40.03511047363281, 103.02671813964844, 152.6258544921875, 18.299110412597656, 102.99959564208984, -59.42109680175781, 17.353790283203125, -40.557830810546875, 99.62085723876953, -50.20942687988281, -9.909004211425781, -1.0793933868408203, 9.072025299072266, 16.186328887939453], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000253.npy"}
|
||||
{"epoch": 0.382464096749811, "step": 254, "batch_size": 64, "mean": 37.42321014404297, "std": 63.00991439819336, "min": -96.07752227783203, "p10": -22.467639350891112, "median": 14.003009796142578, "p90": 132.2045928955078, "max": 151.81997680664062, "pos_frac": 0.671875, "sample": [3.9155426025390625, 6.90283203125, 132.060546875, -16.668853759765625, -9.566879272460938, 103.85108947753906, 134.72511291503906, 123.08377075195312, 17.473833084106445, 0.04728889465332031, -0.8563232421875, 14.229496002197266, 43.72277069091797, 109.01702880859375, 116.67784881591797, 122.09933471679688, 151.81997680664062, 29.315582275390625, -44.06108093261719, 96.14712524414062, -19.281028747558594, 135.49876403808594, -20.132362365722656, 128.87216186523438, -32.877479553222656, -96.07752227783203, -56.63731384277344, 132.26632690429688, 20.4301815032959, 1.0265388488769531, 126.15054321289062, -18.8778076171875, 4.7748260498046875, -0.3503608703613281, -3.249034881591797, 9.50433349609375, -5.056118011474609, 107.07461547851562, 4.3958587646484375, 60.87236022949219, 28.977386474609375, -4.438480377197266, 13.77652359008789, -20.234603881835938, -23.416006088256836, 134.1077880859375, -10.343635559082031, 132.30581665039062, 79.63064575195312, 28.129179000854492, 16.913467407226562, -30.413917541503906, -20.254783630371094, 134.65562438964844, 114.0626220703125, -2.2609329223632812, 44.910545349121094, 3.3544044494628906, 69.37975311279297, 91.542236328125, 4.873802185058594, 71.19735717773438, 10.783279418945312, -84.41609954833984], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000254.npy"}
|
||||
{"epoch": 0.3839758125472411, "step": 255, "batch_size": 64, "mean": 44.759979248046875, "std": 67.82852935791016, "min": -125.67384338378906, "p10": -17.427628326416013, "median": 26.209678649902344, "p90": 131.77767944335938, "max": 179.11578369140625, "pos_frac": 0.75, "sample": [130.53439331054688, 83.91481018066406, 179.11578369140625, 128.84063720703125, 97.90313720703125, 74.60818481445312, -18.082008361816406, -8.673362731933594, 15.077320098876953, 109.90782165527344, -88.87104797363281, -2.3434696197509766, 12.24245834350586, 113.12215423583984, -13.961515426635742, 12.8809814453125, 63.05585479736328, -0.8654403686523438, 10.922393798828125, -118.20658111572266, 123.29191589355469, 2.168548583984375, 32.90901184082031, 8.463966369628906, 150.0798797607422, 110.72240447998047, -30.751739501953125, 19.510345458984375, 159.159423828125, -6.23248291015625, 141.3557586669922, 70.9784164428711, -15.900741577148438, 38.22848892211914, 44.744964599609375, 18.21668815612793, 1.9601764678955078, 100.36079406738281, 139.69216918945312, 10.792501449584961, 4.549985885620117, 109.53016662597656, 96.18502807617188, -8.171035766601562, 13.833742141723633, -1.960275650024414, 48.337181091308594, 154.05650329589844, -85.09130096435547, 98.1468276977539, -0.15752029418945312, 110.07249450683594, 4.617958068847656, 132.31051635742188, 72.06599426269531, 117.94929504394531, 5.668788909912109, 59.48551559448242, 90.72805786132812, -125.67384338378906, 3.618377685546875, 18.081886291503906, 67.23094177246094, -21.649818420410156], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000255.npy"}
|
||||
{"epoch": 0.3854875283446712, "step": 256, "batch_size": 64, "mean": 42.44768524169922, "std": 77.10211181640625, "min": -161.85528564453125, "p10": -33.35103912353515, "median": 32.704376220703125, "p90": 133.13702392578125, "max": 241.48565673828125, "pos_frac": 0.671875, "sample": [-60.068336486816406, 10.754348754882812, 84.75248718261719, 81.01026916503906, 96.02629089355469, -41.471527099609375, -16.631681442260742, 126.75953674316406, 42.958885192871094, -20.442052841186523, -84.57898712158203, 2.782369613647461, -4.142333984375, 151.31439208984375, 41.493499755859375, 152.48675537109375, -7.7490997314453125, -8.249237060546875, -161.85528564453125, 86.86717224121094, 10.18447494506836, 131.08282470703125, -19.677610397338867, 30.31097412109375, -28.65594482421875, -122.53517150878906, 64.14378356933594, 4.434181213378906, 24.53691864013672, 88.58211517333984, 12.576568603515625, 35.0977783203125, 93.22052764892578, 125.96978759765625, 116.96952819824219, 134.01739501953125, 115.74267578125, 114.26789093017578, -9.415374755859375, 136.2220001220703, 49.137916564941406, 50.10438537597656, 183.09320068359375, 10.478363037109375, 124.2791519165039, 11.245109558105469, 119.57637023925781, 23.232574462890625, 66.27169799804688, 241.48565673828125, -35.20393371582031, 96.01714324951172, -11.953277587890625, -11.004714965820312, 128.70211791992188, -16.029556274414062, 77.60305786132812, -10.209663391113281, -11.231851577758789, 142.99200439453125, -123.98574829101562, 26.532867431640625, -29.027618408203125, 85.45391082763672], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000256.npy"}
|
||||
{"epoch": 0.3869992441421013, "step": 257, "batch_size": 64, "mean": 53.843666076660156, "std": 75.50260925292969, "min": -119.12422180175781, "p10": -29.738607025146475, "median": 39.12651824951172, "p90": 137.64754943847657, "max": 267.8374938964844, "pos_frac": 0.78125, "sample": [-0.1179656982421875, 31.13067626953125, -39.29658508300781, 267.8374938964844, 147.01458740234375, 51.427066802978516, 12.521356582641602, 115.44441986083984, 34.34154510498047, 112.5829849243164, 130.322998046875, 30.03636932373047, 11.090108871459961, -5.3934783935546875, 131.18704223632812, 112.4214096069336, 123.64120483398438, 4.755096435546875, -68.38677978515625, 4.576057434082031, 3.451631546020508, 104.04292297363281, 60.6092529296875, 24.139766693115234, 113.4480209350586, -1.6723480224609375, 4.137599945068359, 136.5472412109375, 139.48370361328125, -0.26141357421875, 112.64657592773438, 133.70957946777344, 121.82258605957031, 14.069595336914062, 50.42028045654297, 43.91149139404297, 72.31539916992188, 133.35049438476562, 136.69004821777344, -104.37461853027344, -119.12422180175781, 241.97601318359375, 21.29503631591797, 1.8487014770507812, -33.7403564453125, -66.79165649414062, 48.524253845214844, 5.7681732177734375, 102.86846160888672, 94.7129898071289, 125.99111938476562, 28.116546630859375, 144.51760864257812, 13.638065338134766, -49.99322509765625, 138.0579071044922, -3.399250030517578, -20.40119171142578, 142.14842224121094, 10.907699584960938, 68.27970886230469, 14.65399169921875, 44.95530700683594, -14.439002990722656], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000257.npy"}
|
||||
{"epoch": 0.3885109599395314, "step": 258, "batch_size": 64, "mean": 41.25651550292969, "std": 58.60247802734375, "min": -115.79238891601562, "p10": -3.0894935607910137, "median": 36.549991607666016, "p90": 117.5783187866211, "max": 164.22817993164062, "pos_frac": 0.875, "sample": [63.103057861328125, 15.610942840576172, 51.41441345214844, -3.8470821380615234, 22.70275115966797, 117.92716217041016, 162.89361572265625, 104.0111083984375, 47.283348083496094, -1.321786880493164, 6.656644821166992, 48.318031311035156, -71.35990905761719, 11.085481643676758, 72.31531524658203, 149.90560913085938, 36.807960510253906, -27.126495361328125, 70.92079162597656, -114.6705322265625, 28.237049102783203, 74.26289367675781, -97.65185546875, 22.742820739746094, 18.219345092773438, 24.61341094970703, 11.377151489257812, 73.05996704101562, 13.44449234008789, 30.17833709716797, 66.23675537109375, 0.9526405334472656, 8.7161865234375, 103.81824493408203, 62.83806610107422, 57.93012237548828, 1.9972610473632812, 25.486953735351562, 78.50839233398438, 164.22817993164062, 18.92388153076172, -44.379058837890625, 147.2228546142578, 76.05020141601562, 36.292022705078125, 129.26602172851562, 56.99375915527344, 91.45588684082031, 10.001777648925781, 116.76435089111328, 10.771446228027344, 37.304718017578125, 80.47460174560547, 48.357566833496094, 16.352867126464844, 5.902547836303711, 74.16455078125, -115.79238891601562, 1.8843536376953125, 7.655675888061523, 74.89229583740234, 97.0555191040039, 118.91229248046875, 12.062238693237305], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000258.npy"}
|
||||
{"epoch": 0.3900226757369615, "step": 259, "batch_size": 64, "mean": 59.60941696166992, "std": 78.6446762084961, "min": -122.97604370117188, "p10": -40.269091796874996, "median": 66.9742431640625, "p90": 147.01143035888674, "max": 204.22628784179688, "pos_frac": 0.765625, "sample": [-11.292465209960938, -122.97604370117188, -82.25544738769531, 116.81788635253906, 65.53851318359375, 124.34982299804688, 118.54059600830078, 167.6835174560547, 115.61051940917969, -97.09242248535156, 84.54676818847656, -14.317405700683594, 74.17430114746094, 204.22628784179688, 103.09683227539062, -8.989501953125, -74.69922637939453, 6.27996826171875, 171.20355224609375, 27.0224609375, 61.31449508666992, 120.65746307373047, 131.30043029785156, 121.8262939453125, 73.30985260009766, -4.405902862548828, -93.26947021484375, 17.958770751953125, 4.941902160644531, 17.279314041137695, 121.69518280029297, 115.40997314453125, -7.260520935058594, 36.695884704589844, 104.07999420166016, 114.61692810058594, 173.39306640625, -42.312164306640625, 118.21159362792969, 147.9727020263672, 143.8240509033203, 68.40997314453125, 17.794784545898438, 179.95248413085938, 14.372932434082031, -8.360107421875, -35.501922607421875, -9.429922103881836, -111.61711120605469, 8.822303771972656, 169.62745666503906, 144.0556182861328, 140.08792114257812, 94.43521118164062, 81.25419616699219, 84.11402130126953, 44.98548889160156, 34.140342712402344, 14.876190185546875, 45.208717346191406, 137.7113494873047, 144.76846313476562, 62.618988037109375, 47.966888427734375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000259.npy"}
|
||||
{"epoch": 0.3915343915343915, "step": 260, "batch_size": 64, "mean": 47.838096618652344, "std": 73.84290313720703, "min": -121.61656188964844, "p10": -49.65347824096679, "median": 40.855262756347656, "p90": 141.37220153808593, "max": 152.26690673828125, "pos_frac": 0.6875, "sample": [140.2364044189453, -60.62895202636719, 124.03892517089844, -18.866100311279297, 13.233570098876953, 151.11195373535156, 39.548614501953125, 52.356483459472656, 142.6226043701172, -50.860809326171875, 144.1285400390625, 96.99891662597656, 120.95925903320312, 97.75184631347656, 129.31060791015625, -39.27519607543945, 143.4344482421875, 107.63236999511719, 39.27202606201172, 139.78866577148438, 118.44916534423828, -3.4533233642578125, 138.49859619140625, 140.97381591796875, -12.533782958984375, 49.4196662902832, 120.23503112792969, 136.04885864257812, -7.0452880859375, -5.2376556396484375, 21.54114532470703, 37.31501770019531, 151.4530029296875, 66.38801574707031, -3.4953975677490234, 59.45389938354492, 75.0478515625, -32.07623291015625, -121.61656188964844, 22.684852600097656, 2.700075149536133, 105.37377166748047, -14.576349258422852, -8.365036010742188, 9.927299499511719, 42.16191101074219, 141.54293823242188, -46.83637237548828, 135.39158630371094, -26.3380126953125, -68.94351196289062, 70.43994140625, 4.5723114013671875, 18.57794189453125, 114.02439880371094, 74.79269409179688, 13.4473876953125, 106.75082397460938, 152.26690673828125, -53.21659851074219, 1.2077827453613281, -113.24642181396484, -55.48383712768555, -9.378334045410156], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000260.npy"}
|
||||
{"epoch": 0.3930461073318216, "step": 261, "batch_size": 64, "mean": 39.698997497558594, "std": 67.44759368896484, "min": -99.3521728515625, "p10": -40.38426208496093, "median": 27.941845893859863, "p90": 134.358349609375, "max": 166.84432983398438, "pos_frac": 0.671875, "sample": [1.6121635437011719, 28.136611938476562, 52.77630615234375, -13.662576675415039, -42.69889831542969, 84.0465087890625, -2.7746944427490234, 133.65805053710938, 27.38555908203125, -4.843515396118164, -27.97473907470703, -34.98344421386719, -15.04449462890625, 144.2958984375, 103.67176818847656, -29.313003540039062, -4.576148986816406, 162.0364227294922, 6.338861465454102, 134.65847778320312, 13.530105590820312, 29.959625244140625, -8.118518829345703, -87.75173950195312, 166.84432983398438, 81.55045318603516, 148.30282592773438, 66.18734741210938, -3.9936904907226562, 36.047515869140625, 75.01422119140625, 122.75403594970703, 73.04742431640625, 143.1199951171875, 76.42512512207031, 14.859132766723633, 135.97608947753906, -99.3521728515625, -31.82733917236328, 92.67442321777344, 75.32508087158203, 28.03619956970215, 123.73973846435547, 103.6715316772461, 27.847492218017578, -77.54086303710938, 3.0254745483398438, -50.96685791015625, -87.87946319580078, 16.549663543701172, 25.84246826171875, -43.898345947265625, 113.67884826660156, 124.40956115722656, 18.822494506835938, -13.994277954101562, 55.237667083740234, 21.760147094726562, -6.80584716796875, 125.60198974609375, 46.734642028808594, 124.61154174804688, 39.038818359375, -0.10622978210449219], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000261.npy"}
|
||||
{"epoch": 0.3945578231292517, "step": 262, "batch_size": 64, "mean": 36.91168975830078, "std": 62.90443420410156, "min": -106.57283782958984, "p10": -33.943041992187496, "median": 19.337958335876465, "p90": 121.33411712646486, "max": 166.97592163085938, "pos_frac": 0.703125, "sample": [-16.682876586914062, 55.09510803222656, 18.959136962890625, -2.735940933227539, 33.58671569824219, 109.19384765625, -12.812255859375, 4.056610107421875, 37.340675354003906, 125.92403411865234, -17.10620880126953, 90.78316497802734, -64.97523498535156, 19.613096237182617, 69.68807983398438, -88.25006103515625, 30.956148147583008, 102.73641967773438, 122.37265014648438, 89.36262512207031, -106.57283782958984, -8.290851593017578, 15.967277526855469, -27.596160888671875, 12.917549133300781, 115.35134887695312, 17.065765380859375, 150.2947998046875, -50.76812744140625, 2.420499801635742, 115.75160217285156, 77.72740936279297, 19.062820434570312, 118.91087341308594, -35.37305450439453, 133.134765625, -0.11135673522949219, 5.720607757568359, 166.97592163085938, 1.4096755981445312, -6.23426628112793, 9.54510498046875, -30.606346130371094, 60.69535827636719, 93.02044677734375, 63.812339782714844, 79.06147766113281, 7.42326545715332, -70.47434997558594, 43.06028366088867, 64.63455200195312, -2.4637222290039062, 84.68272399902344, -17.827951431274414, 18.7714900970459, 57.157981872558594, 79.54924011230469, 138.87472534179688, 10.98895263671875, 158.6666717529297, 39.929603576660156, -27.27996063232422, 114.79987335205078, -38.543670654296875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000262.npy"}
|
||||
{"epoch": 0.3960695389266818, "step": 263, "batch_size": 64, "mean": 40.892086029052734, "std": 63.02323913574219, "min": -118.35967254638672, "p10": -31.58862152099609, "median": 38.59196853637695, "p90": 120.87292709350587, "max": 142.68467712402344, "pos_frac": 0.6875, "sample": [54.83253860473633, 67.08000183105469, -12.881477355957031, -35.34748077392578, -6.769502639770508, -6.806396484375, 126.14418029785156, -0.3767852783203125, 123.78106689453125, 99.33026123046875, 77.401123046875, 26.040571212768555, -9.958213806152344, 63.98848342895508, -7.3592376708984375, -91.2253646850586, 104.31640625, -16.63726806640625, 56.26322937011719, -7.958076477050781, 93.17840576171875, 119.44657135009766, 120.93586730957031, 35.904727935791016, 52.355140686035156, 4.945377349853516, 120.70026397705078, 57.90184020996094, 132.05999755859375, 96.57769012451172, 130.99586486816406, 34.05101013183594, -3.6834716796875, 25.45672607421875, -12.783191680908203, 39.760520935058594, -117.92118835449219, 124.45018005371094, 46.94517517089844, 113.10255432128906, 45.57347106933594, 0.9150466918945312, -45.27610778808594, 103.72474670410156, -50.54057312011719, -2.8739166259765625, 37.42341613769531, 120.72606658935547, 89.2928466796875, 18.553977966308594, 21.061767578125, 142.68467712402344, -27.533554077148438, 15.604969024658203, 84.62277221679688, -17.207509994506836, 31.18061065673828, 115.8681869506836, 95.16847229003906, -33.326507568359375, 5.5959930419921875, -118.35967254638672, 119.49220275878906, 46.48397445678711], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000263.npy"}
|
||||
{"epoch": 0.3975812547241119, "step": 264, "batch_size": 64, "mean": 43.957969665527344, "std": 71.3612060546875, "min": -146.9260711669922, "p10": -35.632756042480466, "median": 39.41275215148926, "p90": 134.5755172729492, "max": 198.61329650878906, "pos_frac": 0.71875, "sample": [83.36282348632812, -36.57139587402344, 21.18694305419922, 3.8104171752929688, 5.182781219482422, -17.094491958618164, 15.082870483398438, 10.055992126464844, 123.33720397949219, 14.036941528320312, -93.31047058105469, 1.9329757690429688, -146.9260711669922, 128.88497924804688, -3.8155670166015625, 89.69944763183594, 70.66121673583984, 74.84754943847656, 7.644279479980469, 155.40017700195312, 127.23867797851562, 105.04815673828125, 102.16664123535156, 47.90155029296875, 54.98992919921875, 138.40870666503906, -3.3426952362060547, 32.139556884765625, -10.14837646484375, 125.8612060546875, 19.792762756347656, -33.186988830566406, 134.5802001953125, 12.959625244140625, -8.241744995117188, 53.641326904296875, 134.56459045410156, -16.32550048828125, 101.64215850830078, 118.13613891601562, -57.858978271484375, -4.9777679443359375, 53.392845153808594, 198.61329650878906, 71.0676040649414, 120.53709411621094, -119.15426635742188, 65.79869079589844, -33.442596435546875, -2.01995849609375, -54.472381591796875, 9.774040222167969, 141.27084350585938, 38.359195709228516, 97.91189575195312, -55.59379959106445, 32.282325744628906, 40.46630859375, 62.85752868652344, -0.692596435546875, 48.115753173828125, 169.76629638671875, 109.17149353027344, 136.90280151367188], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000264.npy"}
|
||||
{"epoch": 0.39909297052154197, "step": 265, "batch_size": 64, "mean": 45.45208740234375, "std": 77.9606704711914, "min": -139.5663299560547, "p10": -51.08726387023924, "median": 41.1145076751709, "p90": 143.61704406738284, "max": 165.18325805664062, "pos_frac": 0.734375, "sample": [11.713165283203125, 1.1578617095947266, 6.975799560546875, -57.65557861328125, 2.909208297729492, -26.843772888183594, 115.94783782958984, 46.375465393066406, -123.84519958496094, 135.06610107421875, 43.83629608154297, -5.112037658691406, 97.03250885009766, 61.74176025390625, -139.5663299560547, -34.43669891357422, 89.96947479248047, 0.10247993469238281, 34.0828857421875, -121.40682220458984, 49.47467803955078, 165.18325805664062, 129.8116455078125, 27.059934616088867, 4.658998489379883, -10.84157943725586, 106.38278198242188, -5.631263732910156, 30.591415405273438, -1.8537158966064453, 145.96463012695312, 156.91110229492188, -13.77886962890625, 130.984375, 106.05193328857422, 144.65780639648438, 5.586755752563477, 38.39271926879883, 2.0891571044921875, -20.940902709960938, 82.23072814941406, 127.28425598144531, 132.07789611816406, 146.78216552734375, 119.66374206542969, 109.22454833984375, 139.0157012939453, 45.38642120361328, 29.123008728027344, -75.8841323852539, 50.31184387207031, 94.27143859863281, -35.76119613647461, 151.80372619628906, 149.93756103515625, 131.27310180664062, 141.1885986328125, -85.8397216796875, 22.517393112182617, -12.072330474853516, 94.1651382446289, 17.539772033691406, 107.86139678955078, -101.96661376953125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000265.npy"}
|
||||
{"epoch": 0.40060468631897206, "step": 266, "batch_size": 64, "mean": 43.36137390136719, "std": 73.78817749023438, "min": -160.32623291015625, "p10": -39.50669975280761, "median": 25.112411499023438, "p90": 138.34646301269532, "max": 180.79324340820312, "pos_frac": 0.75, "sample": [-32.92671203613281, 138.89495849609375, 134.7022705078125, 10.595113754272461, 47.371803283691406, -3.155933380126953, 4.844017028808594, -23.164520263671875, 7.862810134887695, 92.92511749267578, 48.164669036865234, 21.7215576171875, 3.0156402587890625, 98.62474060058594, 110.62071228027344, 71.69288635253906, 12.828628540039062, 180.79324340820312, -42.32669448852539, -15.379852294921875, 132.7248077392578, 8.059158325195312, 17.941696166992188, 138.34986877441406, 66.8792495727539, 170.19798278808594, -10.047439575195312, 159.01986694335938, 97.35838317871094, -54.95616149902344, 127.82579040527344, 37.32942199707031, 143.65121459960938, -11.898994445800781, 78.35135650634766, -15.820259094238281, 43.11279296875, 128.40216064453125, 138.33851623535156, -4.833637237548828, 5.149452209472656, -58.30586242675781, -66.11648559570312, 6.308963775634766, -160.32623291015625, 17.96954345703125, 7.2542877197265625, 33.718284606933594, 84.43578338623047, 92.72167205810547, -113.01875305175781, 11.304327011108398, -1.9112701416015625, 28.503265380859375, 132.3611297607422, 80.22406005859375, 165.40817260742188, -102.63605499267578, 64.44127655029297, 16.300003051757812, 125.7037353515625, 4.337858200073242, 126.99490356445312, 16.615665435791016], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000266.npy"}
|
||||
{"epoch": 0.4021164021164021, "step": 267, "batch_size": 64, "mean": 35.831661224365234, "std": 76.09845733642578, "min": -146.7209014892578, "p10": -59.198154067993165, "median": 25.076107025146484, "p90": 134.97875061035157, "max": 159.47262573242188, "pos_frac": 0.703125, "sample": [121.04664611816406, 2.8460540771484375, 1.2363319396972656, -130.27096557617188, 13.379135131835938, -59.83607864379883, 68.70501708984375, 157.40855407714844, 91.52650451660156, 0.8160438537597656, -19.0634822845459, 96.25428771972656, 24.542102813720703, 159.47262573242188, 5.632686614990234, -80.19231414794922, 20.682220458984375, -89.69195556640625, 6.117095947265625, -135.14144897460938, -57.70966339111328, -24.898353576660156, -20.06145477294922, 104.70269012451172, 79.37260437011719, 132.7174072265625, -2.8219871520996094, 41.4284553527832, 135.0516357421875, -23.355506896972656, 40.27909851074219, 0.5922260284423828, 128.1080322265625, 126.6507568359375, -51.36375427246094, 25.610111236572266, 40.31616973876953, -33.78196716308594, -4.6744384765625, 153.07974243164062, 69.556884765625, -0.1447124481201172, 156.0443115234375, 0.0651702880859375, 117.57433319091797, 116.41845703125, 139.37252807617188, 26.880332946777344, -146.7209014892578, 134.80868530273438, 104.22496032714844, -68.25350952148438, -1.1012153625488281, 46.580692291259766, 9.991851806640625, 6.165966033935547, 102.97434997558594, 97.33333587646484, 135.45404052734375, 73.10533142089844, 5.146053314208984, 89.53192138671875, 35.22191619873047, -1.7152156829833984], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000267.npy"}
|
||||
{"epoch": 0.4036281179138322, "step": 268, "batch_size": 64, "mean": 42.50840759277344, "std": 62.98203659057617, "min": -126.27792358398438, "p10": -12.24446105957031, "median": 25.604461669921875, "p90": 139.54608764648438, "max": 196.86102294921875, "pos_frac": 0.734375, "sample": [26.608612060546875, 147.65969848632812, 0.14095306396484375, 5.948455810546875, 108.13124084472656, 140.87539672851562, 38.0877685546875, 0.8636322021484375, -30.349781036376953, 27.051197052001953, 140.952392578125, 1.6266708374023438, -1.042837142944336, 24.600311279296875, 31.237899780273438, 21.958934783935547, 43.65142822265625, -43.96903991699219, 0.5566272735595703, 133.65162658691406, 114.97772979736328, 143.41085815429688, -6.038675308227539, 2.5524139404296875, -5.576728820800781, 60.17706298828125, 44.71931457519531, 108.92193603515625, 29.456466674804688, -16.231094360351562, 133.7154541015625, -4.591030120849609, -1.4111366271972656, 9.319854736328125, -31.508132934570312, 136.44436645507812, -13.128013610839844, -9.759834289550781, -39.074005126953125, 18.083229064941406, 95.40097045898438, 13.738615036010742, 10.510261535644531, 90.69747924804688, 95.77300262451172, 46.411102294921875, 152.6336212158203, -7.52714729309082, 7.43621826171875, -126.27792358398438, -9.117668151855469, -10.182838439941406, 56.8404541015625, 1.0789260864257812, 196.86102294921875, 119.86676788330078, 32.80775451660156, 33.808738708496094, 128.58554077148438, 57.136695861816406, -0.12249374389648438, 148.6994171142578, 86.45500183105469, 6.323436737060547], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000268.npy"}
|
||||
{"epoch": 0.4051398337112623, "step": 269, "batch_size": 64, "mean": 47.38948440551758, "std": 68.34508514404297, "min": -133.54229736328125, "p10": -11.02520580291748, "median": 31.374332427978516, "p90": 135.2846481323242, "max": 214.66940307617188, "pos_frac": 0.78125, "sample": [-9.734344482421875, -25.263458251953125, 20.946884155273438, 52.91286087036133, 96.98928833007812, -13.624214172363281, 3.4031524658203125, 134.61578369140625, 6.7363739013671875, -10.6903076171875, -22.439315795898438, 12.467430114746094, 75.36121368408203, 88.64071655273438, 128.02297973632812, 102.81253051757812, 0.3200874328613281, 83.42900848388672, 187.37008666992188, -122.73503875732422, 10.953437805175781, 32.444488525390625, 124.63836669921875, 43.216461181640625, 24.666702270507812, 135.0559539794922, -11.168733596801758, 185.68560791015625, 37.899658203125, 130.5850830078125, 27.63300895690918, 12.764209747314453, -4.26887321472168, 34.960357666015625, -8.068344116210938, 1.3577938079833984, 5.353404998779297, 72.75706481933594, 1.0252685546875, 135.38265991210938, 138.0868682861328, 98.49104309082031, 32.031898498535156, 6.902645111083984, 124.43726348876953, -56.845699310302734, 111.607177734375, 30.716766357421875, 4.774261474609375, 5.988128662109375, 214.66940307617188, -0.3427238464355469, 155.12525939941406, 47.30296325683594, -2.2656478881835938, -133.54229736328125, -6.875274658203125, 85.13504028320312, 58.23878479003906, 78.27149963378906, 86.42023468017578, 14.477081298828125, 137.85662841796875, 19.850421905517578], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000269.npy"}
|
||||
{"epoch": 0.40665154950869237, "step": 270, "batch_size": 64, "mean": 42.89282989501953, "std": 69.68904876708984, "min": -130.24688720703125, "p10": -18.391938781738276, "median": 37.08374786376953, "p90": 132.97745208740236, "max": 217.36614990234375, "pos_frac": 0.78125, "sample": [86.87393951416016, 136.21588134765625, -5.772483825683594, 171.93075561523438, 65.29966735839844, 142.39474487304688, 8.158487319946289, 70.28599548339844, 106.06924438476562, 47.13929748535156, 146.0544891357422, 122.96399688720703, 23.0965576171875, 123.07361602783203, 6.504329681396484, -97.57342529296875, 91.50535583496094, 121.63652038574219, 8.326282501220703, 55.75340270996094, 22.790924072265625, 83.37060546875, 16.361351013183594, -59.03537368774414, 12.542152404785156, 16.420265197753906, 127.04605865478516, -94.6602783203125, 40.45405197143555, 20.227481842041016, 49.295318603515625, 65.79901123046875, 23.610885620117188, 92.1185531616211, 31.836898803710938, 5.885486602783203, 102.7887954711914, -11.731658935546875, 8.176891326904297, -130.24688720703125, 217.36614990234375, 128.55984497070312, 120.27398681640625, -4.02873420715332, 18.676361083984375, 41.4805908203125, 9.277420043945312, -21.234390258789062, 134.87071228027344, 63.647682189941406, 141.26148986816406, 1.0412101745605469, -68.58250427246094, 46.87626647949219, -1.397064208984375, -8.235115051269531, -8.978462219238281, -126.06388092041016, 79.94631958007812, 9.859893798828125, 35.273292541503906, -11.759552001953125, 38.894203186035156, 55.128326416015625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000270.npy"}
|
||||
{"epoch": 0.40816326530612246, "step": 271, "batch_size": 64, "mean": 31.7574462890625, "std": 83.56893920898438, "min": -128.26060485839844, "p10": -69.9703598022461, "median": 7.5112714767456055, "p90": 148.08461608886722, "max": 260.8228759765625, "pos_frac": 0.65625, "sample": [36.93518829345703, 74.09437561035156, -9.576904296875, -6.933357238769531, 152.70773315429688, 150.38946533203125, 10.926141738891602, 7.039976119995117, 138.60081481933594, -69.55699920654297, 96.20538330078125, -6.004316329956055, 193.7833251953125, -38.140113830566406, 106.1027603149414, 154.5128173828125, 3.3402328491210938, -86.42933654785156, 2.7359561920166016, 152.9661407470703, 7.982566833496094, 3.7940101623535156, 125.48040771484375, -18.76473045349121, -63.18597412109375, 2.60791015625, -31.371749877929688, 104.19918060302734, 3.7820587158203125, -68.75778198242188, 27.992422103881836, 100.97398376464844, 3.92425537109375, -128.26060485839844, 4.3934326171875, 128.21173095703125, 6.341453552246094, 1.5284347534179688, 155.88365173339844, 133.03140258789062, 20.232620239257812, -82.24134063720703, 260.8228759765625, -70.14751434326172, -18.199615478515625, 81.83636474609375, -14.799835205078125, -117.7177963256836, -108.60812377929688, 22.807334899902344, 19.51535415649414, -36.034690856933594, 133.91311645507812, 137.02845764160156, 107.91816711425781, 21.4365234375, -75.34292602539062, 66.97444915771484, -4.537986755371094, 16.442142486572266, -43.84954071044922, -11.753433227539062, 142.70663452148438, 20.58995819091797], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000271.npy"}
|
||||
{"epoch": 0.40967498110355255, "step": 272, "batch_size": 64, "mean": 60.91047668457031, "std": 59.965492248535156, "min": -121.68255615234375, "p10": -0.06492652893066325, "median": 60.66444778442383, "p90": 141.3075103759766, "max": 168.85394287109375, "pos_frac": 0.890625, "sample": [23.991554260253906, -8.676521301269531, -11.678741455078125, 128.45652770996094, 58.34764862060547, 137.50491333007812, 29.518035888671875, 122.27606964111328, 168.85394287109375, 17.871517181396484, 104.66988372802734, 83.50546264648438, 14.871299743652344, 6.594291687011719, 63.66284942626953, 37.89637756347656, 83.37289428710938, 11.536727905273438, 62.98124694824219, 57.948429107666016, 134.16624450683594, 20.64635467529297, 108.85722351074219, 80.20352172851562, 162.06861877441406, 7.688179016113281, 81.0389175415039, -29.01367950439453, 14.549667358398438, 142.93719482421875, 101.21187591552734, 143.98924255371094, 146.20730590820312, -0.39913177490234375, 111.7308120727539, 21.883739471435547, 30.178302764892578, 0.9267921447753906, 48.478729248046875, -121.68255615234375, 103.92623138427734, 147.57681274414062, 144.01010131835938, 99.43952178955078, 2.8139724731445312, 14.626953125, 70.09086608886719, 117.30423736572266, 118.95677185058594, 115.7993392944336, 33.38346862792969, 130.335205078125, -23.254180908203125, -14.526718139648438, 15.607063293457031, 102.62115478515625, 77.49408721923828, 82.46028137207031, 3.6097373962402344, 0.7148857116699219, 1.1036968231201172, 0.8745040893554688, 135.26803588867188, 18.862668991088867], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000272.npy"}
|
||||
{"epoch": 0.41118669690098264, "step": 273, "batch_size": 64, "mean": 44.4174690246582, "std": 87.75820922851562, "min": -180.8701934814453, "p10": -62.613052368164055, "median": 32.066123962402344, "p90": 146.68667297363282, "max": 266.76361083984375, "pos_frac": 0.734375, "sample": [5.941051483154297, 94.59224700927734, 134.58428955078125, -43.209754943847656, 158.6881103515625, -66.79113006591797, -36.791473388671875, 133.59425354003906, -98.01618957519531, 31.78369140625, 122.17500305175781, -46.59819030761719, 71.54885864257812, -3.772359848022461, 15.188514709472656, 28.193851470947266, 151.39511108398438, 29.482234954833984, 132.9033660888672, 99.56598663330078, 97.6114730834961, 10.512744903564453, -1.7264404296875, 4.100574493408203, -180.8701934814453, 144.6229248046875, 33.6153564453125, 4.911186218261719, -52.86420440673828, 9.318347930908203, 101.1597900390625, 266.76361083984375, 151.85137939453125, -133.55361938476562, 18.398605346679688, 110.07421112060547, 77.80377197265625, 91.30534362792969, -26.87120819091797, 119.72605895996094, -105.10002899169922, 18.22484588623047, 133.622802734375, -42.918724060058594, 43.1050910949707, 231.46156311035156, 147.57113647460938, 118.3339614868164, 27.278900146484375, 103.5181884765625, 0.9447002410888672, 38.451927185058594, 0.5938549041748047, -18.197463989257812, 20.75556182861328, 157.22784423828125, -1.9276657104492188, 47.476524353027344, 32.34855651855469, -78.2828598022461, 113.99310302734375, 140.5688018798828, -124.85781860351562, 78.17794799804688], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000273.npy"}
|
||||
{"epoch": 0.4126984126984127, "step": 274, "batch_size": 64, "mean": 49.20020294189453, "std": 77.38175201416016, "min": -164.78475952148438, "p10": -46.6045295715332, "median": 60.12593650817871, "p90": 146.60047607421876, "max": 172.4611358642578, "pos_frac": 0.734375, "sample": [66.23248291015625, -13.212394714355469, -10.20235824584961, -1.5847625732421875, -22.76617431640625, -79.41172790527344, 146.36961364746094, 146.8938446044922, 54.66331481933594, -1.5028247833251953, 106.00279998779297, 126.96854400634766, -8.845954895019531, -47.77886962890625, 104.44404602050781, 172.4611358642578, 141.32264709472656, -121.59385681152344, 6.084293365478516, 27.36193084716797, 147.5869903564453, 131.62879943847656, 72.09841918945312, 126.89786529541016, 60.02952575683594, 5.8265380859375, 151.12301635742188, 136.98085021972656, 73.46281433105469, 24.439056396484375, 9.015323638916016, 11.780670166015625, 136.93716430664062, 3.4150238037109375, -30.90281867980957, 68.99411010742188, -43.864402770996094, -57.87348175048828, 136.23727416992188, 13.341156005859375, -51.981666564941406, 68.989501953125, 41.20658493041992, 60.37852478027344, -22.933717727661133, 62.590721130371094, 28.486801147460938, -164.78475952148438, 64.11418151855469, 60.222347259521484, 147.95358276367188, 112.69507598876953, 4.325588226318359, 90.75254821777344, 20.55731201171875, -26.064416885375977, -105.93707275390625, 9.947437286376953, 146.6994171142578, 140.97998046875, 104.13813781738281, 82.99491119384766, 171.09603881835938, 133.3262481689453], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000274.npy"}
|
||||
{"epoch": 0.41421012849584277, "step": 275, "batch_size": 64, "mean": 46.62141418457031, "std": 66.41350555419922, "min": -116.91302490234375, "p10": -16.5854995727539, "median": 33.03109359741211, "p90": 140.59163208007814, "max": 167.49014282226562, "pos_frac": 0.734375, "sample": [48.91288757324219, 21.910505294799805, -60.42472839355469, 44.97357940673828, 134.3613739013672, 5.446449279785156, 55.890174865722656, -77.1609878540039, 137.30056762695312, 128.03726196289062, 8.694107055664062, 30.042037963867188, -1.4624099731445312, 138.309326171875, -4.915069580078125, -1.0681877136230469, 157.15109252929688, 15.930801391601562, 124.20938873291016, -19.62877655029297, -116.91302490234375, 144.44589233398438, 18.126937866210938, 17.883445739746094, 55.97496795654297, 5.2782745361328125, -63.23641586303711, 27.56121063232422, -2.533336639404297, 46.305511474609375, 67.44509887695312, 167.49014282226562, -3.5562820434570312, 61.013771057128906, 146.97055053710938, 42.634063720703125, 96.94954681396484, 13.593137741088867, -7.3114471435546875, -9.484519958496094, 21.452743530273438, -0.20290374755859375, 73.54080963134766, 2.4876556396484375, 9.361238479614258, 164.07781982421875, 73.74142456054688, 141.56976318359375, -5.037200927734375, 119.08805847167969, 10.056629180908203, 36.02014923095703, -3.237489700317383, 21.645599365234375, 142.93515014648438, 137.78305053710938, -53.44099044799805, 85.1799087524414, 86.93667602539062, 121.82852935791016, 62.010353088378906, 73.806396484375, -58.408416748046875, 125.4286117553711], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000275.npy"}
|
||||
{"epoch": 0.41572184429327286, "step": 276, "batch_size": 64, "mean": 47.07915496826172, "std": 67.28515625, "min": -120.02820587158203, "p10": -27.376524353027342, "median": 39.867563247680664, "p90": 136.59910278320314, "max": 163.25660705566406, "pos_frac": 0.8125, "sample": [129.33827209472656, 13.678153991699219, 55.827674865722656, 37.570106506347656, 1.6664676666259766, 19.477012634277344, 155.37559509277344, -88.2594223022461, 97.78960418701172, 24.890138626098633, 9.436107635498047, 40.47137451171875, 64.7735366821289, 111.32501220703125, -28.2791748046875, 138.7720947265625, 22.18087387084961, 6.9721221923828125, 23.502059936523438, -1.9092140197753906, 39.26375198364258, 136.46084594726562, 120.80732727050781, 152.91909790039062, 51.865966796875, -79.62030792236328, 121.6755599975586, -30.046539306640625, 84.03253173828125, 136.65835571289062, 12.153514862060547, 6.498191833496094, -59.57410430908203, -120.02820587158203, 116.97944641113281, -25.270339965820312, 75.83856201171875, 3.2947769165039062, 108.15084838867188, 117.45245361328125, -115.56094360351562, 41.80162048339844, 140.41897583007812, 0.31748199462890625, -21.971336364746094, 59.584564208984375, 123.63825225830078, 69.22982788085938, 135.6696319580078, 15.257720947265625, -16.051979064941406, 49.01509094238281, 12.531898498535156, 28.351703643798828, 25.335792541503906, 163.25660705566406, 144.3150634765625, 109.53092956542969, 62.941856384277344, 40.880218505859375, 25.807947158813477, 84.08566284179688, 34.0556755065918, -3.4864730834960938], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000276.npy"}
|
||||
{"epoch": 0.41723356009070295, "step": 277, "batch_size": 64, "mean": 24.27852439880371, "std": 76.46260833740234, "min": -140.74951171875, "p10": -61.356396484375, "median": 18.51704216003418, "p90": 124.16207885742189, "max": 164.7461395263672, "pos_frac": 0.609375, "sample": [126.81816101074219, -126.17768859863281, 120.53829193115234, -21.165481567382812, 117.2608413696289, -108.64381408691406, 18.429668426513672, 18.604415893554688, 110.01288604736328, 25.89544677734375, -59.748252868652344, -34.07129669189453, 73.76278686523438, -5.8567047119140625, -50.64312744140625, 71.60505676269531, 25.06496238708496, -15.9805908203125, -2.395782470703125, 37.29084396362305, 88.42009735107422, 48.14765167236328, 8.380386352539062, 21.05657196044922, -11.963581085205078, -51.81787109375, -37.5775146484375, -50.64414978027344, -6.1387176513671875, 114.98081970214844, 0.8887214660644531, 67.5992660522461, -18.145875930786133, 164.7461395263672, 61.303077697753906, 137.810791015625, -22.887001037597656, -140.74951171875, 28.513275146484375, 72.80621337890625, -26.200027465820312, -86.04782104492188, 119.43058776855469, 79.59751892089844, 75.16679382324219, 3.8937339782714844, -36.855735778808594, 116.61766052246094, -131.36465454101562, 158.2386932373047, 69.10961151123047, -35.003013610839844, 131.11602783203125, 61.87467956542969, -4.8796539306640625, 0.6991786956787109, 121.99786376953125, -133.29290771484375, 125.089599609375, 140.19561767578125, -62.04560089111328, 14.928024291992188, 0.5705299377441406, 55.659400939941406], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000277.npy"}
|
||||
{"epoch": 0.41874527588813304, "step": 278, "batch_size": 64, "mean": 45.93433380126953, "std": 72.84137725830078, "min": -180.71104431152344, "p10": -19.503125381469726, "median": 39.22319793701172, "p90": 145.15590820312502, "max": 174.3933868408203, "pos_frac": 0.75, "sample": [-43.263099670410156, -115.0908203125, 42.73193359375, 82.57160949707031, 0.9944305419921875, -15.444242477416992, -145.58668518066406, 78.66433715820312, 155.16265869140625, 174.3933868408203, 50.477935791015625, -10.639175415039062, 147.9942626953125, 138.5330810546875, 7.535758972167969, -12.250364303588867, 22.669803619384766, 113.4117431640625, 106.96540069580078, 124.37515258789062, 76.02851867675781, 61.782127380371094, 46.43138122558594, 16.336631774902344, -11.0341796875, 148.88494873046875, -37.15394592285156, 128.78428649902344, 75.76641845703125, 13.193115234375, 9.933258056640625, 68.73607635498047, -5.87396240234375, -19.82141876220703, 83.814697265625, 23.07665252685547, 107.29820251464844, 35.71446228027344, 109.8062744140625, 64.30323028564453, 133.28489685058594, 162.3665008544922, -18.760440826416016, 95.98025512695312, 77.585205078125, 35.157814025878906, -4.686653137207031, 4.646764755249023, -50.593650817871094, 16.422103881835938, 96.87113952636719, 113.63719177246094, -180.71104431152344, -4.841569900512695, 15.998710632324219, 159.35726928710938, 61.608551025390625, 161.93043518066406, 34.028316497802734, 16.92412567138672, 18.568628311157227, -6.838642120361328, 89.84708404541016, 11.800399780273438], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000278.npy"}
|
||||
{"epoch": 0.42025699168556313, "step": 279, "batch_size": 64, "mean": 62.73311996459961, "std": 65.3265609741211, "min": -122.55096435546875, "p10": -3.878702545166014, "median": 66.21332550048828, "p90": 142.9111099243164, "max": 177.98150634765625, "pos_frac": 0.859375, "sample": [0.1754283905029297, 2.546611785888672, 147.7201385498047, 50.779945373535156, -122.55096435546875, 144.1984100341797, 9.180229187011719, -2.6189193725585938, -49.60968017578125, 129.4579620361328, -2.5562362670898438, 126.11412048339844, 86.12000274658203, 139.90740966796875, 87.3642578125, 177.98150634765625, 79.97879028320312, 67.21781921386719, -48.29785919189453, 9.413372039794922, -73.02062225341797, 110.81916809082031, 74.9147720336914, 71.5847396850586, 14.349754333496094, 65.20883178710938, 24.49839210510254, 21.72064208984375, 28.814964294433594, 108.86358642578125, -4.418609619140625, 111.32311248779297, 164.81353759765625, 41.95972442626953, -25.929122924804688, 131.0274200439453, 97.99497985839844, 65.00442504882812, 8.766548156738281, 154.2227783203125, 132.82052612304688, 10.558040618896484, 131.6742401123047, 30.27105712890625, 44.28607177734375, 94.81965637207031, 2.0011634826660156, 13.675674438476562, 95.50115966796875, 144.22720336914062, 97.57402038574219, -27.136642456054688, 121.20535278320312, 139.88771057128906, 2.1350746154785156, 28.321441650390625, 55.108642578125, 31.412521362304688, 138.56057739257812, 122.15326690673828, 124.69696807861328, 94.10824584960938, 145.87826538085938, 16.138259887695312], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000279.npy"}
|
||||
{"epoch": 0.4217687074829932, "step": 280, "batch_size": 64, "mean": 65.48545837402344, "std": 66.93756103515625, "min": -85.04491424560547, "p10": -8.776237869262694, "median": 53.60224533081055, "p90": 147.27129211425782, "max": 243.36456298828125, "pos_frac": 0.84375, "sample": [119.85787963867188, 148.45431518554688, 6.021879196166992, 123.35177612304688, -13.024909973144531, 142.30284118652344, 2.283205032348633, -12.075614929199219, 109.73717498779297, 122.47024536132812, 49.99192810058594, 41.746063232421875, 115.32476806640625, 159.5137939453125, 124.6995849609375, 129.13095092773438, 108.43900299072266, 38.67647171020508, 117.9193115234375, 56.44330596923828, 16.337631225585938, 8.42742919921875, -0.5513763427734375, -1.0264549255371094, 152.52749633789062, 118.18315124511719, 0.3031349182128906, 86.55878448486328, 129.50650024414062, 243.36456298828125, 11.646244049072266, -45.642547607421875, 26.748336791992188, 126.12324523925781, 8.57320785522461, 70.20965576171875, 164.75350952148438, 73.00364685058594, 146.74853515625, 175.11740112304688, -44.915855407714844, -14.319416046142578, 11.307807922363281, -85.04491424560547, 79.23939514160156, 18.340808868408203, 68.81841278076172, -7.003612518310547, 25.62860107421875, 6.596706390380859, 147.49533081054688, 69.79829406738281, 20.46664810180664, 140.71511840820312, 5.220479965209961, -9.535934448242188, 48.608184814453125, 50.76118469238281, 25.152780532836914, 36.1494140625, 6.409797668457031, 142.34243774414062, 140.03762817382812, 106.62443542480469], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000280.npy"}
|
||||
{"epoch": 0.42328042328042326, "step": 281, "batch_size": 64, "mean": 52.701515197753906, "std": 76.74485778808594, "min": -112.88723754882812, "p10": -47.03465118408202, "median": 51.20952606201172, "p90": 141.5445068359375, "max": 167.4064483642578, "pos_frac": 0.765625, "sample": [-50.58151626586914, 129.8485107421875, 8.766586303710938, 3.6923828125, 126.72943878173828, 80.8885498046875, 41.80738830566406, 130.93145751953125, 112.16622924804688, 140.64703369140625, 25.538902282714844, 144.81326293945312, 116.23455810546875, 104.69666290283203, -35.86231994628906, 5.784379959106445, 23.653993606567383, 124.8560791015625, 2.1325931549072266, 144.06295776367188, 71.89955139160156, 2.8734703063964844, -62.75114440917969, 63.21092987060547, 141.92913818359375, -1.0853767395019531, 7.191368103027344, -19.430728912353516, 96.08251953125, -92.67754364013672, 9.346183776855469, 13.522781372070312, 2.7558517456054688, 82.72982025146484, 136.3044891357422, -109.4591064453125, 60.611663818359375, 167.4064483642578, 10.49884033203125, -98.53334045410156, 148.22390747070312, 155.6803436279297, 145.01470947265625, 81.57613372802734, 139.0755615234375, 24.508018493652344, 140.39447021484375, 17.762630462646484, 139.169677734375, 86.74466705322266, -16.576202392578125, 16.781295776367188, 136.10623168945312, -112.88723754882812, -6.036998748779297, 132.4827423095703, 138.52578735351562, -52.6722412109375, 125.46653747558594, 125.04505920410156, -38.75863265991211, -6.06706428527832, 19.673215866088867, -29.568342208862305], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000281.npy"}
|
||||
{"epoch": 0.42479213907785335, "step": 282, "batch_size": 64, "mean": 48.18358612060547, "std": 82.26641845703125, "min": -123.89663696289062, "p10": -49.48595314025878, "median": 27.593162536621094, "p90": 149.7693054199219, "max": 219.3514404296875, "pos_frac": 0.65625, "sample": [-6.436088562011719, -19.47930145263672, -96.2935562133789, -10.411117553710938, 26.483901977539062, 130.56417846679688, 103.82325744628906, 5.294824600219727, -6.493770599365234, 153.17013549804688, 94.06185150146484, 136.73300170898438, 18.379907608032227, 138.73080444335938, 163.76614379882812, 28.702423095703125, 152.90185546875, 87.2451171875, -16.625457763671875, 117.64292907714844, 50.20575714111328, 22.926910400390625, -10.975381851196289, 192.96624755859375, 24.747774124145508, -7.599235534667969, -19.605087280273438, 141.0071258544922, 135.51022338867188, 22.743236541748047, -41.52742385864258, -52.896751403808594, -123.89663696289062, 133.0840301513672, 6.930593490600586, -100.39208984375, 141.30221557617188, 10.91063117980957, -98.32723999023438, 53.899993896484375, 130.20651245117188, 140.860107421875, -0.9365081787109375, 67.1878662109375, 142.46002197265625, 219.3514404296875, 96.44810485839844, 45.05609130859375, -4.1699066162109375, 65.99836730957031, -34.43316650390625, -10.176177978515625, -91.28564453125, 11.713985443115234, -22.798789978027344, 176.70118713378906, 153.72372436523438, 114.42557525634766, -78.9925308227539, -18.548187255859375, 43.5389404296875, 101.22289276123047, 127.93165588378906, 25.48790740966797], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000282.npy"}
|
||||
{"epoch": 0.42630385487528344, "step": 283, "batch_size": 64, "mean": 50.65044403076172, "std": 81.75411987304688, "min": -136.41720581054688, "p10": -39.115640258789064, "median": 34.11133575439453, "p90": 146.2916687011719, "max": 269.052734375, "pos_frac": 0.8125, "sample": [174.86070251464844, -124.25639343261719, -38.862518310546875, -37.259185791015625, 120.99192810058594, 102.461669921875, 9.741744995117188, 36.89385223388672, 111.9105224609375, 1.3571014404296875, -59.9423828125, 89.72198486328125, 35.35795593261719, 71.12725830078125, 130.37857055664062, 11.999645233154297, 32.297027587890625, 146.42965698242188, -47.09040832519531, 91.5135498046875, 126.8443832397461, 120.02433776855469, 110.60091400146484, 1.4129867553710938, 83.86833190917969, 147.70950317382812, 109.06291198730469, 0.5949478149414062, 187.62472534179688, 19.82685089111328, 10.331779479980469, 134.06459045410156, -107.8016357421875, 27.661773681640625, -14.6219482421875, 122.11614990234375, -132.38458251953125, 35.685760498046875, 81.85963439941406, 130.04571533203125, 145.96969604492188, -26.957067489624023, 123.21148681640625, 131.6242218017578, 15.796138763427734, 7.826961517333984, 30.85395050048828, -136.41720581054688, 7.910436630249023, 2.024017333984375, 122.40701293945312, 4.723102569580078, 269.052734375, 2.6647987365722656, 172.27813720703125, 87.27670288085938, -0.911956787109375, 8.67184066772461, 3.374643325805664, 28.212081909179688, -39.22412109375, 156.79055786132812, 32.864715576171875, 37.44628143310547], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000283.npy"}
|
||||
{"epoch": 0.42781557067271353, "step": 284, "batch_size": 64, "mean": 48.86520767211914, "std": 70.08464813232422, "min": -133.82366943359375, "p10": -8.93490867614746, "median": 28.69501495361328, "p90": 148.80522918701172, "max": 181.5330352783203, "pos_frac": 0.75, "sample": [-39.30488586425781, 11.500640869140625, 31.754697799682617, 38.579071044921875, 175.53598022460938, 114.53970336914062, 3.685689926147461, -1.5258026123046875, 181.5330352783203, 13.075920104980469, -9.398670196533203, 14.271968841552734, -1.3149566650390625, -37.08110046386719, 127.44190979003906, -2.7775421142578125, 7.175783157348633, 81.7700424194336, -7.8527984619140625, 95.24160766601562, 16.102584838867188, 57.31077575683594, 39.05747985839844, -51.88251495361328, -33.647056579589844, 31.639984130859375, 10.402275085449219, 21.31463623046875, 65.4205322265625, -7.469692230224609, 134.5667724609375, 121.68889617919922, 93.8529281616211, 112.61134338378906, 14.727325439453125, 8.047050476074219, 146.20335388183594, 144.58642578125, 177.00833129882812, 28.21540069580078, 149.92031860351562, 177.26573181152344, -5.8287506103515625, 17.951160430908203, -1.0192012786865234, -0.6353359222412109, 99.62777709960938, 8.674915313720703, 95.58277130126953, 139.77426147460938, 6.318058013916016, 48.63918685913086, 153.06597900390625, 10.129981994628906, 26.799453735351562, 89.35050964355469, 172.55303955078125, 69.43883514404297, 78.08425903320312, -133.82366943359375, 87.94083404541016, -116.15923309326172, 29.17462921142578, -2.059263229370117], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000284.npy"}
|
||||
{"epoch": 0.4293272864701436, "step": 285, "batch_size": 64, "mean": 40.06449890136719, "std": 74.60974884033203, "min": -138.61590576171875, "p10": -47.89978523254394, "median": 25.582584381103516, "p90": 144.8079803466797, "max": 162.5154266357422, "pos_frac": 0.65625, "sample": [24.109397888183594, 4.122673034667969, -72.4056396484375, 132.78692626953125, 24.635833740234375, -51.491878509521484, -138.61590576171875, -29.471153259277344, 73.26593017578125, -20.18494987487793, 73.32967376708984, 136.33143615722656, 131.66567993164062, 15.087547302246094, 160.8207244873047, -52.17987060546875, 91.55107879638672, 140.32537841796875, -18.593807220458984, 9.64914321899414, 55.922203063964844, 6.119358062744141, -82.9183349609375, -49.11539840698242, 5.549835205078125, -92.45285034179688, -41.95404052734375, 65.85716247558594, 77.829345703125, -36.46361541748047, -32.40809631347656, 162.5154266357422, -9.589561462402344, 137.76882934570312, 86.72770690917969, 104.61454772949219, 43.14667510986328, 38.29850769042969, 58.692100524902344, 137.72596740722656, 104.78657531738281, -9.192459106445312, -5.586738586425781, 146.72909545898438, 152.61569213867188, 26.529335021972656, 10.393512725830078, 160.7081298828125, -45.0633544921875, -2.041423797607422, 38.812828063964844, 134.7659454345703, 22.12896728515625, 85.34117126464844, -2.2958526611328125, 45.79435729980469, 159.5389404296875, 161.83717346191406, -41.238365173339844, 27.99774932861328, -0.6236572265625, 9.519180297851562, -6.7748870849609375, 118.84199523925781], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000285.npy"}
|
||||
{"epoch": 0.4308390022675737, "step": 286, "batch_size": 64, "mean": 65.30426025390625, "std": 75.48104858398438, "min": -72.81402587890625, "p10": -12.133254623413084, "median": 52.050981521606445, "p90": 158.42323150634766, "max": 288.25140380859375, "pos_frac": 0.84375, "sample": [52.6143798828125, -0.30966758728027344, 2.2581348419189453, 60.66522979736328, 17.188491821289062, 75.64311218261719, 16.8843994140625, 2.827127456665039, 25.131479263305664, -9.444656372070312, 31.268173217773438, 8.042007446289062, 43.01100158691406, 132.2512969970703, 146.6032257080078, 183.36843872070312, 121.09767150878906, 43.531280517578125, -23.24707794189453, 24.282608032226562, 95.45833587646484, -69.41930389404297, 25.158550262451172, 159.78323364257812, -13.285511016845703, -48.70991516113281, 86.0324935913086, 120.23893737792969, -43.01188659667969, 90.03727722167969, 152.32533264160156, 124.4974365234375, 138.96458435058594, 169.3063201904297, 99.15332794189453, 228.85018920898438, 70.07162475585938, 129.22702026367188, -72.81402587890625, 12.940383911132812, 155.24989318847656, 288.25140380859375, 16.494842529296875, 5.0583038330078125, 119.48499298095703, 13.330108642578125, 20.231019973754883, 43.656097412109375, 138.52554321289062, 145.96485900878906, 179.8546142578125, 94.44792175292969, 51.48758316040039, 7.285741806030273, -6.102725982666016, 64.81074523925781, 72.56456756591797, 18.724620819091797, -65.63902282714844, 189.4630584716797, 104.89437866210938, 4.724178314208984, 103.15019226074219, 5.088958740234375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000286.npy"}
|
||||
{"epoch": 0.4323507180650038, "step": 287, "batch_size": 64, "mean": 64.4097671508789, "std": 73.48490142822266, "min": -135.58399963378906, "p10": -3.5793914794921857, "median": 39.981727600097656, "p90": 155.51461944580078, "max": 204.77438354492188, "pos_frac": 0.84375, "sample": [45.05109405517578, -74.817626953125, 11.226951599121094, 122.31390380859375, 204.77438354492188, 153.97946166992188, 36.84820556640625, 55.64649200439453, 63.23133087158203, -4.5993804931640625, 86.07565307617188, 7.710918426513672, 74.92787170410156, 84.80538940429688, 167.8817901611328, 170.71481323242188, 183.66366577148438, 97.39488220214844, 144.44493103027344, 98.98014831542969, 38.653160095214844, 24.89312744140625, 139.39532470703125, 0.5513210296630859, 17.54605484008789, 141.70074462890625, -4.235160827636719, -41.689918518066406, 130.70828247070312, 8.621768951416016, 134.91748046875, 27.517593383789062, -135.58399963378906, 19.249290466308594, 2.189321517944336, 124.25067901611328, 7.2579345703125, 4.850360870361328, 12.007759094238281, 140.57012939453125, 41.31029510498047, -17.412193298339844, 155.19052124023438, 158.63897705078125, 8.092033386230469, 26.317161560058594, 155.6535186767578, 4.762035369873047, 138.8531494140625, 139.58558654785156, -0.8447151184082031, 35.181854248046875, 134.71026611328125, 5.328256607055664, 8.046375274658203, 23.93048858642578, 152.99951171875, 6.525367736816406, 115.4864273071289, -24.839893341064453, -2.0492630004882812, 186.35528564453125, 146.95303344726562, -0.17530059814453125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000287.npy"}
|
||||
{"epoch": 0.43386243386243384, "step": 288, "batch_size": 64, "mean": 49.551292419433594, "std": 75.72702026367188, "min": -110.90819549560547, "p10": -60.49286041259766, "median": 43.76378059387207, "p90": 151.4921081542969, "max": 205.76382446289062, "pos_frac": 0.703125, "sample": [82.84677124023438, 4.2547760009765625, -60.288818359375, -67.36428833007812, 29.793495178222656, 49.403316497802734, 21.23792839050293, -14.669082641601562, 52.36164855957031, 63.188941955566406, 25.443241119384766, 107.39949035644531, 76.12847137451172, 142.53213500976562, -22.18060302734375, 91.926025390625, -60.58030700683594, 113.68594360351562, -23.24102783203125, 61.80027389526367, -110.90819549560547, 127.70460510253906, 205.76382446289062, 145.6395721435547, 8.600982666015625, 167.27264404296875, 31.18363380432129, 93.08943939208984, 22.335105895996094, 24.93926429748535, 153.62661743164062, -74.73896789550781, 82.2496337890625, 81.08857727050781, 170.48118591308594, 87.53378295898438, 57.15650939941406, -74.94432067871094, 33.86390686035156, -16.099327087402344, -77.2072982788086, -67.692626953125, 173.00732421875, -10.149398803710938, 0.5291519165039062, -7.314666748046875, 129.90750122070312, -35.8973388671875, 138.11314392089844, 153.021240234375, 125.96421813964844, -4.1903228759765625, 147.92413330078125, 15.93426513671875, -9.705375671386719, -2.475372314453125, 74.17715454101562, 38.124244689941406, -12.618505477905273, 163.4959716796875, 32.85979461669922, 126.24464416503906, 68.39295959472656, 121.32109069824219], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000288.npy"}
|
||||
{"epoch": 0.43537414965986393, "step": 289, "batch_size": 64, "mean": 46.778343200683594, "std": 80.03875732421875, "min": -172.92092895507812, "p10": -35.69210586547851, "median": 23.580991744995117, "p90": 149.76121368408204, "max": 249.69631958007812, "pos_frac": 0.6875, "sample": [76.9173583984375, 11.687393188476562, 104.30290222167969, -6.601339340209961, -39.208396911621094, -27.4874267578125, -12.153667449951172, 65.8017349243164, 108.24844360351562, 55.19939422607422, -1.3308792114257812, 52.12074279785156, 78.89643096923828, 148.19635009765625, 16.506271362304688, 77.37237548828125, 241.05007934570312, -56.363037109375, 30.655712127685547, 133.3159637451172, 15.121856689453125, 154.63160705566406, 13.016342163085938, 7.146209716796875, 133.36581420898438, 7.196357727050781, 143.08673095703125, 10.318138122558594, -4.756175994873047, 150.48878479003906, -12.253158569335938, 61.736167907714844, -17.017398834228516, -172.92092895507812, 178.25169372558594, -9.14272689819336, 16.19358253479004, 249.69631958007812, 6.5313568115234375, 6.807899475097656, -2.068544387817383, 36.73456573486328, -71.57982635498047, 39.06996154785156, 108.81653594970703, -23.934656143188477, 80.1534423828125, -3.6285667419433594, -91.55253601074219, -58.56687545776367, 125.06504821777344, 38.405670166015625, 4.5953521728515625, 55.32965087890625, 141.3472900390625, 138.4266815185547, 105.93885040283203, 156.69393920898438, 150.43186950683594, 6.511383056640625, -1.5133628845214844, -58.932350158691406, 141.30075073242188, -17.855175018310547], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000289.npy"}
|
||||
{"epoch": 0.436885865457294, "step": 290, "batch_size": 64, "mean": 49.99085235595703, "std": 77.34732818603516, "min": -136.48684692382812, "p10": -18.64650726318359, "median": 26.848833084106445, "p90": 156.89138946533205, "max": 199.33261108398438, "pos_frac": 0.734375, "sample": [68.16459655761719, 25.62397003173828, 142.14328002929688, 199.33261108398438, 44.733978271484375, 8.643133163452148, -52.77808380126953, -3.1697940826416016, 16.788604736328125, 183.65545654296875, 53.615203857421875, 70.68753051757812, 15.061317443847656, -96.7778549194336, 25.928863525390625, -8.197410583496094, 33.17578887939453, -15.670845031738281, 8.21036148071289, -6.269142150878906, 76.82543182373047, -19.921791076660156, 146.060791015625, 164.83438110351562, 6.931526184082031, -2.5244140625, 32.7845344543457, -9.0133056640625, 5.916830062866211, -78.85403442382812, 161.26162719726562, 159.27818298339844, 138.71701049804688, 57.99887466430664, 25.071151733398438, 182.02169799804688, 139.4007568359375, 116.98284912109375, 57.67559051513672, -39.50604248046875, 134.65672302246094, 129.1498260498047, 26.032283782958984, -113.53466033935547, -1.8365554809570312, -5.4603424072265625, -1.2643451690673828, 169.19805908203125, 151.32220458984375, 145.4879913330078, 21.462787628173828, 119.79598999023438, -11.257125854492188, 11.03973388671875, -136.48684692382812, 150.18707275390625, 27.665382385253906, 146.70111083984375, 56.645660400390625, 14.102592468261719, 0.5441856384277344, 41.217559814453125, 78.5885009765625, 10.613487243652344], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000290.npy"}
|
||||
{"epoch": 0.4383975812547241, "step": 291, "batch_size": 64, "mean": 61.95741271972656, "std": 85.01072692871094, "min": -137.02105712890625, "p10": -20.496873092651366, "median": 28.530502319335938, "p90": 162.37441711425782, "max": 359.73931884765625, "pos_frac": 0.734375, "sample": [234.02239990234375, 147.10108947753906, -6.705718994140625, -14.812545776367188, 62.22819519042969, 1.6263427734375, -26.31230926513672, 7.043575286865234, 134.0706024169922, 7.924568176269531, 135.21954345703125, 97.79703521728516, -9.706928253173828, -6.231836318969727, 128.9016876220703, 143.92681884765625, -27.16930389404297, -37.212684631347656, 40.668701171875, 133.9114990234375, -5.51849365234375, 1.3423233032226562, 64.51216125488281, 54.56829071044922, 71.25309753417969, 146.93606567382812, 208.4659423828125, -0.7802505493164062, 162.9029083251953, 5.03228759765625, 99.58696746826172, 171.43634033203125, -0.4708061218261719, 4.219215393066406, -41.1673583984375, 3.6792545318603516, 150.32154846191406, 27.697235107421875, 162.57644653320312, 8.856575012207031, 10.571098327636719, 161.90301513671875, -137.02105712890625, 19.836898803710938, 162.9170684814453, 92.63044738769531, 66.97592163085938, 145.00709533691406, 91.10562896728516, 359.73931884765625, -21.19925308227539, 20.731246948242188, -6.802928924560547, 8.600990295410156, -57.074668884277344, -1.7765388488769531, 20.842208862304688, 151.96701049804688, 29.36376953125, 125.65869140625, 153.18653869628906, 119.69170379638672, 25.5377197265625, -18.857986450195312], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000291.npy"}
|
||||
{"epoch": 0.4399092970521542, "step": 292, "batch_size": 64, "mean": 42.40858459472656, "std": 90.76409912109375, "min": -153.19873046875, "p10": -69.79820556640625, "median": 31.744426727294922, "p90": 142.9062438964844, "max": 303.9539489746094, "pos_frac": 0.6875, "sample": [-145.97509765625, -1.9228286743164062, 139.13253784179688, 1.1758480072021484, 161.91957092285156, 105.21714782714844, 27.985034942626953, 303.9539489746094, 124.9229507446289, -68.81130981445312, -47.41126251220703, 6.321531295776367, 15.762674331665039, -22.326438903808594, -103.5899658203125, -4.431121826171875, 66.8756103515625, 60.43865966796875, 96.0633544921875, 54.70379638671875, 61.421417236328125, 127.03279113769531, 115.85969543457031, -129.72702026367188, -27.1053466796875, 0.33876800537109375, 1.8307685852050781, 140.35592651367188, -125.43736267089844, 54.728668212890625, 34.95073699951172, 117.70127868652344, 4.914125442504883, -14.369110107421875, -7.330862045288086, -109.48485565185547, 145.14930725097656, -153.19873046875, -2.7274932861328125, -2.0712051391601562, -40.455841064453125, 143.99923706054688, 130.0000457763672, -5.747720718383789, 166.92843627929688, 123.26321411132812, 19.3660888671875, 119.98139190673828, 55.31792068481445, 14.255264282226562, 214.24603271484375, 125.71863555908203, 5.120431900024414, 125.64459228515625, 32.92313003540039, 9.643882751464844, 105.75936889648438, 30.565723419189453, 136.07049560546875, 49.44878387451172, 88.12783813476562, -39.235626220703125, -70.22116088867188, 170.59310913085938], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000292.npy"}
|
||||
{"epoch": 0.4414210128495843, "step": 293, "batch_size": 64, "mean": 68.19401550292969, "std": 90.96387481689453, "min": -158.00697326660156, "p10": -24.230877876281735, "median": 53.570491790771484, "p90": 164.5778549194336, "max": 295.536376953125, "pos_frac": 0.78125, "sample": [7.968418121337891, 23.331443786621094, 22.740524291992188, 147.00985717773438, 128.43679809570312, 25.937297821044922, 20.51218032836914, 13.593265533447266, 116.99047088623047, 5.228057861328125, 156.5640106201172, -20.173368453979492, 162.18389892578125, 126.90621185302734, 112.92103576660156, 30.39948272705078, 151.0604705810547, 119.68183898925781, 114.09291076660156, 148.05868530273438, 142.89361572265625, -66.59657287597656, -10.704795837402344, -120.74998474121094, 129.9281768798828, 8.258499145507812, 120.76763916015625, -69.91265869140625, 24.028091430664062, 232.58309936523438, 2.829784393310547, 184.94320678710938, 229.94906616210938, -103.9498062133789, 116.03988647460938, 8.38840103149414, 149.770263671875, -158.00697326660156, -11.206058502197266, -14.922500610351562, 20.7132568359375, 165.6038360595703, -17.197650909423828, -3.196746826171875, -33.5721435546875, 6.695274353027344, 46.916839599609375, 145.4084014892578, 295.536376953125, 62.39149475097656, 150.5355987548828, 34.922882080078125, 59.141929626464844, 4.288204193115234, 147.28192138671875, -25.969810485839844, 186.45948791503906, 68.03523254394531, 105.69392395019531, 154.0206756591797, 47.999053955078125, -9.892322540283203, 161.38027954101562, 183.44741821289062], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000293.npy"}
|
||||
{"epoch": 0.4429327286470144, "step": 294, "batch_size": 64, "mean": 46.266258239746094, "std": 69.16143798828125, "min": -145.38314819335938, "p10": -10.149257469177245, "median": 22.561878204345703, "p90": 147.49505310058595, "max": 249.40103149414062, "pos_frac": 0.75, "sample": [153.6741943359375, 37.44203186035156, 46.725616455078125, 95.4345703125, 138.8341064453125, 13.316688537597656, 157.94891357421875, 1.076171875, -28.460601806640625, 8.052925109863281, 34.02290344238281, -145.38314819335938, -4.153417587280273, 48.168060302734375, 20.21709632873535, 135.09539794921875, 147.39813232421875, 10.543891906738281, 145.79019165039062, 22.238441467285156, -20.22515869140625, 42.854530334472656, 1.8995685577392578, 94.40850067138672, -20.169410705566406, 155.57867431640625, -6.924625396728516, 102.80667114257812, -6.955863952636719, 249.40103149414062, 66.89031982421875, -0.9855766296386719, 4.867700576782227, 213.12680053710938, 23.330007553100586, 21.815719604492188, 23.123870849609375, -7.773735046386719, -4.0926055908203125, 22.88531494140625, 8.736534118652344, 33.79644775390625, 147.53659057617188, -0.4370307922363281, 4.148839950561523, 60.91735076904297, 78.4542007446289, 33.4415168762207, 8.695358276367188, 160.6724853515625, -9.58292007446289, 51.62283706665039, 134.91961669921875, 10.025279998779297, 2.453092575073242, 106.5518798828125, -3.6615447998046875, -10.391973495483398, 13.771865844726562, 7.6876373291015625, 119.9945297241211, -36.14821243286133, 69.98532104492188, -25.992950439453125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000294.npy"}
|
||||
{"epoch": 0.4444444444444444, "step": 295, "batch_size": 64, "mean": 43.430938720703125, "std": 76.56755065917969, "min": -136.25582885742188, "p10": -24.381842994689936, "median": 19.49755096435547, "p90": 152.54831695556643, "max": 212.18630981445312, "pos_frac": 0.71875, "sample": [172.27230834960938, 24.78125, -10.096633911132812, -12.707489013671875, 64.13909149169922, 145.79147338867188, 142.51219177246094, 34.22726821899414, 10.222160339355469, 67.57562255859375, 177.63278198242188, -4.715923309326172, 212.18630981445312, 17.239646911621094, 2.2436256408691406, 16.23084259033203, 2.22784423828125, 21.755455017089844, -20.292861938476562, 117.70123291015625, 158.66860961914062, -124.0069580078125, -2.4292640686035156, 15.404462814331055, 130.81063842773438, 74.49285888671875, 6.75840950012207, -14.864799499511719, 146.2022705078125, -17.55431365966797, -54.59864807128906, -6.666042327880859, 161.11981201171875, 134.3482208251953, 2.146759033203125, -109.44959259033203, 134.1156463623047, 14.022480010986328, 35.92014694213867, 155.26805114746094, 94.48663330078125, -136.25582885742188, 7.7529296875, 10.41766357421875, 132.91856384277344, 67.79534149169922, -68.63805389404297, -43.764671325683594, -20.811933517456055, 101.48468780517578, -5.156614303588867, 4.503076553344727, 29.213912963867188, 21.9271240234375, 55.37053680419922, 5.937488555908203, 110.63648986816406, -25.91180419921875, 55.590187072753906, 16.040390014648438, 158.87088012695312, 81.15504455566406, 117.99312591552734, -12.610153198242188], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000295.npy"}
|
||||
{"epoch": 0.4459561602418745, "step": 296, "batch_size": 64, "mean": 54.6158561706543, "std": 89.60792541503906, "min": -171.15518188476562, "p10": -62.698582458496084, "median": 45.610822677612305, "p90": 165.13912506103517, "max": 227.9658203125, "pos_frac": 0.703125, "sample": [55.32026672363281, 179.30148315429688, 129.1416473388672, 2.916440963745117, -0.38957977294921875, 79.46884155273438, 3.214265823364258, 227.9658203125, -5.048835754394531, 148.6571502685547, 35.9013786315918, 8.80660629272461, -82.7434310913086, 154.37466430664062, -0.5707244873046875, 84.04893493652344, 163.7762908935547, 101.63825225830078, 11.024658203125, 71.47233581542969, -0.8476104736328125, 14.44260025024414, -9.011734008789062, -112.619384765625, -84.83596801757812, -52.98707580566406, 3.966552734375, 88.89058685302734, 33.266517639160156, 19.714927673339844, 93.69821166992188, -116.24343872070312, 111.17706298828125, 23.440093994140625, 224.84942626953125, 19.756134033203125, 17.568405151367188, -66.86065673828125, 137.7318572998047, -94.319580078125, 73.08382415771484, 92.82054138183594, 165.49911499023438, -171.15518188476562, 164.2991485595703, 98.996826171875, 145.25942993164062, 136.12606811523438, 112.35020446777344, 103.00763702392578, 171.18812561035156, -2.6903533935546875, 153.13714599609375, -0.2634754180908203, 204.88433837890625, -24.852508544921875, -3.9176483154296875, 178.18661499023438, 25.1705322265625, -42.80888366699219, -31.931793212890625, 147.97161865234375, 105.82217407226562, 76.17780303955078], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000296.npy"}
|
||||
{"epoch": 0.4474678760393046, "step": 297, "batch_size": 64, "mean": 54.44794464111328, "std": 73.03050994873047, "min": -127.49176025390625, "p10": -22.023241424560542, "median": 37.46393585205078, "p90": 149.0858367919922, "max": 175.41983032226562, "pos_frac": 0.828125, "sample": [143.5564727783203, -64.05534362792969, 34.948524475097656, 123.21724700927734, 43.486785888671875, 37.51226806640625, 14.148155212402344, 148.187744140625, -8.690872192382812, -127.49176025390625, 139.32130432128906, 21.29412841796875, 8.473163604736328, 22.649333953857422, 2.8604793548583984, 152.9419403076172, 26.783981323242188, 79.78207397460938, 23.453948974609375, 23.065444946289062, 36.14497375488281, 37.41560363769531, -95.89738464355469, -66.22543334960938, -102.7568359375, 149.47073364257812, 51.183387756347656, 13.285903930664062, 119.8123779296875, 17.91063690185547, 144.007568359375, 159.53488159179688, -16.834716796875, -0.9838733673095703, 44.0208625793457, 22.190908432006836, 1.0081634521484375, -25.46466064453125, 109.57080078125, 137.5091552734375, 69.5928955078125, 15.9815673828125, 141.2449188232422, 20.947832107543945, 23.453048706054688, 118.4088363647461, 103.3673324584961, 0.3447275161743164, -14.549972534179688, 151.68711853027344, 51.89670181274414, 119.65711975097656, 131.77145385742188, 144.6128387451172, 5.459659576416016, 118.41336822509766, 172.8887176513672, 109.74037170410156, 2.294464111328125, 175.41983032226562, -24.24689483642578, 39.44047164916992, 164.34451293945312, 62.14930725097656], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000297.npy"}
|
||||
{"epoch": 0.4489795918367347, "step": 298, "batch_size": 64, "mean": 40.729461669921875, "std": 82.57907104492188, "min": -153.3341522216797, "p10": -65.40683670043946, "median": 19.02605152130127, "p90": 153.10586853027345, "max": 199.37200927734375, "pos_frac": 0.71875, "sample": [28.64641571044922, -107.38309478759766, 163.30189514160156, 26.040557861328125, 158.36819458007812, 24.800140380859375, 133.16036987304688, -5.5017852783203125, -70.54060363769531, 137.02252197265625, 157.29774475097656, 130.17132568359375, -64.73281860351562, -14.215850830078125, -6.059486389160156, 45.382389068603516, 144.58456420898438, -36.979637145996094, 140.2505645751953, 102.31072998046875, -0.9083213806152344, 4.210842132568359, 151.79345703125, 78.59893798828125, 12.06207275390625, 34.86407470703125, -121.2873764038086, 21.20210838317871, 108.90151977539062, 155.04971313476562, 3.834562301635742, 162.23876953125, -11.033706665039062, 7.611045837402344, 25.747638702392578, 129.1240234375, 3.820737838745117, 145.64772033691406, 7.517114639282227, 0.6237716674804688, 5.506584167480469, -0.0065975189208984375, 9.712997436523438, 29.195541381835938, 9.496681213378906, 199.37200927734375, 146.93527221679688, -7.4258880615234375, -23.92435073852539, 16.675552368164062, 78.4049072265625, 29.70526123046875, -65.6957015991211, -73.99667358398438, 6.233795166015625, 153.66015625, -105.08564758300781, 64.00494384765625, -34.025360107421875, -153.3341522216797, 151.81253051757812, 16.849994659423828, 0.8055343627929688, 146.2653045654297], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000298.npy"}
|
||||
{"epoch": 0.4504913076341648, "step": 299, "batch_size": 64, "mean": 32.70663833618164, "std": 92.09805297851562, "min": -164.0584716796875, "p10": -89.62839431762694, "median": 16.934794425964355, "p90": 148.036572265625, "max": 243.88157653808594, "pos_frac": 0.640625, "sample": [128.22740173339844, 112.80130004882812, -19.407773971557617, -18.894325256347656, 24.363801956176758, 119.89479064941406, 49.95613098144531, 98.84725952148438, 218.71102905273438, -1.9840068817138672, -93.12577056884766, 117.20033264160156, -154.52023315429688, 5.941215515136719, -119.41233825683594, -22.555328369140625, 18.098876953125, 141.34426879882812, 69.79229736328125, -99.18498992919922, 161.417236328125, -40.3807258605957, 19.91214370727539, 114.09954833984375, -121.78034973144531, -133.07980346679688, 5.422027587890625, 24.184818267822266, 177.64271545410156, -6.463283538818359, 178.89892578125, -20.87763214111328, 32.747642517089844, -76.85223388671875, 26.488555908203125, 13.37005615234375, 15.046661376953125, 144.8563232421875, 15.035545349121094, 3.0020217895507812, 26.332733154296875, -78.6092758178711, 149.3995361328125, 64.16217803955078, 243.88157653808594, -75.5323257446289, 15.770711898803711, 131.60037231445312, 99.95101928710938, 100.10055541992188, 1.2300281524658203, 64.26248168945312, 108.15554809570312, -164.0584716796875, -2.593048095703125, 130.0455780029297, -3.5782470703125, 100.73219299316406, 0.05329704284667969, -10.389495849609375, -2.2919445037841797, -5.808845520019531, -81.46784973144531, 173.0924072265625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000299.npy"}
|
||||
{"epoch": 0.4520030234315949, "step": 300, "batch_size": 64, "mean": 44.34907150268555, "std": 85.15448760986328, "min": -141.0426483154297, "p10": -45.575054931640615, "median": 29.88842487335205, "p90": 165.68469848632813, "max": 212.06214904785156, "pos_frac": 0.671875, "sample": [26.061208724975586, 124.6334228515625, 19.085159301757812, 35.05682373046875, -5.339847564697266, -9.3792724609375, 123.51966094970703, -15.6724853515625, 9.555910110473633, -134.6523895263672, 168.5048065185547, 22.333045959472656, 128.76617431640625, 212.06214904785156, 161.4957275390625, 74.49491119384766, 169.66940307617188, -108.69678497314453, -8.653480529785156, 117.56970977783203, -1.2135295867919922, 60.60761260986328, 56.39263916015625, -28.67308807373047, -108.989013671875, 47.44818115234375, 132.30540466308594, -14.481239318847656, 50.10651397705078, 19.639514923095703, 3.2890586853027344, -141.0426483154297, 132.54515075683594, -5.559993743896484, -34.323272705078125, 65.12239074707031, 16.04048728942871, 189.1667938232422, 38.707000732421875, 169.06298828125, 144.95083618164062, 33.715641021728516, 14.084869384765625, 209.65745544433594, 132.49899291992188, -50.397247314453125, -4.044857025146484, 134.53033447265625, -1.700927734375, 68.69650268554688, 83.052734375, 6.288158416748047, 37.56512451171875, 51.13957977294922, -12.752403259277344, -101.7183609008789, 10.028091430664062, 166.3954315185547, 117.07916259765625, -104.38008880615234, -4.689903259277344, 3.2628097534179688, 164.0263214111328, -15.512458801269531], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000300.npy"}
|
||||
{"epoch": 0.45351473922902497, "step": 301, "batch_size": 64, "mean": 39.13589859008789, "std": 84.019775390625, "min": -152.6695556640625, "p10": -72.98289566040039, "median": 24.058879852294922, "p90": 152.27887725830078, "max": 271.2235412597656, "pos_frac": 0.703125, "sample": [134.7777557373047, 26.24219512939453, 63.732994079589844, 21.627887725830078, -5.944091796875, -82.16795349121094, 100.4143295288086, 165.1781768798828, 88.01864624023438, 63.31201934814453, 7.407373428344727, 10.915918350219727, 36.12782287597656, -8.432689666748047, -131.23544311523438, 51.63856506347656, 41.891632080078125, 145.67214965820312, 124.7501449584961, 10.76724624633789, -77.82139587402344, 55.62346267700195, 156.1329345703125, -9.420082092285156, 76.12400817871094, 5.9246673583984375, -5.268857955932617, 151.5033721923828, 58.11695861816406, -76.43457794189453, -33.14999771118164, 44.12244415283203, 54.6084098815918, 208.82644653320312, -10.4669189453125, 0.16449737548828125, -82.11334228515625, 271.2235412597656, -39.67475891113281, 16.254253387451172, -38.93505096435547, 74.79243469238281, 21.875564575195312, -132.4541473388672, 10.546913146972656, 116.03532409667969, 152.89312744140625, 56.632568359375, 15.488983154296875, 147.25442504882812, 18.954832077026367, 33.92987060546875, 121.51065826416016, -22.293712615966797, -64.92897033691406, 6.904895782470703, 152.61123657226562, 150.4393310546875, 46.40220642089844, 10.824775695800781, -13.254348754882812, -152.6695556640625, 164.22354125976562, -1.0571060180664062], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000301.npy"}
|
||||
{"epoch": 0.455026455026455, "step": 302, "batch_size": 64, "mean": 30.10628890991211, "std": 83.74534606933594, "min": -157.66993713378906, "p10": -65.90065612792966, "median": 16.59962272644043, "p90": 133.2327819824219, "max": 249.76568603515625, "pos_frac": 0.65625, "sample": [109.47850036621094, 186.87982177734375, -30.229385375976562, 249.76568603515625, 27.391204833984375, 2.3244571685791016, 126.23931884765625, -38.01630401611328, 185.71859741210938, 16.055633544921875, 41.45506286621094, 17.512290954589844, 5.19636344909668, 3.6716041564941406, 69.24055480957031, 152.48974609375, -0.19927215576171875, 12.937515258789062, -10.179023742675781, 72.36595153808594, 38.78818893432617, -134.03079223632812, -121.18675231933594, 64.33155822753906, 31.517044067382812, -15.601404190063477, 73.41865539550781, 36.105018615722656, 17.143611907958984, -25.487411499023438, -22.053565979003906, 112.02467346191406, 70.70048522949219, 7.74195671081543, -34.622947692871094, 182.3041534423828, 181.01092529296875, -37.75566101074219, -8.397735595703125, 44.888736724853516, 108.10992431640625, 116.53977966308594, 14.833621978759766, 59.66853332519531, 90.65048217773438, 61.17057800292969, -144.12571716308594, -1.0016860961914062, -39.431884765625, 11.880516052246094, 108.58882141113281, 62.44679260253906, -8.562248229980469, -5.964561462402344, 49.74806594848633, -101.89030456542969, -77.24441528320312, 136.22998046875, -35.64324951171875, -123.50569152832031, -157.66993713378906, 10.785369873046875, 117.71205139160156, 12.540594100952148], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000302.npy"}
|
||||
{"epoch": 0.4565381708238851, "step": 303, "batch_size": 64, "mean": 62.58820343017578, "std": 97.97750854492188, "min": -147.70838928222656, "p10": -79.87061920166015, "median": 90.45462036132812, "p90": 169.1375900268555, "max": 287.57415771484375, "pos_frac": 0.6875, "sample": [-6.4030914306640625, 87.59140014648438, 149.74847412109375, 12.03196907043457, 153.60382080078125, 138.2924041748047, 125.81028747558594, -107.17256164550781, 142.8421630859375, 107.31217956542969, 152.6267547607422, -42.52417755126953, 125.27561950683594, -6.662050247192383, -82.92434692382812, -116.27330017089844, -88.22840881347656, 210.57647705078125, -147.70838928222656, -72.74525451660156, -0.2203655242919922, 33.76860046386719, 287.57415771484375, 121.14212036132812, -127.90762329101562, 110.78114318847656, 132.86817932128906, -35.543678283691406, -5.839115142822266, 18.9932861328125, 96.7365493774414, -16.880538940429688, 9.593193054199219, 16.566757202148438, -12.113761901855469, -7.301662445068359, 162.26104736328125, 5.107444763183594, 147.97238159179688, 127.54847717285156, 49.33074951171875, 139.17462158203125, 107.94742584228516, 164.4421844482422, -10.446922302246094, 149.23585510253906, 171.1175537109375, 32.67779541015625, 81.03959655761719, 170.90814208984375, -61.50547790527344, 4.025449752807617, -87.73841857910156, 204.10885620117188, 176.85855102539062, 165.0063018798828, 20.939088821411133, 106.99771118164062, 128.0825653076172, 140.5522003173828, 140.18350219726562, -56.58855438232422, 93.31784057617188, 175.80166625976562], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000303.npy"}
|
||||
{"epoch": 0.4580498866213152, "step": 304, "batch_size": 64, "mean": 63.55152130126953, "std": 76.15451049804688, "min": -144.92745971679688, "p10": -25.195969581604004, "median": 56.92619323730469, "p90": 158.13589782714845, "max": 236.83114624023438, "pos_frac": 0.78125, "sample": [147.8997802734375, 2.9307632446289062, 128.32785034179688, 141.81602478027344, 34.48164367675781, 77.349609375, 57.886940002441406, 50.139503479003906, -25.31397819519043, 35.47005081176758, -7.15509033203125, 161.88739013671875, 173.66256713867188, 75.60545349121094, 156.91177368164062, 42.24268341064453, -6.859630584716797, 101.96868896484375, 154.1800079345703, -144.92745971679688, 116.15452575683594, 163.7475128173828, 12.0462646484375, 29.453582763671875, -24.920616149902344, 16.100479125976562, -25.623762130737305, 45.14133834838867, 51.74812316894531, 153.23146057128906, 120.44564056396484, -21.755311965942383, 236.83114624023438, 81.8739242553711, 6.663932800292969, -67.4910888671875, 70.95784759521484, 136.3619384765625, 64.04574584960938, 27.7589111328125, -0.36258888244628906, 3.3479061126708984, 188.971923828125, 67.07391357421875, 0.8604507446289062, 6.37518310546875, 49.32807159423828, 55.96544647216797, 138.63265991210938, 158.6605224609375, -48.1402587890625, -7.741485595703125, 194.7373046875, -14.090618133544922, 142.18893432617188, -32.51297378540039, 68.07745361328125, 152.56747436523438, 151.29843139648438, 13.8431396484375, 58.26118469238281, -38.72213363647461, 79.40371704101562, 127.99761962890625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000304.npy"}
|
||||
{"epoch": 0.4595616024187453, "step": 305, "batch_size": 64, "mean": 42.259552001953125, "std": 88.92378997802734, "min": -196.16900634765625, "p10": -85.85062332153319, "median": 39.142112731933594, "p90": 158.53179779052735, "max": 186.9888916015625, "pos_frac": 0.734375, "sample": [145.1195831298828, 14.642570495605469, 23.537490844726562, -58.230628967285156, -3.4575958251953125, 1.3907508850097656, 185.35418701171875, -59.53423309326172, 44.55366516113281, 176.39781188964844, 7.239292144775391, 177.5937042236328, -88.57723999023438, 130.29270935058594, -129.70106506347656, -79.48851776123047, 37.13146209716797, 1.7698249816894531, -93.10653686523438, 94.57254028320312, 59.294307708740234, 32.54644012451172, 26.136817932128906, 97.48890686035156, 79.67292022705078, 125.51727294921875, 45.6376953125, 14.087371826171875, 104.33224487304688, -196.16900634765625, -24.648941040039062, 67.96682739257812, 134.79029846191406, 133.37652587890625, -19.758941650390625, 15.967145919799805, 179.34368896484375, -92.82040405273438, 34.32353973388672, 68.02151489257812, 5.336677551269531, 26.274215698242188, 112.46092224121094, 160.56622314453125, 109.78133392333984, -11.647518157958984, 59.008140563964844, -71.21623229980469, 163.1377410888672, 118.75489807128906, 129.84811401367188, 41.15276336669922, 18.96127700805664, 126.06413269042969, 54.66136169433594, 186.9888916015625, -47.226226806640625, 33.25328063964844, -153.316650390625, 153.78480529785156, -8.664663314819336, 62.07098388671875, 112.55833435058594, -90.58743286132812], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000305.npy"}
|
||||
{"epoch": 0.46107331821617537, "step": 306, "batch_size": 64, "mean": 47.468666076660156, "std": 74.74519348144531, "min": -145.74560546875, "p10": -20.665227508544916, "median": 30.438077926635742, "p90": 151.1721420288086, "max": 228.05589294433594, "pos_frac": 0.765625, "sample": [129.796142578125, 68.24999237060547, 0.1599864959716797, 160.76821899414062, 27.4937744140625, 137.70623779296875, 80.84706115722656, 21.83646011352539, 45.46405029296875, 100.78435516357422, 155.3634033203125, 132.27398681640625, -96.1187515258789, 33.382381439208984, 0.11464309692382812, -145.74560546875, 169.4185333251953, 88.67657470703125, 38.7767219543457, -4.233501434326172, -14.370658874511719, 78.31145477294922, 148.89016723632812, 9.942663192749023, 10.231199264526367, 5.196922302246094, -43.93817901611328, 125.81733703613281, 17.911224365234375, -0.050201416015625, -71.02995300292969, 125.20651245117188, -14.315780639648438, 7.80186653137207, 115.04402923583984, 57.968833923339844, 170.3556365966797, 228.05589294433594, 17.963470458984375, 13.329513549804688, 61.00123596191406, 11.949983596801758, 1.5180549621582031, 67.40587615966797, -90.16964721679688, 6.5980682373046875, -4.530878067016602, 34.404571533203125, 17.652366638183594, 43.76806640625, 112.84380340576172, 95.74970245361328, -23.362899780273438, -3.805999755859375, 96.97052001953125, 0.8431129455566406, 11.633888244628906, -1.753143310546875, 83.13553619384766, 223.58843994140625, -50.91699981689453, 59.510406494140625, 152.15013122558594, -1.52606201171875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000306.npy"}
|
||||
{"epoch": 0.46258503401360546, "step": 307, "batch_size": 64, "mean": 23.952377319335938, "std": 90.67818450927734, "min": -180.1347198486328, "p10": -87.55860900878906, "median": 13.111342430114746, "p90": 156.5473175048828, "max": 182.7522735595703, "pos_frac": 0.609375, "sample": [100.57634735107422, -15.903877258300781, 18.998138427734375, -92.86257934570312, -79.04792785644531, 138.62188720703125, 168.9878387451172, -35.88072967529297, 175.950927734375, 92.07585144042969, -68.29815673828125, 25.856746673583984, 10.085216522216797, 157.34597778320312, -180.1347198486328, 55.11408996582031, 70.05039978027344, 10.51141357421875, 146.9087677001953, -105.85183715820312, 99.02740478515625, -57.68511962890625, -31.683792114257812, -94.93063354492188, 23.20641326904297, -88.6703109741211, -13.38299560546875, 75.78385162353516, 96.93986511230469, 0.10674667358398438, -50.436744689941406, 15.711271286010742, 5.924659729003906, -30.623653411865234, -79.8194580078125, 117.2203369140625, -46.732330322265625, -18.699378967285156, 108.14934539794922, -50.74833679199219, 2.456085205078125, -51.881996154785156, -10.293106079101562, 182.7522735595703, 154.68377685546875, -58.26286315917969, 43.86127471923828, 157.56631469726562, 172.476806640625, 49.00598907470703, -169.62811279296875, 49.690711975097656, 134.64349365234375, 35.81627655029297, 95.09574890136719, 75.837646484375, -9.067787170410156, 166.01661682128906, -84.96463775634766, 0.15081024169921875, -128.71005249023438, 16.398841857910156, 135.091796875, 2.4553985595703125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000307.npy"}
|
||||
{"epoch": 0.46409674981103555, "step": 308, "batch_size": 64, "mean": 45.81995391845703, "std": 88.60362243652344, "min": -178.8493194580078, "p10": -53.73299026489258, "median": 25.05431365966797, "p90": 161.2260971069336, "max": 210.27505493164062, "pos_frac": 0.734375, "sample": [169.59030151367188, 116.54117584228516, 115.7218017578125, -52.470584869384766, 154.55569458007812, 68.20321655273438, -22.758399963378906, 29.357650756835938, 36.84735107421875, -36.198036193847656, 161.66744995117188, 139.7254638671875, 160.19627380371094, 26.199485778808594, 17.806060791015625, -112.93238830566406, 12.424507141113281, 137.64947509765625, 148.96176147460938, 52.48928451538086, 103.72418212890625, -17.823429107666016, 209.82296752929688, 88.35975646972656, 56.229530334472656, 92.37551879882812, 19.15110206604004, 2.5117874145507812, 111.78573608398438, -97.32086181640625, 204.03253173828125, -22.72132110595703, 19.747596740722656, -54.27402114868164, 105.39016723632812, 0.9250335693359375, 111.45423126220703, -6.471839904785156, 3.65423583984375, -178.8493194580078, 22.98053741455078, -6.334209442138672, 51.75215148925781, 168.2460174560547, -0.5138397216796875, 81.19337463378906, 75.0047836303711, -81.55632019042969, 123.20706176757812, 17.130054473876953, 145.81373596191406, 21.938629150390625, -30.234451293945312, 112.6093521118164, 210.27505493164062, 23.909141540527344, 187.31280517578125, -127.00679016113281, -45.21800231933594, 10.56199836730957, 12.392950057983398, -136.8227996826172, 18.200210571289062, 2.3546390533447266], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000308.npy"}
|
||||
{"epoch": 0.4656084656084656, "step": 309, "batch_size": 64, "mean": 47.63380432128906, "std": 87.2940902709961, "min": -202.99273681640625, "p10": -43.578618621826166, "median": 27.090845108032227, "p90": 153.15730743408204, "max": 253.3116455078125, "pos_frac": 0.6875, "sample": [140.8380889892578, 177.11917114257812, 20.797142028808594, 137.57284545898438, 29.316518783569336, 14.453163146972656, 114.45481872558594, -26.956985473632812, 51.90821838378906, 16.96422576904297, 70.44815063476562, -1.8547611236572266, -88.75755310058594, -37.647987365722656, 30.647659301757812, -116.2835693359375, 24.865171432495117, 150.0963134765625, -77.59033966064453, -11.21959114074707, 5.4505615234375, 146.12779235839844, 154.4691619873047, 1.5732707977294922, 68.4404296875, 62.4610595703125, -0.2213287353515625, 20.489316940307617, 2.9227371215820312, -8.240188598632812, -3.603517532348633, -18.453140258789062, -17.055419921875, 175.4503631591797, 119.46983337402344, 14.660335540771484, 24.202190399169922, -2.4329795837402344, -6.712503433227539, 83.3217544555664, -37.04487609863281, 253.3116455078125, 140.6773681640625, 10.879642486572266, 199.9988250732422, -202.99273681640625, 111.18340301513672, 95.0047607421875, 63.80316925048828, 91.18273162841797, -39.721168518066406, 11.185592651367188, 103.91229248046875, 45.50331115722656, 159.92877197265625, 145.12921142578125, 96.8330078125, 118.1899185180664, -69.18418884277344, -94.052490234375, 139.4617919921875, -45.2318115234375, 81.26421356201172, 227.8508758544922], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000309.npy"}
|
||||
{"epoch": 0.4671201814058957, "step": 310, "batch_size": 64, "mean": 26.756664276123047, "std": 80.21725463867188, "min": -141.8821563720703, "p10": -71.31149139404296, "median": 10.263044357299805, "p90": 143.57850189208986, "max": 253.05807495117188, "pos_frac": 0.578125, "sample": [164.6331329345703, -16.657546997070312, 19.400569915771484, 82.36257934570312, 44.01426696777344, -46.03728485107422, -63.58808898925781, 30.028228759765625, 145.02279663085938, 7.725519180297852, -82.42591857910156, -35.894775390625, 53.38355255126953, 159.36041259765625, -13.202880859375, -8.21164321899414, 4.356697082519531, 56.597137451171875, 190.99545288085938, -15.54736328125, 10.545612335205078, -74.62152099609375, -94.63167572021484, 253.05807495117188, -35.38471221923828, -15.1553955078125, 120.15206146240234, -8.1988525390625, 41.39192199707031, 57.94886016845703, -141.8821563720703, 57.693817138671875, -16.976119995117188, 137.69451904296875, -37.24864196777344, 46.070404052734375, -1.3697357177734375, 141.21144104003906, -4.089408874511719, 131.00759887695312, 28.154624938964844, 26.712295532226562, -62.74632263183594, -87.36494445800781, 117.95479583740234, 15.247268676757812, -17.81621551513672, 99.94002532958984, 8.885719299316406, -27.5361328125, 31.513038635253906, 144.59295654296875, 117.76634979248047, -85.09664916992188, 1.3996334075927734, -27.30852699279785, 76.28997802734375, 157.410888671875, 22.24518394470215, -39.500946044921875, -107.57208251953125, 81.96862030029297, -16.224367141723633, 9.980476379394531], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000310.npy"}
|
||||
{"epoch": 0.46863189720332576, "step": 311, "batch_size": 64, "mean": 52.9659423828125, "std": 93.13359069824219, "min": -200.79725646972656, "p10": -23.51016693115234, "median": 32.746110916137695, "p90": 152.60576782226565, "max": 489.8673400878906, "pos_frac": 0.78125, "sample": [118.30392456054688, 3.5367202758789062, 24.066978454589844, -3.6778335571289062, -58.754669189453125, -24.647315979003906, 32.658206939697266, 140.3167724609375, 4.98979377746582, 98.60489654541016, 25.453025817871094, -200.79725646972656, 6.7433319091796875, 129.94271850585938, 79.82647705078125, -20.3503475189209, 120.37640380859375, 42.066871643066406, 146.79129028320312, 60.12806701660156, 14.592735290527344, 82.9761962890625, 0.6646289825439453, 62.46034240722656, 166.6982421875, 35.70253372192383, -11.557384490966797, -6.227741241455078, -43.1680908203125, 73.1434326171875, -69.42182922363281, 212.7198486328125, 232.9303436279297, 90.1309585571289, 14.6485595703125, 50.88508605957031, 9.967002868652344, 155.09768676757812, 19.90020751953125, 80.72454833984375, 489.8673400878906, 6.6274261474609375, 16.894271850585938, 142.3062286376953, -46.27616882324219, 2.1505279541015625, 44.10174560546875, 1.7887344360351562, 42.239715576171875, 31.943504333496094, 175.18191528320312, 141.77145385742188, 60.53150939941406, -18.180463790893555, 45.97529602050781, 126.73823547363281, 3.6824474334716797, -81.18522644042969, -20.85681915283203, 32.834014892578125, -2.4464797973632812, 31.390602111816406, 156.53208923339844, 107.76296997070312], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000311.npy"}
|
||||
{"epoch": 0.47014361300075586, "step": 312, "batch_size": 64, "mean": 58.67571258544922, "std": 73.46771240234375, "min": -102.92060852050781, "p10": -33.35341720581053, "median": 42.419424057006836, "p90": 164.4765838623047, "max": 215.6845245361328, "pos_frac": 0.859375, "sample": [26.070125579833984, 10.054718017578125, -58.9217529296875, 39.634674072265625, 23.89618682861328, 30.253929138183594, 45.07558059692383, -16.208602905273438, 22.475265502929688, 16.290809631347656, 167.87362670898438, 69.05653381347656, 171.7938232421875, 58.65003204345703, 5.81781005859375, 124.4450454711914, 6.1365509033203125, 9.70203971862793, 17.12943458557129, 171.83645629882812, 156.07968139648438, 5.965675354003906, 21.98763656616211, 46.40485382080078, 49.419830322265625, 165.23854064941406, 22.672683715820312, -54.02738952636719, 95.02790832519531, 146.4674072265625, 162.6986846923828, -102.92060852050781, 54.887977600097656, 143.85861206054688, -66.98761749267578, 101.97604370117188, 130.24578857421875, 9.460182189941406, 128.35740661621094, 215.6845245361328, 173.1285400390625, 2.9513206481933594, 29.826141357421875, 120.53129577636719, 159.92715454101562, 75.84042358398438, 9.300941467285156, 144.69891357421875, 60.15135192871094, -45.704524993896484, -10.177047729492188, 128.15435791015625, 138.4929656982422, 53.22157287597656, 25.83880615234375, 87.16763305664062, 39.763267517089844, 31.61326789855957, -91.67242431640625, 165.49404907226562, 79.3939437866211, -40.701194763183594, 26.844520568847656, 17.600135803222656], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000312.npy"}
|
||||
{"epoch": 0.47165532879818595, "step": 313, "batch_size": 64, "mean": 31.566078186035156, "std": 82.02605438232422, "min": -170.8195343017578, "p10": -76.71177215576172, "median": 20.30817413330078, "p90": 136.05796203613284, "max": 230.119384765625, "pos_frac": 0.703125, "sample": [-13.827386856079102, 14.470390319824219, 97.15612030029297, -3.4188575744628906, 34.223594665527344, 79.3736801147461, 5.623386383056641, 59.44140625, 157.79660034179688, -73.02687072753906, 3.0393295288085938, 80.73912811279297, 130.1072998046875, 35.018310546875, 4.115058898925781, -28.3272705078125, 11.872344970703125, -78.291015625, -58.13471984863281, 7.139076232910156, 58.37268829345703, 115.60414123535156, 3.257692337036133, 184.29933166503906, -138.43115234375, -116.74372863769531, -65.40104675292969, 73.99508666992188, 105.373046875, 51.905784606933594, 180.4458770751953, 83.77928161621094, -114.12031555175781, -2.9347400665283203, 188.07362365722656, 23.537277221679688, 2.3629608154296875, 24.947322845458984, -21.753707885742188, 166.343017578125, 80.34478759765625, 8.253070831298828, 81.51167297363281, -19.999351501464844, 230.119384765625, 10.6483154296875, 17.079071044921875, -21.63378143310547, 54.09388732910156, 34.63445281982422, 112.8487319946289, 90.98054504394531, 34.53724670410156, -2.960132598876953, 1.959075927734375, -170.8195343017578, 89.68791961669922, -10.630603790283203, 138.60824584960938, -88.69297790527344, -92.79019165039062, 37.035919189453125, 13.566062927246094, 123.84513854980469], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000313.npy"}
|
||||
{"epoch": 0.47316704459561604, "step": 314, "batch_size": 64, "mean": 60.09541320800781, "std": 77.40853118896484, "min": -83.92184448242188, "p10": -23.96100692749023, "median": 40.53284454345703, "p90": 161.27807006835937, "max": 250.70809936523438, "pos_frac": 0.734375, "sample": [26.48388671875, 13.561344146728516, 44.39227294921875, 151.6283721923828, 91.08514404296875, 78.96186065673828, -9.243240356445312, 132.99612426757812, -25.296653747558594, 86.27021789550781, 20.275588989257812, -12.68197250366211, 218.77944946289062, 30.639907836914062, -12.683853149414062, 250.70809936523438, 99.20989990234375, 96.54862213134766, 152.1859130859375, -20.29418182373047, 11.097274780273438, -20.844497680664062, 145.68431091308594, 160.92037963867188, 181.91094970703125, -76.554443359375, 48.037635803222656, 194.60269165039062, 18.353836059570312, -29.959686279296875, 132.7616729736328, 41.23736572265625, 33.35853958129883, -6.37506103515625, 4.424619674682617, 58.57017135620117, 53.977294921875, -26.675338745117188, 67.60679626464844, 124.66279602050781, -11.933034896850586, -5.203544616699219, 163.51303100585938, 38.9192008972168, 128.25216674804688, 6.739166259765625, 120.73136901855469, 148.51553344726562, -20.032752990722656, 168.4444580078125, 79.32795715332031, 152.84353637695312, 119.37347412109375, 12.807418823242188, 135.2388916015625, 32.805877685546875, -62.42292785644531, 22.617881774902344, 39.82832336425781, -3.8817214965820312, 32.20966339111328, -60.421173095703125, -83.92184448242188, 161.43136596679688], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000314.npy"}
|
||||
{"epoch": 0.47467876039304613, "step": 315, "batch_size": 64, "mean": 44.43128967285156, "std": 78.44254302978516, "min": -147.95376586914062, "p10": -47.792305374145506, "median": 40.293636322021484, "p90": 142.5764343261719, "max": 197.43167114257812, "pos_frac": 0.71875, "sample": [-37.807655334472656, 86.39483642578125, 139.30697631835938, -3.285585403442383, -45.23859405517578, 27.868112564086914, -129.5546875, -2.232461929321289, -120.26544189453125, 83.7863540649414, -147.95376586914062, 39.47608184814453, -7.730091094970703, 165.8587646484375, -6.115108489990234, 35.028785705566406, 54.41841125488281, -43.961585998535156, 25.724071502685547, 2.5230274200439453, 37.088218688964844, 111.46488952636719, 1.1772899627685547, 65.76799011230469, 41.11119079589844, 88.74261474609375, -42.915985107421875, -24.479095458984375, 26.662273406982422, 103.26415252685547, 93.57398986816406, 87.43011474609375, 154.00100708007812, 7.844970703125, 46.018516540527344, 132.86865234375, 137.2079620361328, -75.52253723144531, -58.371978759765625, 47.22862243652344, 130.42974853515625, 14.751007080078125, 116.308349609375, 149.9267578125, -14.263336181640625, 91.24412536621094, 76.50811004638672, 168.26876831054688, -48.88675308227539, 28.779296875, 110.69964599609375, 3.5389633178710938, 197.43167114257812, 3.1234588623046875, 110.45179748535156, 20.365135192871094, 116.57810974121094, 70.82669067382812, 143.97763061523438, -76.6298599243164, 128.137451171875, -26.026113510131836, 183.34707641601562, 48.31158447265625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000315.npy"}
|
||||
{"epoch": 0.47619047619047616, "step": 316, "batch_size": 64, "mean": 67.00627136230469, "std": 81.50523376464844, "min": -105.54832458496094, "p10": -22.269221687316893, "median": 47.20731735229492, "p90": 170.66334381103516, "max": 248.18212890625, "pos_frac": 0.75, "sample": [5.5467071533203125, -8.90460205078125, 134.47459411621094, -51.36730194091797, -76.82373046875, -21.97699737548828, 111.613037109375, 36.76654052734375, 48.103553771972656, 140.71182250976562, 155.92034912109375, 248.18212890625, 133.9756622314453, 201.40768432617188, 33.597442626953125, -35.55949401855469, -105.54832458496094, 7.258049011230469, -9.885477066040039, 132.92442321777344, 11.193513870239258, 181.68307495117188, -18.274703979492188, 2.6927566528320312, 39.36546325683594, 84.92941284179688, 117.86181640625, 18.008346557617188, -9.584980010986328, 26.45758819580078, 156.10035705566406, 172.62278747558594, 105.35739135742188, 238.17686462402344, -5.560373306274414, 41.101348876953125, 25.1441650390625, 158.17733764648438, 115.39495086669922, 125.06684875488281, 149.77047729492188, 73.6049575805664, 92.56051635742188, -23.84756851196289, 186.0399932861328, 46.273948669433594, -1.9223213195800781, 203.1065673828125, 8.056503295898438, 54.404380798339844, -61.3702392578125, -22.394460678100586, 19.36534881591797, -19.29106903076172, 64.15007019042969, 46.31108093261719, -12.06893539428711, 147.11453247070312, 148.5339813232422, 99.90335845947266, 69.72357940673828, 45.04032897949219, 166.09130859375, 142.91519165039062], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000316.npy"}
|
||||
{"epoch": 0.47770219198790626, "step": 317, "batch_size": 64, "mean": 62.080379486083984, "std": 76.49890899658203, "min": -197.447021484375, "p10": -32.73036727905273, "median": 65.31198310852051, "p90": 158.25046691894534, "max": 186.20181274414062, "pos_frac": 0.8125, "sample": [136.65533447265625, 140.88804626464844, 61.86416244506836, 129.6439208984375, 100.65631103515625, 186.20181274414062, -18.68688201904297, 68.75980377197266, -72.63951110839844, 86.8835220336914, 8.914886474609375, 38.86892318725586, -34.358131408691406, 2.505828857421875, 57.7559814453125, 9.319446563720703, 116.54920959472656, 20.7943115234375, 134.10476684570312, 172.44137573242188, -35.04875183105469, -5.211204528808594, 48.82618713378906, -73.0233154296875, 76.20873260498047, 78.50294494628906, 154.103271484375, 101.29545593261719, 117.7454605102539, -197.447021484375, 146.12588500976562, 47.58000183105469, 89.99800872802734, 170.3824920654297, 159.27557373046875, -79.81460571289062, 71.87120056152344, -0.7900962829589844, 97.09178924560547, 178.0213623046875, 30.051776885986328, 108.00196838378906, 183.54495239257812, -9.693092346191406, 40.3983154296875, 9.696880340576172, 11.977897644042969, 145.44357299804688, 43.41047668457031, 6.76043701171875, 103.56248474121094, 16.85147476196289, 29.962919235229492, 155.85855102539062, -29.126487731933594, 80.10984802246094, -34.27488708496094, 88.23622131347656, 161.792724609375, 125.46209716796875, 0.750152587890625, 121.1107177734375, 30.852493286132812, 59.58619689941406], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000317.npy"}
|
||||
{"epoch": 0.47921390778533635, "step": 318, "batch_size": 64, "mean": 45.35422134399414, "std": 101.59390258789062, "min": -161.83306884765625, "p10": -95.82469177246094, "median": 31.321334838867188, "p90": 170.67458648681642, "max": 350.64825439453125, "pos_frac": 0.640625, "sample": [20.763301849365234, 116.65531158447266, -9.9530029296875, -161.83306884765625, 173.23553466796875, 145.99278259277344, 82.44241333007812, -58.37152099609375, 62.895538330078125, 165.93844604492188, 7.516727447509766, -95.44791412353516, 116.67485046386719, 110.53985595703125, 161.81005859375, -11.515140533447266, 27.128768920898438, -40.77593231201172, 60.17583465576172, 7.060636520385742, -11.481803894042969, 86.0905532836914, 174.80462646484375, 0.9043502807617188, 13.103004455566406, 130.90057373046875, 193.6925506591797, -128.6699676513672, 157.97711181640625, 128.11778259277344, -16.493919372558594, 20.215347290039062, -77.5278549194336, 65.60244750976562, 19.549209594726562, 350.64825439453125, 248.4321746826172, -58.90448760986328, 172.70436096191406, 29.012107849121094, -98.46224975585938, 50.21136474609375, 86.32050323486328, -135.70797729492188, -11.479278564453125, 109.86482238769531, -22.89238739013672, 202.57884216308594, -7.225921630859375, 110.36192321777344, -1.293313980102539, 159.32656860351562, 84.43165588378906, 82.74636840820312, -125.19230651855469, -26.809921264648438, 80.61456298828125, -109.52288818359375, 56.15812683105469, -5.776092529296875, 133.7058563232422, -95.98616790771484, -26.542293548583984, 33.63056182861328], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000318.npy"}
|
||||
{"epoch": 0.48072562358276644, "step": 319, "batch_size": 64, "mean": 64.24567413330078, "std": 90.62066650390625, "min": -131.08445739746094, "p10": -41.40685997009277, "median": 61.61906051635742, "p90": 169.96553955078124, "max": 322.87823486328125, "pos_frac": 0.734375, "sample": [124.05217742919922, 177.904296875, 7.656150817871094, 211.80377197265625, 32.568275451660156, 203.026123046875, -2.6490554809570312, 68.99703979492188, 44.75593566894531, 72.7064208984375, 145.71185302734375, 142.80902099609375, -8.484870910644531, 23.59949493408203, 38.21055603027344, 105.56805419921875, 8.698295593261719, 206.5338592529297, 8.249557495117188, 94.10601806640625, 9.698169708251953, -21.840301513671875, 71.58940124511719, 96.50004577636719, 261.50537109375, -17.978363037109375, 154.1954803466797, 116.1399917602539, 170.05227661132812, 158.26548767089844, -65.88798522949219, -108.75346374511719, 112.14814758300781, 322.87823486328125, -10.557636260986328, 103.17747497558594, -58.096221923828125, 88.96817016601562, 91.24491119384766, -42.511600494384766, 160.89732360839844, 54.24108123779297, -21.86559295654297, -8.425666809082031, 22.342857360839844, 34.550636291503906, -131.08445739746094, -21.050460815429688, -72.25439453125, 123.79753112792969, -24.36620330810547, 40.2703857421875, 39.94010925292969, -38.829132080078125, 48.326133728027344, 128.7874298095703, 104.07154846191406, 141.2992706298828, 139.3391876220703, 23.042306900024414, 73.60823059082031, -101.20105743408203, 169.76315307617188, 89.96223449707031], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000319.npy"}
|
||||
{"epoch": 0.48223733938019653, "step": 320, "batch_size": 64, "mean": 52.74368667602539, "std": 89.73084259033203, "min": -154.2203826904297, "p10": -31.743913269042967, "median": 24.256553649902344, "p90": 152.92092742919922, "max": 395.7449951171875, "pos_frac": 0.671875, "sample": [108.05787658691406, -67.59297180175781, 2.1361160278320312, 85.16114807128906, 30.3143310546875, 94.51310729980469, 162.18069458007812, 1.4357833862304688, 102.84793853759766, -9.068029403686523, -89.88606262207031, 147.65963745117188, -2.0253658294677734, -1.6270599365234375, -29.386070251464844, 148.89544677734375, 110.6155776977539, 6.784996032714844, -16.224716186523438, 135.05287170410156, -24.046728134155273, 2.224681854248047, -154.2203826904297, 6.72166633605957, 27.598831176757812, 395.7449951171875, 188.70480346679688, 129.85714721679688, 154.64613342285156, -11.639328002929688, 57.6683349609375, -16.550992965698242, 5.901973724365234, 61.46435546875, -1.4118824005126953, 145.79769897460938, 15.651538848876953, -47.832176208496094, -8.324291229248047, -11.512821197509766, 59.726261138916016, 191.55616760253906, 9.279388427734375, 64.02470397949219, 15.071182250976562, 67.23707580566406, 193.2334747314453, 61.834083557128906, -12.626235961914062, -11.163846969604492, 3.9935760498046875, -80.0792007446289, 225.34201049804688, -32.754417419433594, -22.866718292236328, 95.60487365722656, 20.914276123046875, 114.92153930664062, 130.96090698242188, 111.11638641357422, 105.53144836425781, -37.358917236328125, 130.10678100585938, 135.70230102539062], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000320.npy"}
|
||||
{"epoch": 0.4837490551776266, "step": 321, "batch_size": 64, "mean": 55.908905029296875, "std": 83.24577331542969, "min": -151.1898956298828, "p10": -59.804413604736325, "median": 49.832305908203125, "p90": 168.88433532714845, "max": 203.97442626953125, "pos_frac": 0.765625, "sample": [119.1585693359375, 116.77437591552734, 6.5676422119140625, 146.38107299804688, 149.88848876953125, -45.46143341064453, 170.26048278808594, 77.0499038696289, 165.67332458496094, 48.714256286621094, 163.8543701171875, 45.47603225708008, 76.21824645996094, 128.76077270507812, -127.44796752929688, 17.23769760131836, -4.436363220214844, 108.57123565673828, 55.900634765625, 193.261474609375, 26.300003051757812, 40.93305969238281, -75.81986236572266, -2.6562042236328125, 98.6501693725586, -109.55043029785156, -59.28545379638672, -112.09426879882812, 94.22333526611328, 35.64076232910156, 42.90058898925781, 111.9703369140625, 114.74195861816406, 91.13320922851562, 31.308868408203125, 92.34683227539062, 71.96653747558594, -0.9491291046142578, 59.7754020690918, 185.67242431640625, 5.494026184082031, 5.075771331787109, 42.50596237182617, 93.0078353881836, -7.1106414794921875, -151.1898956298828, 144.3389892578125, 170.650146484375, 170.42996215820312, 82.30332946777344, 12.926795959472656, -1.391592025756836, 50.950355529785156, 170.9083709716797, 203.97442626953125, -5.876182556152344, 37.37987518310547, 48.60285949707031, 165.2930145263672, -66.05298614501953, 24.28949737548828, 4.2471771240234375, 87.82891082763672, -60.026824951171875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000321.npy"}
|
||||
{"epoch": 0.4852607709750567, "step": 322, "batch_size": 64, "mean": 69.25843811035156, "std": 74.67259216308594, "min": -133.565185546875, "p10": -15.495320892333982, "median": 59.49514389038086, "p90": 163.69366607666015, "max": 186.57412719726562, "pos_frac": 0.828125, "sample": [-12.144674301147461, 111.14173126220703, 164.54891967773438, 25.960479736328125, 109.28498840332031, 12.361690521240234, 186.57412719726562, 131.97702026367188, 149.55735778808594, 30.660612106323242, 116.0976333618164, 69.98486328125, 160.06317138671875, 164.41053771972656, 139.31375122070312, -0.33260345458984375, 36.807708740234375, 36.91943359375, 4.206470489501953, 42.3653564453125, 90.8469009399414, -13.326774597167969, -102.84335327148438, 152.19720458984375, 161.72903442382812, 23.747112274169922, 47.2073974609375, 54.753578186035156, 0.781982421875, -26.1156005859375, -16.424697875976562, 154.02366638183594, 42.60948181152344, 148.18841552734375, 176.52914428710938, 99.59820556640625, 18.095375061035156, 165.07000732421875, 22.766708374023438, -10.2611083984375, 83.09706115722656, 90.49219512939453, 3.4904403686523438, 154.9954071044922, 14.558891296386719, 151.19857788085938, 162.02096557617188, 155.00918579101562, -24.358322143554688, -133.565185546875, 78.10432434082031, 30.446029663085938, 31.121280670166016, 56.37071990966797, 167.6240997314453, 62.61956787109375, 56.09929656982422, 90.78334045410156, 127.89462280273438, -69.81723022460938, 169.83082580566406, -30.159828186035156, 109.48287963867188, 26.26951789855957], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000322.npy"}
|
||||
{"epoch": 0.48677248677248675, "step": 323, "batch_size": 64, "mean": 44.99114227294922, "std": 94.26055908203125, "min": -162.09683227539062, "p10": -69.80100708007811, "median": 24.35330581665039, "p90": 174.88481140136722, "max": 249.78359985351562, "pos_frac": 0.671875, "sample": [-3.6194229125976562, -94.557373046875, 74.1385498046875, 26.36071014404297, -41.06800079345703, 166.541259765625, 54.93980407714844, 8.80063247680664, 17.576988220214844, 178.37535095214844, 41.970611572265625, -36.7020263671875, 163.53262329101562, 0.17893218994140625, 126.16951751708984, -26.265888214111328, -49.21368408203125, 143.03692626953125, 107.73450469970703, 22.099456787109375, 69.94355773925781, 166.74021911621094, 77.58241271972656, 27.864013671875, 188.4415283203125, -47.26396942138672, -103.37062072753906, 18.098527908325195, -3.9495487213134766, -78.6241455078125, 249.78359985351562, 1.2614288330078125, 16.156143188476562, -79.02569580078125, 184.03076171875, 36.027503967285156, -98.1409912109375, 159.85394287109375, 125.54220581054688, 0.6415348052978516, 25.301109313964844, 24.70514678955078, -13.687850952148438, 2.1946487426757812, 198.74600219726562, 213.04811096191406, 153.26211547851562, -149.21026611328125, 244.7978515625, 6.988212585449219, -162.09683227539062, -10.295890808105469, 150.9995574951172, 32.573394775390625, 24.00146484375, 101.4940185546875, 132.76113891601562, 117.3055419921875, -10.642837524414062, -2.9148826599121094, 45.30515670776367, -26.273635864257812, -8.64306640625, -1.9070510864257812], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000323.npy"}
|
||||
{"epoch": 0.48828420256991684, "step": 324, "batch_size": 64, "mean": 58.11041259765625, "std": 77.96807098388672, "min": -166.34146118164062, "p10": -21.982798194885245, "median": 52.51810836791992, "p90": 155.40052642822266, "max": 243.25399780273438, "pos_frac": 0.796875, "sample": [69.22785186767578, 56.82896423339844, 75.55494689941406, 20.61817169189453, -88.98211669921875, 173.56285095214844, 64.01768493652344, 75.77001953125, -49.622955322265625, -3.6050186157226562, 2.8821334838867188, 123.19180297851562, 71.17265319824219, 58.34030532836914, 115.47018432617188, 3.3771400451660156, 154.507080078125, -28.774681091308594, 124.88762664794922, -6.507659912109375, 76.0294189453125, 145.99560546875, 0.8723335266113281, 119.17180633544922, 42.47936248779297, 10.592193603515625, -12.084602355957031, 12.804901123046875, 231.55685424804688, 16.371097564697266, 9.557235717773438, -166.34146118164062, 70.65284729003906, 131.67474365234375, 39.52784729003906, 88.24845886230469, 31.074119567871094, -26.224882125854492, 105.42533874511719, 93.72669219970703, 40.75921630859375, -65.49607849121094, 243.25399780273438, 27.11096954345703, 201.53256225585938, 17.829402923583984, 7.852848052978516, 155.78343200683594, 89.4146728515625, 133.0023193359375, 193.63059997558594, 2.138050079345703, 5.139854431152344, 130.2753448486328, 138.70730590820312, -10.643203735351562, 48.207252502441406, -6.3301849365234375, -41.97119903564453, 81.07769775390625, 159.4144744873047, 134.66163635253906, -2.5952377319335938, 3.2837047576904297], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000324.npy"}
|
||||
{"epoch": 0.4897959183673469, "step": 325, "batch_size": 64, "mean": 59.03926467895508, "std": 100.42520141601562, "min": -243.77537536621094, "p10": -44.06515045166015, "median": 45.90712547302246, "p90": 151.7736053466797, "max": 387.37982177734375, "pos_frac": 0.765625, "sample": [31.813861846923828, 115.97050476074219, 5.349407196044922, 138.47329711914062, -31.293663024902344, 86.96720886230469, 102.81816101074219, 69.0766830444336, 36.709442138671875, 95.58663177490234, 72.81634521484375, 140.46449279785156, 1.7441234588623047, -28.967208862304688, 125.42680358886719, -15.931407928466797, 144.33094787597656, 137.92100524902344, 129.2515869140625, -132.68841552734375, 189.792236328125, -17.28540802001953, 126.86189270019531, 2.2848644256591797, -1.0213584899902344, 253.23367309570312, 32.244510650634766, 152.13134765625, -55.39627456665039, 76.94599151611328, -91.82823181152344, 49.3707389831543, 118.13272857666016, 108.78598022460938, -97.25174713134766, 99.562744140625, -96.45652770996094, 103.28436279296875, 59.15522003173828, 387.37982177734375, 86.40467834472656, 27.710905075073242, 21.203887939453125, 294.7122802734375, -243.77537536621094, 12.791679382324219, 46.598236083984375, 24.285484313964844, 150.93887329101562, 108.57091522216797, 164.64065551757812, -10.24114990234375, 25.88612174987793, -12.1558837890625, 38.22582244873047, -39.217498779296875, 28.566268920898438, 45.21601486206055, 263.64678955078125, 56.768951416015625, 41.27162170410156, 28.876182556152344, 37.9638671875, -46.14271545410156], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000325.npy"}
|
||||
{"epoch": 0.491307634164777, "step": 326, "batch_size": 64, "mean": 22.577842712402344, "std": 85.61376190185547, "min": -148.38632202148438, "p10": -86.55392150878906, "median": 18.722086906433105, "p90": 138.2199859619141, "max": 213.4132080078125, "pos_frac": 0.578125, "sample": [-40.371665954589844, -14.825252532958984, 9.26047134399414, 24.23055648803711, 20.3155517578125, -148.38632202148438, 194.96327209472656, 213.4132080078125, 1.675628662109375, 75.6729965209961, -30.453125, 92.22868347167969, 27.049510955810547, 24.947296142578125, 68.87435913085938, 128.9693603515625, 17.12862205505371, 93.81846618652344, -17.763755798339844, 63.10015869140625, 142.18453979492188, -14.707534790039062, 3.031219482421875, 82.21426391601562, -59.27661895751953, -49.018341064453125, -100.97491455078125, -16.58050537109375, -88.99652099609375, -9.190521240234375, 171.42242431640625, -138.55401611328125, 164.54751586914062, -20.518512725830078, 42.51624298095703, 20.478988647460938, 3.2851181030273438, -56.7318115234375, -55.048484802246094, -27.9488525390625, 127.02415466308594, 156.60195922851562, -80.85452270507812, -16.077945709228516, 121.23799133300781, 24.72107696533203, 32.44528579711914, -103.1445541381836, 99.09455108642578, 163.51329040527344, -12.794872283935547, -15.344621658325195, -130.0637664794922, 105.7715072631836, -70.7587890625, -8.901100158691406, -124.7584228515625, 96.8509521484375, 65.35308837890625, 52.231842041015625, 88.03433227539062, -71.68937683105469, 29.276901245117188, 121.23138427734375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000326.npy"}
|
||||
{"epoch": 0.4928193499622071, "step": 327, "batch_size": 64, "mean": 46.449100494384766, "std": 91.23785400390625, "min": -172.9596710205078, "p10": -70.95796012878417, "median": 36.74678993225098, "p90": 163.05148468017578, "max": 237.385009765625, "pos_frac": 0.71875, "sample": [24.26714324951172, -108.12813568115234, -25.33692169189453, 214.90628051757812, 18.214065551757812, 72.77545166015625, -120.87564086914062, 91.57781219482422, -75.76639556884766, 86.76472473144531, 177.29931640625, 34.26199722290039, -147.25979614257812, 136.19078063964844, 52.345909118652344, 162.51760864257812, 153.90908813476562, 15.75954818725586, 102.74455261230469, -41.72297286987305, 1.2267627716064453, 22.06363296508789, 71.91161346435547, 219.65316772460938, 163.28028869628906, 9.757278442382812, 51.9537353515625, 84.22189331054688, 64.69487762451172, 39.302276611328125, 29.257308959960938, 19.42544937133789, -89.49098205566406, 159.390869140625, 116.31863403320312, 16.97991180419922, 58.68125915527344, 52.63691711425781, -14.255790710449219, -2.313365936279297, 186.5326385498047, 14.476882934570312, -172.9596710205078, 7.3844146728515625, 145.47195434570312, 237.385009765625, -16.277587890625, 150.75253295898438, 190.36761474609375, 39.23158264160156, 141.78500366210938, -3.932699203491211, 23.63817596435547, 8.438972473144531, -59.738277435302734, -27.6063175201416, 121.4692153930664, 78.2387924194336, 56.69769287109375, -21.31460189819336, -28.835174560546875, -101.91472625732422, 126.30377197265625, -21.99298858642578], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000327.npy"}
|
||||
{"epoch": 0.4943310657596372, "step": 328, "batch_size": 64, "mean": 57.80853271484375, "std": 77.45532989501953, "min": -110.32816314697266, "p10": -24.71194229125976, "median": 43.35427284240723, "p90": 156.0927505493164, "max": 240.51220703125, "pos_frac": 0.765625, "sample": [132.26882934570312, 4.268983840942383, 7.125036239624023, 28.921043395996094, 152.0604705810547, 10.138097763061523, -10.429622650146484, 129.541259765625, 71.71924591064453, 30.71930694580078, 60.89086151123047, 204.78082275390625, 135.45254516601562, 37.375057220458984, 101.9195556640625, -73.3563232421875, 165.49081420898438, 5.264455795288086, 87.63111877441406, -4.616588592529297, 70.0828857421875, -20.43280029296875, 2.512392044067383, 41.207698822021484, -39.08583068847656, -26.545860290527344, 240.51220703125, 62.381370544433594, -95.9676513671875, 47.93291473388672, -110.32816314697266, 129.27978515625, 97.21318054199219, 1.491668701171875, 156.52293395996094, 3.1579208374023438, 36.672515869140625, -3.6840057373046875, 233.51461791992188, 17.075408935546875, 126.54756164550781, 187.3611602783203, -3.70953369140625, 18.41878318786621, -18.729629516601562, 147.52752685546875, -10.668500900268555, 46.89906311035156, -30.936935424804688, 28.892215728759766, 87.01334381103516, -40.0748291015625, 82.25714874267578, 58.8839111328125, 35.018836975097656, 108.51007080078125, 103.15064239501953, -3.6947994232177734, 148.69725036621094, 45.50084686279297, 138.36309814453125, 3.6030006408691406, 165.14871215820312, 155.0889892578125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000328.npy"}
|
||||
{"epoch": 0.4958427815570673, "step": 329, "batch_size": 64, "mean": 57.826499938964844, "std": 89.23344421386719, "min": -140.40850830078125, "p10": -25.55212059020996, "median": 27.17682647705078, "p90": 180.05191345214845, "max": 253.96038818359375, "pos_frac": 0.765625, "sample": [4.349580764770508, 6.569080352783203, 75.44911193847656, 22.022064208984375, -88.15435791015625, 19.806907653808594, 253.96038818359375, -19.4093017578125, 178.11322021484375, -22.570762634277344, 126.94033813476562, 66.02813720703125, -3.5716304779052734, 116.28768920898438, 1.7154293060302734, 21.543434143066406, 34.95387268066406, 212.17385864257812, -56.0047607421875, 33.60316467285156, 15.962387084960938, 12.572725296020508, 180.88278198242188, 122.72491455078125, -44.060935974121094, 173.56436157226562, 182.78514099121094, 149.19680786132812, 32.0614013671875, 43.93861389160156, 163.42523193359375, 118.52786254882812, -22.060073852539062, 235.95025634765625, 2.5593929290771484, 22.292251586914062, 200.24667358398438, 7.097379684448242, 160.68101501464844, 85.92527770996094, 229.88973999023438, 162.00791931152344, 3.0358848571777344, -88.03770446777344, -17.65607452392578, 136.838623046875, 4.6206817626953125, 37.63936233520508, 8.851974487304688, 41.87819290161133, -26.829845428466797, 115.75286865234375, -17.254318237304688, -15.489940643310547, 171.53274536132812, 153.94259643554688, 4.863515853881836, 122.72504425048828, -14.103286743164062, -140.40850830078125, 47.09748840332031, 7.63604736328125, 1.303976058959961, -57.01963806152344], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000329.npy"}
|
||||
{"epoch": 0.4973544973544973, "step": 330, "batch_size": 64, "mean": 57.41804885864258, "std": 99.51565551757812, "min": -177.28411865234375, "p10": -64.95229034423828, "median": 55.43199157714844, "p90": 180.16872711181645, "max": 215.065673828125, "pos_frac": 0.6875, "sample": [-15.025466918945312, -60.3121337890625, -19.731557846069336, -140.58294677734375, 44.93125915527344, -177.28411865234375, 148.4499969482422, 6.107032775878906, -61.21288299560547, 154.60800170898438, 156.28875732421875, 62.3319091796875, -97.27742767333984, -14.392341613769531, 46.56052017211914, 199.40890502929688, 183.94830322265625, -100.76737213134766, 215.065673828125, 158.2887420654297, 142.90499877929688, 4.722572326660156, 168.78146362304688, -3.5712833404541016, 155.22015380859375, 201.9721221923828, 12.900787353515625, 13.477264404296875, -66.55489349365234, 19.823280334472656, 105.28346252441406, 146.64556884765625, -26.25347900390625, 77.45895385742188, -77.809814453125, 184.6766357421875, 98.97249603271484, 189.165283203125, 161.7029266357422, 12.13760757446289, 11.827415466308594, 102.17372131347656, 48.532073974609375, 88.68026733398438, 149.853515625, 152.42691040039062, 3.0325756072998047, 85.1221923828125, -25.72637939453125, -46.84785461425781, -8.567840576171875, 76.4322509765625, 161.45953369140625, 154.98907470703125, -47.49345397949219, 38.74232482910156, 66.50422668457031, -52.222137451171875, 188.63421630859375, -142.12193298339844, 142.32412719726562, 152.08111572265625, -7.4893341064453125, 171.34971618652344], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000330.npy"}
|
||||
{"epoch": 0.4988662131519274, "step": 331, "batch_size": 64, "mean": 71.51943969726562, "std": 87.96726989746094, "min": -177.1209259033203, "p10": -21.120959472656242, "median": 65.97774124145508, "p90": 171.4195297241211, "max": 321.06103515625, "pos_frac": 0.828125, "sample": [181.44032287597656, 9.553176879882812, 46.51143264770508, 116.42646789550781, 131.75189208984375, 196.47824096679688, 136.1915740966797, 131.85882568359375, 109.2149658203125, -0.597686767578125, 18.468151092529297, 216.39845275878906, 172.0778350830078, 164.1050567626953, 89.63600158691406, 18.885658264160156, 189.19866943359375, 63.26997375488281, 28.69994354248047, 169.88348388671875, 124.398681640625, 3.2611770629882812, 25.496795654296875, 36.109619140625, -12.629646301269531, 147.64578247070312, 92.51199340820312, 321.06103515625, 50.81123352050781, 51.798004150390625, 82.68556213378906, 68.68550872802734, 166.70359802246094, 119.50936889648438, -24.760093688964844, 7.112068176269531, 87.51461029052734, -122.36846923828125, -82.08518981933594, 117.25301361083984, -43.04583740234375, 180.9424591064453, 53.34175109863281, -2.73187255859375, 30.390039443969727, -177.1209259033203, 10.832572937011719, 11.06329345703125, -36.253501892089844, 54.935157775878906, 43.52246856689453, 132.4365692138672, 129.88333129882812, 3.3529815673828125, 62.18712615966797, 10.665657043457031, -0.8181562423706055, 157.77581787109375, 106.74166870117188, -114.67848205566406, 160.1355743408203, 105.20603942871094, 84.36056518554688, 163.95278930664062], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000331.npy"}
|
||||
{"epoch": 0.5003779289493575, "step": 332, "batch_size": 64, "mean": 59.41328811645508, "std": 84.23377227783203, "min": -124.38233947753906, "p10": -15.756678771972656, "median": 40.25319290161133, "p90": 179.1894424438477, "max": 250.023681640625, "pos_frac": 0.734375, "sample": [-22.974498748779297, -5.716697692871094, -6.174125671386719, -2.287109375, 2.6426868438720703, 170.755615234375, 41.43794250488281, 30.127824783325195, 54.48283386230469, 56.24982833862305, -25.07959747314453, 250.023681640625, 204.6489715576172, 94.39517211914062, 148.16415405273438, 39.523582458496094, 25.21856689453125, -1.4986801147460938, 127.77494812011719, 44.0704345703125, 75.9275131225586, 33.68061828613281, -113.50459289550781, 159.5715789794922, 111.5594482421875, 133.43429565429688, -124.38233947753906, 2.2549514770507812, 17.683242797851562, -10.123390197753906, -5.950691223144531, 167.33694458007812, 61.82403564453125, 2.1378021240234375, -15.647186279296875, 22.50543212890625, 5.534677505493164, 206.42071533203125, -95.5757064819336, 10.994409561157227, 99.17764282226562, 98.93185424804688, 129.27516174316406, -6.59807014465332, 33.571014404296875, 3.6689529418945312, 154.570556640625, 138.83889770507812, 157.4627685546875, 57.653839111328125, 238.47499084472656, 9.566665649414062, 81.98600769042969, -4.919761657714844, 24.68689727783203, -15.803604125976562, 186.71554565429688, -7.049770355224609, 40.98280334472656, 182.80393981933594, 239.22286987304688, 68.09391784667969, -33.74686050415039, 53.41705322265625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000332.npy"}
|
||||
{"epoch": 0.5018896447467877, "step": 333, "batch_size": 64, "mean": 68.53299713134766, "std": 85.40159606933594, "min": -137.26226806640625, "p10": -23.675155258178705, "median": 62.790584564208984, "p90": 182.09623107910159, "max": 294.9146423339844, "pos_frac": 0.796875, "sample": [6.849555969238281, -30.078298568725586, 33.10530090332031, 134.1453857421875, 35.6094970703125, 0.35448455810546875, 28.560161590576172, 149.38815307617188, 8.9200439453125, -4.75555419921875, 77.86968231201172, 91.82150268554688, 92.34392547607422, 143.53512573242188, 155.09197998046875, 53.969879150390625, 15.58038330078125, 194.59486389160156, 246.53701782226562, 184.56874084472656, -7.5604705810546875, 99.80616760253906, 73.39302825927734, 65.49494934082031, 80.16513061523438, -83.38887786865234, -25.619068145751953, 49.4542236328125, -0.91143798828125, 63.36492156982422, 192.80332946777344, 163.6865234375, -32.60966491699219, 95.64200592041016, 128.5693817138672, 63.48012924194336, 224.46359252929688, 12.810478210449219, 20.30899429321289, 62.21624755859375, -3.059114456176758, 37.99320983886719, -137.26226806640625, 8.251287460327148, 56.206764221191406, -86.21333312988281, 294.9146423339844, 16.95183563232422, -19.139358520507812, 65.1659927368164, -15.700965881347656, -26.67254638671875, 111.68560791015625, 176.32704162597656, 5.683969497680664, 0.11469268798828125, 154.41964721679688, 36.612159729003906, 161.3270263671875, 79.21296691894531, 242.71240234375, 89.9460220336914, 150.1329345703125, 122.92011260986328], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000333.npy"}
|
||||
{"epoch": 0.5034013605442177, "step": 334, "batch_size": 64, "mean": 49.529945373535156, "std": 92.34193420410156, "min": -159.0748291015625, "p10": -66.53589324951172, "median": 31.534027099609375, "p90": 177.7447021484375, "max": 282.9754638671875, "pos_frac": 0.65625, "sample": [-70.27046966552734, -159.0748291015625, 71.2866439819336, 123.14936065673828, 84.80079650878906, -27.131423950195312, -101.25228881835938, 183.72146606445312, 13.806716918945312, -152.62869262695312, -4.9134674072265625, 282.9754638671875, -12.369937896728516, 117.36473083496094, 5.845695495605469, 122.3075180053711, -54.577392578125, -14.312204360961914, 69.96817016601562, 102.66346740722656, 132.09320068359375, 189.05380249023438, 46.408714294433594, 102.94204711914062, 72.30216979980469, -70.15190124511719, 28.497703552246094, -1.351522445678711, 125.30528259277344, 9.449405670166016, -2.980226516723633, 2.2046051025390625, 177.89706420898438, -58.098541259765625, 117.98303985595703, 168.8148193359375, 50.519195556640625, 28.81719970703125, 57.30047607421875, -27.15576934814453, 17.93628692626953, -33.618797302246094, -19.28863525390625, -40.22454833984375, 186.7552490234375, -0.11309051513671875, -5.384056091308594, 4.449714660644531, -77.88175964355469, 122.03291320800781, 137.88076782226562, 66.49405670166016, -8.492156982421875, 107.42322540283203, -75.45465087890625, 34.2508544921875, 74.53446960449219, 191.56301879882812, 23.24530029296875, 25.3637638092041, 177.38919067382812, 154.07022094726562, 211.91485595703125, 163.86016845703125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000334.npy"}
|
||||
{"epoch": 0.5049130763416477, "step": 335, "batch_size": 64, "mean": 53.92637634277344, "std": 94.44384765625, "min": -139.83279418945312, "p10": -46.475450897216795, "median": 31.981792449951172, "p90": 178.29115142822266, "max": 282.7897033691406, "pos_frac": 0.703125, "sample": [80.50163269042969, 85.9050521850586, -35.77516174316406, 243.06277465820312, 17.271881103515625, 190.56094360351562, 39.08396911621094, 4.124208450317383, 282.7897033691406, 119.4585189819336, 189.64852905273438, 150.32080078125, 76.62162780761719, 71.55061340332031, 179.00643920898438, 165.81044006347656, 23.326683044433594, 73.83795166015625, 46.974891662597656, 176.6221466064453, -46.87240982055664, 3.922943115234375, 37.25305938720703, 193.03517150878906, -96.63856506347656, -12.75067138671875, 148.30490112304688, -12.119888305664062, -39.4031982421875, 73.88944244384766, 156.08363342285156, 1.3556747436523438, 14.81707763671875, 83.16791534423828, 162.8585205078125, -88.33695983886719, 176.42453002929688, -2.7214088439941406, -3.1552658081054688, 176.55068969726562, 10.2353515625, -15.909074783325195, 6.2978057861328125, 51.43524932861328, 113.54838562011719, 108.32947540283203, -38.999839782714844, -139.83279418945312, 24.750019073486328, 0.6579685211181641, 26.710525512695312, -45.54921340942383, -11.352317810058594, -135.03924560546875, -109.689208984375, -66.2186279296875, 187.4633331298828, 157.53912353515625, 1.3479461669921875, 86.99392700195312, -10.771919250488281, 5.738471984863281, -2.7485427856445312, 139.98245239257812], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000335.npy"}
|
||||
{"epoch": 0.5064247921390779, "step": 336, "batch_size": 64, "mean": 57.68294906616211, "std": 100.77291870117188, "min": -119.31135559082031, "p10": -43.635927581787094, "median": 29.695037841796875, "p90": 192.71974182128906, "max": 381.58770751953125, "pos_frac": 0.671875, "sample": [-77.90524291992188, -4.08917236328125, 210.20745849609375, 53.41838073730469, 381.58770751953125, 193.1136016845703, 122.830810546875, -0.30632781982421875, 176.50827026367188, -2.022878646850586, 15.672426223754883, -4.303689956665039, 191.8007354736328, -10.531570434570312, 180.59815979003906, 15.344432830810547, 15.028427124023438, 54.92362976074219, 223.16567993164062, -119.31135559082031, -12.802833557128906, 78.72640991210938, 228.03115844726562, 42.590248107910156, -20.279632568359375, 134.2334442138672, 34.247074127197266, -111.4911117553711, -24.018402099609375, -19.581398010253906, 212.76866149902344, -109.40438079833984, 57.85649871826172, 38.74864196777344, 6.017993927001953, 7.167280197143555, -3.307342529296875, -5.884193420410156, 51.10999298095703, 148.44764709472656, 13.192703247070312, -27.863609313964844, 124.00401306152344, 12.187644958496094, 152.10801696777344, 46.592987060546875, 69.30964660644531, -4.738037109375, 18.428495407104492, 142.95118713378906, 20.365623474121094, 0.42657470703125, 128.72573852539062, 170.3346405029297, 313.2386474609375, 25.775711059570312, -59.78326416015625, -16.971435546875, 143.69989013671875, -50.39549255371094, 43.668060302734375, 122.09724426269531, 33.61436462402344, -78.16587829589844], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000336.npy"}
|
||||
{"epoch": 0.5079365079365079, "step": 337, "batch_size": 64, "mean": 46.51734924316406, "std": 91.95816802978516, "min": -161.96920776367188, "p10": -63.10958633422851, "median": 31.27747344970703, "p90": 168.6347137451172, "max": 253.89193725585938, "pos_frac": 0.703125, "sample": [17.60912322998047, -96.02501678466797, -161.96920776367188, -10.625770568847656, 33.50056457519531, 140.71441650390625, -12.559946060180664, 141.58926391601562, 47.29498291015625, -10.943473815917969, 215.34637451171875, 12.684532165527344, 74.2569808959961, -145.60836791992188, -81.86834716796875, -30.67290496826172, 29.05438232421875, 237.46401977539062, 25.28839874267578, -17.139816284179688, 7.473594665527344, 2.4272937774658203, -7.0995635986328125, -93.76519775390625, 144.275634765625, 97.66883850097656, 8.743865966796875, 15.815269470214844, 34.9478759765625, 228.74806213378906, 6.863639831542969, 4.111045837402344, 161.26187133789062, -12.593284606933594, 116.09429931640625, 110.64816284179688, 195.4592742919922, -33.70951843261719, -8.268928527832031, 86.84707641601562, 143.88931274414062, 7.384498596191406, 170.2112274169922, 47.484161376953125, 112.16952514648438, -31.815383911132812, 84.30902862548828, 55.78358459472656, 82.88595581054688, 43.90687561035156, 97.47386169433594, 253.89193725585938, -5.158653259277344, -59.54437255859375, 18.754379272460938, 170.07102966308594, 6.808359146118164, -64.63753509521484, 109.8022689819336, 114.73258209228516, -124.36271667480469, 60.6990966796875, 165.28330993652344, 43.74863815307617], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000337.npy"}
|
||||
{"epoch": 0.509448223733938, "step": 338, "batch_size": 64, "mean": 48.56975555419922, "std": 86.48896789550781, "min": -188.97813415527344, "p10": -63.747180938720696, "median": 38.80610466003418, "p90": 167.91599884033207, "max": 218.13613891601562, "pos_frac": 0.75, "sample": [-65.69882202148438, 40.67171096801758, 209.09530639648438, 160.3441619873047, 150.77947998046875, 136.34844970703125, -107.71282958984375, 110.32393646240234, 42.88079833984375, 179.3402862548828, 194.9534912109375, 26.85547637939453, 4.534523010253906, 78.6790771484375, 2.8862762451171875, -87.87029266357422, 11.762619018554688, -124.91973876953125, -1.61322021484375, 1.7355422973632812, -69.97515106201172, 150.07839965820312, 42.246585845947266, -6.659883499145508, 32.33055877685547, 86.42984008789062, -188.97813415527344, 102.23432159423828, 90.3348388671875, 83.75395965576172, -15.130683898925781, 118.31060791015625, 8.909366607666016, 58.3144645690918, 87.42841339111328, 116.33638000488281, 1.78802490234375, 29.174468994140625, 218.13613891601562, 26.064346313476562, 41.622642517089844, 157.66262817382812, -10.88665771484375, 126.877197265625, 17.68514633178711, 30.500228881835938, -59.19335174560547, -42.319854736328125, 67.45868682861328, -23.56760025024414, 36.94049835205078, 187.1372833251953, 6.014438629150391, 7.4203338623046875, 89.75056457519531, 108.53480529785156, 27.381423950195312, 173.7967529296875, 115.54117584228516, -85.82229614257812, -42.773704528808594, 78.16948699951172, 171.16107177734375, -5.129467010498047], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000338.npy"}
|
||||
{"epoch": 0.5109599395313681, "step": 339, "batch_size": 64, "mean": 58.057308197021484, "std": 88.03205871582031, "min": -159.50064086914062, "p10": -49.23199157714842, "median": 47.95565605163574, "p90": 174.01166229248048, "max": 230.77236938476562, "pos_frac": 0.796875, "sample": [57.47705078125, 52.4515380859375, 29.98302459716797, 10.9791259765625, -21.29938507080078, 167.86065673828125, -19.3521728515625, 73.86978149414062, 21.543750762939453, -101.72772979736328, 50.23846435546875, 19.433517456054688, 14.670402526855469, 27.047199249267578, 33.48257827758789, 41.41001892089844, -95.06036376953125, 186.6104736328125, 173.7271270751953, 12.736749649047852, 167.0753173828125, 79.53755950927734, 139.45240783691406, 104.85454559326172, 64.8682632446289, 174.13360595703125, -0.585968017578125, 33.49626541137695, -87.2695083618164, 79.18377685546875, 47.4404411315918, 136.26657104492188, 203.56341552734375, 186.01348876953125, 70.49996185302734, 2.554363250732422, 153.51925659179688, -56.33686828613281, 154.76089477539062, 84.25852966308594, 194.011962890625, 48.47087097167969, -9.806665420532227, -17.41730499267578, 46.05438232421875, 2.02545166015625, 57.51611328125, 166.77816772460938, -121.30663299560547, -73.829833984375, 13.466279983520508, 4.631359100341797, 132.94839477539062, 141.94815063476562, -159.50064086914062, 230.77236938476562, -32.65394592285156, 41.590576171875, 7.0620269775390625, 150.42935180664062, 135.72157287597656, 69.62802124023438, 0.8214550018310547, 212.93809509277344], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000339.npy"}
|
||||
{"epoch": 0.5124716553287982, "step": 340, "batch_size": 64, "mean": 53.96831512451172, "std": 101.36837768554688, "min": -170.69578552246094, "p10": -63.58469848632812, "median": 37.53986358642578, "p90": 190.4011383056641, "max": 228.38233947753906, "pos_frac": 0.640625, "sample": [88.90087890625, 170.66293334960938, -7.296581268310547, 119.373291015625, 11.098526000976562, 193.6020965576172, 1.4920520782470703, 1.058462142944336, -54.849754333496094, 221.41229248046875, -17.215797424316406, 70.69435119628906, 108.30349731445312, 159.96192932128906, 208.019775390625, 132.73194885253906, 37.397003173828125, -92.92445373535156, 120.3499755859375, 128.86541748046875, -23.595232009887695, -15.118402481079102, -69.78659057617188, 130.96380615234375, -19.17633056640625, 215.53482055664062, -17.236907958984375, 150.2591552734375, -44.660621643066406, 177.7542724609375, -145.2421417236328, 138.6621551513672, -17.638885498046875, 44.927490234375, 228.38233947753906, 34.865631103515625, 147.42161560058594, -0.7871952056884766, -57.71600341796875, 169.8328857421875, 133.14828491210938, -51.02050018310547, 8.572944641113281, 182.93223571777344, 71.10366821289062, -4.830507278442383, -152.71160888671875, -104.91641235351562, 4.153602600097656, 209.77505493164062, 37.68272399902344, 87.98072814941406, 85.08820343017578, -18.1068058013916, -170.69578552246094, -66.099853515625, 167.240478515625, 30.44934844970703, 139.46336364746094, 208.61660766601562, 46.0554084777832, 15.790313720703125, -13.763542175292969, -21.21955108642578], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000340.npy"}
|
||||
{"epoch": 0.5139833711262283, "step": 341, "batch_size": 64, "mean": 63.324546813964844, "std": 99.0156021118164, "min": -141.6121826171875, "p10": -57.213802337646484, "median": 62.83527374267578, "p90": 194.76907501220703, "max": 256.023681640625, "pos_frac": 0.703125, "sample": [-61.10080337524414, 85.15992736816406, 197.02426147460938, 67.21914672851562, 157.2994842529297, 83.31816101074219, 158.65304565429688, 253.572509765625, 194.10659790039062, -31.310028076171875, 160.13026428222656, 73.64323425292969, -141.6121826171875, -9.930675506591797, 58.45140075683594, 8.56527328491211, 185.22073364257812, 5.305717468261719, 148.40646362304688, 178.24667358398438, 122.11991882324219, 195.66458129882812, 196.57652282714844, 104.8389663696289, -119.54019165039062, 159.32386779785156, 132.2158660888672, 19.40542221069336, -4.451194763183594, -42.546905517578125, 19.890796661376953, -4.9987030029296875, -50.92265319824219, 7.954620361328125, 21.354537963867188, 175.02023315429688, 83.27835083007812, -0.8349933624267578, 157.69671630859375, -98.23397064208984, -54.739967346191406, -85.124755859375, 74.94930267333984, 256.023681640625, 6.236114501953125, 195.05299377441406, 11.30474853515625, 112.04103088378906, 146.90841674804688, 219.58926391601562, 99.11488342285156, 15.60797119140625, 167.71063232421875, -40.127777099609375, -5.5742950439453125, 132.6179656982422, 97.96106719970703, 42.333641052246094, -7.036958694458008, -58.274017333984375, -24.19536590576172, 0.2286376953125, 23.977815628051758, -117.99496459960938], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000341.npy"}
|
||||
{"epoch": 0.5154950869236583, "step": 342, "batch_size": 64, "mean": 60.264991760253906, "std": 96.02432250976562, "min": -159.7568359375, "p10": -55.322284698486314, "median": 54.61497497558594, "p90": 182.34885864257814, "max": 313.4954833984375, "pos_frac": 0.6875, "sample": [58.70807647705078, 31.5347900390625, 2.244964599609375, -22.153152465820312, 10.875242233276367, 187.5965118408203, 34.95390701293945, 174.25961303710938, -74.10340881347656, -24.71917724609375, -15.360017776489258, 64.92146301269531, -2.307462692260742, 154.99267578125, -20.66663360595703, -45.21217346191406, 6.491025924682617, -159.7568359375, -2.157693862915039, -59.655189514160156, 7.184307098388672, 170.81739807128906, 138.07225036621094, 190.54666137695312, -76.32613372802734, 83.07523345947266, 50.77021789550781, 102.10765075683594, 145.93084716796875, 57.12803268432617, 91.94281005859375, 73.09574890136719, 55.11512756347656, 126.0040283203125, 62.217315673828125, -7.33302116394043, 142.71250915527344, 37.83415222167969, -101.9501953125, 172.6831512451172, 48.33091735839844, 182.7070770263672, 136.03338623046875, 68.82647705078125, -75.55763244628906, 199.7364501953125, -1.5766830444335938, -3.782215118408203, 313.4954833984375, 227.6716766357422, 136.87796020507812, -121.85295867919922, 13.440666198730469, 176.81236267089844, -21.748287200927734, 181.5130157470703, 89.47225952148438, -5.268577575683594, -1.0997848510742188, 87.32017517089844, 55.56788635253906, 54.11482238769531, 18.39825439453125, 275.41204833984375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000342.npy"}
|
||||
{"epoch": 0.5170068027210885, "step": 343, "batch_size": 64, "mean": 46.320106506347656, "std": 92.14629364013672, "min": -190.363525390625, "p10": -39.23381958007812, "median": 29.91925048828125, "p90": 157.32745056152345, "max": 373.04986572265625, "pos_frac": 0.765625, "sample": [94.05995178222656, 12.322700500488281, 10.118232727050781, -49.26243591308594, -3.8185272216796875, 87.97846221923828, -33.378753662109375, 0.08535194396972656, -0.9620361328125, -20.21363067626953, -161.16046142578125, -41.743133544921875, 89.60585021972656, 63.64691925048828, 373.04986572265625, -8.072032928466797, 169.0574951171875, 217.02700805664062, 158.12411499023438, 130.8612060546875, 87.56866455078125, 5.986509323120117, 99.57963562011719, 133.98367309570312, 12.363765716552734, 94.4373779296875, 29.017410278320312, -19.41802978515625, 76.81814575195312, 8.536333084106445, -123.30484771728516, 44.634857177734375, 124.61997985839844, 30.157264709472656, 75.52719116210938, -15.016273498535156, 124.4848403930664, 158.82711791992188, -156.7508544921875, 167.45741271972656, -61.065948486328125, 76.33990478515625, 13.638298034667969, 235.8524169921875, 98.54928588867188, 29.681236267089844, 107.04412841796875, 8.675350189208984, 34.08943176269531, 155.46856689453125, 125.9635238647461, 18.36986541748047, 3.1898422241210938, 68.88614654541016, 51.615577697753906, 31.486343383789062, 11.871307373046875, 46.303619384765625, 25.606414794921875, 29.02448272705078, -190.363525390625, 10.041519165039062, -25.154361724853516, 12.537065505981445], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000343.npy"}
|
||||
{"epoch": 0.5185185185185185, "step": 344, "batch_size": 64, "mean": 66.63272094726562, "std": 103.0145263671875, "min": -159.696533203125, "p10": -67.25014877319335, "median": 61.16084671020508, "p90": 175.7673065185547, "max": 359.6590270996094, "pos_frac": 0.71875, "sample": [-114.03912353515625, 158.93804931640625, 60.20996856689453, 69.16339111328125, 185.13189697265625, -2.858186721801758, 27.981868743896484, -1.3764457702636719, 32.64533996582031, 135.48611450195312, 88.22998046875, -1.31256103515625, 12.528974533081055, 89.04251098632812, 111.25767517089844, 45.88711166381836, 68.13986206054688, 359.6590270996094, -47.716156005859375, 149.79974365234375, -101.13876342773438, -86.49604797363281, -71.86492156982422, 212.88702392578125, 2.0147743225097656, -36.971046447753906, -56.48234558105469, 2.2134475708007812, 259.97100830078125, 139.1407470703125, -86.50164794921875, 213.08038330078125, -29.15314483642578, 135.48715209960938, 62.111724853515625, 176.36013793945312, 150.1913604736328, -0.3632965087890625, 119.1964111328125, -159.696533203125, -4.337688446044922, 19.11422348022461, -9.359352111816406, 9.120634078979492, -24.695167541503906, 10.592473983764648, 173.95013427734375, 164.81594848632812, 108.28672790527344, 136.9925994873047, 145.34669494628906, 56.60462951660156, 137.00949096679688, 6.769502639770508, 171.39111328125, 21.266923904418945, 172.51144409179688, 91.40713500976562, 47.50688171386719, 174.384033203125, -131.90447998046875, 130.13645935058594, 162.0221710205078, 224.7762908935547], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000344.npy"}
|
||||
{"epoch": 0.5200302343159486, "step": 345, "batch_size": 64, "mean": 59.94282531738281, "std": 95.35880279541016, "min": -147.937255859375, "p10": -38.39407806396484, "median": 50.12553405761719, "p90": 176.1851608276367, "max": 254.99041748046875, "pos_frac": 0.75, "sample": [91.98243713378906, 9.873531341552734, 218.44790649414062, -132.61978149414062, 12.509220123291016, 12.826202392578125, -17.354385375976562, -38.796539306640625, 10.982017517089844, -37.45500183105469, 163.6823272705078, 149.67686462402344, 159.38076782226562, 127.3591537475586, -87.89103698730469, 176.44805908203125, -118.31403350830078, 113.09220886230469, 47.76183319091797, 175.48489379882812, 245.36732482910156, 2.3809261322021484, 178.2018280029297, 93.50376892089844, 136.04164123535156, -25.759613037109375, 23.915550231933594, 29.27840232849121, -24.61614990234375, 52.489234924316406, 110.77165985107422, 232.740478515625, 134.77267456054688, 26.769739151000977, 98.84317016601562, -147.937255859375, 166.76136779785156, 254.99041748046875, 114.64292907714844, -33.40981674194336, 172.89698791503906, 53.228782653808594, -27.3143310546875, 187.5269012451172, 92.46212768554688, -35.792964935302734, 68.2845230102539, -56.81365203857422, 15.584716796875, 35.676551818847656, 38.19679260253906, 38.06526184082031, -14.684150695800781, 1.59027099609375, 3.259796142578125, 133.55560302734375, 2.3353633880615234, -133.38217163085938, -3.084613800048828, 79.66791534423828, 101.99260711669922, 130.11752319335938, 70.5743408203125, 175.5717315673828], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000345.npy"}
|
||||
{"epoch": 0.5215419501133787, "step": 346, "batch_size": 64, "mean": 73.20783996582031, "std": 99.07064056396484, "min": -198.2408447265625, "p10": -35.76640396118164, "median": 80.4500617980957, "p90": 186.6819061279297, "max": 275.40325927734375, "pos_frac": 0.78125, "sample": [140.14305114746094, 165.57553100585938, 156.59518432617188, 159.90631103515625, 88.38236999511719, 89.49986267089844, 36.000274658203125, 102.17337799072266, 162.78134155273438, 198.9829559326172, 169.932861328125, 11.939353942871094, 202.65728759765625, 167.2506103515625, -73.31086730957031, 178.15892028808594, -137.10171508789062, 82.95658874511719, -26.97443389892578, 93.33340454101562, 100.64422607421875, 134.24432373046875, 146.6890411376953, 49.886993408203125, 165.50750732421875, 275.40325927734375, -198.2408447265625, 57.61054229736328, 17.970439910888672, -143.1632537841797, 39.013092041015625, -36.024864196777344, 169.4669647216797, 0.0930633544921875, 49.8785514831543, 71.60167694091797, 18.71668243408203, -21.88311767578125, 153.38478088378906, 187.19424438476562, 228.7916259765625, 149.89712524414062, -35.163330078125, 240.4484405517578, 22.76689910888672, 77.94353485107422, 101.78076171875, 84.34397888183594, -91.41114807128906, 108.77476501464844, 61.59416198730469, -10.779541015625, 6.797666549682617, -3.094268798828125, -60.322166442871094, 37.082305908203125, 13.680213928222656, 156.82957458496094, -3.7276382446289062, 214.68136596679688, -15.948345184326172, 2.1807422637939453, 5.79290771484375, 185.4864501953125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000346.npy"}
|
||||
{"epoch": 0.5230536659108088, "step": 347, "batch_size": 64, "mean": 55.977928161621094, "std": 87.86829376220703, "min": -133.69534301757812, "p10": -34.67994918823242, "median": 37.565542221069336, "p90": 179.60633087158206, "max": 221.0574951171875, "pos_frac": 0.671875, "sample": [27.235750198364258, -15.375152587890625, 27.58580780029297, 146.0200653076172, -105.5252685546875, 37.70924377441406, 134.8126220703125, 129.64230346679688, 131.38523864746094, 143.05825805664062, -5.0194091796875, 164.49923706054688, 49.268798828125, -93.23409271240234, -97.45790100097656, 221.0574951171875, 38.019866943359375, 39.38138198852539, -8.894287109375, 27.065025329589844, 54.34257507324219, 6.96473503112793, 83.19775390625, -22.410125732421875, -32.932342529296875, 26.589702606201172, 115.12798309326172, 153.55877685546875, 213.7464599609375, -33.49578857421875, 10.396522521972656, -12.166341781616211, -35.18744659423828, 175.20767211914062, -5.079072952270508, 65.88143920898438, -2.494617462158203, 121.09523010253906, 76.543701171875, 104.40863800048828, 37.42184066772461, 42.77326202392578, 4.4263763427734375, 147.7445526123047, 5.870004653930664, 197.90609741210938, -38.981468200683594, -17.382720947265625, 215.992919921875, 197.83103942871094, 215.3375701904297, 166.1476593017578, -36.8037109375, 100.65861511230469, -133.69534301757812, 113.28718566894531, 32.40845489501953, 0.85528564453125, -4.870723724365234, -4.998414993286133, -24.594547271728516, -9.77523422241211, 181.49147033691406, 139.00698852539062], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000347.npy"}
|
||||
{"epoch": 0.5245653817082389, "step": 348, "batch_size": 64, "mean": 61.29752731323242, "std": 97.46875762939453, "min": -164.83001708984375, "p10": -49.784232330322254, "median": 44.70072937011719, "p90": 197.9577560424805, "max": 314.21478271484375, "pos_frac": 0.6875, "sample": [184.985107421875, 174.74981689453125, 40.29267883300781, 3.21441650390625, 215.36721801757812, 93.92720031738281, -56.338531494140625, -14.292167663574219, -164.83001708984375, -19.352699279785156, 229.85684204101562, -14.637557983398438, 49.10877990722656, 73.33320617675781, 153.8431396484375, 160.80795288085938, 27.510482788085938, 210.5840301513672, 58.80805587768555, 223.93560791015625, 34.7577018737793, 153.86114501953125, 172.562744140625, 24.65380859375, -5.114860534667969, 173.3722686767578, -14.0362548828125, 164.89109802246094, -71.73441314697266, -10.220603942871094, 63.71397399902344, -0.8882522583007812, 3.3641223907470703, 314.21478271484375, 83.68587493896484, -102.0198745727539, 58.9881706237793, -97.13619995117188, -55.96350860595703, 116.70291137695312, 190.30882263183594, 179.27728271484375, 94.42581176757812, 0.813323974609375, 58.76695251464844, -35.36592102050781, -64.78582000732422, -25.084022521972656, 20.690689086914062, 94.39241027832031, -20.99142074584961, 201.23587036132812, -15.198087692260742, 60.663002014160156, 92.91030883789062, 20.15625762939453, 55.42106628417969, 30.659515380859375, -24.408111572265625, 8.701772689819336, -1.3049163818359375, 122.25321960449219, 221.30862426757812, 19.666866302490234], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000348.npy"}
|
||||
{"epoch": 0.5260770975056689, "step": 349, "batch_size": 64, "mean": 65.11505126953125, "std": 99.506103515625, "min": -119.73689270019531, "p10": -41.07706298828124, "median": 49.78950119018555, "p90": 209.57366180419928, "max": 289.5904235839844, "pos_frac": 0.6875, "sample": [-105.83651733398438, 61.171142578125, 214.06837463378906, 51.94244384765625, 35.82300567626953, -11.180351257324219, 134.51483154296875, 199.08599853515625, 115.1911849975586, 3.6644668579101562, -2.669015884399414, 51.10227966308594, 153.53878784179688, 46.65702819824219, 48.476722717285156, 146.50588989257812, 45.19340515136719, -113.02362823486328, -119.73689270019531, 183.1290740966797, 82.36941528320312, 177.45809936523438, -28.871490478515625, -3.7870521545410156, 164.43260192871094, -4.3462371826171875, 25.747886657714844, 71.16798400878906, -44.033424377441406, 289.5904235839844, 115.84164428710938, 147.67919921875, -2.425811767578125, 23.753385543823242, 1.7071609497070312, 175.35174560546875, 58.251731872558594, 243.60289001464844, -100.3733901977539, 54.38019943237305, 235.1214599609375, 14.898078918457031, 10.448089599609375, 103.03033447265625, 92.97309875488281, 194.58419799804688, 253.8270263671875, -11.306720733642578, -46.68400573730469, -2.981424331665039, 71.88365173339844, -15.013076782226562, -7.165702819824219, 6.606361389160156, 105.8428726196289, 240.20635986328125, -6.4342803955078125, -81.09098815917969, -25.35527801513672, 163.80740356445312, 2.5525875091552734, -34.17888641357422, 71.77959442138672, 244.89723205566406], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000349.npy"}
|
||||
{"epoch": 0.527588813303099, "step": 350, "batch_size": 64, "mean": 88.0762939453125, "std": 93.32093048095703, "min": -102.89408111572266, "p10": -18.806930923461913, "median": 84.73885726928711, "p90": 215.6404479980469, "max": 324.66436767578125, "pos_frac": 0.8125, "sample": [120.91132354736328, -27.40076446533203, 72.50859069824219, -3.3220977783203125, 126.3934097290039, 20.850542068481445, 242.26327514648438, 116.69380187988281, 107.00048828125, 150.739501953125, 21.495712280273438, -54.020050048828125, 177.85458374023438, 142.71151733398438, 103.31844329833984, 115.24505615234375, 210.02459716796875, 11.434589385986328, -12.611839294433594, -6.10624885559082, 155.0914306640625, 141.54336547851562, -102.89408111572266, 194.9486083984375, 48.81743621826172, 12.870979309082031, -102.0257568359375, 324.66436767578125, 17.966039657592773, 36.06192398071289, 171.251953125, -3.7581787109375, 219.5263671875, 26.694679260253906, 139.2408447265625, 227.52130126953125, 109.55886840820312, 218.0472412109375, 229.51943969726562, 257.4339904785156, 79.6712417602539, 203.728759765625, 173.561279296875, -25.281471252441406, 52.17741012573242, 8.084030151367188, 11.934700012207031, 21.203725814819336, -16.391300201416016, 88.18854522705078, 104.48725891113281, 53.49753189086914, 1.9307384490966797, 60.60254669189453, 81.32494354248047, -76.81478881835938, 88.15277099609375, 70.96025085449219, -19.842201232910156, 62.543914794921875, 185.27584838867188, 152.20289611816406, 137.44908142089844, 180.1695556640625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000350.npy"}
|
||||
{"epoch": 0.5291005291005291, "step": 351, "batch_size": 64, "mean": 61.5703125, "std": 106.94042205810547, "min": -210.54078674316406, "p10": -67.45227432250977, "median": 56.17592239379883, "p90": 193.49668731689457, "max": 283.05120849609375, "pos_frac": 0.71875, "sample": [-13.828897476196289, 199.61630249023438, 173.1390380859375, 196.23245239257812, -4.755245208740234, 178.5415802001953, -8.52481460571289, 35.602935791015625, 202.73098754882812, 64.70204162597656, -19.283477783203125, 199.599609375, 14.229089736938477, -36.113433837890625, 24.779949188232422, 5.576255798339844, -18.346725463867188, 17.629791259765625, 112.06902313232422, 138.15274047851562, 149.14505004882812, -117.69135284423828, 186.93089294433594, 17.972427368164062, 154.7171173095703, 219.01657104492188, 85.4466552734375, 152.0013427734375, -196.5660400390625, 85.75970458984375, 170.7556915283203, -68.7654037475586, 70.32089233398438, 2.3910369873046875, 145.60098266601562, -78.77113342285156, -71.9882583618164, 283.05120849609375, 47.649803161621094, -2.3185081481933594, 115.0702133178711, 176.05471801757812, 37.18809509277344, -48.331886291503906, 89.41615295410156, 42.153717041015625, 118.75569152832031, 82.37831115722656, 5.4290008544921875, 129.46536254882812, -53.434181213378906, 15.987468719482422, 158.36419677734375, 126.76758575439453, 187.1132354736328, 13.26397705078125, -0.0732574462890625, 214.4792022705078, -181.15447998046875, 18.627519607543945, -210.54078674316406, 135.6719970703125, 135.82843017578125, -64.3883056640625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000351.npy"}
|
||||
{"epoch": 0.5306122448979592, "step": 352, "batch_size": 64, "mean": 68.71488189697266, "std": 103.62076568603516, "min": -132.18569946289062, "p10": -41.483145523071286, "median": 59.641719818115234, "p90": 204.46248779296877, "max": 425.4741516113281, "pos_frac": 0.703125, "sample": [150.55897521972656, 225.718017578125, 170.7248077392578, 3.3638458251953125, 209.59039306640625, -112.08279418945312, -4.940273284912109, 7.9882049560546875, 19.726543426513672, 118.12374877929688, 115.45972442626953, 133.87844848632812, -132.18569946289062, -2.49639892578125, 107.14778137207031, 63.89923095703125, -92.76376342773438, -12.608367919921875, -75.52129364013672, 29.957366943359375, 151.568115234375, 125.80116271972656, 45.55644226074219, 187.35592651367188, 181.4352569580078, 146.31332397460938, 59.85948944091797, -129.38702392578125, 62.9563102722168, -43.683170318603516, 16.839092254638672, -5.115129470825195, 39.187957763671875, 111.33816528320312, 27.955352783203125, 103.61807250976562, 59.4239501953125, 207.06582641601562, 205.72988891601562, 82.52483367919922, 171.19415283203125, -36.349754333496094, 425.4741516113281, 5.50474739074707, 61.09239959716797, 174.13267517089844, -30.38878631591797, 43.94073486328125, -4.144411087036133, 64.8289794921875, -22.454071044921875, 218.13009643554688, 81.87725830078125, 149.75897216796875, 244.80850219726562, -7.7833099365234375, -50.171470642089844, 201.50521850585938, -14.077701568603516, -15.826461791992188, 4.264324188232422, 153.90138244628906, 28.61244010925293, -9.959989547729492], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000352.npy"}
|
||||
{"epoch": 0.5321239606953893, "step": 353, "batch_size": 64, "mean": 61.34170150756836, "std": 87.89678192138672, "min": -193.0448455810547, "p10": -25.210343933105467, "median": 45.80473327636719, "p90": 183.92694396972658, "max": 226.93795776367188, "pos_frac": 0.734375, "sample": [61.087867736816406, 226.93795776367188, -65.11775207519531, 58.81303024291992, -120.70942687988281, 32.663265228271484, 141.97052001953125, -22.82270050048828, 205.23887634277344, 42.34751892089844, -11.014118194580078, -12.188743591308594, 131.56088256835938, 30.487396240234375, 159.44468688964844, 30.21946907043457, 35.02001953125, 172.55224609375, 121.3282470703125, 170.79745483398438, 189.23216247558594, 178.46176147460938, 2.9370861053466797, 137.82803344726562, -26.233619689941406, -11.19076156616211, 197.7412109375, 21.16802215576172, 107.34636688232422, 80.88381958007812, 69.80838012695312, -17.874591827392578, 130.9766082763672, -0.7023200988769531, -193.0448455810547, 190.91384887695312, -55.51020812988281, -46.05778884887695, 163.9852294921875, 164.20538330078125, 101.84954833984375, 160.08978271484375, 186.2691650390625, -8.723068237304688, -57.83099365234375, 87.43284606933594, 192.58006286621094, 13.528039932250977, 8.45541763305664, 29.00786590576172, 47.272552490234375, 44.96690368652344, -20.150711059570312, 51.30223083496094, 1.7502250671386719, 46.64256286621094, 5.67724609375, -0.9630889892578125, -13.812881469726562, 20.160438537597656, 117.7872314453125, 17.548614501953125, 151.95394897460938, 69.58439636230469], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000353.npy"}
|
||||
{"epoch": 0.5336356764928194, "step": 354, "batch_size": 64, "mean": 73.95189666748047, "std": 98.55117797851562, "min": -138.6548614501953, "p10": -58.47489891052246, "median": 80.97872161865234, "p90": 205.21455841064454, "max": 246.6551971435547, "pos_frac": 0.78125, "sample": [213.75979614257812, 35.50883865356445, -65.40901947021484, 159.6822967529297, -57.54468536376953, 24.74357032775879, 45.925148010253906, 118.36109161376953, 205.92930603027344, 10.39559555053711, 144.81101989746094, 179.19610595703125, 157.01783752441406, 90.66875457763672, 73.03252410888672, 0.4831504821777344, 211.13482666015625, -83.06100463867188, 170.50437927246094, 230.6964111328125, 74.38310241699219, 175.16702270507812, 144.47874450683594, 102.73294067382812, -123.1731948852539, -8.160799026489258, -119.64910888671875, 154.45175170898438, 93.1948471069336, 60.17803192138672, 32.988067626953125, 112.18009948730469, 203.54681396484375, 95.6422119140625, 91.08001708984375, 2.3497695922851562, 77.24830627441406, -138.6548614501953, 176.5908966064453, 17.011032104492188, 137.88519287109375, 165.40907287597656, 6.648612976074219, -49.1610107421875, 222.49176025390625, -113.49189758300781, 246.6551971435547, 176.7151641845703, 15.730440139770508, -20.73175811767578, 52.791419982910156, 84.70913696289062, 108.71762084960938, 214.42002868652344, 59.99984359741211, 41.045841217041016, -43.06739044189453, 49.64921188354492, -33.88356018066406, 152.4827423095703, 109.2972183227539, 150.0426025390625, -31.98236083984375, -58.87356185913086], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000354.npy"}
|
||||
{"epoch": 0.5351473922902494, "step": 355, "batch_size": 64, "mean": 85.13143920898438, "std": 107.55074310302734, "min": -159.52517700195312, "p10": -38.4166851043701, "median": 76.02662658691406, "p90": 222.33889923095705, "max": 298.9481201171875, "pos_frac": 0.734375, "sample": [157.4650115966797, 219.53805541992188, -44.9127197265625, -11.2178955078125, 298.9481201171875, 156.9459228515625, 67.31788635253906, -12.673429489135742, -23.807701110839844, 121.97196197509766, 89.12752532958984, 29.874114990234375, 128.12269592285156, -44.42095947265625, 183.93719482421875, 166.56063842773438, -1.3908424377441406, 102.73983764648438, -132.2400360107422, 223.70358276367188, -11.356216430664062, 38.0903434753418, 44.27494812011719, 207.75331115722656, 158.40383911132812, 27.539213180541992, 112.89766693115234, 42.37758255004883, 159.7742462158203, 258.37774658203125, -56.16967010498047, 114.46208190917969, -24.40671157836914, 223.99151611328125, 23.685150146484375, -156.46194458007812, 84.60870361328125, -18.092315673828125, 67.44454956054688, 180.82859802246094, -47.160926818847656, 10.7633056640625, 157.25177001953125, 58.47010803222656, -9.236623764038086, 274.48779296875, 292.0874328613281, 40.266788482666016, 179.380126953125, -159.52517700195312, 172.66082763671875, 185.9758758544922, 223.5392608642578, 5.318714141845703, 178.55625915527344, -6.103799819946289, -1.8465347290039062, 29.86538314819336, 218.72125244140625, 24.938194274902344, 183.19378662109375, 44.31896209716797, 91.31322479248047, 147.56475830078125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000355.npy"}
|
||||
{"epoch": 0.5366591080876795, "step": 356, "batch_size": 64, "mean": 65.9458236694336, "std": 111.95003509521484, "min": -195.73123168945312, "p10": -89.04108963012695, "median": 60.62194633483887, "p90": 203.8854187011719, "max": 343.88397216796875, "pos_frac": 0.6875, "sample": [171.1171417236328, 204.6165771484375, 135.33566284179688, -51.44859313964844, 169.6258087158203, 43.40186309814453, 0.145904541015625, 343.88397216796875, 143.7867431640625, 2.989351272583008, -81.25895690917969, -149.0501708984375, 159.80384826660156, -92.37628936767578, 182.86077880859375, 209.59005737304688, 32.20594024658203, 19.541484832763672, 179.16864013671875, -6.464818954467773, 202.17938232421875, -195.73123168945312, 223.73468017578125, 62.889739990234375, 10.117645263671875, 225.3134307861328, -41.9851188659668, -102.88859558105469, 116.4783935546875, 87.95584869384766, 107.52971649169922, 15.967681884765625, -16.420412063598633, 7.251075744628906, -104.93599700927734, 60.024627685546875, 181.44161987304688, -111.07789611816406, 36.89544677734375, 166.57638549804688, -17.485736846923828, 25.601882934570312, 171.9154052734375, -8.329269409179688, 45.713134765625, 146.96798706054688, 75.71318054199219, -6.220024108886719, -37.98291778564453, 168.32582092285156, -8.412303924560547, 91.6179428100586, 173.284423828125, 210.1593017578125, -13.249717712402344, 159.55661010742188, 61.21926498413086, -29.46198272705078, 231.39239501953125, -16.003326416015625, -124.84683227539062, 188.51011657714844, 64.29209899902344, 119.46392822265625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000356.npy"}
|
||||
{"epoch": 0.5381708238851096, "step": 357, "batch_size": 64, "mean": 52.3343505859375, "std": 106.12026977539062, "min": -216.53079223632812, "p10": -56.14656867980956, "median": 49.35960578918457, "p90": 192.02677459716796, "max": 287.71063232421875, "pos_frac": 0.703125, "sample": [196.93495178222656, -14.202253341674805, 207.96258544921875, -9.671005249023438, -216.53079223632812, 165.24691772460938, 84.72894287109375, 48.18495559692383, 57.57447814941406, 47.71893310546875, 26.30633544921875, 2.3947677612304688, -49.55858612060547, 17.122661590576172, 192.17564392089844, 94.20858764648438, 165.1685333251953, 133.2915496826172, 112.95667266845703, 0.3706321716308594, 170.34957885742188, 40.785064697265625, 80.16265869140625, 7.430742263793945, 37.85015106201172, 139.006103515625, -11.618314743041992, 54.74833679199219, -120.54119873046875, 85.85716247558594, -93.73887634277344, -23.5131778717041, -41.47395706176758, 61.012489318847656, -5.663246154785156, 199.79983520507812, 205.9708709716797, 50.53425598144531, 183.62184143066406, 186.62966918945312, -24.145286560058594, 77.05413818359375, -0.838348388671875, 6.530174255371094, 233.02969360351562, 80.64103698730469, 2.5300064086914062, -26.906097412109375, -58.96998977661133, 51.79791259765625, 53.73774337768555, 287.71063232421875, 161.36679077148438, 47.14609146118164, 117.79724884033203, 60.68695068359375, -8.032011032104492, 2.90753173828125, 160.87640380859375, -139.8507537841797, -210.33619689941406, -184.09970092773438, 191.67941284179688, -2.5095157623291016], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000357.npy"}
|
||||
{"epoch": 0.5396825396825397, "step": 358, "batch_size": 64, "mean": 67.28176879882812, "std": 114.71636199951172, "min": -191.9139404296875, "p10": -69.43978958129883, "median": 41.63240051269531, "p90": 202.30762329101566, "max": 399.4039306640625, "pos_frac": 0.75, "sample": [126.37202453613281, 61.317535400390625, 65.27870178222656, 197.236083984375, -73.35136413574219, 218.43112182617188, 160.46798706054688, -63.83061981201172, -12.831058502197266, 8.615264892578125, -18.586566925048828, 149.66073608398438, 70.16747283935547, 204.48114013671875, 166.24806213378906, 83.91529846191406, -91.56729125976562, -114.91285705566406, 88.94003295898438, 24.18294906616211, -50.02825927734375, 42.19102096557617, 5.638832092285156, 108.24065399169922, 182.90921020507812, 28.05628204345703, 21.727798461914062, 88.08106994628906, -71.84371948242188, 41.07378005981445, -97.39710235595703, 55.708030700683594, -191.9139404296875, 4.681051254272461, -63.80939483642578, 1.8272705078125, 164.72930908203125, 26.3465576171875, 34.163047790527344, -16.38572883605957, 193.10850524902344, 318.94158935546875, 399.4039306640625, 30.54488754272461, 264.880859375, 182.95774841308594, 137.1480712890625, -24.427303314208984, 15.4830322265625, 103.5625, -4.992015838623047, -43.04203796386719, -167.86611938476562, 171.88900756835938, 25.205917358398438, 174.5788116455078, 17.15896987915039, 133.58018493652344, 21.937957763671875, 96.06459045410156, 178.70419311523438, 218.77371215820312, 32.4473876953125, 265.7581787109375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000358.npy"}
|
||||
{"epoch": 0.5411942554799698, "step": 359, "batch_size": 64, "mean": 37.35707092285156, "std": 108.9959716796875, "min": -230.29275512695312, "p10": -90.02613754272461, "median": 21.40774154663086, "p90": 181.43710174560547, "max": 317.9493408203125, "pos_frac": 0.640625, "sample": [-33.791290283203125, 133.31820678710938, -9.7559814453125, 14.599777221679688, 73.47051239013672, 59.38048553466797, -20.013614654541016, 20.438453674316406, -27.278961181640625, -220.9149169921875, 268.50347900390625, 17.936250686645508, 46.46684265136719, 53.06298828125, -3.0298614501953125, 58.416717529296875, 179.55404663085938, -90.14482879638672, -8.5682373046875, -142.72268676757812, 41.31914138793945, -36.20851516723633, 72.84164428710938, -89.74919128417969, 52.8450927734375, 145.06578063964844, 76.15156555175781, 20.203441619873047, 76.39431762695312, -163.95008850097656, 317.9493408203125, 262.326171875, -7.488922119140625, 204.96273803710938, 98.07681274414062, 47.35790252685547, -6.190605163574219, 152.0782470703125, 68.94548034667969, -55.57855987548828, -96.85347747802734, 125.09495544433594, 156.67083740234375, 95.00456237792969, 16.83722686767578, 209.58572387695312, -80.62653350830078, -163.35247802734375, -18.067642211914062, 98.37551879882812, 19.263629913330078, -4.336286544799805, -230.29275512695312, 0.11478233337402344, 59.041175842285156, -7.9852447509765625, 96.02565002441406, 16.62721824645996, 22.377029418945312, 217.9519805908203, 37.63271713256836, 182.24412536621094, 10.749568939208984, -17.508743286132812], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000359.npy"}
|
||||
{"epoch": 0.5427059712773998, "step": 360, "batch_size": 64, "mean": 67.91548919677734, "std": 101.03427124023438, "min": -145.94496154785156, "p10": -33.548660087585446, "median": 43.50065803527832, "p90": 192.18491363525393, "max": 319.32696533203125, "pos_frac": 0.703125, "sample": [3.6172637939453125, 167.92791748046875, 69.91604614257812, 76.70973205566406, 90.36040496826172, 174.00994873046875, -7.719566345214844, 22.04336929321289, 30.175067901611328, 259.7193298339844, 0.6439380645751953, -35.26579284667969, -47.97166442871094, -145.94496154785156, 170.0579833984375, -95.906005859375, 319.32696533203125, -19.72549819946289, -5.830013275146484, 140.51414489746094, 91.79083251953125, -7.201179504394531, -43.84320068359375, 1.159515380859375, -6.575469970703125, -2.2480392456054688, 141.4222869873047, -21.346900939941406, 111.6823959350586, -16.890213012695312, 64.91104888916016, 13.877662658691406, 88.1402587890625, -83.05242919921875, 69.91993713378906, 215.008056640625, -24.95074462890625, 127.27912902832031, 172.205322265625, -8.083892822265625, 263.0854187011719, 15.184906005859375, 188.3502197265625, -8.816640853881836, 285.4154357910156, 32.465110778808594, 168.88864135742188, 34.787681579589844, 52.2136344909668, 121.00970458984375, 119.95491790771484, 120.56689453125, 201.49258422851562, -29.542016983032227, 16.187273025512695, 131.79396057128906, 19.585466384887695, -140.67367553710938, 124.96002197265625, 176.38912963867188, 13.456451416015625, 26.87543487548828, 169.26950073242188, 193.82835388183594], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000360.npy"}
|
||||
{"epoch": 0.54421768707483, "step": 361, "batch_size": 64, "mean": 73.8453369140625, "std": 96.79232025146484, "min": -122.84901428222656, "p10": -40.54407691955565, "median": 62.59201431274414, "p90": 191.5699447631836, "max": 268.1335754394531, "pos_frac": 0.71875, "sample": [115.80379486083984, 141.90341186523438, 68.29411315917969, -8.195762634277344, 191.42172241210938, 104.44314575195312, -73.47105407714844, -104.07215881347656, 104.43539428710938, 22.494281768798828, 191.6334686279297, -28.930660247802734, -9.181427001953125, 160.6768798828125, 110.59957885742188, 114.37957000732422, -21.687610626220703, 12.43729019165039, 31.280181884765625, -7.6353759765625, 268.1335754394531, 71.00762176513672, -11.818305969238281, 182.39117431640625, 158.88787841796875, 198.0548095703125, 173.0333709716797, -122.84901428222656, 174.69448852539062, 170.03305053710938, 23.86600112915039, 11.109695434570312, 171.5572052001953, 230.97360229492188, 166.96035766601562, 159.58377075195312, 171.4892120361328, 143.09176635742188, -118.46403503417969, -9.780860900878906, -3.010042190551758, 1.3458633422851562, 193.7196807861328, 225.5720672607422, -9.620349884033203, 178.10015869140625, 180.4718017578125, 16.896263122558594, -49.47455596923828, -45.52125549316406, 60.9967041015625, 29.310562133789062, 49.06877136230469, 14.302623748779297, 207.58535766601562, 29.74842643737793, 175.5477294921875, -5.540351867675781, 43.89469909667969, 64.18732452392578, 83.51323699951172, 60.93293762207031, -15.073200225830078, -89.43702697753906], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000361.npy"}
|
||||
{"epoch": 0.54572940287226, "step": 362, "batch_size": 64, "mean": 85.74800872802734, "std": 131.30645751953125, "min": -192.88772583007812, "p10": -65.69952087402342, "median": 92.49052810668945, "p90": 195.45683746337892, "max": 646.6310424804688, "pos_frac": 0.734375, "sample": [-75.78045654296875, 4.993070602416992, -77.01248168945312, 2.541616439819336, -17.36229705810547, -192.88772583007812, 46.395469665527344, 104.31027221679688, 646.6310424804688, 186.85731506347656, 47.603538513183594, -12.85000228881836, 59.29033660888672, -15.158279418945312, -4.43408203125, 1.3297538757324219, 113.61918640136719, 65.55491638183594, 149.659912109375, 230.23358154296875, 5.043664932250977, 85.89291381835938, 132.88397216796875, 162.14102172851562, 126.48567199707031, -4.5256195068359375, 166.2947998046875, -42.177337646484375, 9.736808776855469, 151.88034057617188, 361.792724609375, 16.79001808166504, -8.431777954101562, 195.7160186767578, 45.89038848876953, 344.23199462890625, 172.612060546875, 57.57403564453125, 119.43746185302734, -39.33818054199219, -2.439403533935547, 150.10556030273438, 188.40939331054688, 194.85208129882812, 48.19721984863281, 116.41769409179688, 159.17822265625, 244.01600646972656, 156.57884216308594, 149.5848388671875, -108.7291030883789, 121.002685546875, 185.85577392578125, -118.22331237792969, 99.08814239501953, 115.22918701171875, 165.1476593017578, -152.8133544921875, -16.44414520263672, 169.96730041503906, 232.95416259765625, -101.64653015136719, 101.80874633789062, 66.309326171875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000362.npy"}
|
||||
{"epoch": 0.54724111866969, "step": 363, "batch_size": 64, "mean": 77.198486328125, "std": 111.18279266357422, "min": -166.52813720703125, "p10": -20.758812713623033, "median": 42.32039451599121, "p90": 221.41214904785159, "max": 443.6802673339844, "pos_frac": 0.828125, "sample": [112.21388244628906, -166.52813720703125, 192.774169921875, -3.8032302856445312, 129.46725463867188, 39.69536590576172, 14.73089599609375, 257.23419189453125, 50.35667419433594, -64.61807250976562, 6.2049102783203125, 170.54098510742188, 443.6802673339844, 321.9192810058594, 29.496116638183594, 222.5211181640625, 33.00009536743164, 117.6939697265625, 41.74858093261719, 178.9580841064453, 42.892208099365234, 166.88514709472656, 16.582002639770508, 273.41021728515625, 164.1088409423828, 21.007204055786133, 11.291290283203125, 16.057334899902344, -26.062355041503906, 36.703453063964844, -6.489522933959961, -162.34483337402344, 139.13406372070312, -151.2929229736328, 78.13035583496094, 150.658203125, 25.521177291870117, 156.9984130859375, 11.529806137084961, -5.472660064697266, 31.240177154541016, 18.153221130371094, 12.79730224609375, 242.86607360839844, 10.214597702026367, 218.82455444335938, 131.8614501953125, 67.02394104003906, 181.45590209960938, 57.66321563720703, 131.08998107910156, -8.383880615234375, 159.76275634765625, 69.10885620117188, -69.53645324707031, 92.0164794921875, 236.1162872314453, -42.63528060913086, 113.38032531738281, 16.225666046142578, 6.91908073425293, 135.08651733398438, 30.739242553710938, 12.179624557495117], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000363.npy"}
|
||||
{"epoch": 0.5487528344671202, "step": 364, "batch_size": 64, "mean": 87.3396224975586, "std": 129.47384643554688, "min": -195.08457946777344, "p10": -47.07382354736327, "median": 69.15262985229492, "p90": 246.5240234375, "max": 419.03594970703125, "pos_frac": 0.71875, "sample": [258.68511962890625, -16.389007568359375, 30.38233184814453, 208.44873046875, 247.26858520507812, 20.91114044189453, 81.86919403076172, 164.843994140625, 23.047767639160156, 268.7218933105469, 289.22576904296875, 144.00140380859375, -16.88711166381836, -69.03360748291016, 165.1383056640625, 179.6363983154297, 81.71964263916016, -19.014209747314453, 419.03594970703125, 4.686767578125, 244.78671264648438, 65.06109619140625, 192.6865692138672, 221.32862854003906, -168.8819122314453, -56.060691833496094, 230.71051025390625, -2.5356807708740234, 1.151071548461914, 188.7787322998047, 197.49440002441406, 73.2441635131836, -22.66106414794922, 138.21514892578125, 34.94734191894531, -7.6066741943359375, 146.28208923339844, 325.1217041015625, -36.35680389404297, 36.57733154296875, 154.08824157714844, 92.36687469482422, -23.147621154785156, -142.02772521972656, 376.62762451171875, 13.86679458618164, -23.192138671875, 187.27529907226562, 7.911415100097656, 148.12843322753906, 21.379226684570312, 170.70091247558594, -148.97669982910156, 88.45858001708984, -195.08457946777344, 21.906234741210938, 62.04405975341797, 179.95640563964844, -4.415071487426758, 10.101837158203125, -9.356571197509766, -51.666831970214844, 183.87229919433594, 200.33712768554688], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000364.npy"}
|
||||
{"epoch": 0.5502645502645502, "step": 365, "batch_size": 64, "mean": 77.98208618164062, "std": 98.528076171875, "min": -135.15196228027344, "p10": -9.152231025695798, "median": 66.45542907714844, "p90": 187.83334503173828, "max": 530.8749389648438, "pos_frac": 0.765625, "sample": [65.6855239868164, 38.880577087402344, 109.80792236328125, -2.3227462768554688, 84.83354949951172, 2.524005889892578, 188.15731811523438, 155.92205810546875, 145.4580841064453, 90.864501953125, 199.41722106933594, 54.6258544921875, -135.15196228027344, -10.716796875, 39.9862060546875, 187.07740783691406, -0.5834465026855469, -17.043460845947266, 36.492225646972656, 9.346992492675781, 144.4798583984375, 110.30570220947266, 165.1646728515625, 5.888605117797852, -1.5238475799560547, -99.153564453125, 69.46529388427734, 179.05210876464844, 85.00598907470703, 115.22428131103516, -0.3689422607421875, 1.794656753540039, 212.9858856201172, 116.56483459472656, 172.50929260253906, 120.86489868164062, 157.02944946289062, -4.577119827270508, 96.73489379882812, 67.22533416748047, 185.1222686767578, 49.270111083984375, 29.914505004882812, -50.407012939453125, 192.38575744628906, 158.9315185546875, 6.379964828491211, -5.501577377319336, 530.8749389648438, 48.87903594970703, 72.8263931274414, 9.87837028503418, 153.82431030273438, 12.991607666015625, -2.956737518310547, 86.652587890625, -1.7608489990234375, -42.9528923034668, 58.38018798828125, 18.501800537109375, 188.54539489746094, 207.46026611328125, -14.138648986816406, 139.81884765625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000365.npy"}
|
||||
{"epoch": 0.5517762660619804, "step": 366, "batch_size": 64, "mean": 64.93423461914062, "std": 124.55628204345703, "min": -247.64476013183594, "p10": -110.9188247680664, "median": 46.15084457397461, "p90": 202.67527770996097, "max": 336.23602294921875, "pos_frac": 0.703125, "sample": [36.7652702331543, 306.6270751953125, 95.57593536376953, 128.7722930908203, -23.048011779785156, 170.5445556640625, -161.89117431640625, -35.25688934326172, -71.59768676757812, 7.396751403808594, -73.48460388183594, 169.87310791015625, -4.419258117675781, 111.35997009277344, 124.5592041015625, 12.513671875, -119.9495849609375, 83.44281005859375, 4.151456832885742, 179.76531982421875, 99.94125366210938, 46.301307678222656, 67.00041198730469, -9.80294418334961, 46.00038146972656, 152.37289428710938, 39.7988166809082, 193.68179321289062, -247.64476013183594, -39.71641540527344, 243.60104370117188, 11.173164367675781, -0.49405860900878906, 317.052490234375, -142.44200134277344, 10.905065536499023, -113.12701416015625, -112.56120300292969, 14.628765106201172, 26.602947235107422, -0.099945068359375, -62.015071868896484, 4.289411544799805, 46.84088134765625, -123.61849975585938, 44.77076721191406, 16.192319869995117, 336.23602294921875, 229.9980926513672, 180.7528533935547, 75.4796142578125, 183.6507110595703, 169.16148376464844, -107.08660888671875, 194.99232482910156, -30.2179012298584, 155.63812255859375, 244.22581481933594, 205.9679718017578, 176.95269775390625, 186.09109497070312, 125.10090637207031, 185.37445068359375, 172.14112854003906], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000366.npy"}
|
||||
{"epoch": 0.5532879818594104, "step": 367, "batch_size": 64, "mean": 64.94355010986328, "std": 101.34310150146484, "min": -197.14981079101562, "p10": -70.78546295166015, "median": 69.73983764648438, "p90": 181.18622436523438, "max": 248.39697265625, "pos_frac": 0.765625, "sample": [1.4475975036621094, 248.39697265625, 98.19532012939453, -21.8251953125, 101.09297943115234, -4.08380126953125, 190.80783081054688, 150.88613891601562, 69.06260681152344, 6.941486358642578, 48.65751647949219, 4.655181884765625, 13.924415588378906, 84.97393798828125, 179.03488159179688, 10.917804718017578, 70.41706848144531, -14.128074645996094, -80.18567657470703, 188.90249633789062, -36.54502868652344, 142.81639099121094, -19.72239875793457, -112.98970031738281, 177.57318115234375, -176.3585968017578, -78.82429504394531, 145.06961059570312, 6.539569854736328, 35.2220458984375, 196.14727783203125, 181.575927734375, 161.71014404296875, 170.99249267578125, 169.52297973632812, 2.7202911376953125, 46.30451965332031, -73.9254150390625, 4.321012496948242, 152.3017120361328, 80.63829803466797, -197.14981079101562, 13.311212539672852, 224.29295349121094, 166.24261474609375, -47.78692626953125, 77.79188537597656, 33.03922653198242, 147.71890258789062, 30.027141571044922, 45.35734558105469, 95.26016235351562, -130.860595703125, 141.05828857421875, 63.1707763671875, 211.84902954101562, 75.55307006835938, -63.45890808105469, 161.2987518310547, 180.27691650390625, -4.93511962890625, 107.59123229980469, 130.77734375, 172.78021240234375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000367.npy"}
|
||||
{"epoch": 0.5547996976568406, "step": 368, "batch_size": 64, "mean": 57.044639587402344, "std": 103.28266143798828, "min": -191.66639709472656, "p10": -84.81496505737303, "median": 52.33685874938965, "p90": 193.11709289550782, "max": 262.0933532714844, "pos_frac": 0.6875, "sample": [-191.66639709472656, 187.61021423339844, 21.306320190429688, 13.343629837036133, 45.697723388671875, 190.75234985351562, 173.39584350585938, 86.2053451538086, -56.67223358154297, -38.40055847167969, 182.52780151367188, -9.196300506591797, -17.55413818359375, 46.33329772949219, 194.13055419921875, 124.43830871582031, -14.257415771484375, -112.70530700683594, -105.17547607421875, 115.21796417236328, 60.04905700683594, 27.590665817260742, 73.82413482666016, -67.29326629638672, 185.6779022216797, 23.335430145263672, -92.32426452636719, 204.78379821777344, 213.9346466064453, 50.35854721069336, -38.23313903808594, 262.0933532714844, 128.11785888671875, -0.4670982360839844, 66.73503112792969, 155.09617614746094, 77.76558685302734, 89.02153015136719, 70.04402923583984, 137.2589111328125, 34.52226257324219, -16.78126335144043, 133.87796020507812, 6.84259033203125, -59.164093017578125, -4.382057189941406, 30.505313873291016, -48.046478271484375, 213.05398559570312, -23.651260375976562, 76.47102355957031, 198.9766845703125, 239.29385375976562, 137.23202514648438, 39.86537170410156, 168.95973205566406, 64.35367584228516, -112.91372680664062, -113.71378326416016, 54.31517028808594, 104.12460327148438, 18.059938430786133, -121.23570251464844, 167.59060668945312], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000368.npy"}
|
||||
{"epoch": 0.5563114134542706, "step": 369, "batch_size": 64, "mean": 54.98382568359375, "std": 101.28605651855469, "min": -171.5729217529297, "p10": -71.37548255920409, "median": 41.69139289855957, "p90": 194.60481872558594, "max": 309.5675048828125, "pos_frac": 0.703125, "sample": [63.14427947998047, -108.94815826416016, 78.3971176147461, 190.9403076171875, -139.6156463623047, 32.754974365234375, 27.043437957763672, -18.706253051757812, 267.71661376953125, 41.76517868041992, 40.74799346923828, 159.53402709960938, 170.98512268066406, 109.27852630615234, 187.6065673828125, -0.36130523681640625, 90.71693420410156, 109.62605285644531, 8.348030090332031, 26.015621185302734, -29.747772216796875, -171.5729217529297, 115.8880615234375, -87.52931213378906, 95.23387145996094, 117.35751342773438, -58.11818313598633, 106.75244140625, 80.2635498046875, -14.799718856811523, 98.7574462890625, -19.626983642578125, 18.967744827270508, 201.46731567382812, 77.23979187011719, 187.35098266601562, 206.6501007080078, 20.046417236328125, 75.14163208007812, -88.9725570678711, -54.623748779296875, 41.61760711669922, -9.351757049560547, 49.50371551513672, -17.221149444580078, 78.27887725830078, 309.5675048828125, 0.482513427734375, 257.2081298828125, -39.663963317871094, 4.429832458496094, 52.102081298828125, 3.945476531982422, 204.07315063476562, -0.174102783203125, -5.397665023803711, 1.3022212982177734, 196.17532348632812, 26.83319854736328, 87.29708862304688, -77.05718231201172, 74.734375, -116.86136627197266, 184.0255889892578], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000369.npy"}
|
||||
{"epoch": 0.5578231292517006, "step": 370, "batch_size": 64, "mean": 80.09278869628906, "std": 116.04756927490234, "min": -280.6058349609375, "p10": -25.951753997802733, "median": 79.09668350219727, "p90": 214.6130142211914, "max": 301.8949890136719, "pos_frac": 0.765625, "sample": [118.58208465576172, 212.27923583984375, 149.8605194091797, 79.93229675292969, 301.8949890136719, 11.443595886230469, 209.6768341064453, -25.309219360351562, 183.1201171875, 33.52800369262695, 66.62828063964844, 90.08894348144531, 5.099632263183594, 226.55270385742188, 250.42506408691406, 163.5479736328125, 161.63792419433594, -145.57723999023438, -21.501388549804688, 132.46340942382812, 116.4029541015625, 157.95823669433594, 140.0394744873047, 111.55918884277344, 197.75987243652344, 24.230682373046875, 108.82527160644531, 71.85319519042969, -21.122901916503906, -9.019378662109375, -2.5763397216796875, -25.571510314941406, 2.1318130493164062, -113.38041687011719, 69.90176391601562, 78.26107025146484, 284.427734375, 52.40605163574219, 177.05136108398438, 186.2473602294922, 237.2681884765625, -280.6058349609375, 35.60536193847656, 128.31129455566406, 20.38203239440918, 215.6132049560547, 170.68824768066406, 167.1685333251953, 181.49819946289062, 243.98556518554688, -183.96128845214844, -14.411575317382812, 199.36740112304688, -3.6997547149658203, -26.114715576171875, 50.32850646972656, 66.97859191894531, 8.953689575195312, -106.02851104736328, 5.371694564819336, 13.480567932128906, 173.60977172851562, -96.73692321777344, 107.12650299072266], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000370.npy"}
|
||||
{"epoch": 0.5593348450491308, "step": 371, "batch_size": 64, "mean": 69.7497787475586, "std": 104.24437713623047, "min": -136.33847045898438, "p10": -77.78280181884763, "median": 81.5377082824707, "p90": 212.9661529541016, "max": 274.6344299316406, "pos_frac": 0.71875, "sample": [226.2234649658203, -97.50304412841797, 236.66259765625, 37.91693878173828, 79.1517333984375, 49.631134033203125, 152.49072265625, 14.040155410766602, -22.09752655029297, 5.76136589050293, 20.168212890625, 141.55575561523438, -97.28163146972656, 205.60562133789062, -3.5564804077148438, 274.6344299316406, -5.82530403137207, 163.09030151367188, 7.3120574951171875, 177.61517333984375, 104.51878356933594, -111.73600769042969, 1.0342540740966797, -8.422569274902344, 49.625396728515625, 156.36648559570312, 135.87408447265625, 260.3352966308594, 127.90736389160156, 102.90790557861328, 247.71087646484375, 216.12066650390625, -26.166770935058594, -22.118804931640625, 123.97282409667969, 34.1641845703125, -35.42631530761719, -136.33847045898438, -55.26564025878906, 195.25254821777344, -105.14482879638672, 50.31584548950195, 91.91693878173828, 91.1715087890625, 64.84246826171875, -38.45252227783203, 134.64315795898438, 147.58270263671875, -47.62718963623047, 86.98049926757812, 148.17726135253906, -23.794570922851562, 3.7797698974609375, 137.21621704101562, 227.99374389648438, 6.269187927246094, 188.83816528320312, 108.56696319580078, 97.33384704589844, 83.9236831665039, 131.90806579589844, 143.72543334960938, -104.65947723388672, -87.43301391601562], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000371.npy"}
|
||||
{"epoch": 0.5608465608465608, "step": 372, "batch_size": 64, "mean": 41.982093811035156, "std": 123.35909271240234, "min": -207.04022216796875, "p10": -117.98215179443359, "median": 16.116865158081055, "p90": 206.7002685546875, "max": 283.0057067871094, "pos_frac": 0.671875, "sample": [-3.9253692626953125, 17.156200408935547, 202.53179931640625, 86.87179565429688, 93.30645751953125, 147.81314086914062, 197.58265686035156, -54.34149932861328, 176.786865234375, 234.23062133789062, 234.98739624023438, -12.862380981445312, 144.3484344482422, 248.74063110351562, 42.535438537597656, 165.04122924804688, 194.080078125, -192.4403076171875, 1.9269943237304688, 20.29651641845703, -0.8734397888183594, -11.284543991088867, 18.436859130859375, 211.63433837890625, 13.442920684814453, -60.727516174316406, -112.80746459960938, -14.336441040039062, 208.48675537109375, -94.68555450439453, 104.61263275146484, 51.834442138671875, -120.19987487792969, 4.093784332275391, -140.9144744873047, 10.958486557006836, -189.08847045898438, 3.3604965209960938, 150.06683349609375, 7.852987289428711, 276.6484069824219, 52.57331466674805, 283.0057067871094, 17.761322021484375, -65.14823913574219, 100.21513366699219, -207.04022216796875, 3.4686737060546875, -4.69633674621582, 5.253120422363281, 15.077529907226562, -181.37596130371094, 156.3083038330078, 9.144004821777344, 4.82594108581543, -9.753122329711914, -179.56692504882812, 115.94426727294922, 193.3648681640625, 197.05213928222656, -109.3238525390625, 60.15437316894531, -56.75519561767578, 25.18731689453125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000372.npy"}
|
||||
{"epoch": 0.562358276643991, "step": 373, "batch_size": 64, "mean": 54.94411849975586, "std": 124.59658813476562, "min": -219.97869873046875, "p10": -113.29620361328124, "median": 53.79439735412598, "p90": 208.924772644043, "max": 365.75970458984375, "pos_frac": 0.671875, "sample": [-156.49063110351562, 179.27188110351562, 109.51918029785156, 365.75970458984375, 168.76437377929688, -197.71591186523438, -61.750709533691406, 157.00784301757812, 130.67868041992188, 69.79947662353516, 148.3260955810547, -154.4222412109375, -30.138351440429688, -23.450881958007812, -120.72172546386719, -11.16568374633789, 91.79486846923828, 148.99356079101562, -219.97869873046875, -75.22552490234375, 64.93510437011719, -21.57903289794922, 128.65481567382812, -175.58795166015625, 211.2279510498047, 179.54806518554688, -71.77015686035156, -3.6709747314453125, 63.812950134277344, 156.73512268066406, 201.76002502441406, 0.3041839599609375, 16.987621307373047, 203.55068969726562, -89.98877716064453, -12.051319122314453, 33.8582763671875, 184.3968505859375, -5.325080871582031, 173.2705078125, 138.4344940185547, 42.884033203125, -138.26527404785156, -74.60107421875, 74.11848449707031, 166.494384765625, -95.96998596191406, 226.61651611328125, 49.330711364746094, 65.01535034179688, 15.258346557617188, 216.9125518798828, 3.793426513671875, 240.83285522460938, 22.93604850769043, 58.25808334350586, 73.08262634277344, 123.16883850097656, 8.30462646484375, 275.55950927734375, 31.046281814575195, -4.771259307861328, 21.433364868164062, 218.62631225585938], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000373.npy"}
|
||||
{"epoch": 0.563869992441421, "step": 374, "batch_size": 64, "mean": 87.072021484375, "std": 128.74435424804688, "min": -298.150146484375, "p10": -41.23802337646483, "median": 82.42154693603516, "p90": 253.31983337402352, "max": 444.3794250488281, "pos_frac": 0.796875, "sample": [444.3794250488281, 227.90980529785156, 173.67942810058594, -30.570083618164062, 28.797657012939453, 187.22760009765625, 223.54266357421875, 262.3538513183594, 215.50039672851562, 62.18341827392578, 90.65770721435547, -29.866226196289062, 153.4541778564453, 282.5856018066406, 109.15596771240234, 41.32366943359375, 147.16412353515625, 7.568817138671875, 33.50260925292969, 57.392127990722656, 8.194435119628906, 175.39207458496094, 38.76773452758789, -204.30245971679688, 147.31988525390625, 152.39263916015625, 89.91474914550781, 234.08993530273438, -18.373214721679688, 142.2909698486328, 81.83198547363281, 117.82080078125, 0.0496978759765625, 42.980934143066406, 111.01759338378906, 63.16701889038086, -45.80999755859375, -298.150146484375, 298.031982421875, 123.93844604492188, -7.105335235595703, -159.14065551757812, 28.774099349975586, 286.85931396484375, 266.379638671875, 93.21514129638672, -193.80718994140625, -56.05242919921875, 77.76031494140625, 76.91519927978516, 97.320556640625, 152.2865447998047, 261.56121826171875, 78.26666259765625, -0.9108543395996094, -0.2359333038330078, 54.618125915527344, 192.920166015625, 181.84335327148438, 182.48876953125, -100.13316345214844, 12.868072509765625, 83.0111083984375, 14.398849487304688], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000374.npy"}
|
||||
{"epoch": 0.5653817082388511, "step": 375, "batch_size": 64, "mean": 65.92269897460938, "std": 123.38801574707031, "min": -193.05825805664062, "p10": -75.51368255615235, "median": 41.523590087890625, "p90": 223.86344299316414, "max": 390.3948669433594, "pos_frac": 0.671875, "sample": [-104.31562042236328, -82.8532485961914, 111.8948745727539, 154.9095458984375, 157.68734741210938, 13.449544906616211, 250.35931396484375, -7.48748779296875, 60.69757080078125, 128.17547607421875, -15.104774475097656, 231.71852111816406, -70.86375427246094, 169.70826721191406, 6.452922821044922, 4.328102111816406, 13.26801872253418, -154.5831756591797, 156.70404052734375, -6.1447601318359375, 22.970130920410156, 261.4983825683594, 186.06390380859375, -193.05825805664062, 198.14801025390625, 180.68551635742188, 4.332786560058594, -72.96142578125, 110.2558364868164, 178.69842529296875, 28.853317260742188, 2.3450374603271484, 3.470247268676758, -69.0952377319336, -18.9954833984375, 101.7479248046875, 140.415283203125, 54.19386291503906, 205.53492736816406, 181.748779296875, 259.7204284667969, 16.406415939331055, 132.44764709472656, -187.30186462402344, -73.47154235839844, 83.55583190917969, 260.60699462890625, -8.17120361328125, 390.3948669433594, 125.95893859863281, 236.51345825195312, 157.4273223876953, -26.976150512695312, 148.60890197753906, -65.39962768554688, -9.610877990722656, 143.94793701171875, -127.02482604980469, -32.75762176513672, 183.95208740234375, -13.651519775390625, 148.5492401123047, 26.86383819580078, -76.38888549804688], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000375.npy"}
|
||||
{"epoch": 0.5668934240362812, "step": 376, "batch_size": 64, "mean": 81.8962173461914, "std": 134.01837158203125, "min": -173.0171356201172, "p10": -81.42352485656735, "median": 78.00669860839844, "p90": 231.4058990478516, "max": 453.3071594238281, "pos_frac": 0.65625, "sample": [47.1624755859375, -29.530624389648438, -9.409822463989258, 59.306182861328125, 13.321540832519531, -15.091182708740234, -30.62188720703125, 123.88035583496094, -36.7987060546875, 38.6141357421875, 71.50570678710938, 134.3970184326172, -173.0171356201172, 0.8301620483398438, -111.02662658691406, -11.369171142578125, 236.89340209960938, 92.1248779296875, -6.449520111083984, 453.3071594238281, 127.31227111816406, 283.81591796875, 6.374725341796875, 166.9871063232422, 267.9173583984375, 5.112518310546875, 142.92535400390625, 160.40530395507812, 162.20953369140625, 159.9764404296875, 168.1602783203125, 45.56043243408203, 197.29037475585938, 84.75932312011719, 84.5076904296875, 225.80662536621094, 452.0835876464844, -60.683650970458984, 197.1252899169922, -90.31204223632812, 114.50533294677734, -130.1592559814453, 224.27743530273438, 91.4078369140625, 150.48609924316406, 393.3371276855469, -138.51031494140625, 151.23910522460938, -52.32276916503906, 172.60855102539062, -103.67708587646484, -8.769111633300781, 54.17597961425781, 89.36221313476562, 197.7900390625, -6.188104629516602, 233.8055877685547, -12.568767547607422, 185.044921875, 173.03555297851562, -51.778961181640625, -28.002641677856445, -0.6311092376708984, -92.47221374511719], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000376.npy"}
|
||||
{"epoch": 0.5684051398337112, "step": 377, "batch_size": 64, "mean": 54.798828125, "std": 100.71382904052734, "min": -183.28648376464844, "p10": -55.209159088134754, "median": 49.75811004638672, "p90": 174.6550277709961, "max": 272.884521484375, "pos_frac": 0.6875, "sample": [151.96063232421875, -32.14722442626953, 153.92813110351562, 163.09762573242188, 130.32281494140625, 20.144081115722656, -153.8018798828125, -114.39962768554688, 61.69892120361328, 25.046730041503906, 172.10931396484375, -6.052406311035156, 111.84940338134766, -2.02984619140625, 172.5749053955078, 121.57880401611328, -121.6417007446289, 94.78385162353516, -58.758766174316406, -23.58758544921875, 225.7415313720703, -12.786300659179688, 66.16874694824219, 49.60552215576172, 158.52200317382812, -183.28648376464844, 23.35711669921875, 272.884521484375, 0.7563133239746094, -72.98075866699219, 158.27401733398438, -6.622842788696289, 169.60723876953125, 1.2504425048828125, -46.92674255371094, -36.98448944091797, 113.89620971679688, 24.805137634277344, 218.5392608642578, 38.13764953613281, 4.812469482421875, 78.08440399169922, 0.7530765533447266, 68.51747131347656, 142.75527954101562, 200.2689666748047, 92.71609497070312, -16.570987701416016, -7.034576416015625, 70.8599853515625, 175.5465087890625, 63.90191650390625, -30.44927406311035, 179.01373291015625, 17.6422119140625, -12.147516250610352, 177.24526977539062, 49.91069793701172, -177.2367401123047, 19.73572540283203, 117.93090057373047, -26.728553771972656, 122.92616271972656, 166.03765869140625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000377.npy"}
|
||||
{"epoch": 0.5699168556311414, "step": 378, "batch_size": 64, "mean": 55.714988708496094, "std": 105.49913024902344, "min": -189.9314727783203, "p10": -30.837348937988278, "median": 23.23347282409668, "p90": 191.14490051269533, "max": 274.68572998046875, "pos_frac": 0.640625, "sample": [148.5602569580078, 126.22884368896484, -1.077413558959961, 4.933685302734375, 159.69342041015625, 193.47412109375, -87.03608703613281, 28.31308364868164, 101.94792175292969, 210.6717071533203, -173.59971618652344, 4.806379318237305, 274.68572998046875, -16.308452606201172, 206.15309143066406, 173.1326904296875, 107.83993530273438, -0.8837852478027344, 185.71005249023438, -12.51437759399414, 95.37019348144531, -15.664886474609375, -6.549365997314453, 88.21260833740234, 174.93942260742188, -32.48674011230469, 32.57952117919922, 5.456268310546875, -135.76206970214844, 16.469974517822266, 60.15081787109375, 24.594467163085938, -82.69952392578125, -1.8738079071044922, 144.65435791015625, 21.252037048339844, -15.885604858398438, 13.727920532226562, 149.88140869140625, 6.937826156616211, 258.349853515625, 139.19313049316406, -20.644012451171875, 165.89718627929688, 123.17416381835938, 177.80430603027344, -4.049633026123047, 183.01806640625, 87.1511459350586, 47.22241973876953, 14.57686996459961, -26.98876953125, 21.872478485107422, -151.1495361328125, -9.77322769165039, 92.42254638671875, -16.326032638549805, 256.3173522949219, -8.941717147827148, 51.76355743408203, -24.131267547607422, -189.9314727783203, -14.77336311340332, 235.669189453125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000378.npy"}
|
||||
{"epoch": 0.5714285714285714, "step": 379, "batch_size": 64, "mean": 80.19296264648438, "std": 121.54928588867188, "min": -181.8527374267578, "p10": -48.7426975250244, "median": 40.86650085449219, "p90": 206.37339477539064, "max": 379.26409912109375, "pos_frac": 0.8125, "sample": [102.49043273925781, 20.298873901367188, 168.71261596679688, -24.293853759765625, 4.59796142578125, 172.4623565673828, 34.93561553955078, -146.08074951171875, 107.3912353515625, 8.648857116699219, 57.589622497558594, 132.13644409179688, -87.32174682617188, 137.09039306640625, 3.6803970336914062, 376.90887451171875, 13.627349853515625, -56.0016975402832, 169.15023803710938, 243.10606384277344, 379.26409912109375, 190.54681396484375, 194.3223419189453, 173.26666259765625, 177.95440673828125, -108.56201171875, 150.1295166015625, -181.8527374267578, 21.83264923095703, -63.83477783203125, 297.47113037109375, 33.838233947753906, -1.3415336608886719, 31.76761245727539, -177.11190795898438, 279.0325622558594, 188.21945190429688, 40.51603698730469, 38.76728820800781, -5.010631561279297, -31.805030822753906, -16.931922912597656, 136.85226440429688, 208.20648193359375, 202.09619140625, 17.978382110595703, 184.83848571777344, 361.7834777832031, 14.95663833618164, 95.27162170410156, 84.24418640136719, 14.752593994140625, 53.1873779296875, 26.61543846130371, 159.55088806152344, 2.7401580810546875, 41.21696472167969, 63.24510955810547, 35.50197982788086, 5.289911270141602, 172.10565185546875, 7.6751861572265625, 175.06106567382812, 19.572174072265625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000379.npy"}
|
||||
{"epoch": 0.5729402872260015, "step": 380, "batch_size": 64, "mean": 85.70262145996094, "std": 114.60368347167969, "min": -195.0770263671875, "p10": -52.80879516601561, "median": 86.42996978759766, "p90": 202.31587677001954, "max": 342.1642761230469, "pos_frac": 0.8125, "sample": [4.964069366455078, 152.27133178710938, 105.67781066894531, 81.27127075195312, -118.73843383789062, -150.52621459960938, 180.1754150390625, 270.60491943359375, -3.333904266357422, 181.1712188720703, 159.12290954589844, 60.5566291809082, 0.7628822326660156, 72.89128112792969, 18.205326080322266, -34.196075439453125, 16.107982635498047, 48.359222412109375, 146.99842834472656, 191.8244171142578, 131.2501983642578, 121.54549407958984, 32.80426025390625, -195.0770263671875, 186.98208618164062, -25.825355529785156, 183.19873046875, 77.62201690673828, -60.785675048828125, -1.6169147491455078, 134.3498992919922, 277.11053466796875, 112.41606903076172, 14.10141372680664, 317.9117431640625, 13.81766128540039, 178.71847534179688, 137.90090942382812, 40.06776428222656, 5.638618469238281, 168.45974731445312, -152.71485900878906, 184.72337341308594, 200.16854858398438, 32.24183654785156, 168.91326904296875, -70.04508209228516, 91.58866882324219, 46.997711181640625, 176.77096557617188, 47.116188049316406, -0.9749393463134766, -145.5338592529297, 246.07464599609375, 203.2361602783203, 199.07171630859375, 8.315439224243164, 342.1642761230469, 167.1865234375, 220.450439453125, 99.06201934814453, 69.94232177734375, 13.900651931762695, 101.55062866210938], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000380.npy"}
|
||||
{"epoch": 0.5744520030234316, "step": 381, "batch_size": 64, "mean": 34.12318420410156, "std": 125.15888977050781, "min": -327.50921630859375, "p10": -102.35578384399413, "median": 10.16884994506836, "p90": 184.1152084350586, "max": 405.9857482910156, "pos_frac": 0.578125, "sample": [49.129730224609375, 54.07875061035156, -3.2960948944091797, -149.37936401367188, -275.93487548828125, 18.09567642211914, 8.564506530761719, 25.229488372802734, 203.50363159179688, 96.18826293945312, -14.679725646972656, -4.056369781494141, 32.9638671875, 20.166805267333984, 68.1463623046875, 405.9857482910156, 157.91452026367188, -3.450237274169922, 69.45088195800781, -3.6787185668945312, 36.5667724609375, -51.555442810058594, -3.0053272247314453, 328.98553466796875, -138.98089599609375, 108.80683135986328, -104.69635009765625, -48.51280212402344, 159.22372436523438, -1.7157726287841797, 82.7486801147461, 179.188232421875, -1.5557403564453125, 35.84181213378906, -5.663105010986328, 204.87266540527344, -3.7411155700683594, -6.173160552978516, -174.4019775390625, 28.632247924804688, 184.58395385742188, -11.08929443359375, -49.774085998535156, 1.7540950775146484, 69.23230743408203, 203.0572052001953, 162.85491943359375, 130.08892822265625, -0.21692466735839844, 5.080467224121094, 0.216705322265625, 143.2064971923828, 257.88629150390625, -47.24809265136719, 3.425586700439453, -45.22193908691406, 156.80703735351562, -327.50921630859375, -127.09227752685547, 88.82716369628906, 11.773193359375, -96.89446258544922, -92.6936264038086, 183.02146911621094], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000381.npy"}
|
||||
{"epoch": 0.5759637188208617, "step": 382, "batch_size": 64, "mean": 63.80790710449219, "std": 102.09278106689453, "min": -188.43406677246094, "p10": -43.11510620117187, "median": 36.632240295410156, "p90": 196.63435821533204, "max": 315.6812744140625, "pos_frac": 0.703125, "sample": [-63.04632568359375, -8.116308212280273, 315.6812744140625, 27.078168869018555, 242.11886596679688, -58.532073974609375, 259.483642578125, 9.184600830078125, 248.93753051757812, 128.8326873779297, 90.29081726074219, -33.81422424316406, -30.368820190429688, 59.951751708984375, 191.48248291015625, 72.12678527832031, -38.76020812988281, 36.522369384765625, -188.43406677246094, 173.19287109375, -2.495450973510742, 36.74211120605469, -1.7010650634765625, 13.534385681152344, -36.42289733886719, 8.433013916015625, 103.3426284790039, 50.43208312988281, -78.2004623413086, 197.6305389404297, 126.27888488769531, 32.21876907348633, -58.1024284362793, -44.98149108886719, 194.3099365234375, -11.915374755859375, -53.213802337646484, 14.644485473632812, 48.840248107910156, 183.82003784179688, 58.580718994140625, 283.0781555175781, -24.020034790039062, 16.042526245117188, 21.5660400390625, 154.06512451171875, -18.972583770751953, -33.10980224609375, 75.111328125, 31.87195587158203, 81.40618896484375, 161.4334259033203, -6.931110382080078, 25.288646697998047, 169.60313415527344, 1.6132011413574219, 124.91873168945312, 52.63857650756836, 189.29208374023438, 248.6220703125, 52.99808883666992, 151.5403289794922, 97.54116821289062, 12.522193908691406], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000382.npy"}
|
||||
{"epoch": 0.5774754346182918, "step": 383, "batch_size": 64, "mean": 52.359596252441406, "std": 127.82490539550781, "min": -295.3905334472656, "p10": -102.23295593261719, "median": 44.73037338256836, "p90": 212.64033203125004, "max": 397.3662414550781, "pos_frac": 0.625, "sample": [-8.675600051879883, 158.8397216796875, 162.8475341796875, 47.5966796875, -33.144935607910156, -18.2586669921875, 153.32241821289062, 47.198814392089844, 41.99956512451172, -119.37290954589844, -9.26662826538086, -101.1343994140625, 94.74313354492188, 66.8695297241211, 179.35858154296875, 36.98193359375, 222.24038696289062, -56.362953186035156, 223.9638671875, 198.88949584960938, 83.52116394042969, -102.70376586914062, 242.8805694580078, -152.31329345703125, 177.76071166992188, -26.110137939453125, 66.70382690429688, -8.985549926757812, -8.179525375366211, 397.3662414550781, 200.6083984375, -70.54916381835938, -84.61031341552734, 217.796875, 20.155216217041016, 126.94253540039062, 136.91885375976562, -295.3905334472656, 12.823089599609375, 42.261932373046875, -162.28756713867188, 68.32150268554688, -38.49627685546875, -3.762144088745117, 169.03834533691406, -152.37384033203125, 4.339515686035156, 109.75128173828125, -47.64068603515625, 298.5553894042969, 35.59099578857422, 219.62591552734375, 55.099510192871094, 40.04176330566406, -22.877410888671875, 176.34149169921875, 112.12881469726562, 105.11044311523438, -39.98827362060547, -166.28591918945312, 187.30609130859375, -94.41244506835938, 179.9951629638672, 52.35979461669922], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000383.npy"}
|
||||
{"epoch": 0.5789871504157218, "step": 384, "batch_size": 64, "mean": 66.59786987304688, "std": 103.2125244140625, "min": -298.7729797363281, "p10": -47.67302169799802, "median": 62.11140823364258, "p90": 198.08052520751954, "max": 256.8943786621094, "pos_frac": 0.8125, "sample": [-60.16204833984375, 210.52490234375, 6.284950256347656, 153.11026000976562, 15.543811798095703, 74.62381744384766, 171.78594970703125, 44.32361602783203, 205.81570434570312, -1.6538963317871094, 100.4071044921875, 107.21134948730469, 65.7266845703125, 28.844764709472656, -18.531959533691406, 41.22657775878906, 20.72754669189453, 250.70156860351562, -13.880035400390625, 195.9401397705078, -63.63550567626953, 124.46371459960938, 22.56305694580078, -15.339340209960938, -132.8929901123047, 115.1622314453125, 256.8943786621094, 92.9586410522461, -145.54898071289062, 152.2609405517578, 198.99783325195312, 143.43887329101562, 0.0077667236328125, 59.015052795410156, 70.39874267578125, 48.17579650878906, 186.510498046875, 11.152507781982422, -298.7729797363281, 215.15618896484375, 41.523162841796875, 120.55712890625, 177.9638671875, -97.6018295288086, 254.45147705078125, 144.96231079101562, 85.40077209472656, 17.642595291137695, 6.239849090576172, 157.47979736328125, 127.05545043945312, 15.612136840820312, 81.2659912109375, 90.08918762207031, 26.222557067871094, 6.501255035400391, -113.17850494384766, 65.207763671875, 28.846023559570312, -7.580280303955078, 143.226318359375, 143.25479125976562, 58.64337158203125, 48.94147491455078], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000384.npy"}
|
||||
{"epoch": 0.5804988662131519, "step": 385, "batch_size": 64, "mean": 85.55187225341797, "std": 106.57611846923828, "min": -125.39683532714844, "p10": -59.114642333984364, "median": 77.70602798461914, "p90": 221.854167175293, "max": 327.709716796875, "pos_frac": 0.78125, "sample": [-118.77336883544922, 5.779201507568359, 179.3037567138672, 181.43402099609375, 67.79777526855469, 3.557403564453125, 101.48675537109375, 148.19383239746094, 35.29686737060547, 55.183135986328125, -125.39683532714844, 129.77517700195312, 87.6142807006836, 182.0208740234375, 42.799400329589844, 149.380615234375, 187.6968994140625, 119.48735046386719, 15.657386779785156, 42.35954666137695, 18.027618408203125, -4.9342193603515625, -14.519317626953125, 222.35519409179688, 211.91226196289062, 321.3111572265625, 4.445270538330078, 128.8656768798828, 44.50458526611328, 189.43016052246094, 143.7710723876953, -68.55561828613281, 109.06108093261719, 10.514236450195312, 58.57486343383789, 238.13865661621094, 155.23806762695312, 140.2974395751953, 140.15444946289062, 156.38677978515625, -3.1393470764160156, -71.03006744384766, 230.30050659179688, 89.13883972167969, 8.700416564941406, 327.709716796875, 267.95013427734375, -19.024337768554688, -17.699947357177734, 43.73345947265625, -64.74429321289062, 12.79007339477539, -45.978790283203125, -66.98739624023438, 166.05442810058594, 113.9874496459961, 23.76880645751953, 198.08668518066406, -76.27071380615234, 228.8987579345703, -18.725116729736328, 30.262001037597656, 201.2198028564453, 220.6851043701172], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000385.npy"}
|
||||
{"epoch": 0.582010582010582, "step": 386, "batch_size": 64, "mean": 57.493988037109375, "std": 110.54379272460938, "min": -230.18345642089844, "p10": -65.82039070129395, "median": 59.568580627441406, "p90": 207.39331665039063, "max": 264.3594970703125, "pos_frac": 0.71875, "sample": [220.01815795898438, -111.14990234375, -60.59275436401367, 38.26239013671875, 95.13761901855469, 3.9354248046875, 2.6961288452148438, 89.18666076660156, 108.53575134277344, 83.4952621459961, 25.807022094726562, 118.783203125, -158.6567840576172, -15.240447998046875, 78.164794921875, 154.44573974609375, 178.92337036132812, 128.97091674804688, 152.77691650390625, 175.71629333496094, -1.843912124633789, 71.71590423583984, 114.56348419189453, -85.98475646972656, -60.37810516357422, 165.78697204589844, 141.28515625, -23.7623291015625, 84.98784637451172, 128.75955200195312, 75.56751251220703, 205.14202880859375, 11.261268615722656, 124.6204833984375, -25.76669692993164, -47.04925537109375, 102.12647247314453, 35.09600830078125, 115.54054260253906, 235.070068359375, 30.112396240234375, -36.9483642578125, 33.77014923095703, 203.70169067382812, 208.358154296875, -68.06080627441406, -137.58270263671875, 65.4682388305664, 1.943338394165039, 243.02346801757812, -13.268104553222656, -47.29759216308594, -223.90301513671875, 218.99598693847656, 264.3594970703125, 53.668922424316406, 48.73368835449219, 138.3179473876953, 29.368118286132812, 210.22955322265625, 33.55537414550781, -41.2612419128418, 18.55982208251953, -230.18345642089844], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000386.npy"}
|
||||
{"epoch": 0.5835222978080121, "step": 387, "batch_size": 64, "mean": 80.18566131591797, "std": 104.44783020019531, "min": -181.26039123535156, "p10": -30.79741325378417, "median": 58.68577575683594, "p90": 205.66472778320315, "max": 371.37066650390625, "pos_frac": 0.78125, "sample": [131.05599975585938, 17.547164916992188, 144.72744750976562, -73.4932632446289, -16.273880004882812, 280.4013671875, 70.79983520507812, -33.9443244934082, 7.773496627807617, 371.37066650390625, 13.793855667114258, -1.814035415649414, 53.15440368652344, 62.734130859375, 198.39024353027344, 6.560050964355469, 44.66941833496094, 208.78236389160156, 197.8155059814453, 93.086181640625, 267.6132507324219, 188.83367919921875, 175.82626342773438, -51.93796920776367, 7.863853454589844, 164.80462646484375, -46.23662567138672, 11.31706428527832, 162.6708984375, 126.12631225585938, -16.988292694091797, 16.944793701171875, 21.923011779785156, 191.5327911376953, -23.454620361328125, 104.13347625732422, 69.20912170410156, 44.32926940917969, 33.37765884399414, -11.904617309570312, -1.9996204376220703, 82.3265380859375, 75.47069549560547, -181.26039123535156, -82.3525390625, 29.443782806396484, 180.11131286621094, 226.89968872070312, 90.06500244140625, 176.46142578125, -10.474559783935547, 49.05781555175781, 27.833953857421875, 104.24301147460938, 86.70426940917969, 49.74946975708008, -63.01634979248047, 54.637420654296875, 152.3576202392578, 267.80792236328125, 294.6932678222656, 110.54740142822266, 22.078794479370117, 177.37582397460938], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000387.npy"}
|
||||
{"epoch": 0.5850340136054422, "step": 388, "batch_size": 64, "mean": 48.54740905761719, "std": 126.82772064208984, "min": -219.19395446777344, "p10": -92.8024429321289, "median": 13.649772644042969, "p90": 192.42189331054686, "max": 416.13031005859375, "pos_frac": 0.609375, "sample": [258.5332946777344, 97.52346801757812, 75.07086181640625, 245.31654357910156, 56.822959899902344, 226.3092498779297, 41.75010681152344, 85.16242980957031, 46.383323669433594, -24.182960510253906, -0.7056427001953125, 201.17947387695312, -37.402557373046875, 121.05265045166016, 10.721389770507812, 416.13031005859375, 1.4928245544433594, 333.890869140625, 16.578155517578125, 173.91046142578125, 191.13003540039062, -29.692550659179688, 185.92660522460938, 78.98736572265625, 182.2796173095703, -196.49676513671875, 146.8782958984375, 77.010986328125, 5.581476211547852, 9.287355422973633, -6.42064094543457, -11.671728134155273, -2.231536865234375, 170.15975952148438, -0.47102928161621094, 22.92059326171875, 56.085689544677734, -21.39306640625, 7.557687759399414, -125.64743041992188, 66.0361557006836, 157.96194458007812, -219.19395446777344, 192.42823791503906, 192.40708923339844, -0.05413818359375, 42.53368377685547, -6.75860595703125, -3.962188720703125, 192.38523864746094, -24.185760498046875, 182.99488830566406, -69.79971313476562, 9.416793823242188, -205.48049926757812, -89.9833984375, -94.01060485839844, 135.266845703125, -62.95175552368164, -7.656028747558594, -23.16259765625, -147.43356323242188, -200.62664794921875, 5.544633865356445], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000388.npy"}
|
||||
{"epoch": 0.5865457294028723, "step": 389, "batch_size": 64, "mean": 79.76494598388672, "std": 110.5688705444336, "min": -180.10952758789062, "p10": -40.720320510864255, "median": 78.30721282958984, "p90": 228.08948974609382, "max": 352.8174743652344, "pos_frac": 0.671875, "sample": [-2.4323272705078125, -9.520063400268555, 134.55284118652344, 108.94439697265625, 161.5891571044922, 38.313690185546875, 152.48709106445312, 158.73995971679688, 262.3890380859375, 137.0057373046875, 5.141815185546875, -48.36924743652344, 84.95094299316406, 170.32493591308594, 51.08518981933594, 156.17079162597656, 127.5030517578125, 199.72265625, 122.8828125, -11.42221450805664, 177.7061309814453, 235.35101318359375, -5.237863540649414, 71.66348266601562, 132.995849609375, -5.104438781738281, 211.14593505859375, -34.75243377685547, 100.53804016113281, 90.7216567993164, -73.72161865234375, 158.62896728515625, -24.822837829589844, 200.51162719726562, 208.0628662109375, 70.16741943359375, 246.87960815429688, 105.72136688232422, 65.38668823242188, -180.10952758789062, 290.3825988769531, -12.560138702392578, -6.630746841430664, -40.27811813354492, 36.30194854736328, -92.89053344726562, 238.14566040039062, -39.80404281616211, -75.03553771972656, 1.52447509765625, 17.285072326660156, 0.6135101318359375, -99.89675903320312, -4.405570983886719, -40.90983581542969, 237.79156494140625, 191.77992248535156, 140.33030700683594, 89.67713928222656, 352.8174743652344, -26.366777420043945, 195.46224975585938, 5.21153450012207, -5.380893707275391], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000389.npy"}
|
||||
{"epoch": 0.5880574452003023, "step": 390, "batch_size": 64, "mean": 61.533573150634766, "std": 103.03948211669922, "min": -185.66200256347656, "p10": -44.6884162902832, "median": 25.310744285583496, "p90": 203.64956970214845, "max": 295.5864562988281, "pos_frac": 0.71875, "sample": [-47.041717529296875, 84.57941436767578, 124.50480651855469, -2.82647705078125, 145.48773193359375, 22.54230499267578, 174.989013671875, 169.28643798828125, -38.15966796875, 22.82489776611328, 1.078481674194336, -48.79618835449219, 171.839599609375, 26.719655990600586, -10.418357849121094, 275.1082458496094, 161.57260131835938, -105.54696655273438, 137.099365234375, -110.04865264892578, 74.08659362792969, 65.1911849975586, 2.5287246704101562, 215.22433471679688, 275.627685546875, 179.73434448242188, 41.675811767578125, -36.43461608886719, 3.714710235595703, 146.94842529296875, -72.57196807861328, 5.172781944274902, -4.555219650268555, 118.44397735595703, -10.25674057006836, -39.19738006591797, -185.66200256347656, 213.72500610351562, 196.619384765625, -3.6348304748535156, 157.02114868164062, 11.300447463989258, -16.628677368164062, 32.65118408203125, 2.9323692321777344, 85.74795532226562, 8.735313415527344, -19.250572204589844, 295.5864562988281, 23.901832580566406, 224.93814086914062, 151.15963745117188, 21.97911834716797, 21.916128158569336, 8.985191345214844, 122.295654296875, -1.119272232055664, -126.71208190917969, 116.02001953125, 1.3126544952392578, 117.63511657714844, 75.75065612792969, 74.15290069580078, 206.66250610351562], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000390.npy"}
|
||||
{"epoch": 0.5895691609977324, "step": 391, "batch_size": 64, "mean": 76.68891906738281, "std": 120.05994415283203, "min": -198.30101013183594, "p10": -67.50263137817382, "median": 70.37997817993164, "p90": 214.5053451538086, "max": 386.94085693359375, "pos_frac": 0.75, "sample": [96.70759582519531, -181.06956481933594, 188.06549072265625, 27.13494873046875, -9.688026428222656, 217.49032592773438, -14.30630874633789, 209.4012451171875, 183.36007690429688, 386.94085693359375, 58.832122802734375, -142.0210418701172, 201.21665954589844, 52.06603240966797, 135.02723693847656, 236.73074340820312, -72.16991424560547, 90.5482406616211, -51.59324645996094, 47.375694274902344, -17.668060302734375, 128.61618041992188, 144.29251098632812, 69.02727508544922, 71.73268127441406, -114.23516845703125, -198.30101013183594, -21.2208251953125, 61.93601989746094, -34.58538055419922, 368.93292236328125, 217.97808837890625, 139.0606689453125, 8.908699035644531, 113.09796142578125, 9.391120910644531, -11.620817184448242, 52.133880615234375, 185.0256805419922, 212.04632568359375, 143.92556762695312, 85.94931030273438, -177.9442138671875, -56.6123046875, 55.60517120361328, 80.51902770996094, 215.13912963867188, 62.99573516845703, 167.81570434570312, 159.73248291015625, 84.48040771484375, -48.21650695800781, 214.7736358642578, 139.31539916992188, 213.87933349609375, 171.7750244140625, 158.29946899414062, 21.659887313842773, -88.4811782836914, 3.840576171875, 65.18074035644531, 31.405094146728516, 130.0330352783203, 28.42236328125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000391.npy"}
|
||||
{"epoch": 0.5910808767951625, "step": 392, "batch_size": 64, "mean": 78.31180572509766, "std": 127.3968505859375, "min": -286.9374084472656, "p10": -34.081506347656244, "median": 63.850006103515625, "p90": 229.21379699707032, "max": 493.91937255859375, "pos_frac": 0.765625, "sample": [-193.53204345703125, -3.5564422607421875, 4.209220886230469, 55.858917236328125, 66.91802978515625, -24.276199340820312, 181.52357482910156, -55.596832275390625, 103.842041015625, 188.77711486816406, 81.81663513183594, -286.9374084472656, 159.96859741210938, 184.69537353515625, 266.82281494140625, 48.268951416015625, 37.70814514160156, 176.41357421875, 236.8205108642578, 41.363677978515625, 82.30474090576172, -27.3548583984375, 54.09467315673828, -128.2703094482422, 102.46368408203125, 221.418212890625, 196.13050842285156, 33.2288818359375, 60.781982421875, -133.34632873535156, 102.31343078613281, 282.2860107421875, 117.75312805175781, 43.718109130859375, -36.96435546875, 55.35346984863281, 7.568277359008789, 134.57797241210938, 100.41134643554688, 162.2940673828125, 253.39797973632812, 231.25653076171875, -21.699739456176758, 191.40234375, 1.6587295532226562, 2.6621170043945312, 7.770393371582031, 69.30148315429688, 224.44741821289062, -2.3460235595703125, 5.028709411621094, 493.91937255859375, 8.472766876220703, 8.748199462890625, -26.643096923828125, -1.123891830444336, 365.14453125, -95.06661987304688, 67.60562133789062, 144.66226196289062, -1.3737239837646484, 132.5900421142578, 68.54930114746094, 181.71990966796875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000392.npy"}
|
||||
{"epoch": 0.5925925925925926, "step": 393, "batch_size": 64, "mean": 82.41268157958984, "std": 107.1650390625, "min": -214.26353454589844, "p10": -21.879794883728028, "median": 78.08541870117188, "p90": 217.8937484741211, "max": 328.5438232421875, "pos_frac": 0.703125, "sample": [86.16558837890625, -10.070802688598633, 102.19932556152344, 201.4310760498047, 186.74400329589844, 163.23748779296875, 23.14389419555664, 58.10649871826172, 98.07347869873047, 77.92210388183594, -8.496259689331055, -2.2359848022460938, 232.99514770507812, 151.89352416992188, 14.695762634277344, -100.6356430053711, -2.1933975219726562, 214.9054718017578, 24.065391540527344, -2.3111534118652344, -22.504425048828125, 10.915918350219727, 297.3081970214844, 229.5986328125, 107.53192138671875, 205.12437438964844, -21.81254768371582, 159.598876953125, 133.06764221191406, 155.97718811035156, 60.56904602050781, -111.05953979492188, 286.96258544921875, 209.9664306640625, -1.5352115631103516, 78.24873352050781, -59.66606140136719, 219.1744384765625, -16.592254638671875, -17.319181442260742, 118.44599151611328, 41.16120910644531, 161.49111938476562, -0.21110153198242188, 50.99413299560547, 167.33956909179688, 150.21585083007812, -21.908615112304688, 91.72557067871094, 259.3678283691406, 184.03318786621094, 328.5438232421875, 90.32319641113281, -15.462776184082031, 24.02344512939453, 176.7734375, 43.42494201660156, 122.67033386230469, -214.26353454589844, -45.75199890136719, 7.684600830078125, 109.916015625, 33.37855529785156, -2.6932525634765625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000393.npy"}
|
||||
{"epoch": 0.5941043083900227, "step": 394, "batch_size": 64, "mean": 71.82258605957031, "std": 100.01547241210938, "min": -163.86961364746094, "p10": -43.83128814697265, "median": 70.55657958984375, "p90": 184.8549011230469, "max": 284.5919494628906, "pos_frac": 0.75, "sample": [18.937103271484375, 185.05262756347656, 184.39353942871094, 23.232284545898438, -20.822364807128906, 76.7447509765625, -10.238174438476562, 47.188690185546875, 275.4301452636719, 78.67459106445312, 270.1346435546875, 136.8626708984375, 71.9293212890625, 62.48378372192383, -101.83039855957031, 27.14821434020996, -40.030479431152344, -17.284685134887695, -110.14022827148438, 30.748355865478516, 187.42153930664062, -45.441226959228516, 58.78062057495117, 29.397903442382812, 133.57240295410156, 212.6562957763672, 171.57554626464844, 141.98199462890625, 180.57276916503906, -8.803642272949219, -140.8634033203125, 66.02545166015625, 188.84373474121094, 101.63192749023438, 56.80558776855469, 111.49954223632812, -10.513341903686523, 153.89122009277344, 133.12026977539062, 20.44247055053711, 119.90011596679688, -40.074764251708984, -130.91943359375, 72.98897552490234, 131.5584259033203, 284.5919494628906, -163.86961364746094, 165.78762817382812, -4.957191467285156, 84.62986755371094, -74.87826538085938, 145.4952392578125, 59.90929412841797, 159.79190063476562, 44.98033142089844, 112.02971649169922, 106.86076354980469, 164.19290161132812, 9.324508666992188, 69.183837890625, -12.996902465820312, 182.4696044921875, 28.79553985595703, 150.63876342773438], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000394.npy"}
|
||||
{"epoch": 0.5956160241874527, "step": 395, "batch_size": 64, "mean": 70.3055648803711, "std": 111.27969360351562, "min": -150.82847595214844, "p10": -52.609505844116214, "median": 52.9982795715332, "p90": 208.31217956542974, "max": 394.1062316894531, "pos_frac": 0.734375, "sample": [139.4495391845703, 265.2991943359375, 137.42738342285156, 13.10687255859375, 186.81600952148438, 120.60295104980469, -3.946889877319336, 119.7828598022461, 214.37677001953125, -52.653564453125, 63.69427490234375, 140.17056274414062, 267.86505126953125, 180.33343505859375, 36.636924743652344, 188.17440795898438, -24.87158966064453, 149.7748565673828, 226.73651123046875, 13.73052978515625, -3.7673263549804688, 59.546241760253906, 0.67413330078125, 85.20828247070312, 186.34410095214844, -133.95413208007812, 37.779014587402344, 194.16146850585938, 61.2890510559082, -22.803443908691406, 87.88043975830078, 394.1062316894531, 128.4159698486328, 62.829246520996094, -150.82847595214844, 19.07196044921875, -74.18527221679688, 34.1065559387207, 128.73916625976562, -82.70115661621094, -7.794525146484375, 1.5268993377685547, 132.66366577148438, 23.18294906616211, 1.8527870178222656, 67.01449584960938, -19.44243621826172, 261.68731689453125, -65.70957946777344, -4.732980728149414, 7.354766845703125, 349.2982177734375, -101.86859130859375, 10.27297592163086, 60.89366149902344, 178.18087768554688, -52.5067024230957, 46.4503173828125, 120.49607849121094, -31.183998107910156, 17.516109466552734, -40.059814453125, 15.745256423950195, 134.30026245117188], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000395.npy"}
|
||||
{"epoch": 0.5971277399848829, "step": 396, "batch_size": 64, "mean": 70.58039855957031, "std": 103.08235931396484, "min": -203.94119262695312, "p10": -26.392331314086913, "median": 54.92537307739258, "p90": 215.6065841674805, "max": 346.62078857421875, "pos_frac": 0.71875, "sample": [168.79736328125, 346.62078857421875, 218.0914764404297, 157.34835815429688, 209.80850219726562, 134.74371337890625, 16.90591049194336, 184.86370849609375, 58.10865020751953, 75.43269348144531, 109.88220977783203, -11.113555908203125, 172.75900268554688, -25.986316680908203, 54.93152618408203, -3.6228103637695312, 90.14892578125, 56.210960388183594, 4.160976409912109, 40.56975555419922, 40.04252624511719, -29.235244750976562, 231.7342529296875, 68.38311004638672, -16.023841857910156, 56.731849670410156, 120.47023010253906, -203.94119262695312, 229.7447509765625, 124.26239013671875, -0.24995803833007812, -147.70277404785156, -0.4882621765136719, 200.51063537597656, 86.24978637695312, 58.704750061035156, 145.21641540527344, 54.919219970703125, 234.35232543945312, 335.2041015625, 3.4869956970214844, 8.989561080932617, 4.415142059326172, 135.43531799316406, -6.538627624511719, 49.30902099609375, -8.557197570800781, 47.210121154785156, 77.57620239257812, -38.61259460449219, -26.56633758544922, -30.448497772216797, -3.905202865600586, 7.6635284423828125, 6.251857757568359, 182.83987426757812, 76.81546783447266, 219.29559326171875, 15.288135528564453, 28.1245174407959, -7.7111663818359375, -42.909000396728516, -21.023277282714844, 193.1691436767578], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000396.npy"}
|
||||
{"epoch": 0.5986394557823129, "step": 397, "batch_size": 64, "mean": 40.21295928955078, "std": 117.55915832519531, "min": -282.785888671875, "p10": -84.31422424316406, "median": 32.15547752380371, "p90": 191.88574066162113, "max": 429.9638977050781, "pos_frac": 0.65625, "sample": [-6.066179275512695, 59.44055938720703, -85.50408935546875, 41.08837890625, -101.5110092163086, 78.82389831542969, 163.07232666015625, 103.6878433227539, 209.28468322753906, -38.92408752441406, 1.0084991455078125, 88.59725952148438, 53.970420837402344, 157.9798126220703, 90.7132797241211, 109.76770782470703, -185.508056640625, -70.68231201171875, 68.24766540527344, 17.949783325195312, 140.7032012939453, -1.0894927978515625, 40.720176696777344, 82.88229370117188, -1.9680862426757812, -282.785888671875, -81.53787231445312, 25.242843627929688, 202.99392700195312, 20.794021606445312, -22.44799041748047, 48.73290252685547, 199.45819091796875, 83.77774047851562, 182.05580139160156, -48.3282470703125, 39.068111419677734, 225.35751342773438, 138.92227172851562, 16.44389533996582, 47.999847412109375, 249.221435546875, 167.13308715820312, 22.033355712890625, -200.53305053710938, 98.69081115722656, 429.9638977050781, -50.7392578125, -44.66657257080078, 39.988037109375, -7.959239959716797, 121.65168762207031, -11.301719665527344, 9.937171936035156, 0.4243602752685547, -45.36634826660156, -0.8186721801757812, 23.902877807617188, -51.47174072265625, 134.05018615722656, -165.39895629882812, -153.65289306640625, 0.010606765747070312, 196.09857177734375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000397.npy"}
|
||||
{"epoch": 0.600151171579743, "step": 398, "batch_size": 64, "mean": 95.13571166992188, "std": 119.2037582397461, "min": -99.1190185546875, "p10": -30.476065063476558, "median": 83.46660614013672, "p90": 199.33441162109375, "max": 593.1195678710938, "pos_frac": 0.765625, "sample": [177.97674560546875, 82.53453063964844, 165.2187957763672, 118.99342346191406, 213.87969970703125, 35.2523078918457, 200.5149383544922, 326.27685546875, -32.92597961425781, 199.51669311523438, 169.72154235839844, 44.80314636230469, -18.146759033203125, -18.548011779785156, 195.7926025390625, 61.22685241699219, 156.11712646484375, -7.119873046875, 133.44595336914062, -99.1190185546875, 163.72109985351562, 25.67580223083496, 198.90908813476562, 322.41693115234375, -1.0390167236328125, 191.5938720703125, 176.5231475830078, 155.62535095214844, 37.00110626220703, 85.5256576538086, 0.003986358642578125, 112.24493408203125, 22.220855712890625, 190.68182373046875, 3.7517318725585938, -3.856630325317383, 27.761428833007812, 195.26463317871094, 63.04237365722656, 109.16224670410156, -50.199737548828125, -61.935020446777344, 182.5356903076172, 75.28877258300781, 113.16802978515625, 153.35125732421875, 30.897817611694336, 0.6696395874023438, -22.497913360595703, 84.398681640625, -24.759597778320312, 593.1195678710938, 46.462196350097656, 2.4600448608398438, 121.72477722167969, -75.49271392822266, 334.7770080566406, -51.893882751464844, 156.64398193359375, 19.378931045532227, -71.09483337402344, 163.19996643066406, 189.24520874023438, -2.4041881561279297], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000398.npy"}
|
||||
{"epoch": 0.6016628873771731, "step": 399, "batch_size": 64, "mean": 108.39849090576172, "std": 142.4736328125, "min": -135.44635009765625, "p10": -41.035072708129874, "median": 88.16083908081055, "p90": 246.90373382568362, "max": 665.8743896484375, "pos_frac": 0.734375, "sample": [225.3638153076172, 76.8520736694336, 177.63037109375, 407.80499267578125, -0.4823760986328125, 261.7486572265625, 220.81979370117188, 53.188560485839844, -8.112842559814453, 2.1816177368164062, -115.38824462890625, 55.3798828125, 187.864990234375, -33.9180793762207, 98.41673278808594, 77.02128601074219, 97.13310241699219, 239.10678100585938, 118.0724105834961, 350.7577819824219, -101.42813110351562, -87.46414184570312, -135.44635009765625, 148.56106567382812, 423.8281555175781, -9.906997680664062, 6.257389068603516, 665.8743896484375, 34.08495330810547, -10.423118591308594, 39.05542755126953, -1.818002700805664, 131.42356872558594, 57.04642105102539, -3.2140960693359375, -30.99970245361328, 94.53490447998047, 227.43983459472656, 93.70446014404297, 250.2452850341797, 45.404945373535156, 5.816072463989258, 233.4634246826172, -97.70343017578125, 233.99929809570312, 162.72747802734375, 72.38275909423828, 82.61721801757812, -62.84002685546875, -44.08521270751953, 189.10391235351562, 210.4921112060547, 184.47299194335938, 203.63522338867188, 168.9895477294922, 275.6411437988281, 137.5081787109375, 217.86849975585938, 62.43754577636719, -14.666828155517578, 216.6461181640625, -23.258529663085938, 143.26571655273438, 50.788597106933594], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000399.npy"}
|
||||
{"epoch": 0.6031746031746031, "step": 400, "batch_size": 64, "mean": 63.28759002685547, "std": 109.66353607177734, "min": -275.57647705078125, "p10": -53.39511566162109, "median": 68.83644104003906, "p90": 190.28423156738282, "max": 270.08721923828125, "pos_frac": 0.71875, "sample": [-45.15762710571289, 75.66007232666016, 85.36259460449219, -25.819839477539062, 190.9241485595703, 139.8291015625, 68.0294418334961, 166.9152069091797, 70.47793579101562, 172.33326721191406, 40.38149642944336, -53.64546203613281, 90.28892517089844, 3.992450714111328, 1.9948158264160156, -78.0570068359375, 142.21527099609375, 188.7910919189453, 173.532470703125, 156.6651611328125, 70.74826049804688, 80.12767028808594, 157.8013916015625, 29.00310707092285, -275.57647705078125, 108.49746704101562, 13.077306747436523, 22.598739624023438, 16.28274917602539, -10.190629959106445, -113.29473876953125, -10.578634262084961, 180.65953063964844, -35.209716796875, 192.37940979003906, -5.238941192626953, -23.333969116210938, 60.33159255981445, 186.54556274414062, 140.3323211669922, 250.61697387695312, 148.900390625, -181.20950317382812, 184.43350219726562, 45.71898651123047, 41.93873596191406, 3.5628662109375, -52.81097412109375, -22.553199768066406, 270.08721923828125, 104.71688842773438, 231.74844360351562, -76.08326721191406, 85.46761322021484, -9.659543991088867, 212.4014434814453, 238.32440185546875, 61.626556396484375, 30.82756805419922, 114.93963623046875, -181.64163208007812, -17.088909149169922, 69.64344024658203, 146.82244873046875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000400.npy"}
|
||||
{"epoch": 0.6046863189720333, "step": 401, "batch_size": 64, "mean": 87.16850280761719, "std": 117.32716369628906, "min": -129.51358032226562, "p10": -32.51890411376952, "median": 58.652156829833984, "p90": 241.7427642822267, "max": 434.4801025390625, "pos_frac": 0.75, "sample": [-26.197349548339844, 163.25025939941406, 210.35476684570312, 75.892578125, -20.880229949951172, 3.260883331298828, 37.832122802734375, 49.966285705566406, -41.74570083618164, 8.441139221191406, 256.4114074707031, -11.511093139648438, -96.0933837890625, -22.69318389892578, 0.28766632080078125, -35.22814178466797, 178.07574462890625, 434.4801025390625, 187.44529724121094, 203.05880737304688, 82.53417205810547, 19.85196304321289, 66.25868225097656, 43.66596603393555, -96.71136474609375, -101.6438217163086, 171.23046875, 123.93414306640625, 116.39949798583984, 79.1558609008789, 185.44357299804688, 51.045631408691406, -2.7890148162841797, 255.19476318359375, -6.443351745605469, 204.33139038085938, 8.493444442749023, 191.72122192382812, 27.104280471801758, 289.23095703125, 9.923377990722656, 192.15562438964844, 165.52957153320312, -3.240938186645508, 285.2755432128906, 8.238780975341797, -129.51358032226562, 281.9599914550781, 210.15798950195312, 144.49681091308594, 174.509765625, -3.892671585083008, 8.825321197509766, 23.582061767578125, -70.82131958007812, 109.88591766357422, 102.83566284179688, 27.02836799621582, 181.4521026611328, 202.78421020507812, 67.3193359375, -5.962604522705078, 14.96524429321289, 318.8736572265625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000401.npy"}
|
||||
{"epoch": 0.6061980347694633, "step": 402, "batch_size": 64, "mean": 64.57801818847656, "std": 122.94330596923828, "min": -210.88319396972656, "p10": -83.05018310546875, "median": 52.275726318359375, "p90": 233.26793670654303, "max": 330.453369140625, "pos_frac": 0.734375, "sample": [109.79954528808594, 237.93765258789062, 158.99212646484375, 180.38116455078125, 167.80345153808594, 93.42890167236328, 283.124267578125, 71.45797729492188, 51.12226867675781, 110.0195083618164, -16.173797607421875, 255.32232666015625, 44.056739807128906, -43.294219970703125, 97.7720947265625, 76.22868347167969, 10.963836669921875, 3.7032108306884766, -2.4515228271484375, 53.42918395996094, 127.84687042236328, 0.8529167175292969, 6.303153991699219, -138.6858367919922, 330.453369140625, 11.182193756103516, 141.69052124023438, 299.7809753417969, 103.27015686035156, 24.68195343017578, 222.37193298339844, 185.73184204101562, -13.928359985351562, 131.28128051757812, -15.218154907226562, 58.19213104248047, 188.04751586914062, 167.9824676513672, 12.445877075195312, 103.35684204101562, -49.45856475830078, -138.6919708251953, -208.95367431640625, 149.06289672851562, 73.8543701171875, -0.6560077667236328, 47.18171691894531, -80.41168212890625, 27.683326721191406, -12.683055877685547, 169.6070098876953, 245.18161010742188, 160.85568237304688, 289.53863525390625, -129.28428649902344, 1.6091556549072266, -51.35472869873047, -210.88319396972656, -200.14686584472656, 37.14764404296875, 45.7436637878418, 147.8642578125, -84.18096923828125, 13.104728698730469], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000402.npy"}
|
||||
{"epoch": 0.6077097505668935, "step": 403, "batch_size": 64, "mean": 85.62864685058594, "std": 121.94348907470703, "min": -166.2835693359375, "p10": -41.26740531921386, "median": 68.83489227294922, "p90": 219.98156890869143, "max": 497.407958984375, "pos_frac": 0.796875, "sample": [-11.601043701171875, 9.72833251953125, 352.1790466308594, 47.805206298828125, 185.75503540039062, 64.83985900878906, 266.20330810546875, 173.9132843017578, 26.977828979492188, -29.239089965820312, 94.5997085571289, -105.40365600585938, 172.75767517089844, -46.42239761352539, 162.7834014892578, 88.65426635742188, 73.67762756347656, 18.084543228149414, 72.82992553710938, 10.968557357788086, 9.60380744934082, -28.745952606201172, 497.407958984375, 209.83920288085938, -24.856170654296875, 137.57730102539062, 228.24136352539062, 9.818605422973633, 217.3064422607422, 12.612648010253906, -4.008434295654297, 79.18170166015625, 10.964950561523438, 24.245445251464844, 20.19524574279785, -17.30068588256836, 63.81798553466797, 196.5871124267578, 5.384233474731445, 216.69320678710938, 33.9852294921875, 0.8876895904541016, 186.44842529296875, 4.279155731201172, 121.40022277832031, -83.008056640625, 150.71957397460938, 187.48455810546875, 121.21501159667969, 107.10662841796875, 283.9338684082031, 207.0658416748047, -166.2835693359375, -67.48725891113281, 304.1295166015625, -153.19041442871094, 221.1280517578125, 50.60741424560547, 93.30905151367188, 183.03994750976562, -72.12002563476562, 23.08509063720703, 86.31302642822266, 162.52748107910156], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000403.npy"}
|
||||
{"epoch": 0.6092214663643235, "step": 404, "batch_size": 64, "mean": 90.55796813964844, "std": 136.4097442626953, "min": -153.57180786132812, "p10": -78.21093902587889, "median": 69.08023452758789, "p90": 241.16399841308595, "max": 415.95477294921875, "pos_frac": 0.71875, "sample": [-127.79611206054688, 132.94989013671875, 144.43856811523438, -12.359817504882812, 80.09188842773438, 17.79735565185547, 146.91891479492188, 19.670082092285156, -57.04479217529297, -83.93281555175781, 241.53915405273438, -148.03176879882812, -153.57180786132812, -130.00733947753906, -1.7310600280761719, 347.76617431640625, -37.11445617675781, 2.3053016662597656, 204.62628173828125, 286.8971252441406, 146.01332092285156, 141.2376708984375, 168.51348876953125, 240.28863525390625, 26.37030029296875, 162.6673583984375, -127.44959259033203, 395.6204833984375, 218.37583923339844, 190.6240997314453, 20.430709838867188, 132.80807495117188, -5.463010787963867, 173.10690307617188, -44.70600128173828, -6.432456970214844, 57.98859405517578, 200.00180053710938, 193.57528686523438, -43.495662689208984, 57.487953186035156, -54.179840087890625, 165.85755920410156, 90.86708068847656, 180.54190063476562, 5.584552764892578, 188.05111694335938, 202.97442626953125, 216.6608428955078, 58.068580627441406, -0.11738967895507812, 407.85382080078125, 415.95477294921875, 124.64299011230469, 185.9759063720703, 179.28909301757812, 248.56655883789062, -64.85989379882812, -112.75770568847656, 7.4300689697265625, 55.17584228515625, 40.56875991821289, 29.092849731445312, 53.49378967285156], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000404.npy"}
|
||||
{"epoch": 0.6107331821617535, "step": 405, "batch_size": 64, "mean": 66.47544860839844, "std": 108.53825378417969, "min": -182.52210998535156, "p10": -29.950511360168456, "median": 21.85187339782715, "p90": 218.19895935058597, "max": 301.4509582519531, "pos_frac": 0.703125, "sample": [246.38119506835938, -19.629390716552734, -94.87596130371094, 301.4509582519531, 37.07018280029297, 21.82745361328125, 205.13706970214844, 51.510345458984375, -11.820667266845703, -52.13872528076172, -8.473526000976562, 6.402975082397461, 11.74388313293457, -0.3346977233886719, 59.63407897949219, -18.95940399169922, 283.8866882324219, -27.774625778198242, -8.751029968261719, 44.950042724609375, 172.69190979003906, 212.73587036132812, 1.7203216552734375, 137.79527282714844, -0.3836250305175781, 8.862064361572266, 190.84994506835938, 198.88722229003906, 10.148887634277344, 145.42971801757812, 20.35665512084961, -11.675098419189453, 220.540283203125, 242.27777099609375, -25.629913330078125, 123.2718276977539, 21.876293182373047, -182.52210998535156, 61.222686767578125, 108.9508056640625, 3.5000457763671875, -44.215057373046875, 91.33606719970703, 42.56285095214844, 11.696876525878906, 15.64845085144043, 6.291404724121094, 7.320350646972656, 260.005615234375, 171.82240295410156, -26.102031707763672, -74.34329223632812, 136.64385986328125, 22.608840942382812, 195.8400421142578, 204.91104125976562, 18.493804931640625, 28.58321762084961, 153.84817504882812, -88.43630981445312, -22.60578155517578, 301.3608093261719, 183.8965606689453, -30.883033752441406], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000405.npy"}
|
||||
{"epoch": 0.6122448979591837, "step": 406, "batch_size": 64, "mean": 98.25647735595703, "std": 125.2013168334961, "min": -175.98594665527344, "p10": -31.39689254760742, "median": 85.69726943969727, "p90": 251.0331756591797, "max": 417.9566650390625, "pos_frac": 0.75, "sample": [207.01031494140625, 64.36454772949219, -23.211233139038086, -17.591064453125, 218.3614044189453, 142.4193115234375, 183.55369567871094, 248.73388671875, -0.8926067352294922, 298.1776123046875, 159.21878051757812, 222.6773681640625, -8.813030242919922, -50.50941467285156, 261.20098876953125, 417.9566650390625, 301.0804138183594, -1.0581035614013672, 52.947784423828125, 210.32699584960938, 96.22933959960938, 33.139549255371094, 79.20610046386719, 86.17328643798828, 43.05577850341797, 145.10044860839844, -28.123226165771484, 163.4410858154297, -8.399042129516602, 71.94508361816406, -1.9780311584472656, 5.128623962402344, -175.98594665527344, -10.281558990478516, 53.52154541015625, 223.88751220703125, 147.40768432617188, 200.96714782714844, -149.14407348632812, 252.01858520507812, 135.64425659179688, 55.30509948730469, 63.37696075439453, 270.633544921875, 33.06126403808594, 170.37298583984375, -32.79989242553711, 7.111368179321289, 174.17526245117188, 153.266845703125, 37.949737548828125, -147.32997131347656, 247.43946838378906, -103.93955993652344, 3.857391357421875, -165.4079132080078, 148.533447265625, 214.24176025390625, 31.11065673828125, 85.22125244140625, 255.6582489013672, 245.9591064453125, 149.3817138671875, 142.32701110839844], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000406.npy"}
|
||||
{"epoch": 0.6137566137566137, "step": 407, "batch_size": 64, "mean": 74.74103546142578, "std": 136.78553771972656, "min": -318.66888427734375, "p10": -83.76420135498046, "median": 68.65702819824219, "p90": 244.2045608520508, "max": 384.7788391113281, "pos_frac": 0.703125, "sample": [-2.2723007202148438, -9.652219772338867, 72.57772827148438, 79.6980972290039, -24.88556671142578, 164.57069396972656, 261.43896484375, -25.39227294921875, -155.403564453125, 147.35818481445312, 135.38722229003906, 60.48917770385742, 5.285858154296875, 33.067474365234375, -156.86334228515625, 22.607337951660156, 117.63671112060547, 190.17919921875, 246.222900390625, 185.85853576660156, 170.75704956054688, 29.685144424438477, 93.93815612792969, 64.736328125, 196.13906860351562, 78.43743896484375, -5.678306579589844, 23.764434814453125, 171.73736572265625, 256.4623718261719, 139.70477294921875, -0.24741363525390625, 227.30941772460938, 214.67437744140625, 268.7000427246094, 17.785247802734375, -197.79904174804688, 263.9457702636719, -41.41078186035156, 27.72315216064453, -76.23558044433594, 137.11029052734375, 234.5344696044922, 97.56893157958984, 239.49510192871094, 368.53619384765625, 169.45046997070312, 35.21971893310547, 384.7788391113281, 0.20981216430664062, 42.13233947753906, -318.66888427734375, -225.05886840820312, -86.99075317382812, -98.37324523925781, 15.318721771240234, -0.3139972686767578, 189.1818389892578, -25.51736068725586, -16.582427978515625, -2.0935134887695312, 99.3499755859375, 75.6342544555664, 196.4664306640625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000407.npy"}
|
||||
{"epoch": 0.6152683295540439, "step": 408, "batch_size": 64, "mean": 103.10829162597656, "std": 110.06849670410156, "min": -142.83438110351562, "p10": -10.96079921722412, "median": 100.38168716430664, "p90": 221.7721374511719, "max": 382.1929016113281, "pos_frac": 0.828125, "sample": [97.10218048095703, 195.37474060058594, 109.20118713378906, 121.10171508789062, 5.5775909423828125, 8.704355239868164, -142.83438110351562, 10.526081085205078, 180.70411682128906, 96.82715606689453, 69.9798355102539, 272.06671142578125, 2.2220325469970703, 207.01377868652344, 178.7792205810547, 382.1929016113281, 36.421688079833984, 29.05504608154297, 45.347999572753906, 60.429595947265625, 165.8326873779297, 10.563621520996094, -8.835176467895508, 177.94056701660156, 298.0684509277344, -133.2400360107422, 101.6808090209961, 265.33282470703125, 141.28790283203125, 174.4363250732422, 200.44871520996094, -43.73704147338867, 148.22923278808594, 18.316679000854492, 40.201934814453125, 195.0032196044922, 55.72073745727539, -12.970151901245117, 310.699462890625, 189.2989044189453, 39.86863708496094, 195.72398376464844, 224.32794189453125, 215.80859375, 119.11253356933594, 21.119802474975586, 181.74429321289062, -4.438478469848633, 40.393890380859375, -11.871780395507812, 181.5726318359375, 99.08256530761719, 285.94256591796875, -133.17787170410156, 194.92901611328125, 154.5772705078125, 43.91936111450195, -2.955883026123047, -4.7186279296875, 89.11677551269531, 212.330322265625, 123.88288879394531, -38.15238952636719, 110.7197036743164], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000408.npy"}
|
||||
{"epoch": 0.6167800453514739, "step": 409, "batch_size": 64, "mean": 46.11394500732422, "std": 98.75579833984375, "min": -212.62200927734375, "p10": -51.927127838134766, "median": 26.239072799682617, "p90": 192.93738708496093, "max": 242.45941162109375, "pos_frac": 0.609375, "sample": [40.30024719238281, 2.4779300689697266, 29.944534301757812, -39.49009704589844, -7.244174957275391, -21.277591705322266, -3.276090621948242, 27.52817153930664, 0.5692787170410156, -50.813270568847656, -43.88271713256836, 192.4556884765625, -16.182313919067383, 199.74200439453125, 91.97926330566406, -27.1905517578125, -23.47259521484375, -1.530975341796875, 19.42813491821289, 228.32125854492188, 161.26864624023438, -46.563358306884766, 145.92984008789062, 229.55081176757812, 44.09999084472656, -105.67626953125, -52.40907287597656, 174.3483428955078, 28.207242965698242, -23.096771240234375, -93.47309875488281, 95.26603698730469, 137.21185302734375, -15.337459564208984, 193.14382934570312, -0.7149791717529297, 157.87313842773438, 145.13101196289062, 17.275405883789062, 20.87879753112793, -60.371768951416016, 55.05609893798828, 116.39447021484375, -29.31106185913086, 2.3645401000976562, -52.40449523925781, 34.67103576660156, 87.23179626464844, -212.62200927734375, 206.23593139648438, -0.7095298767089844, 151.1895751953125, -2.5443115234375, 217.49435424804688, -176.0388946533203, 24.949974060058594, 111.55841064453125, 58.034523010253906, 66.0673599243164, 60.62211608886719, 242.45941162109375, -23.09840202331543, 145.97802734375, 116.78520202636719], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000409.npy"}
|
||||
{"epoch": 0.618291761148904, "step": 410, "batch_size": 64, "mean": 82.12518310546875, "std": 112.5008773803711, "min": -142.98046875, "p10": -52.77948989868162, "median": 74.90809631347656, "p90": 218.12803497314454, "max": 453.080078125, "pos_frac": 0.796875, "sample": [115.26761627197266, 97.10008239746094, 53.97544479370117, -11.68105697631836, 186.33737182617188, 148.54129028320312, 109.68263244628906, 24.278507232666016, -116.29083251953125, -140.3292236328125, 119.43248748779297, 218.4508819580078, 9.531105041503906, -127.4345474243164, 171.16439819335938, 92.46095275878906, 199.83346557617188, -142.98046875, 132.3369140625, 201.04605102539062, 453.080078125, 50.63618469238281, 63.46575927734375, 135.03302001953125, 29.40704345703125, 165.80908203125, 18.770370483398438, 77.37420654296875, 43.624366760253906, -29.25011444091797, 162.57315063476562, -61.93116760253906, -8.759929656982422, 156.29840087890625, 36.23930358886719, 240.84774780273438, 272.7855224609375, 87.12965393066406, 50.23565673828125, 75.9996337890625, -31.425575256347656, -86.29898071289062, 164.19171142578125, 115.89274597167969, 217.37472534179688, 55.65931701660156, 180.75071716308594, -5.147216796875, 22.385730743408203, 228.799560546875, 73.81655883789062, 46.834320068359375, 297.3745422363281, 88.99525451660156, 90.46372985839844, 30.261451721191406, 13.29385757446289, 72.95556640625, 8.831588745117188, 159.31817626953125, 245.3819580078125, 41.95525360107422, -126.55206298828125, -9.192283630371094], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000410.npy"}
|
||||
{"epoch": 0.6198034769463341, "step": 411, "batch_size": 64, "mean": 83.36807250976562, "std": 117.00282287597656, "min": -247.80609130859375, "p10": -50.44164199829101, "median": 77.15477752685547, "p90": 225.35853424072272, "max": 470.2189636230469, "pos_frac": 0.734375, "sample": [-31.883522033691406, 157.06271362304688, 158.896728515625, 198.8409423828125, 117.2601089477539, -35.90673828125, 129.68775939941406, -247.80609130859375, 207.33251953125, 23.743453979492188, 112.57295989990234, 240.5728759765625, 115.66812896728516, 144.66812133789062, 94.28619384765625, 30.372642517089844, 118.06283569335938, -80.32779693603516, 135.38479614257812, 63.24296569824219, -56.76616668701172, 233.08396911621094, 250.52090454101562, 34.80389404296875, -99.39607238769531, 196.67738342285156, 8.591339111328125, 10.004886627197266, -71.56924438476562, -29.927322387695312, 73.83560180664062, -31.981470108032227, 31.715980529785156, 111.732421875, 148.45852661132812, 58.695125579833984, -7.633188247680664, 172.06246948242188, 52.77960205078125, 470.2189636230469, 274.33929443359375, 95.48802947998047, -16.59100341796875, 200.47579956054688, 253.57687377929688, 193.75796508789062, -49.3988037109375, 80.00163269042969, -18.297107696533203, -90.87734985351562, 31.368616104125977, -17.776077270507812, 187.82601928710938, 158.6394500732422, 46.142784118652344, 98.9975357055664, 146.74151611328125, 56.42053985595703, 289.92181396484375, 74.30792236328125, -50.888572692871094, -0.7852783203125, 24.106002807617188, 160.4196319580078], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000411.npy"}
|
||||
{"epoch": 0.6213151927437641, "step": 412, "batch_size": 64, "mean": 72.9715576171875, "std": 99.94523620605469, "min": -150.00946044921875, "p10": -55.95219497680664, "median": 68.01679992675781, "p90": 201.88857116699222, "max": 299.18316650390625, "pos_frac": 0.71875, "sample": [-60.90247344970703, 162.79049682617188, 114.0212173461914, 205.87835693359375, 15.305244445800781, 161.84127807617188, 79.88777160644531, 188.1967315673828, -124.86238098144531, -5.285320281982422, 33.805381774902344, 4.586265563964844, 97.11708068847656, 179.16737365722656, 19.759666442871094, -58.573585510253906, 177.68508911132812, -150.00946044921875, 56.13098907470703, -3.019195556640625, 223.31556701660156, 210.8126678466797, 35.61505126953125, 77.23735046386719, 182.97482299804688, -70.31810760498047, -49.83561706542969, 65.17631530761719, 26.72690200805664, 83.43534851074219, -9.90546989440918, 299.18316650390625, 8.019338607788086, 255.5397186279297, -78.67111206054688, 56.56559753417969, 146.76910400390625, -9.574520111083984, 185.84274291992188, -10.839134216308594, -5.707941055297852, 1.9568939208984375, 186.42852783203125, -41.35293197631836, 84.73841857910156, 244.06964111328125, 101.44662475585938, 156.68353271484375, 209.9788360595703, -67.45040893554688, 116.55472564697266, 124.93037414550781, 70.85728454589844, 38.465126037597656, 38.02366256713867, 159.0925750732422, -18.12932777404785, 118.36503601074219, -43.67110061645508, 81.89889526367188, 192.57907104492188, -1.1283130645751953, 136.38052368164062, 63.5794677734375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000412.npy"}
|
||||
{"epoch": 0.6228269085411943, "step": 413, "batch_size": 64, "mean": 70.00959777832031, "std": 140.19088745117188, "min": -241.40133666992188, "p10": -108.86865463256834, "median": 64.05964279174805, "p90": 215.17395935058596, "max": 513.0689086914062, "pos_frac": 0.703125, "sample": [17.91135025024414, -65.49626159667969, 126.10022735595703, 130.7554473876953, -20.654014587402344, -24.442642211914062, -5.1122283935546875, 206.17916870117188, -36.34129333496094, 366.7577819824219, -241.40133666992188, 191.28109741210938, 151.9676513671875, -54.04681396484375, -26.012100219726562, 36.011383056640625, -144.0157928466797, 172.0372772216797, -2.6120758056640625, 10.945138931274414, 248.41482543945312, 116.43113708496094, 79.27131652832031, 41.58328628540039, 161.19723510742188, 48.84796905517578, 513.0689086914062, 82.52685546875, 30.941978454589844, 16.900672912597656, 129.0059051513672, 199.4598846435547, 116.97796630859375, 29.420257568359375, 201.4944305419922, -188.98178100585938, 188.88438415527344, 29.696731567382812, 116.88361358642578, 98.62960815429688, -123.5386734008789, 219.02886962890625, 193.015625, -120.9023666381836, 79.29119873046875, -22.449419021606445, -236.7865447998047, 143.5780029296875, 133.7828369140625, -3.0728721618652344, -75.98330688476562, -158.90451049804688, 368.71661376953125, 4.225704193115234, 84.06405639648438, 145.73011779785156, 7.220760345458984, 230.95639038085938, 25.023624420166016, 40.55730438232422, -80.78999328613281, 163.81695556640625, 255.1439208984375, 158.42282104492188], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000413.npy"}
|
||||
{"epoch": 0.6243386243386243, "step": 414, "batch_size": 64, "mean": 82.96546936035156, "std": 117.24591827392578, "min": -268.7430419921875, "p10": -39.90658111572265, "median": 75.0953369140625, "p90": 222.91265411376955, "max": 452.98638916015625, "pos_frac": 0.75, "sample": [182.12991333007812, 16.195083618164062, 143.5990447998047, 226.885986328125, 91.45816040039062, 67.26924133300781, -268.7430419921875, 105.46255493164062, 96.92826080322266, 53.87355041503906, 135.19869995117188, 168.1418914794922, 168.97735595703125, -29.849769592285156, 264.8597717285156, 56.38930892944336, -149.03555297851562, -84.46784210205078, 66.62031555175781, 167.2911376953125, -35.832252502441406, 107.22322082519531, 225.88656616210938, -49.64234161376953, 215.97352600097656, 179.416015625, 59.37727737426758, -10.790061950683594, 129.7964324951172, 158.54574584960938, 197.63571166992188, 73.73237609863281, 165.0015411376953, -17.254653930664062, 256.8255615234375, 161.25271606445312, 163.111572265625, 29.4266357421875, 41.715423583984375, 167.7581787109375, 10.399429321289062, -40.51426696777344, 4.8961029052734375, -4.507375717163086, 133.15982055664062, -76.34339904785156, 14.357383728027344, 76.45829772949219, -150.51629638671875, 85.4784927368164, 452.98638916015625, -16.790740966796875, -4.2674560546875, 55.1245231628418, 193.7233123779297, 234.00204467773438, 61.59546661376953, 46.03924560546875, 169.9067840576172, 156.2080078125, -38.4886474609375, -6.927701950073242, 10.359874725341797, 245.10714721679688], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000414.npy"}
|
||||
{"epoch": 0.6258503401360545, "step": 415, "batch_size": 64, "mean": 63.36273956298828, "std": 119.46244812011719, "min": -205.86871337890625, "p10": -62.76224060058591, "median": 45.91215896606445, "p90": 198.16784973144533, "max": 475.36199951171875, "pos_frac": 0.703125, "sample": [-13.359834671020508, 133.107666015625, 2.988962173461914, 32.658504486083984, -11.361583709716797, 25.492942810058594, 141.17276000976562, 211.34646606445312, -71.79037475585938, 131.00494384765625, 92.84322357177734, -22.81561279296875, 189.52944946289062, 199.13233947753906, -26.52204132080078, 195.91737365722656, 236.7813720703125, 121.17379760742188, 118.30381774902344, -2.949106216430664, -13.959075927734375, 77.63334655761719, 150.84188842773438, 0.39530181884765625, 67.99198913574219, 178.967529296875, -161.41732788085938, -7.543159484863281, 45.96208190917969, 274.3697204589844, -90.9326171875, 191.6452178955078, 62.33119201660156, 18.745498657226562, -205.86871337890625, 475.36199951171875, 93.8628158569336, 118.32425689697266, 176.45333862304688, -41.69659423828125, 162.379150390625, -200.02796936035156, -86.53767395019531, -195.13250732421875, 49.37983703613281, 47.55390167236328, -18.70879364013672, 84.85433959960938, -1.3396835327148438, 11.953187942504883, 3.7913036346435547, 35.243873596191406, 37.44471740722656, 28.765804290771484, 19.583229064941406, 155.56564331054688, -3.645170211791992, 185.7137451171875, 174.59939575195312, 0.01235198974609375, -7.157218933105469, 45.86223602294922, 203.4976348876953, 227.44015502929688], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000415.npy"}
|
||||
{"epoch": 0.6273620559334845, "step": 416, "batch_size": 64, "mean": 62.57478332519531, "std": 123.70858764648438, "min": -193.6041259765625, "p10": -55.44602737426758, "median": 41.43099594116211, "p90": 197.74830322265626, "max": 507.5016174316406, "pos_frac": 0.625, "sample": [-2.3890380859375, -160.149169921875, -13.292987823486328, 145.8159942626953, 15.066322326660156, 188.02496337890625, -86.70198059082031, 14.587326049804688, 197.01425170898438, 224.75814819335938, 72.6585693359375, 285.9708251953125, 507.5016174316406, 15.6778564453125, -17.851581573486328, 93.74799346923828, -1.7086849212646484, 149.59344482421875, -193.6041259765625, 22.613800048828125, -15.494417190551758, 171.402587890625, 38.092681884765625, 170.90335083007812, -0.7159614562988281, 187.7139434814453, 180.61038208007812, 198.06289672851562, 267.284912109375, 44.769309997558594, 121.97734832763672, 181.9486083984375, -56.0057373046875, -54.140037536621094, 2.5960865020751953, -47.590721130371094, -89.666748046875, -133.84666442871094, -4.027864456176758, 1.7807197570800781, -0.8402290344238281, 50.10881805419922, -35.2119140625, -18.733352661132812, 109.00955200195312, -47.971797943115234, 193.02569580078125, -191.05743408203125, -17.118066787719727, 176.98297119140625, 30.756927490234375, -0.27176475524902344, -41.57035827636719, 160.02137756347656, -25.09722137451172, 46.06730270385742, 51.32202911376953, 83.20567321777344, 81.7350082397461, 66.18560028076172, 231.77706909179688, 158.18731689453125, 220.5816650390625, 100.7029037475586], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000416.npy"}
|
||||
{"epoch": 0.6288737717309146, "step": 417, "batch_size": 64, "mean": 74.52841186523438, "std": 138.6715850830078, "min": -498.3720703125, "p10": -59.24393386840819, "median": 74.5403060913086, "p90": 211.45907135009767, "max": 445.5635986328125, "pos_frac": 0.765625, "sample": [33.577613830566406, 11.349922180175781, -48.86492156982422, -216.9710693359375, 209.90335083007812, 132.18130493164062, -498.3720703125, 198.0755615234375, 104.07673645019531, 83.63211059570312, 445.5635986328125, 88.93620300292969, -6.713706970214844, 69.09688568115234, 99.96125030517578, -90.03657531738281, 193.82701110839844, 209.8853759765625, -17.070877075195312, 85.29487609863281, 69.74325561523438, 169.80642700195312, 134.86599731445312, 58.63896179199219, 91.30289459228516, -84.28520202636719, 212.1258087158203, 66.23890686035156, 27.492168426513672, 236.8685302734375, -63.302574157714844, 227.38720703125, 197.0662384033203, -173.07762145996094, 266.2042236328125, -49.75975036621094, 57.79681396484375, -34.968238830566406, 230.49392700195312, 123.1955795288086, 191.274658203125, 8.232551574707031, 201.17694091796875, 68.85709381103516, -157.17311096191406, 79.33735656738281, 7.918100357055664, 64.3852767944336, 0.680908203125, 128.64324951171875, 58.02203369140625, 182.23513793945312, 20.067581176757812, -49.773773193359375, 115.84996795654297, -18.7099609375, 98.71379089355469, 15.942432403564453, 131.80970764160156, -0.3029041290283203, 173.10340881347656, 68.71424865722656, 183.99087524414062, 345.65716552734375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000417.npy"}
|
||||
{"epoch": 0.6303854875283447, "step": 418, "batch_size": 64, "mean": 63.16754913330078, "std": 140.6312255859375, "min": -335.0693359375, "p10": -97.9407585144043, "median": 49.83390235900879, "p90": 229.19599456787114, "max": 513.4403076171875, "pos_frac": 0.640625, "sample": [131.43544006347656, 163.55450439453125, 139.994384765625, 164.61459350585938, 86.90617370605469, 232.676513671875, 27.335250854492188, 129.34378051757812, 221.0747833251953, -80.77210235595703, 182.08889770507812, 219.70079040527344, -3.2517547607421875, 256.5881042480469, 78.25069427490234, -128.7363739013672, 208.5006561279297, -17.42121124267578, 141.14141845703125, -14.107002258300781, -0.9377975463867188, 513.4403076171875, 113.1110610961914, -51.13256072998047, 239.67990112304688, 40.85889434814453, -119.8824462890625, 12.004676818847656, 58.80891036987305, -88.93113708496094, -335.0693359375, -263.90826416015625, 220.3026885986328, 100.58056640625, 195.79283142089844, -24.518173217773438, 18.054515838623047, -5.2258758544921875, 9.235496520996094, -22.785423278808594, -101.8020248413086, -44.175331115722656, -105.53318786621094, -6.655092239379883, 142.92544555664062, -32.34270477294922, -73.90312957763672, 13.705947875976562, 29.39472007751465, -29.953697204589844, 2.6404342651367188, 319.63427734375, 271.9521789550781, 265.66217041015625, 15.896446228027344, 98.78518676757812, -124.45381164550781, 82.32209777832031, 107.81971740722656, 133.46109008789062, 153.58734130859375, -53.38508605957031, 159.41412353515625, 69.32972717285156], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000418.npy"}
|
||||
{"epoch": 0.6318972033257747, "step": 419, "batch_size": 64, "mean": 94.15660095214844, "std": 131.4395294189453, "min": -173.31275939941406, "p10": -43.72005577087401, "median": 60.919029235839844, "p90": 245.20095672607422, "max": 475.6505126953125, "pos_frac": 0.796875, "sample": [-5.967800140380859, 402.05438232421875, 7.844490051269531, 50.778541564941406, -114.77578735351562, 6.713348388671875, 211.08551025390625, 30.977828979492188, -21.940696716308594, 32.91263198852539, -173.31275939941406, 177.65679931640625, 364.46966552734375, 140.33851623535156, -75.412353515625, 125.58561706542969, 123.12646484375, 87.98200988769531, 246.67919921875, 45.40873718261719, 60.074920654296875, 112.77942657470703, -48.91716003417969, 94.1002197265625, 15.2401123046875, 61.76313781738281, 323.82763671875, -64.25955200195312, 475.6505126953125, 72.32962036132812, 51.408573150634766, -133.6751708984375, -31.59347915649414, 210.48385620117188, 394.09564208984375, 140.7602081298828, 155.63043212890625, 6.2544708251953125, -7.9834747314453125, 113.04611206054688, 4.087697982788086, 209.941162109375, 44.659584045410156, 23.533554077148438, 185.97059631347656, -10.483261108398438, 110.41503143310547, 51.11136245727539, 3.559894561767578, 221.83734130859375, 233.12091064453125, 2.8658485412597656, 54.2656364440918, 76.34805297851562, 162.652587890625, 45.861427307128906, 213.42095947265625, 134.856201171875, 303.50262451171875, -8.463348388671875, 120.36131286621094, 36.152130126953125, -98.52716827392578, 241.75172424316406], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000419.npy"}
|
||||
{"epoch": 0.6334089191232048, "step": 420, "batch_size": 64, "mean": 45.62171936035156, "std": 137.15640258789062, "min": -336.492431640625, "p10": -116.71776123046874, "median": 11.993670463562012, "p90": 225.95945892333987, "max": 448.6060791015625, "pos_frac": 0.578125, "sample": [-11.555938720703125, -31.433414459228516, 157.94857788085938, 33.87358856201172, 163.98239135742188, -164.80824279785156, 7.8878326416015625, 6.521724700927734, -3.9178390502929688, -186.58802795410156, -45.092681884765625, 82.32752227783203, -13.842910766601562, 153.23983764648438, -8.828155517578125, 210.25650024414062, 313.4739074707031, 112.65483093261719, 219.00653076171875, 244.50457763671875, 16.09950828552246, -93.02899169921875, 4.98375129699707, 52.89733123779297, 111.46971130371094, 190.81277465820312, -174.7611541748047, -120.09844207763672, 137.65545654296875, 3.8355064392089844, -132.71737670898438, 40.133087158203125, -108.82950592041016, 73.55088806152344, -8.911548614501953, -122.98265075683594, -83.78580474853516, 6.735557556152344, 318.424560546875, 32.29821014404297, -35.83899688720703, -336.492431640625, -50.282737731933594, -33.78950500488281, 199.9278106689453, 86.51097869873047, -16.99081039428711, 228.9392852783203, -15.010086059570312, 160.93377685546875, 248.31686401367188, 41.53252410888672, 128.26998901367188, -0.07950019836425781, 33.44907760620117, 158.1409149169922, 448.6060791015625, 95.2030258178711, -26.466861724853516, 55.42411804199219, -4.910224914550781, 265.1493835449219, -63.32676696777344, -30.817134857177734], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000420.npy"}
|
||||
{"epoch": 0.6349206349206349, "step": 421, "batch_size": 64, "mean": 68.93891906738281, "std": 159.37168884277344, "min": -384.2965087890625, "p10": -140.78052520751953, "median": 77.35415267944336, "p90": 247.83626708984374, "max": 487.4422302246094, "pos_frac": 0.703125, "sample": [-141.6768035888672, 311.1297912597656, 13.267570495605469, 187.85162353515625, 252.44125366210938, 14.792694091796875, 113.17604064941406, 77.76237487792969, 14.582160949707031, 487.4422302246094, 344.4976806640625, -101.44049072265625, 218.6654052734375, 65.71691131591797, -30.801029205322266, 168.65745544433594, 57.36579895019531, -66.31742858886719, 234.99472045898438, 144.1427001953125, 152.63270568847656, 185.91879272460938, -4.997869491577148, 14.467277526855469, 90.453125, -16.011798858642578, 172.31005859375, -209.68789672851562, -231.02493286132812, -92.88320922851562, 3.8173694610595703, 188.81874084472656, 188.23126220703125, 24.761831283569336, 182.66366577148438, 246.7689208984375, 280.11419677734375, 137.12887573242188, 91.38268280029297, 1.9663658142089844, 185.23585510253906, 4.899662017822266, 88.0382080078125, -222.15380859375, -212.86500549316406, 154.8795166015625, -12.121932983398438, 59.17438507080078, -173.60891723632812, -28.582199096679688, 396.03363037109375, 248.293701171875, -17.83704376220703, 97.30598449707031, 77.07839965820312, -384.2965087890625, -14.897514343261719, 124.3638916015625, 166.2820587158203, -138.689208984375, 200.6616973876953, 77.6299057006836, -49.86225128173828, 14.04720687866211], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000421.npy"}
|
||||
{"epoch": 0.636432350718065, "step": 422, "batch_size": 64, "mean": 86.11872100830078, "std": 114.36446380615234, "min": -191.97708129882812, "p10": -47.49416389465332, "median": 68.53842544555664, "p90": 234.3546615600587, "max": 392.7339782714844, "pos_frac": 0.796875, "sample": [260.4463806152344, 392.7339782714844, -49.7451171875, -8.032176971435547, 166.61143493652344, 1.8909072875976562, 157.62608337402344, -133.42323303222656, 156.08694458007812, 300.45562744140625, 46.12159729003906, 18.240129470825195, 31.54254150390625, 94.041748046875, 326.51959228515625, 61.4209098815918, 177.83201599121094, 244.93711853027344, -78.90254211425781, 147.52297973632812, 87.34825134277344, -191.97708129882812, 25.01034164428711, 5.200519561767578, 171.98313903808594, 174.21224975585938, 158.42953491210938, 170.1394805908203, 322.48724365234375, 115.36215209960938, 63.02276611328125, 101.36308288574219, 37.10585403442383, -5.783403396606445, 67.01688385009766, 73.53407287597656, 166.91917419433594, 61.21422576904297, 158.80926513671875, 53.51202392578125, -48.508995056152344, -45.126224517822266, 197.58355712890625, 122.77108764648438, -25.8160400390625, -40.77062225341797, 14.244338989257812, 147.03683471679688, -39.126548767089844, 28.415088653564453, 92.04983520507812, 285.0180358886719, 23.33172035217285, -62.51927947998047, 6.162059783935547, 99.59294128417969, 55.46428680419922, -67.3669662475586, 11.380109786987305, 69.78060150146484, 209.66226196289062, 77.57274627685547, 67.29624938964844, 204.63438415527344], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000422.npy"}
|
||||
{"epoch": 0.6379440665154951, "step": 423, "batch_size": 64, "mean": 96.80520629882812, "std": 127.44925689697266, "min": -131.5261688232422, "p10": -28.361384582519527, "median": 74.658203125, "p90": 266.6605255126953, "max": 402.64068603515625, "pos_frac": 0.734375, "sample": [95.54574584960938, -13.359268188476562, 263.97479248046875, -131.5261688232422, 208.01524353027344, 3.3397045135498047, 46.69723129272461, -24.49506378173828, 194.94876098632812, 48.151344299316406, -117.52906799316406, 181.79458618164062, -55.723541259765625, -55.398048400878906, -12.018096923828125, 33.458953857421875, 42.999969482421875, 27.399185180664062, 206.5857696533203, -126.74845123291016, 195.1798858642578, 239.75042724609375, -11.660629272460938, -4.850542068481445, 71.9139404296875, 343.86944580078125, 240.79391479492188, 16.961078643798828, -117.99564361572266, 160.31541442871094, 364.984130859375, 173.5711669921875, 195.39903259277344, 158.5177459716797, 85.0174331665039, 77.4024658203125, 92.18792724609375, 316.76824951171875, 222.99647521972656, 267.8115539550781, 68.77042388916016, 10.745979309082031, 25.2347412109375, 210.31802368164062, 146.60791015625, 128.18936157226562, 199.47320556640625, 101.42982482910156, -19.320999145507812, 80.17817687988281, 402.64068603515625, 99.51097106933594, -5.378805160522461, 19.65812873840332, 145.33612060546875, 318.71295166015625, -22.095626831054688, -0.6813507080078125, -30.01837921142578, 62.58772277832031, 329.03009033203125, 4.442962646484375, 22.521930694580078, -7.407346725463867], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000423.npy"}
|
||||
{"epoch": 0.6394557823129252, "step": 424, "batch_size": 64, "mean": 93.35157775878906, "std": 126.2649154663086, "min": -161.29067993164062, "p10": -69.14241561889648, "median": 64.92608261108398, "p90": 236.37505340576175, "max": 545.3099365234375, "pos_frac": 0.8125, "sample": [-22.014190673828125, 269.39764404296875, 79.75135803222656, 158.4436492919922, -71.08038330078125, 166.54718017578125, 545.3099365234375, 242.84255981445312, -149.71006774902344, -88.37318420410156, 33.32722854614258, 37.72399139404297, 120.3946304321289, -161.29067993164062, 195.1712646484375, 25.006057739257812, 230.3402557373047, -64.62049102783203, 53.146514892578125, 43.876792907714844, -0.9047756195068359, 94.35482025146484, 108.45235443115234, 40.06078338623047, -79.009033203125, 225.2420196533203, 308.0319519042969, 19.149490356445312, 238.96139526367188, 22.766435623168945, 171.74432373046875, 115.0213394165039, 11.459007263183594, 62.952667236328125, 43.597991943359375, 137.1781768798828, 159.74220275878906, 280.904296875, 229.39227294921875, 132.7462158203125, 156.36410522460938, 51.316505432128906, 5.306613922119141, 211.36532592773438, 36.933387756347656, 218.97708129882812, 147.10877990722656, -0.8077983856201172, -124.9577407836914, 19.47481346130371, 189.360595703125, 38.43638610839844, 353.715087890625, 220.40066528320312, 32.700767517089844, 91.42485809326172, 35.35536193847656, -72.78060913085938, 38.458404541015625, 176.02232360839844, 16.75115203857422, 66.89949798583984, -13.355056762695312, 113.99691772460938], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000424.npy"}
|
||||
{"epoch": 0.6409674981103552, "step": 425, "batch_size": 64, "mean": 68.4725341796875, "std": 120.46753692626953, "min": -161.41050720214844, "p10": -82.58480377197264, "median": 50.59103584289551, "p90": 221.8499252319336, "max": 402.902587890625, "pos_frac": 0.65625, "sample": [140.97171020507812, 153.561767578125, -1.4444808959960938, -4.010986328125, 146.43991088867188, -63.30741500854492, 297.0612487792969, -18.028614044189453, 109.65645599365234, 134.29615783691406, 311.5902099609375, -106.79839324951172, 17.67470932006836, -111.13925170898438, -77.46189880371094, 168.47369384765625, 122.2658920288086, 225.38616943359375, 23.575927734375, 54.44325637817383, -159.27220153808594, 78.05039978027344, -5.881168365478516, 113.97703552246094, 188.18824768066406, 35.42646026611328, -0.7470645904541016, 254.6796875, 22.56237030029297, 215.93907165527344, 185.00596618652344, 0.7849693298339844, 88.16629028320312, -6.935546875, 207.67416381835938, 18.125938415527344, 70.259033203125, 196.30316162109375, 53.28348159790039, 26.850536346435547, 264.4435729980469, -151.6614990234375, 90.43362426757812, 154.87301635742188, 47.898590087890625, -161.41050720214844, 93.39314270019531, 64.39518737792969, 12.673492431640625, -27.254562377929688, 197.14694213867188, -2.800487518310547, -92.98957061767578, 224.38314819335938, -14.502998352050781, -2.516986846923828, 153.62457275390625, 29.248428344726562, -2.6172714233398438, 402.902587890625, -53.45615768432617, -31.21807098388672, 166.3870391845703, -84.78033447265625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000425.npy"}
|
||||
{"epoch": 0.6424792139077853, "step": 426, "batch_size": 64, "mean": 41.34447479248047, "std": 147.93284606933594, "min": -401.07757568359375, "p10": -104.64128570556637, "median": 13.156717300415039, "p90": 198.61081085205078, "max": 495.75341796875, "pos_frac": 0.609375, "sample": [-8.505611419677734, 0.7843513488769531, 135.2403564453125, 166.2117919921875, -19.66895294189453, 154.24916076660156, -14.346542358398438, 204.96124267578125, 35.053245544433594, 104.39393615722656, -55.48896026611328, -34.20918273925781, -18.222885131835938, -171.602783203125, 266.89093017578125, 4.6972198486328125, 111.52568054199219, 21.34649658203125, 84.3618392944336, -11.568672180175781, -212.73849487304688, -30.39654541015625, -71.23341369628906, 135.00225830078125, -24.915931701660156, 27.691848754882812, -3.365388870239258, -345.26837158203125, 153.07992553710938, -24.667945861816406, 316.029541015625, 118.61658477783203, 193.48464965820312, -52.278778076171875, -401.07757568359375, 34.715145111083984, -15.923355102539062, 1.4054698944091797, 10.334579467773438, 495.75341796875, 195.66738891601562, 341.0714111328125, -117.46859741210938, -153.45840454101562, 7.747516632080078, 252.91757202148438, -29.479965209960938, 8.601116180419922, 15.97885513305664, -69.44183349609375, 94.44660949707031, 135.92172241210938, -74.71089172363281, 151.5572052001953, 110.085693359375, -46.851707458496094, 32.23930740356445, 165.5352783203125, 51.562225341796875, -182.85723876953125, 199.29312133789062, 90.30415344238281, 197.0187530517578, 10.016841888427734], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000426.npy"}
|
||||
{"epoch": 0.6439909297052154, "step": 427, "batch_size": 64, "mean": 78.59584045410156, "std": 138.65780639648438, "min": -183.69857788085938, "p10": -82.06679763793944, "median": 67.28898620605469, "p90": 268.47888183593756, "max": 550.274658203125, "pos_frac": 0.6875, "sample": [162.1058807373047, 24.56819725036621, -20.079700469970703, -74.00824737548828, -165.05767822265625, 147.2461395263672, 198.7671356201172, 43.86882400512695, -47.99800109863281, -139.29977416992188, 276.05706787109375, -13.782215118408203, -85.52046203613281, -90.01248168945312, 75.63282012939453, 280.2166442871094, 124.43372344970703, 54.72230529785156, -41.512290954589844, -70.90696716308594, -0.4889240264892578, 67.31248474121094, 253.54022216796875, 44.43305969238281, 11.885955810546875, 160.11256408691406, 168.36825561523438, 147.653564453125, -0.31031036376953125, 49.481346130371094, 210.1080322265625, 54.602508544921875, 290.8710632324219, 95.54352569580078, 76.63216400146484, 203.92352294921875, 50.863807678222656, -183.69857788085938, 311.8421325683594, 225.33050537109375, -15.923194885253906, 67.26548767089844, 135.18609619140625, 64.34222412109375, -151.20433044433594, -54.63152313232422, 112.62641906738281, 144.84246826171875, 92.55555725097656, -112.11248779296875, 194.39122009277344, 410.3505554199219, -3.3662872314453125, 86.66436004638672, 85.13214874267578, 92.31336975097656, 117.5323486328125, 26.940547943115234, 28.243881225585938, -64.23208618164062, 135.90072631835938, -65.28694152832031, 550.274658203125, 274.88116455078125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000427.npy"}
|
||||
{"epoch": 0.6455026455026455, "step": 428, "batch_size": 64, "mean": 77.555419921875, "std": 144.69129943847656, "min": -327.8014221191406, "p10": -58.439805984497056, "median": 66.39163017272949, "p90": 239.89495849609378, "max": 594.747314453125, "pos_frac": 0.71875, "sample": [-162.28041076660156, 110.7615737915039, 243.64450073242188, -0.8297119140625, 4.656822204589844, 10.316879272460938, 52.393009185791016, -155.71240234375, -64.2752685546875, 140.14300537109375, 226.22540283203125, -104.27220916748047, 140.705322265625, -32.76873016357422, 179.73733520507812, 137.17095947265625, 213.92799377441406, 29.75640106201172, -9.228599548339844, 100.51411437988281, 254.72434997558594, -8.039249420166016, 10.282051086425781, -9.34075927734375, 251.40745544433594, 124.06195068359375, 29.92498016357422, -239.02415466308594, 153.7404022216797, -327.8014221191406, 223.52224731445312, -19.01873016357422, 39.200679779052734, 63.315425872802734, 135.53616333007812, 253.01370239257812, 172.49676513671875, 126.72346496582031, 97.24994659423828, 22.768083572387695, 291.59478759765625, 594.747314453125, 88.4513168334961, 167.05068969726562, -172.60301208496094, -20.99652099609375, 93.22990417480469, 45.21356964111328, 0.8105220794677734, 12.30047607421875, 21.419036865234375, -37.23418426513672, -6.618400573730469, -0.5889549255371094, 179.64999389648438, 208.4212188720703, 231.14602661132812, 69.46783447265625, 85.08927917480469, 174.3016357421875, 366.0496826171875, 180.66900634765625, 21.470130920410156, -44.823726654052734], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000428.npy"}
|
||||
{"epoch": 0.6470143613000756, "step": 429, "batch_size": 64, "mean": 56.733734130859375, "std": 120.28500366210938, "min": -256.8996276855469, "p10": -105.9431900024414, "median": 83.70665740966797, "p90": 193.22520751953127, "max": 261.6546630859375, "pos_frac": 0.71875, "sample": [202.5635528564453, -207.31207275390625, 3.4851646423339844, -112.29180145263672, 86.03174591064453, 132.04254150390625, 86.86894226074219, 57.09435272216797, -31.685848236083984, 156.0032196044922, 53.10713195800781, -256.8996276855469, 139.4000701904297, -3.3045196533203125, -14.385971069335938, 114.24380493164062, -213.34115600585938, 82.67337036132812, 3.58978271484375, 151.84130859375, 213.7598419189453, -66.96690368652344, 164.83547973632812, 154.1378173828125, 169.73391723632812, 177.71670532226562, 216.57424926757812, 26.546966552734375, 108.62212371826172, -34.53708267211914, -109.15928649902344, -98.43896484375, 23.04529571533203, -10.799888610839844, 88.4378890991211, 17.84564208984375, -89.3486328125, 102.34484100341797, -32.038841247558594, 8.318550109863281, 195.66355895996094, 143.3453369140625, 20.35051727294922, 187.5357208251953, 165.9933319091797, -218.115234375, -168.23162841796875, 23.131343841552734, -1.4481430053710938, 183.31185913085938, 22.470970153808594, 101.73250579833984, 49.96729278564453, -34.373226165771484, 90.25439453125, 84.73994445800781, 250.66818237304688, 127.20056915283203, 261.6546630859375, 234.15988159179688, 34.04826354980469, 173.4917755126953, 136.1879119873047, 106.86522674560547], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000429.npy"}
|
||||
{"epoch": 0.6485260770975056, "step": 430, "batch_size": 64, "mean": 97.28965759277344, "std": 124.4889144897461, "min": -362.6983337402344, "p10": -28.67735557556152, "median": 126.46996307373047, "p90": 252.12441711425782, "max": 293.9927978515625, "pos_frac": 0.796875, "sample": [3.01727294921875, 8.09657096862793, -34.320556640625, 94.20673370361328, 25.799842834472656, 212.83119201660156, -0.021343231201171875, 168.31031799316406, 71.17265319824219, 162.98007202148438, -141.174560546875, 143.22445678710938, 238.41671752929688, 17.9725284576416, 195.9876708984375, 146.11331176757812, 180.14639282226562, 138.75926208496094, 239.65359497070312, -19.49147605895996, 144.44703674316406, 293.9927978515625, 253.0955810546875, 134.9089813232422, -0.5068435668945312, 194.2203826904297, 94.26765441894531, 271.9924621582031, 179.32383728027344, 215.63568115234375, 135.11105346679688, 45.247955322265625, 49.287200927734375, 270.48260498046875, -26.643207550048828, 262.4819030761719, 258.1691589355469, 180.98483276367188, -154.07334899902344, 172.43466186523438, 3.111663818359375, 193.40921020507812, 48.6751823425293, 62.596229553222656, 22.256515502929688, 172.6589813232422, 66.6791763305664, 0.38471031188964844, -362.6983337402344, -15.628684997558594, -29.54913330078125, 195.98773193359375, 0.19414520263671875, 167.81365966796875, 118.03094482421875, 249.85836791992188, 260.346435546875, 170.24362182617188, 181.28802490234375, -127.19718933105469, 63.666656494140625, -53.10055923461914, 24.262176513671875, -13.2928466796875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000430.npy"}
|
||||
{"epoch": 0.6500377928949358, "step": 431, "batch_size": 64, "mean": 76.16828918457031, "std": 129.3199005126953, "min": -244.87420654296875, "p10": -59.03395729064939, "median": 51.13718223571777, "p90": 243.48044586181646, "max": 355.0622863769531, "pos_frac": 0.734375, "sample": [-182.91091918945312, 175.9698944091797, 44.306827545166016, 99.5835952758789, 29.26797103881836, 217.73995971679688, 323.774658203125, 168.51742553710938, 350.72308349609375, -4.47746467590332, 20.373886108398438, 203.9143829345703, 224.41049194335938, -7.4964599609375, 83.17105865478516, 30.705801010131836, 331.7701110839844, 2.6811294555664062, 355.0622863769531, 89.82850646972656, 104.4972152709961, -244.87420654296875, 25.018287658691406, 82.95930480957031, -27.174240112304688, 198.86288452148438, 54.14332580566406, -89.93876647949219, 8.6038818359375, -0.5561885833740234, 88.1421890258789, -36.508846282958984, 249.03369140625, 228.77987670898438, -171.76902770996094, -29.18633270263672, 269.2711181640625, 116.98408508300781, -116.04220581054688, -34.52250671386719, -100.70292663574219, 28.572509765625, 52.634056091308594, 191.49696350097656, 9.175971984863281, -5.8506317138671875, 230.5228729248047, 182.00796508789062, 2.0812454223632812, 260.7890319824219, 44.66693115234375, 58.58815383911133, 12.639129638671875, 49.64030838012695, 80.9003677368164, 21.048503875732422, 47.734405517578125, -68.68757629394531, -21.58181381225586, 222.22335815429688, 190.316162109375, 113.38275909423828, 71.65785217285156, -31.124740600585938], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000431.npy"}
|
||||
{"epoch": 0.6515495086923658, "step": 432, "batch_size": 64, "mean": 77.10441589355469, "std": 134.0049591064453, "min": -292.8966064453125, "p10": -77.16284103393555, "median": 70.75236129760742, "p90": 237.35305786132815, "max": 346.9263000488281, "pos_frac": 0.71875, "sample": [-67.42329406738281, -55.98039245605469, -64.98393249511719, -292.8966064453125, 4.922941207885742, 12.562318801879883, 29.267807006835938, 69.21392059326172, -94.12792205810547, 196.16485595703125, 157.30502319335938, -78.72197723388672, 123.23614501953125, 22.3802490234375, 222.74099731445312, -264.9167785644531, 71.82735443115234, 90.97285461425781, 267.61279296875, 346.9263000488281, 24.76207733154297, 152.27935791015625, 220.77957153320312, -85.19912719726562, -12.655948638916016, 227.32943725585938, -39.32927703857422, 246.30960083007812, 171.90432739257812, -12.002523422241211, -73.52485656738281, 69.6773681640625, 19.353708267211914, -2.6778717041015625, 9.380706787109375, 99.61268615722656, 199.90342712402344, 241.64889526367188, 23.60101318359375, 153.17581176757812, -31.28342056274414, 39.75307846069336, 182.50244140625, 46.39482116699219, 246.51156616210938, 77.27153015136719, 97.71446990966797, 210.96897888183594, 60.78274154663086, 249.75364685058594, 157.1156005859375, 134.73345947265625, 183.68905639648438, -193.25245666503906, 226.11273193359375, 294.3795471191406, -58.977134704589844, 146.0385284423828, 191.9801025390625, 34.58992004394531, -110.71953582763672, 192.9124755859375, -1.73382568359375, 227.0331573486328], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000432.npy"}
|
||||
{"epoch": 0.6530612244897959, "step": 433, "batch_size": 64, "mean": 86.51754760742188, "std": 117.51759338378906, "min": -247.79745483398438, "p10": -41.54510269165039, "median": 77.76366424560547, "p90": 220.42500305175784, "max": 330.2152404785156, "pos_frac": 0.71875, "sample": [6.332294464111328, 143.76266479492188, -17.697940826416016, 59.7480583190918, -39.533714294433594, 87.62088012695312, -1.131662368774414, 330.2152404785156, -6.228351593017578, -68.62939453125, 27.482101440429688, 33.17974090576172, 116.08349609375, -11.561805725097656, 207.60330200195312, -24.33631134033203, 116.44937133789062, 216.14944458007812, 191.9373779296875, -2.482990264892578, 67.90644836425781, 62.706298828125, -15.229408264160156, 209.88526916503906, -247.79745483398438, 107.76751708984375, 159.89154052734375, 60.70166778564453, 194.54678344726562, 173.13363647460938, 19.71056365966797, 93.8038558959961, -53.296424865722656, 26.94391632080078, 19.69442367553711, -30.672027587890625, -58.053707122802734, 11.676620483398438, 250.22006225585938, 205.7061309814453, 285.71966552734375, 203.49166870117188, 248.59423828125, 61.173095703125, 250.7753448486328, -181.89239501953125, 5.433774948120117, 175.00286865234375, -87.8820571899414, 196.3488311767578, -11.627639770507812, 148.084228515625, 248.05303955078125, -41.625831604003906, 107.04679107666016, 189.30963134765625, 201.66465759277344, 105.71759033203125, 222.25738525390625, 211.12786865234375, 183.30320739746094, 43.22093963623047, 190.97549438476562, -41.35673522949219], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000433.npy"}
|
||||
{"epoch": 0.654572940287226, "step": 434, "batch_size": 64, "mean": 99.90454864501953, "std": 141.84347534179688, "min": -288.85107421875, "p10": -69.29861450195311, "median": 117.85530853271484, "p90": 235.12913970947267, "max": 514.9658813476562, "pos_frac": 0.765625, "sample": [-217.66262817382812, 177.31060791015625, 179.22779846191406, 74.08566284179688, 218.43197631835938, 236.71707153320312, -116.32997131347656, 245.43243408203125, -149.20704650878906, 157.52369689941406, -14.962224960327148, -74.95570373535156, -288.85107421875, 101.27162170410156, 195.4014892578125, -56.09873962402344, 162.84104919433594, 117.96434020996094, 514.9658813476562, 208.47801208496094, 178.70477294921875, -1.082021713256836, 337.413330078125, 10.092857360839844, -173.50506591796875, 23.790267944335938, 221.2476806640625, 161.782958984375, 13.230915069580078, 157.17926025390625, -21.99576187133789, 149.1913299560547, 104.87724304199219, 224.21746826171875, 244.46533203125, 225.33355712890625, -36.35942840576172, 188.8404541015625, 305.5392150878906, 22.59814453125, 226.9693603515625, 12.426528930664062, 83.07435607910156, 182.8269500732422, 50.068214416503906, -1.0065536499023438, 51.685203552246094, 197.92169189453125, 117.74627685546875, 192.1324005126953, 13.54507827758789, 262.38348388671875, -151.40591430664062, -1.4299468994140625, 8.682670593261719, 122.3760757446289, 231.42396545410156, 225.25955200195312, 177.96170043945312, -3.264556884765625, 89.67353820800781, 33.580238342285156, 230.39501953125, 33.71912384033203], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000434.npy"}
|
||||
{"epoch": 0.656084656084656, "step": 435, "batch_size": 64, "mean": 72.20620727539062, "std": 130.68954467773438, "min": -248.93276977539062, "p10": -93.40625305175782, "median": 58.37833786010742, "p90": 245.44827728271486, "max": 318.6737060546875, "pos_frac": 0.6875, "sample": [158.77725219726562, 105.00611877441406, -40.848960876464844, 118.01903533935547, -141.73072814941406, 44.36560821533203, -37.75239562988281, -0.6764144897460938, 125.36766815185547, 30.582229614257812, 248.35012817382812, 196.57266235351562, 2.40728759765625, 163.48403930664062, 196.5394287109375, 231.4078826904297, 30.946983337402344, 9.516067504882812, -30.96124267578125, 256.33050537109375, 47.477210998535156, -23.11200714111328, 203.31719970703125, 246.09788513183594, 61.40106201171875, 27.908203125, -18.453216552734375, 157.61587524414062, 223.6499481201172, 310.35418701171875, -37.53224182128906, 202.72781372070312, -190.298095703125, 216.579833984375, 78.59780883789062, 94.09831237792969, -55.005699157714844, -90.881591796875, 201.21502685546875, 243.93252563476562, 318.6737060546875, 260.80780029296875, 208.46376037597656, -5.008247375488281, 0.0231781005859375, 57.781044006347656, 26.99203872680664, -120.75760650634766, -33.04513168334961, 252.3406982421875, -159.01699829101562, 58.97563171386719, -46.335357666015625, 122.3227767944336, 138.56065368652344, 94.67869567871094, 171.1842498779297, -94.48825073242188, -144.51788330078125, 54.40834045410156, 9.730804443359375, -248.93276977539062, 142.92727661132812, -9.964218139648438], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000435.npy"}
|
||||
{"epoch": 0.6575963718820862, "step": 436, "batch_size": 64, "mean": 58.51536560058594, "std": 133.23995971679688, "min": -244.6226806640625, "p10": -114.10124511718749, "median": 35.51475524902344, "p90": 217.93391876220704, "max": 438.16119384765625, "pos_frac": 0.71875, "sample": [151.9696807861328, -47.015682220458984, 160.03244018554688, 15.486963272094727, 201.6627197265625, 12.765657424926758, 115.51319885253906, 181.9067840576172, -167.6396484375, 43.94878387451172, 19.036605834960938, 177.9358673095703, 23.692039489746094, 126.6705322265625, 12.167896270751953, 212.31954956054688, 217.24798583984375, 109.75416564941406, -244.6226806640625, 10.723487854003906, 19.644649505615234, 99.53289031982422, -52.89519500732422, -28.885963439941406, 16.365358352661133, -14.314521789550781, 26.623687744140625, 9.629716873168945, 274.728271484375, 29.092628479003906, 57.136409759521484, -156.78134155273438, -198.17056274414062, 218.22789001464844, -61.254600524902344, 105.14054107666016, 219.81985473632812, -86.50443267822266, 86.0069580078125, -47.6943359375, 180.6292724609375, 192.07843017578125, 67.65185546875, 22.026931762695312, -120.30209350585938, 173.43463134765625, -148.46112060546875, 242.27781677246094, 123.0045166015625, 438.16119384765625, 170.10430908203125, 2.70550537109375, 262.66650390625, 171.26873779296875, -49.312652587890625, 22.755306243896484, 201.88424682617188, -96.74351501464844, 44.46906280517578, -2.1848888397216797, -99.63259887695312, 233.6123809814453, -178.05184936523438, 41.93688201904297], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000436.npy"}
|
||||
{"epoch": 0.6591080876795162, "step": 437, "batch_size": 64, "mean": 79.81681823730469, "std": 159.63009643554688, "min": -372.2251892089844, "p10": -100.06283721923828, "median": 80.47158432006836, "p90": 280.693505859375, "max": 496.6833801269531, "pos_frac": 0.65625, "sample": [496.6833801269531, 314.8742980957031, 189.21905517578125, 171.55380249023438, 98.5802230834961, 1.2043571472167969, 229.15541076660156, 171.8759307861328, 288.47589111328125, 101.37117004394531, 202.12271118164062, -88.7559814453125, -372.2251892089844, 52.11485290527344, -142.88815307617188, -75.95584869384766, 102.88969421386719, 154.3014678955078, 225.64816284179688, -239.19256591796875, 247.03160095214844, 306.7823486328125, 216.17193603515625, -24.837657928466797, -57.2286262512207, 88.87785339355469, 432.14337158203125, -35.008419036865234, -186.69786071777344, -94.17423248291016, -36.41986083984375, -102.5865249633789, 57.00145721435547, -7.353538513183594, -28.72075653076172, 205.53134155273438, 111.13475036621094, -65.88418579101562, 101.63468170166016, 76.33590698242188, 107.01910400390625, -17.85584259033203, -38.70662307739258, 285.6075439453125, 31.239898681640625, 203.46609497070312, 84.60726165771484, 180.36233520507812, 132.2812042236328, -127.77777862548828, 39.829864501953125, 20.806377410888672, 212.05661010742188, 269.2274169921875, -17.032127380371094, 14.801399230957031, 56.66700744628906, 225.2666473388672, -141.25247192382812, -22.428348541259766, -7.289583206176758, 124.74543762207031, 57.00657653808594, 350.8420715332031], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000437.npy"}
|
||||
{"epoch": 0.6606198034769464, "step": 438, "batch_size": 64, "mean": 97.44377899169922, "std": 142.1068115234375, "min": -166.25494384765625, "p10": -37.3934600830078, "median": 78.93513870239258, "p90": 219.08970947265632, "max": 739.9859619140625, "pos_frac": 0.71875, "sample": [279.29058837890625, 14.744132995605469, 739.9859619140625, -90.2261734008789, -5.595844268798828, -60.04393768310547, -28.592548370361328, 177.96351623535156, 12.99656867980957, 393.18768310546875, -108.474853515625, 267.6219787597656, -18.51728057861328, 24.53759002685547, 198.1036376953125, -0.2571067810058594, 122.79601287841797, -166.25494384765625, 179.03404235839844, 112.3764419555664, 180.45413208007812, 195.00135803222656, -14.738784790039062, 66.95706176757812, 47.384647369384766, 191.11143493652344, 121.01234436035156, 62.08332443237305, 194.15574645996094, 152.00946044921875, 136.02615356445312, -64.4605712890625, 90.91321563720703, 27.326215744018555, 43.7847900390625, 194.2838134765625, 186.1026153564453, -15.897672653198242, 135.69285583496094, 36.54234313964844, 178.5439453125, 24.924362182617188, 110.37312316894531, 465.66400146484375, 111.38502502441406, 44.90757751464844, 166.40847778320312, 105.40630340576172, 20.119583129882812, -7.429538726806641, -41.165279388427734, -5.69944953918457, 33.398887634277344, 168.4129638671875, 228.083740234375, -109.36066436767578, 192.341552734375, 109.58060455322266, -2.4425735473632812, 235.52149963378906, -10.73476791381836, -2.489103317260742, 167.96705627441406, 42.2646484375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000438.npy"}
|
||||
{"epoch": 0.6621315192743764, "step": 439, "batch_size": 64, "mean": 88.71385192871094, "std": 130.96109008789062, "min": -347.06109619140625, "p10": -55.97562255859375, "median": 69.26818084716797, "p90": 280.207977294922, "max": 378.5474853515625, "pos_frac": 0.765625, "sample": [202.867431640625, 25.58517074584961, 48.887718200683594, -54.346458435058594, 91.42074584960938, 198.1611785888672, 140.3762969970703, 2.015869140625, -56.67383575439453, 292.66461181640625, 257.122802734375, 45.57415771484375, 225.6448974609375, 20.045089721679688, 108.08332824707031, 348.5391540527344, 188.741455078125, 72.7421646118164, 251.41836547851562, -67.36759948730469, 10.271678924560547, 160.09210205078125, 58.63384246826172, -3.2215957641601562, 198.63038635253906, 106.92339324951172, 309.7312927246094, -91.24577331542969, 40.64744567871094, 238.67454528808594, -28.56660270690918, 290.10162353515625, -52.64044952392578, -70.0296630859375, 111.8027114868164, -69.77183532714844, 7.517875671386719, 23.984268188476562, -347.06109619140625, 15.562816619873047, 115.90447998046875, 65.36952209472656, 256.64373779296875, 32.242218017578125, 66.73434448242188, 236.2642364501953, 72.56808471679688, 291.0314636230469, 76.81002807617188, 5.171234130859375, 77.16468811035156, 101.83213806152344, -58.57691192626953, 204.77902221679688, 54.242218017578125, 378.5474853515625, 298.57989501953125, -40.7890625, -17.772682189941406, -3.4666290283203125, 39.340572357177734, 71.80201721191406, 123.79197692871094, -22.067237854003906], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000439.npy"}
|
||||
{"epoch": 0.6636432350718064, "step": 440, "batch_size": 64, "mean": 88.80775451660156, "std": 126.75874328613281, "min": -168.93511962890625, "p10": -49.69663619995116, "median": 64.77323913574219, "p90": 222.5273666381836, "max": 499.1231384277344, "pos_frac": 0.6875, "sample": [-33.61548614501953, 159.1294708251953, 41.531982421875, 313.978759765625, 215.90896606445312, 499.1231384277344, 207.69129943847656, -11.036445617675781, 213.13497924804688, 185.62115478515625, 48.464744567871094, 202.89268493652344, 72.30451202392578, 196.92800903320312, 117.38633728027344, 86.95223999023438, 217.860107421875, -90.67842864990234, -168.93511962890625, 55.087181091308594, 161.01263427734375, -7.705896377563477, 29.886680603027344, 242.16229248046875, 64.83769989013672, 179.85836791992188, 145.3776092529297, 262.77227783203125, -6.573266983032227, 64.70877838134766, -8.402627944946289, 60.333335876464844, 164.24908447265625, -17.474536895751953, -1.1050033569335938, -20.356002807617188, 40.11565399169922, -12.798883438110352, 25.948083877563477, 227.20069885253906, 210.28411865234375, 221.2786865234375, 36.83979797363281, 204.8868408203125, -15.640308380126953, -10.121788024902344, 225.44261169433594, 15.187614440917969, 207.06138610839844, -79.1056137084961, -54.333255767822266, -166.87847900390625, 94.30831909179688, -36.3675537109375, 150.51303100585938, 44.893741607666016, -38.87785720825195, -129.3753204345703, 223.06251525878906, -129.5838623046875, 54.08955383300781, 209.54095458984375, 199.43551635742188, 123.37885284423828], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000440.npy"}
|
||||
{"epoch": 0.6651549508692366, "step": 441, "batch_size": 64, "mean": 95.17347717285156, "std": 112.73004150390625, "min": -128.7654571533203, "p10": -52.47607498168944, "median": 105.41965103149414, "p90": 245.33036499023444, "max": 301.0235595703125, "pos_frac": 0.78125, "sample": [202.28050231933594, 131.82217407226562, 229.51388549804688, 258.66192626953125, 6.473840713500977, 185.63389587402344, 188.7552032470703, 14.695091247558594, 166.54611206054688, -36.006439208984375, -79.51609802246094, 158.3365936279297, -61.88074493408203, 226.7906494140625, 90.5504150390625, 142.9794921875, 186.39691162109375, -75.88302612304688, -35.23225784301758, 197.36131286621094, 161.3936004638672, 106.22003936767578, 301.0235595703125, 177.10308837890625, 0.1295318603515625, 199.33642578125, 2.156482696533203, -16.27527618408203, 270.58209228515625, 298.94671630859375, 108.31822967529297, 128.38424682617188, -128.7654571533203, 66.41886901855469, 144.27847290039062, 3.543060302734375, -70.04560852050781, 215.9854278564453, 104.6192626953125, 102.5408935546875, 0.493255615234375, 169.81680297851562, 69.67745208740234, 5.056285858154297, 83.82121276855469, 64.18087768554688, 165.04983520507812, -11.091751098632812, 191.8282470703125, -45.54425811767578, 297.67572021484375, 3.1218128204345703, 259.7857666015625, 252.10885620117188, -55.44685363769531, 34.636741638183594, 131.18600463867188, -13.044395446777344, -32.275779724121094, 115.64820098876953, -104.15837860107422, 27.117652893066406, 190.0064697265625, 17.279138565063477], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000441.npy"}
|
||||
{"epoch": 0.6666666666666666, "step": 442, "batch_size": 64, "mean": 97.03296661376953, "std": 129.19522094726562, "min": -202.39639282226562, "p10": -28.723806762695308, "median": 76.923095703125, "p90": 247.88950805664078, "max": 424.8147888183594, "pos_frac": 0.78125, "sample": [-17.521629333496094, 424.8147888183594, 23.097915649414062, 26.071014404296875, 74.88441467285156, 364.6178894042969, 278.91363525390625, 195.8266143798828, 40.91292953491211, 174.76229858398438, 39.246585845947266, 207.2873077392578, 162.70187377929688, 200.27426147460938, 187.84030151367188, 174.81790161132812, 9.295921325683594, 26.410369873046875, 94.16680908203125, -30.158218383789062, -88.83724975585938, 113.05499267578125, 143.80889892578125, -33.9697265625, 32.80351257324219, -16.810646057128906, -147.16033935546875, 138.990234375, 36.05425262451172, 356.2635498046875, 1.1718292236328125, 100.7706527709961, 263.8902893066406, 197.58828735351562, 385.0713195800781, 120.73865509033203, 71.94231414794922, -3.0788230895996094, 206.49781799316406, 104.40989685058594, 41.397727966308594, 13.110343933105469, 16.21467399597168, 16.285198211669922, -7.549230575561523, -202.39639282226562, 24.680923461914062, 121.46676635742188, -52.483734130859375, -25.376846313476562, 202.25733947753906, 198.27215576171875, -145.23086547851562, 100.34197235107422, 348.60858154296875, -9.135000228881836, 78.96177673339844, 46.075714111328125, 210.55435180664062, 197.49078369140625, 185.4765167236328, 194.64340209960938, -21.428627014160156, 36.409690856933594], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000442.npy"}
|
||||
{"epoch": 0.6681783824640968, "step": 443, "batch_size": 64, "mean": 115.7318115234375, "std": 148.46823120117188, "min": -221.82603454589844, "p10": -28.015577697753898, "median": 123.06002426147461, "p90": 244.25904846191415, "max": 760.8665771484375, "pos_frac": 0.796875, "sample": [16.964954376220703, 181.08311462402344, 81.72468566894531, 191.56167602539062, 45.84400177001953, 92.51837158203125, 175.81422424316406, -37.6207160949707, 122.60372924804688, 158.65267944335938, -91.27542877197266, -17.446809768676758, 99.40946960449219, 208.24252319335938, 88.31419372558594, 274.3092956542969, -0.23923301696777344, 176.08441162109375, 113.02526092529297, 217.56057739257812, 760.8665771484375, -197.25900268554688, -31.89000701904297, 175.91131591796875, 18.795766830444336, 335.3174743652344, 182.01504516601562, 123.51631927490234, 189.45306396484375, 199.914794921875, 213.21939086914062, 153.05267333984375, 189.51226806640625, 196.24722290039062, 134.79501342773438, 222.01202392578125, 194.1988983154297, 170.9990234375, 159.7020721435547, 253.79348754882812, 15.902511596679688, 209.97215270996094, -1.479410171508789, 78.88105010986328, 192.11068725585938, 195.73460388183594, -18.975242614746094, 25.887359619140625, 15.9637451171875, 8.272634506225586, 10.162174224853516, -104.47834014892578, 417.9329833984375, 5.6164093017578125, 26.193111419677734, 3.554950714111328, -221.82603454589844, -81.52362060546875, 315.6741943359375, 106.39046478271484, -5.107723236083984, -8.477180480957031, 133.4507598876953, 345.69970703125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000443.npy"}
|
||||
{"epoch": 0.6696900982615268, "step": 444, "batch_size": 64, "mean": 63.75556564331055, "std": 154.30447387695312, "min": -229.07479858398438, "p10": -91.02962875366211, "median": 24.71764087677002, "p90": 242.85809783935557, "max": 552.3927612304688, "pos_frac": 0.640625, "sample": [202.08273315429688, -5.978527069091797, -84.95731353759766, 1.5322513580322266, 85.4976806640625, -18.12105941772461, 139.68374633789062, -48.53108215332031, 29.004711151123047, 161.75025939941406, 9.516380310058594, 72.6380844116211, -165.54075622558594, -229.07479858398438, 219.08250427246094, 8.95184326171875, 118.28804016113281, 92.58883666992188, 93.68296813964844, -93.63204956054688, 382.8040466308594, 343.7554626464844, -0.02118682861328125, -26.449264526367188, -32.2479248046875, 99.72315216064453, 23.356857299804688, -23.462888717651367, -27.50433349609375, -48.340362548828125, 306.95263671875, 17.066997528076172, 252.43341064453125, 38.16676330566406, 80.26605224609375, 187.1893310546875, 63.75646209716797, 219.0189208984375, -7.358863830566406, -210.31922912597656, 80.06414794921875, 476.2201843261719, -216.18319702148438, -73.8243637084961, 19.018861770629883, 199.4056396484375, 258.4911804199219, 552.3927612304688, 220.5157012939453, 105.58226013183594, 26.07842445373535, 133.10585021972656, -6.704975128173828, 85.51185607910156, 207.84201049804688, 0.2086925506591797, -19.207706451416016, 17.973901748657227, -38.81849670410156, -150.05404663085938, -214.84716796875, 174.68211364746094, -1.3416767120361328, 16.99385643005371], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000444.npy"}
|
||||
{"epoch": 0.671201814058957, "step": 445, "batch_size": 64, "mean": 99.18186950683594, "std": 120.07893371582031, "min": -171.81072998046875, "p10": -64.39206504821776, "median": 132.38064575195312, "p90": 231.95484619140626, "max": 327.4560546875, "pos_frac": 0.75, "sample": [-11.03909683227539, 296.12152099609375, 16.30352020263672, -109.90367126464844, -115.596435546875, 63.73265075683594, 195.98114013671875, 230.15992736816406, 153.97344970703125, 160.57455444335938, -15.69305419921875, -171.81072998046875, -32.447784423828125, 36.76622009277344, -130.45770263671875, 164.77223205566406, 118.42964172363281, 36.46058654785156, -22.09234619140625, 138.94244384765625, -6.761116027832031, 239.43112182617188, -167.66912841796875, 80.97718811035156, -96.47722625732422, 230.50173950195312, -1.6563034057617188, 195.1302947998047, 132.94894409179688, -40.614105224609375, -53.91990280151367, -10.781553268432617, 147.43405151367188, 245.22274780273438, 234.966552734375, 210.29051208496094, 165.1050262451172, 195.92230224609375, 180.33413696289062, 31.25965118408203, 232.57760620117188, 105.87023162841797, 21.741100311279297, 276.34063720703125, 194.8769073486328, 208.56698608398438, 327.4560546875, 26.8738956451416, 214.27711486816406, -68.88013458251953, 216.828125, 13.959602355957031, 70.4217529296875, 150.44052124023438, 146.97360229492188, 168.81622314453125, 180.83123779296875, 201.00146484375, 104.14969635009766, 103.44063568115234, 131.81234741210938, 145.44590759277344, 210.8837890625, 48.11174774169922], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000445.npy"}
|
||||
{"epoch": 0.672713529856387, "step": 446, "batch_size": 64, "mean": 98.93746948242188, "std": 123.89524841308594, "min": -162.1167755126953, "p10": -36.71811180114746, "median": 106.28647994995117, "p90": 212.73301086425784, "max": 429.6489562988281, "pos_frac": 0.828125, "sample": [13.934392929077148, 13.735160827636719, -162.1167755126953, 166.46871948242188, 16.258705139160156, 182.7292022705078, 103.21282958984375, -47.779563903808594, 429.6489562988281, 24.596939086914062, 41.14810562133789, 116.27279663085938, 139.51979064941406, 177.250244140625, 161.34744262695312, 208.38900756835938, -72.45406341552734, 157.88076782226562, 139.75531005859375, 150.61880493164062, 0.10240554809570312, 189.05960083007812, 9.227691650390625, 203.94207763671875, 214.5947265625, 151.61512756347656, 19.613727569580078, 200.02517700195312, 285.2271728515625, -36.165306091308594, 204.9632568359375, 1.6232585906982422, -136.9130859375, -37.06224822998047, -25.343460083007812, -9.636629104614258, 21.44477081298828, -36.95502853393555, 109.3601303100586, 144.45596313476562, 75.3408203125, 256.34014892578125, 251.3533172607422, 192.2054443359375, 377.87823486328125, 4.657007217407227, 396.4066162109375, 204.56446838378906, -15.849748611450195, 17.673145294189453, 56.44164276123047, 46.114898681640625, 180.6103515625, 184.5008544921875, 205.18028259277344, 201.17657470703125, 17.25284194946289, 19.671775817871094, 2.5543899536132812, 123.97000122070312, 6.9523468017578125, 53.806270599365234, -152.70932006835938, 192.3092803955078], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000446.npy"}
|
||||
{"epoch": 0.674225245653817, "step": 447, "batch_size": 64, "mean": 116.06871032714844, "std": 134.54408264160156, "min": -221.34878540039062, "p10": -74.70301437377928, "median": 123.13753509521484, "p90": 269.63489990234376, "max": 344.58270263671875, "pos_frac": 0.75, "sample": [330.671630859375, 258.4879150390625, 201.0349578857422, -15.113571166992188, 84.27970886230469, -64.85945129394531, 262.2379455566406, 103.52139282226562, -98.91663360595703, -53.75236511230469, 92.31383514404297, 214.83782958984375, 4.934150695800781, 89.56961059570312, 239.11257934570312, 190.05177307128906, 116.48995208740234, 74.72090148925781, 128.61888122558594, 107.96397399902344, 41.274635314941406, -28.598304748535156, 87.6422119140625, -78.92168426513672, 106.98696899414062, 289.8520202636719, 255.99501037597656, 285.29888916015625, 229.91439819335938, -142.52972412109375, 111.21563720703125, 66.39106750488281, 202.96664428710938, 272.8050231933594, 120.79129028320312, 192.00503540039062, 256.1292724609375, 221.12290954589844, 232.48727416992188, -27.54458236694336, -0.7864837646484375, 184.49339294433594, 252.70315551757812, -120.37419891357422, -22.45071029663086, -130.5511474609375, 30.913291931152344, 151.133056640625, 313.19732666015625, 186.51869201660156, -28.633148193359375, 181.47470092773438, -94.00172424316406, 256.4107971191406, 89.93424987792969, 139.55209350585938, 125.48377990722656, -221.34878540039062, 169.81637573242188, 298.20166015625, 219.90509033203125, -51.96165466308594, 344.58270263671875, 192.69598388671875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000447.npy"}
|
||||
{"epoch": 0.6757369614512472, "step": 448, "batch_size": 64, "mean": 91.39349365234375, "std": 134.99098205566406, "min": -203.2507781982422, "p10": -45.841741180419895, "median": 68.62153244018555, "p90": 252.4152557373047, "max": 565.79150390625, "pos_frac": 0.8125, "sample": [272.253662109375, -181.1869659423828, 207.62173461914062, 99.08644104003906, 66.18579864501953, 140.77871704101562, 8.596969604492188, 225.8719940185547, 77.55372619628906, 43.665016174316406, 198.60667419433594, 63.420494079589844, 116.9969482421875, 78.23731994628906, 235.2244873046875, 266.1658935546875, -10.028339385986328, 253.48947143554688, 53.111297607421875, -11.59384536743164, 25.244613647460938, 177.8306884765625, 20.611061096191406, 105.14884948730469, 181.87652587890625, 1.9312782287597656, 207.0081024169922, 70.31231689453125, 18.80646514892578, -175.8912811279297, -73.4441909790039, 171.43601989746094, 42.528846740722656, 37.19847106933594, 18.90778350830078, 110.85380554199219, -120.03729248046875, 4.583818435668945, 299.95941162109375, -20.59330940246582, -0.8014621734619141, -203.2507781982422, 91.87782287597656, 30.063003540039062, 270.0509338378906, 249.90875244140625, 179.05142211914062, 165.3209228515625, 66.93074798583984, 16.02092933654785, -56.662498474121094, 56.24951171875, 152.60520935058594, 253.6464385986328, 3.7938461303710938, 565.79150390625, -13.789321899414062, -199.53228759765625, 200.87457275390625, 232.4876708984375, 194.34356689453125, 60.21195983886719, 3.1853561401367188, 222.47634887695312], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000448.npy"}
|
||||
{"epoch": 0.6772486772486772, "step": 449, "batch_size": 64, "mean": 46.31485366821289, "std": 134.2377166748047, "min": -339.2032470703125, "p10": -118.35002822875975, "median": 36.800649642944336, "p90": 208.46462554931642, "max": 312.29022216796875, "pos_frac": 0.65625, "sample": [13.539667129516602, 136.80474853515625, -158.822021484375, 131.92303466796875, 109.3877182006836, -1.3331146240234375, -37.22538757324219, 192.4442138671875, 32.404457092285156, 239.73342895507812, 196.48541259765625, 151.81378173828125, 18.00341033935547, -48.014564514160156, 27.815170288085938, -53.294189453125, 268.0519104003906, 1.1878013610839844, 0.0818634033203125, 92.101806640625, -63.145137786865234, 53.28430938720703, 182.585693359375, 209.3110809326172, -195.14581298828125, -26.363582611083984, 205.5086669921875, -339.2032470703125, 164.60714721679688, 28.49432373046875, 68.05891418457031, -83.29379272460938, -45.677886962890625, -124.72383117675781, 312.29022216796875, 33.43280029296875, 303.08917236328125, -20.247264862060547, 98.08854675292969, 54.471649169921875, 51.20069122314453, 98.80079650878906, -68.93438720703125, 54.2755126953125, -141.36138916015625, 40.16849899291992, 10.935760498046875, 238.40921020507812, 27.349376678466797, 165.11972045898438, 206.48956298828125, -19.833786010742188, 76.69043731689453, -47.85034942626953, -238.37933349609375, 291.19183349609375, 168.09336853027344, -103.47782135009766, 179.22998046875, -59.594932556152344, 68.60627746582031, -170.44888305664062, 45.12227249145508, -36.163047790527344], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000449.npy"}
|
||||
{"epoch": 0.6787603930461074, "step": 450, "batch_size": 64, "mean": 80.59041595458984, "std": 152.00538635253906, "min": -205.3540496826172, "p10": -79.03229827880858, "median": 32.00970268249512, "p90": 284.750765991211, "max": 517.58447265625, "pos_frac": 0.703125, "sample": [-1.184213638305664, -26.545120239257812, -34.08867645263672, -41.20520782470703, 517.58447265625, 239.22463989257812, 124.57220458984375, 141.46121215820312, 220.7379913330078, 217.191650390625, 132.75286865234375, 63.386619567871094, 56.250396728515625, 71.9359130859375, 11.730094909667969, 288.5123291015625, 9.241012573242188, 116.58639526367188, 174.9460906982422, 357.821044921875, 1.1262168884277344, 65.10858154296875, -56.68438720703125, -63.909645080566406, 49.98820495605469, -37.6310920715332, -85.51343536376953, 289.2076721191406, 156.963623046875, 20.878137588500977, 28.779884338378906, 275.9737854003906, 63.976356506347656, 176.69039916992188, -205.3540496826172, 236.4012451171875, 225.6605224609375, -12.16733169555664, 426.2613830566406, 17.164440155029297, 35.23952102661133, 0.36293792724609375, 156.20938110351562, -135.5394744873047, 5.4769134521484375, 3.701478958129883, 197.47659301757812, -23.30126953125, 338.37677001953125, -53.745723724365234, 15.344232559204102, -144.6810302734375, 14.95219612121582, 113.33772277832031, 445.4473876953125, -42.384857177734375, -98.67091369628906, 205.9783935546875, -191.0981903076172, 199.9789276123047, -91.51307678222656, 20.13361358642578, -28.91887092590332, 1.7916030883789062], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000450.npy"}
|
||||
{"epoch": 0.6802721088435374, "step": 451, "batch_size": 64, "mean": 71.15837860107422, "std": 123.68836975097656, "min": -213.85760498046875, "p10": -53.40788192749023, "median": 40.21910858154297, "p90": 209.30331573486328, "max": 386.9183654785156, "pos_frac": 0.6875, "sample": [160.1936798095703, 11.729339599609375, 91.41175842285156, 113.47380065917969, 319.050048828125, -37.440521240234375, -185.59353637695312, 35.82342529296875, -1.141134262084961, 256.8180236816406, 209.98768615722656, 177.61343383789062, 207.70645141601562, 82.92745971679688, 147.0350799560547, 142.62469482421875, 74.48600006103516, -28.827850341796875, 44.61479187011719, 124.78311157226562, -0.4772014617919922, -183.39886474609375, -146.82815551757812, 386.9183654785156, 193.1381072998047, -89.73287963867188, 233.5982666015625, -213.85760498046875, -65.22012329101562, -2.6320533752441406, 18.677236557006836, 349.3144226074219, 17.59122085571289, 3.951871871948242, 184.84439086914062, 148.0986328125, 203.80799865722656, 137.07913208007812, -17.57177734375, 2.5379867553710938, -0.41558837890625, 114.51390075683594, 3.921905517578125, 19.31529998779297, 156.46490478515625, 140.00534057617188, -0.3273582458496094, 108.08641052246094, 240.60548400878906, 151.40621948242188, -25.965911865234375, 124.49140167236328, 101.51065063476562, -4.537971496582031, 12.917655944824219, 8.200088500976562, 201.0645294189453, -9.192302703857422, -10.779497146606445, 13.573963165283203, 205.60113525390625, 1.9254722595214844, -54.49664306640625, -50.86743927001953], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000451.npy"}
|
||||
{"epoch": 0.6817838246409675, "step": 452, "batch_size": 64, "mean": 79.87801361083984, "std": 140.19859313964844, "min": -247.1478271484375, "p10": -100.6080787658691, "median": 61.83943176269531, "p90": 247.27458953857425, "max": 373.93841552734375, "pos_frac": 0.765625, "sample": [10.628318786621094, 19.517662048339844, 23.680143356323242, 73.82638549804688, 28.676612854003906, 169.0481719970703, 29.865257263183594, 249.69361877441406, 149.1396942138672, 232.3594970703125, 28.464000701904297, 335.88909912109375, -116.88055419921875, 238.4911651611328, 235.18429565429688, 121.24491119384766, 373.93841552734375, 217.27633666992188, 180.9440155029297, 310.6791687011719, 65.7125015258789, -41.05329895019531, 352.2645263671875, 124.46136474609375, 9.404186248779297, 7.216070175170898, 190.15625, 113.86392211914062, -13.665977478027344, 51.75147247314453, -8.611080169677734, 306.52825927734375, -156.37789916992188, 241.63018798828125, -51.20100402832031, 136.33888244628906, 57.96636199951172, 166.20899963378906, -247.1478271484375, 13.38918685913086, 149.49859619140625, -219.29339599609375, 76.3467025756836, 71.42901611328125, 36.4598388671875, -183.62796020507812, 237.35595703125, 168.94448852539062, 26.854080200195312, 17.05718994140625, 174.93527221679688, -11.131874084472656, 9.452669143676758, 194.6599578857422, 14.282817840576172, 25.211624145507812, -62.63896942138672, 253.77500915527344, 189.9169158935547, -142.49961853027344, -50.953582763671875, 87.20536041259766, -158.4813690185547, -23.066848754882812], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000452.npy"}
|
||||
{"epoch": 0.6832955404383976, "step": 453, "batch_size": 64, "mean": 71.43656921386719, "std": 133.3302764892578, "min": -235.86825561523438, "p10": -80.92660369873046, "median": 73.39375305175781, "p90": 235.03025512695314, "max": 425.87408447265625, "pos_frac": 0.71875, "sample": [142.74026489257812, 94.37794494628906, 76.06422424316406, 27.27172088623047, 222.1266632080078, 73.7987060546875, -46.03858947753906, 201.65228271484375, 250.00840759277344, -64.5530014038086, -174.10903930664062, -21.8226318359375, 87.95455932617188, 11.159217834472656, 170.98178100585938, 179.17999267578125, 425.87408447265625, 72.98880004882812, 242.18588256835938, 99.92591857910156, -14.539239883422852, -70.91410827636719, 8.530410766601562, 17.235610961914062, 19.455963134765625, -235.86825561523438, -178.50149536132812, 204.24635314941406, 1.4151229858398438, 206.0359344482422, 231.81069946289062, 29.3741455078125, 91.33403015136719, -82.65927124023438, -22.385601043701172, 47.53971862792969, 173.94288635253906, 210.12762451171875, -170.15493774414062, 170.7563934326172, 79.68334197998047, 321.447265625, -10.869319915771484, -38.51836013793945, -76.88371276855469, 2.3715362548828125, -148.56948852539062, 4.014514923095703, 199.50143432617188, 136.30601501464844, -15.17300033569336, 8.155818939208984, 54.1429443359375, 149.2386932373047, 123.7882080078125, -120.24249267578125, -26.037593841552734, 156.69091796875, 242.86817932128906, 86.89166259765625, 295.4562683105469, 195.6913299560547, 236.41006469726562, 7.02734375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000453.npy"}
|
||||
{"epoch": 0.6848072562358276, "step": 454, "batch_size": 64, "mean": 88.31023406982422, "std": 132.96583557128906, "min": -314.5915222167969, "p10": -37.85782623291015, "median": 71.0382080078125, "p90": 235.2948944091797, "max": 567.3463134765625, "pos_frac": 0.75, "sample": [195.71653747558594, -14.5386962890625, -22.439529418945312, 270.9073486328125, -314.5915222167969, -10.144495010375977, 159.49710083007812, 90.7516860961914, 21.111835479736328, 53.956695556640625, 66.11317443847656, 202.85972595214844, -15.9215087890625, -223.82388305664062, 222.56069946289062, 166.69259643554688, 192.79803466796875, 55.64466857910156, 276.6312561035156, 75.96324157714844, -28.004302978515625, -169.67697143554688, 11.324907302856445, 283.9048767089844, 232.61322021484375, 567.3463134765625, -4.356283187866211, 63.425018310546875, -66.08209991455078, 55.980186462402344, -53.08320617675781, 156.98648071289062, 245.59561157226562, 34.44097137451172, 183.62014770507812, 155.72857666015625, 257.0986328125, -59.42648696899414, 236.44418334960938, 111.14704895019531, 65.85556030273438, 133.65846252441406, 34.82653045654297, 1.2940549850463867, 10.500381469726562, 63.23021697998047, 189.30905151367188, 126.315673828125, -1.0188350677490234, 22.379150390625, 145.42774963378906, 133.64456176757812, -8.832233428955078, 194.57516479492188, 133.26296997070312, 48.82366943359375, 172.26878356933594, 148.7538299560547, -42.08076477050781, 224.322509765625, 114.58795928955078, 94.48429870605469, 8.974565505981445, -27.48016357421875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000454.npy"}
|
||||
{"epoch": 0.6863189720332578, "step": 455, "batch_size": 64, "mean": 79.60262298583984, "std": 132.0699462890625, "min": -145.309326171875, "p10": -72.35886535644529, "median": 45.481632232666016, "p90": 241.3853317260742, "max": 424.3926696777344, "pos_frac": 0.65625, "sample": [-78.73065185546875, 110.22579956054688, 33.690895080566406, 311.832275390625, 68.79662322998047, 4.372289657592773, 371.2568359375, 153.43898010253906, -93.02274322509766, 67.58489227294922, 161.4661407470703, -145.309326171875, 112.48977661132812, 219.5701904296875, -14.057861328125, 409.0240173339844, -52.63652801513672, 29.767303466796875, -37.13097381591797, -96.94913482666016, 308.61962890625, 19.913606643676758, 169.28750610351562, -9.391117095947266, -6.981437683105469, 339.12255859375, 46.47312927246094, 187.91986083984375, 206.74688720703125, 44.490135192871094, 27.471240997314453, 19.15079689025879, 241.9700927734375, 39.814247131347656, 27.688810348510742, -29.86023712158203, -115.72071838378906, 180.75531005859375, 188.46307373046875, 165.80364990234375, -26.68051528930664, 65.83350372314453, 143.75411987304688, -57.491363525390625, 159.7481689453125, 240.02088928222656, 76.99767303466797, 114.2032699584961, -23.266036987304688, -1.706146240234375, 230.3858642578125, -13.880073547363281, 2.177104949951172, 202.0192413330078, 63.43013000488281, -5.034820556640625, 77.55662536621094, -43.896202087402344, -88.54203796386719, -30.397544860839844, 92.28495025634766, -11.598873138427734, -83.15859985351562, 424.3926696777344], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000455.npy"}
|
||||
{"epoch": 0.6878306878306878, "step": 456, "batch_size": 64, "mean": 81.46484375, "std": 141.68231201171875, "min": -221.14756774902344, "p10": -55.16574554443359, "median": 55.330034255981445, "p90": 224.64600067138684, "max": 621.8887939453125, "pos_frac": 0.734375, "sample": [176.78614807128906, 4.513206481933594, 10.805810928344727, 139.11318969726562, 621.8887939453125, -2.848217010498047, -89.75155639648438, 142.76768493652344, -19.75924301147461, 159.2821044921875, 36.00151062011719, 128.60311889648438, 395.8048400878906, 143.87454223632812, 18.081192016601562, 19.024826049804688, 188.58901977539062, 159.88314819335938, -5.480539321899414, 236.48715209960938, -135.5313262939453, 39.43379211425781, 97.9511947631836, -49.88072204589844, 197.0166473388672, -11.353151321411133, -221.14756774902344, -157.5266571044922, 176.39425659179688, 24.1805419921875, 1.4487724304199219, 87.24690246582031, 18.181550979614258, 163.8741912841797, 35.13226318359375, 73.671875, 57.84492492675781, 14.291946411132812, 89.43197631835938, -0.4104042053222656, -17.37567710876465, 52.81514358520508, 87.37244415283203, 177.6826934814453, -132.00994873046875, -151.14132690429688, 124.84556579589844, -9.063140869140625, -22.932823181152344, -13.920024871826172, 243.6955108642578, 196.9152069091797, 19.136734008789062, 331.11541748046875, 68.61138916015625, 128.0186309814453, 355.22515869140625, 10.289009094238281, 186.38485717773438, 144.9422607421875, 7.431398391723633, -57.430755615234375, 143.39735412597656, 375.8271484375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000456.npy"}
|
||||
{"epoch": 0.6893424036281179, "step": 457, "batch_size": 64, "mean": 80.48719787597656, "std": 124.77316284179688, "min": -377.9749755859375, "p10": -32.29573726654052, "median": 74.93122482299805, "p90": 218.45941162109375, "max": 409.4427795410156, "pos_frac": 0.8125, "sample": [71.4159927368164, 1.4798126220703125, 28.216068267822266, 169.36366271972656, 114.64369201660156, -101.30387878417969, 20.790149688720703, 78.44645690917969, 55.352882385253906, 29.8226318359375, 20.045026779174805, 408.828369140625, 194.66905212402344, 103.51488494873047, -9.243976593017578, -109.79150390625, 229.5918731689453, 80.52793884277344, -0.47859764099121094, 218.68710327148438, 150.0693359375, 34.47282409667969, 282.921875, 103.02812194824219, 23.31668472290039, 49.360870361328125, -23.293214797973633, 217.92813110351562, 45.018646240234375, -2.987598419189453, -127.83847045898438, -125.29370880126953, 29.71831512451172, 187.0470733642578, 49.8173713684082, -101.64460754394531, -377.9749755859375, 125.63778686523438, 92.55535125732422, 11.275787353515625, 127.92130279541016, 24.76909637451172, 208.10174560546875, 166.16517639160156, -6.362945556640625, 10.427680969238281, 231.4602813720703, 139.41131591796875, 7.056007385253906, 206.91317749023438, 113.02677154541016, 161.65826416015625, 409.4427795410156, 139.7801513671875, 93.40695190429688, 204.70709228515625, 191.14004516601562, 23.312667846679688, 29.64906120300293, 98.01005554199219, 229.86007690429688, -36.153961181640625, 85.93330383300781, 43.831512451171875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000457.npy"}
|
||||
{"epoch": 0.690854119425548, "step": 458, "batch_size": 64, "mean": 95.124267578125, "std": 143.5191650390625, "min": -271.9814453125, "p10": -41.7327953338623, "median": 83.23966217041016, "p90": 254.37956848144538, "max": 431.5953369140625, "pos_frac": 0.796875, "sample": [192.20831298828125, 151.1353759765625, 6.258642196655273, 216.4735107421875, 84.32819366455078, 62.46084976196289, 112.70254516601562, -252.09115600585938, -0.1117095947265625, 24.575820922851562, -22.239517211914062, -105.50382995605469, -29.639144897460938, 348.32904052734375, 230.66864013671875, 98.79351806640625, 90.45870208740234, 212.8382568359375, 22.626243591308594, 200.58030700683594, 36.005393981933594, 73.71647644042969, 82.15113067626953, 3.5522079467773438, 212.74879455566406, 65.26506042480469, 424.985107421875, 41.092247009277344, 31.548866271972656, 215.80972290039062, 285.6897277832031, 109.66136169433594, 103.91912841796875, -212.21240234375, 141.6475067138672, 0.9506397247314453, 61.50884246826172, 61.438232421875, 163.41195678710938, 113.56031799316406, 189.3458709716797, 75.22126770019531, -271.9814453125, -14.98895263671875, 412.46435546875, -1.3148155212402344, 185.31903076171875, -94.23117065429688, 218.46253967285156, 260.66448974609375, -33.00040054321289, 431.5953369140625, 46.74656677246094, 37.88219451904297, -172.38189697265625, 212.0290985107422, 112.42205047607422, 239.71475219726562, -45.475250244140625, 103.76832580566406, 21.803783416748047, 72.30377197265625, 276.02264404296875, 164.25790405273438], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000458.npy"}
|
||||
{"epoch": 0.6923658352229781, "step": 459, "batch_size": 64, "mean": 104.49907684326172, "std": 147.9705352783203, "min": -505.70440673828125, "p10": -27.73248748779296, "median": 113.19328308105469, "p90": 256.5902313232422, "max": 409.5417785644531, "pos_frac": 0.78125, "sample": [64.96343231201172, 170.31045532226562, -17.140159606933594, 190.85528564453125, 34.039207458496094, -505.70440673828125, -5.373884201049805, 254.24826049804688, 116.07235717773438, 189.81967163085938, 233.07608032226562, -11.188663482666016, 128.0437469482422, -144.6153564453125, 208.51174926757812, 334.68719482421875, 111.69291687011719, 118.76058959960938, 29.9833984375, 196.9906005859375, 206.3865966796875, 282.60504150390625, 132.81817626953125, 3.2930831909179688, 6.363077163696289, 206.57266235351562, 85.82402801513672, 155.83865356445312, 20.514026641845703, 155.31295776367188, 409.5417785644531, 68.89322662353516, 3.9840450286865234, 25.54811668395996, -34.537574768066406, 191.90264892578125, 53.203330993652344, -155.7819366455078, 187.1314239501953, -15.305217742919922, -7.030694961547852, 235.99530029296875, 215.47283935546875, 1.63970947265625, 404.87939453125, -128.96383666992188, 197.23406982421875, 67.25824737548828, -47.620361328125, 105.09569549560547, -32.272056579589844, -13.338821411132812, -13.298933029174805, 361.7523193359375, 175.73985290527344, 114.69364929199219, 240.3628692626953, 44.74578857421875, 235.52528381347656, 257.59393310546875, 246.17869567871094, 51.25444793701172, 268.6285705566406, 18.27466583251953], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000459.npy"}
|
||||
{"epoch": 0.6938775510204082, "step": 460, "batch_size": 64, "mean": 105.89602661132812, "std": 145.5498046875, "min": -157.404296875, "p10": -57.27115325927734, "median": 71.9473762512207, "p90": 287.6997100830079, "max": 533.978759765625, "pos_frac": 0.75, "sample": [309.960205078125, 157.97708129882812, 357.8813781738281, -128.87168884277344, 1.6343002319335938, 184.65110778808594, 363.6011962890625, 68.41696166992188, 489.144775390625, 18.30724334716797, -58.195701599121094, 23.578521728515625, -57.41559600830078, 50.9395751953125, -35.897796630859375, 63.248756408691406, 22.32875633239746, 214.58920288085938, -98.06871032714844, 200.1600341796875, 145.39462280273438, 215.0661163330078, 75.47779083251953, 297.13177490234375, 196.21334838867188, 106.52826690673828, 1.7466087341308594, 25.543914794921875, 207.44650268554688, 49.04505157470703, 217.07742309570312, 301.6781005859375, 196.35536193847656, 231.361328125, 11.243364334106445, 115.20443725585938, -2.279773712158203, 12.344131469726562, 139.77468872070312, -13.421632766723633, 140.26841735839844, 219.08116149902344, -49.135475158691406, 18.388246536254883, 175.6505584716797, 212.18125915527344, -93.28582763671875, -9.097793579101562, 217.24761962890625, -122.61669921875, 5.027910232543945, 32.83514404296875, 248.1648406982422, 157.1238555908203, -28.739700317382812, 265.6915588378906, 62.775978088378906, 533.978759765625, -157.404296875, -21.505416870117188, 170.61294555664062, -24.07537841796875, -56.934120178222656, 204.21121215820312], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000460.npy"}
|
||||
{"epoch": 0.6953892668178382, "step": 461, "batch_size": 64, "mean": 106.24911499023438, "std": 138.3647003173828, "min": -181.2144317626953, "p10": -43.55821914672851, "median": 95.66997909545898, "p90": 298.3597412109376, "max": 486.9557189941406, "pos_frac": 0.765625, "sample": [264.20245361328125, 76.30335235595703, 48.89152526855469, 28.2646484375, 194.11959838867188, 311.32269287109375, 54.978660583496094, 182.05633544921875, 118.84723663330078, -181.2144317626953, 89.39830017089844, 135.87086486816406, 152.3763885498047, 7.87523078918457, 159.42776489257812, -73.07151794433594, -176.87828063964844, 216.9639434814453, 7.699546813964844, 4.792943954467773, 199.46241760253906, 201.20816040039062, 233.56591796875, -41.317596435546875, 76.44343566894531, 194.20530700683594, -26.947120666503906, -4.562553405761719, -44.51848602294922, 160.75030517578125, 230.03909301757812, -26.83386993408203, -8.724908828735352, 112.0774154663086, 195.90701293945312, 322.1025085449219, 486.9557189941406, 335.4837341308594, 42.25151062011719, 337.9423522949219, 323.0286865234375, 199.04193115234375, 124.68212890625, 1.3405685424804688, 10.884038925170898, -48.83790588378906, 129.55706787109375, -2.8658504486083984, 192.54995727539062, 13.019416809082031, 384.48504638671875, 10.523490905761719, 101.94165802001953, 52.906402587890625, 257.6980285644531, -10.62432861328125, 268.11285400390625, -130.5984344482422, 53.09319305419922, -19.442771911621094, 172.8824462890625, 7.800621032714844, 192.61004638671875, -81.56271362304688], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000461.npy"}
|
||||
{"epoch": 0.6969009826152683, "step": 462, "batch_size": 64, "mean": 103.15957641601562, "std": 123.41915893554688, "min": -214.89443969726562, "p10": -45.460795593261714, "median": 106.15761184692383, "p90": 238.58465270996095, "max": 395.2270202636719, "pos_frac": 0.8125, "sample": [226.89317321777344, 241.2188720703125, 231.73541259765625, 230.40940856933594, 27.823135375976562, -43.082733154296875, 229.05776977539062, 8.671146392822266, 7.308250427246094, 12.345939636230469, -102.16024017333984, 203.62863159179688, 172.29299926757812, 104.1578598022461, 206.43777465820312, 256.237060546875, -46.47996520996094, 281.84381103515625, 0.9163436889648438, 152.243896484375, 188.84324645996094, 108.15736389160156, 84.12294006347656, 146.83316040039062, 66.20050048828125, 18.55645751953125, 212.2398223876953, 282.832763671875, -41.77102279663086, 155.97837829589844, -74.91107177734375, 81.70329284667969, -53.12798309326172, 200.55340576171875, 232.43814086914062, -73.642578125, -28.648130416870117, 110.47297668457031, 34.63513946533203, 187.44619750976562, 226.90512084960938, 253.01036071777344, 58.324562072753906, 91.1181411743164, 93.41215515136719, 211.40219116210938, -24.436477661132812, 59.09992599487305, 202.0756072998047, 171.59564208984375, 125.58126831054688, -213.5602264404297, 258.2184143066406, 200.02682495117188, 162.2550811767578, 18.894187927246094, 45.10601806640625, 395.2270202636719, 26.830772399902344, 162.1041259765625, -214.89443969726562, 28.09925079345703, -1.056467056274414, 26.461841583251953], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000462.npy"}
|
||||
{"epoch": 0.6984126984126984, "step": 463, "batch_size": 64, "mean": 57.469078063964844, "std": 135.31076049804688, "min": -314.43157958984375, "p10": -92.15112991333007, "median": 62.181949615478516, "p90": 220.72391052246095, "max": 306.34564208984375, "pos_frac": 0.6875, "sample": [25.199420928955078, 282.52294921875, 64.48776245117188, -314.43157958984375, 108.91452026367188, 189.47573852539062, -28.673852920532227, 214.24935913085938, 223.49871826171875, 197.39820861816406, -39.65095901489258, -164.9080352783203, -10.724220275878906, 82.15697479248047, 156.8193359375, -269.896240234375, 38.809608459472656, -26.135757446289062, 3.1681060791015625, -96.27257537841797, 19.126386642456055, 61.87372589111328, -60.041717529296875, 264.75390625, 85.12956237792969, -146.57952880859375, 62.41075134277344, -0.04030609130859375, 12.432357788085938, 18.935771942138672, 68.62464904785156, -23.26471710205078, 191.91885375976562, 201.31028747558594, 88.24661254882812, 152.66131591796875, 100.37613677978516, 38.888832092285156, 30.862770080566406, 306.34564208984375, 63.43203353881836, 61.953147888183594, 182.7376708984375, -258.2696533203125, -82.534423828125, -62.253448486328125, 291.433837890625, 101.01202392578125, 243.07664489746094, 168.85520935058594, -33.40677261352539, -175.30389404296875, -33.32157897949219, 73.72044372558594, 147.7983856201172, 198.68521118164062, 37.234928131103516, -80.19941711425781, 168.65280151367188, 145.46109008789062, -3.2072830200195312, 143.7589569091797, 21.88325309753418, 246.84329223632812], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000463.npy"}
|
||||
{"epoch": 0.6999244142101285, "step": 464, "batch_size": 64, "mean": 97.25220489501953, "std": 152.9383087158203, "min": -289.1579284667969, "p10": -81.87089767456054, "median": 102.24160766601562, "p90": 289.5133911132813, "max": 390.63909912109375, "pos_frac": 0.765625, "sample": [-68.44944763183594, 68.4555892944336, 205.78244018554688, 7.7215576171875, 208.19679260253906, -1.962188720703125, 36.746253967285156, 189.11447143554688, 99.24253845214844, 155.98114013671875, 213.73422241210938, -64.74425506591797, 218.67413330078125, 162.86585998535156, 84.09304809570312, -289.1579284667969, 321.50482177734375, 150.10284423828125, 105.24067687988281, 173.5592803955078, 8.253494262695312, 364.60113525390625, 0.48830223083496094, 113.75929260253906, 18.784561157226562, 351.60858154296875, 125.86042022705078, 293.92864990234375, 165.3903350830078, 255.23109436035156, 229.71913146972656, 15.898460388183594, 192.3446807861328, 390.63909912109375, -260.24456787109375, -159.4761505126953, 216.00390625, 77.5899429321289, -211.35675048828125, -112.45150756835938, 141.88230895996094, 16.002723693847656, 85.88316345214844, 339.715087890625, -87.6229476928711, 8.479911804199219, 41.67479705810547, 97.86727905273438, -0.174530029296875, 357.7432861328125, 163.614013671875, 279.21112060546875, -226.57159423828125, -42.359683990478516, 65.53995513916016, -4.296588897705078, -11.333641052246094, 206.09884643554688, 152.86727905273438, 249.85821533203125, -40.886634826660156, 183.77584838867188, 161.75711059570312, 32.141510009765625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000464.npy"}
|
||||
{"epoch": 0.7014361300075586, "step": 465, "batch_size": 64, "mean": 88.72981262207031, "std": 142.56610107421875, "min": -262.3365478515625, "p10": -80.96494522094726, "median": 82.36301040649414, "p90": 246.6945556640625, "max": 421.1006164550781, "pos_frac": 0.640625, "sample": [-58.132232666015625, 421.1006164550781, 247.7120361328125, -262.3365478515625, 227.21875, 236.51901245117188, -85.98493957519531, 218.2869110107422, 185.6375274658203, 110.1419448852539, 96.77578735351562, 215.30276489257812, 196.33724975585938, 304.3187255859375, -33.38672637939453, 87.601806640625, 65.10953521728516, 215.94039916992188, 368.708984375, -9.741523742675781, -44.80137634277344, -134.91700744628906, 244.3204345703125, 306.29681396484375, -12.634963989257812, 237.8303680419922, -69.25162506103516, -105.76467895507812, -28.789649963378906, 106.12946319580078, -17.284095764160156, -63.94727325439453, 217.92083740234375, 146.40151977539062, 173.5840301513672, -1.8720703125, -52.938201904296875, -6.346809387207031, 6.55457878112793, 197.6140594482422, -91.50010681152344, -5.659553527832031, 72.57431030273438, 74.86353302001953, -19.83325958251953, -179.57931518554688, -97.11373901367188, 132.74205017089844, 76.851806640625, 258.01318359375, 44.42748260498047, 377.5056457519531, 19.629798889160156, 146.2485809326172, 13.368919372558594, -41.51404571533203, 77.12421417236328, -22.268835067749023, 152.1958770751953, 130.5843505859375, 216.77175903320312, 186.7447509765625, 128.04855346679688, 183.2476043701172], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000465.npy"}
|
||||
{"epoch": 0.7029478458049887, "step": 466, "batch_size": 64, "mean": 39.02485656738281, "std": 147.48289489746094, "min": -608.2811279296875, "p10": -84.45676498413084, "median": 20.922332763671875, "p90": 210.93446350097662, "max": 365.9407653808594, "pos_frac": 0.640625, "sample": [-0.5003738403320312, 196.01828002929688, -1.230072021484375, -91.31430053710938, 23.238229751586914, 12.284307479858398, 53.04936218261719, -17.681427001953125, -28.194923400878906, -608.2811279296875, 113.14289855957031, 58.15803527832031, 217.28836059570312, -165.31341552734375, 110.85991668701172, -110.95260620117188, 104.54585266113281, 365.9407653808594, 98.98405456542969, 269.7037048339844, 44.40103530883789, 5.5131072998046875, 196.10870361328125, 0.5258979797363281, -5.592437744140625, -17.125144958496094, 103.90657043457031, -97.75431823730469, 337.52777099609375, 127.72512817382812, -14.782285690307617, 242.68540954589844, -42.919151306152344, 39.29600524902344, 168.71388244628906, 27.224483489990234, 132.28298950195312, 66.27912139892578, -52.2659797668457, 20.17589569091797, -72.44095611572266, -367.18170166015625, -66.94058990478516, -73.30033874511719, 159.56704711914062, 14.24652099609375, 42.75242614746094, 127.6351547241211, 287.5190734863281, -6.320384979248047, -57.096466064453125, 182.27371215820312, -89.23809051513672, 30.76129150390625, -22.15825653076172, -0.10490226745605469, 6.507291793823242, 15.307905197143555, 32.91864013671875, 195.93894958496094, 233.27182006835938, 0.15597152709960938, 20.681068420410156, 21.163597106933594], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000466.npy"}
|
||||
{"epoch": 0.7044595616024187, "step": 467, "batch_size": 64, "mean": 91.3807601928711, "std": 154.7257080078125, "min": -208.66757202148438, "p10": -65.44852676391602, "median": 50.41411590576172, "p90": 263.45115356445314, "max": 732.779296875, "pos_frac": 0.765625, "sample": [-6.9934844970703125, 234.90003967285156, 118.15280151367188, 139.9494171142578, 42.102813720703125, 101.15369415283203, 2.2817840576171875, -48.46665573120117, -84.72592163085938, 68.32866668701172, 215.63925170898438, -63.03775405883789, 214.75999450683594, -65.8299789428711, 217.04637145996094, 116.52196502685547, 138.43026733398438, 71.51998901367188, -158.54417419433594, 19.184616088867188, 195.26242065429688, -64.5584716796875, -48.45338439941406, 260.10919189453125, 264.8834228515625, 52.85974884033203, 38.212806701660156, 93.19685363769531, -208.66757202148438, 378.2309265136719, -6.127372741699219, 87.38093566894531, 13.025274276733398, 11.755645751953125, -10.200180053710938, 11.899826049804688, 185.08734130859375, 418.1520080566406, 132.44554138183594, 332.4632568359375, 0.7525787353515625, -0.43663597106933594, 35.59168243408203, 4.013750076293945, 9.661802291870117, -112.46720886230469, 172.54180908203125, 4.134422302246094, 191.47483825683594, 175.5340118408203, 732.779296875, 40.025482177734375, -66.62500762939453, 2.3596630096435547, 47.968482971191406, 107.28626251220703, 107.41753387451172, 223.6788330078125, 6.120674133300781, 384.70263671875, -126.54830932617188, 303.3568115234375, 160.68096923828125, 35.032386779785156], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000467.npy"}
|
||||
{"epoch": 0.7059712773998488, "step": 468, "batch_size": 64, "mean": 95.58055114746094, "std": 156.08914184570312, "min": -449.65435791015625, "p10": -72.70026397705077, "median": 104.90014266967773, "p90": 293.35289306640624, "max": 341.8144226074219, "pos_frac": 0.796875, "sample": [5.330324172973633, 25.150575637817383, 179.89492797851562, 300.4324951171875, 34.17145538330078, -79.91480255126953, 7.038053512573242, 142.8242950439453, 294.84271240234375, 62.336669921875, 108.35993957519531, 308.66522216796875, 319.6373291015625, -188.82984924316406, 48.48042297363281, -12.935821533203125, 67.59325408935547, -17.245391845703125, -321.46295166015625, 164.19515991210938, 220.00840759277344, 182.29376220703125, 212.86453247070312, 331.48175048828125, 107.6789779663086, 181.1480255126953, 202.97689819335938, 2.7946949005126953, -55.86634063720703, -202.60073852539062, 273.73468017578125, -449.65435791015625, 325.801025390625, -1.5207939147949219, 5.706764221191406, -100.58911895751953, 200.2967529296875, 220.06463623046875, -131.24937438964844, 189.80319213867188, 26.88414764404297, 6.177947998046875, 217.558837890625, 289.87664794921875, 169.45115661621094, -0.6369743347167969, 33.65502166748047, 102.31480407714844, 19.31926727294922, -1.8949832916259766, 212.173095703125, 34.03699493408203, 152.41354370117188, 75.26290893554688, 341.8144226074219, 169.66229248046875, 0.384857177734375, 287.3464050292969, 199.17947387695312, 207.0755157470703, 204.9633331298828, 107.48548126220703, 64.08735656738281, 34.826377868652344], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000468.npy"}
|
||||
{"epoch": 0.7074829931972789, "step": 469, "batch_size": 64, "mean": 87.96936798095703, "std": 123.8318099975586, "min": -161.46316528320312, "p10": -36.77486228942871, "median": 47.37755584716797, "p90": 226.25650177001953, "max": 590.843994140625, "pos_frac": 0.765625, "sample": [5.169830322265625, 205.9712677001953, 27.204666137695312, 176.83575439453125, 46.120521545410156, 0.3563575744628906, 154.52223205566406, 232.61033630371094, 32.6132926940918, -161.46316528320312, 201.55728149414062, 111.53269958496094, 313.2508544921875, 10.889961242675781, -15.308906555175781, 216.31890869140625, 141.26397705078125, 48.63459014892578, 106.2994155883789, 181.501708984375, 224.38491821289062, 12.120773315429688, 85.57047271728516, -19.74908447265625, 53.43560791015625, -99.71369171142578, 63.29395294189453, -0.049774169921875, 3.374349594116211, 42.79820251464844, -3.5932159423828125, 0.8411235809326172, 105.0101318359375, 227.05860900878906, 192.7593231201172, 27.49505615234375, -2.8541183471679688, -64.45284271240234, 182.79701232910156, 198.3048553466797, 243.475341796875, 95.41516876220703, 28.631458282470703, 210.491455078125, -35.60601043701172, 12.94476318359375, 131.16668701171875, -0.1149444580078125, 175.908935546875, 28.473526000976562, -118.46640014648438, -37.27579879760742, 296.64599609375, 20.4578857421875, 590.843994140625, -2.8842086791992188, 151.776123046875, 41.83850860595703, 185.4126434326172, -37.677696228027344, 191.89129638671875, 16.531522750854492, -63.721771240234375, 239.168212890625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000469.npy"}
|
||||
{"epoch": 0.708994708994709, "step": 470, "batch_size": 64, "mean": 93.8465805053711, "std": 137.52102661132812, "min": -242.1945343017578, "p10": -74.32852706909178, "median": 86.3245620727539, "p90": 256.3618591308594, "max": 430.76300048828125, "pos_frac": 0.765625, "sample": [302.55780029296875, 90.54048919677734, 324.3320617675781, -242.1945343017578, 65.77632141113281, 290.0839538574219, 67.32548522949219, 32.18341827392578, 220.26589965820312, 111.7099609375, 228.79681396484375, -90.958251953125, 229.68783569335938, 19.689376831054688, 272.4199523925781, 49.05738067626953, 82.10863494873047, -145.13473510742188, -50.430145263671875, 182.11216735839844, -52.587833404541016, 285.9170227050781, 9.503763198852539, -191.265869140625, 232.57180786132812, -7.731199264526367, 228.05621337890625, 195.11207580566406, 1.8113861083984375, 12.957151412963867, 161.4359588623047, 4.206298828125, 158.79934692382812, 90.82151794433594, -55.39190673828125, 26.572349548339844, 158.18844604492188, 0.7008266448974609, 190.12539672851562, -10.640274047851562, 430.76300048828125, 227.8183135986328, 251.2451171875, 0.367095947265625, -82.44422149658203, -48.76158142089844, 28.621675491333008, 233.15261840820312, 211.0750732421875, 98.92626190185547, 179.16477966308594, 258.55474853515625, -87.46147918701172, 176.23941040039062, 22.277206420898438, 222.05368041992188, 139.71505737304688, 127.20669555664062, -10.339706420898438, -104.33121490478516, 236.50146484375, 26.87276268005371, -29.274356842041016, 19.14654541015625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000470.npy"}
|
||||
{"epoch": 0.7105064247921391, "step": 471, "batch_size": 64, "mean": 101.13079833984375, "std": 135.23062133789062, "min": -306.0777587890625, "p10": -22.47076969146728, "median": 102.11898040771484, "p90": 225.8768508911133, "max": 523.9156494140625, "pos_frac": 0.8125, "sample": [0.6389923095703125, 308.3349609375, 167.4341583251953, 185.52870178222656, 101.14364624023438, -124.04354858398438, 193.39523315429688, 63.99272155761719, 222.82920837402344, 300.2730712890625, 78.53072357177734, 123.50971984863281, 206.24002075195312, 53.842567443847656, -131.91644287109375, 2.749612808227539, 114.30850982666016, 227.1829833984375, 208.17538452148438, -25.6455078125, 42.568302154541016, 28.990848541259766, 215.4490966796875, 60.42774963378906, 215.64755249023438, -260.88995361328125, 73.52652740478516, 181.478271484375, 205.5601348876953, 45.824066162109375, 200.52163696289062, 146.9418487548828, 40.61069869995117, 49.972869873046875, 158.03155517578125, 132.457763671875, 207.63803100585938, -142.7285919189453, 54.66657257080078, 200.09881591796875, 40.332435607910156, 91.10641479492188, 205.24839782714844, 268.7306823730469, -15.063047409057617, 5.7020721435546875, -11.745162963867188, 180.03807067871094, 16.549942016601562, 156.2680206298828, -44.13915252685547, 229.79641723632812, 132.826904296875, 57.044044494628906, 213.6258544921875, -9.393970489501953, -2.09619140625, 149.2727508544922, -306.0777587890625, 523.9156494140625, 53.64720153808594, 103.09431457519531, -8.477325439453125, 308.86627197265625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000471.npy"}
|
||||
{"epoch": 0.7120181405895691, "step": 472, "batch_size": 64, "mean": 79.50518798828125, "std": 160.8025360107422, "min": -334.67755126953125, "p10": -97.50502929687498, "median": 68.3372802734375, "p90": 249.31105957031252, "max": 660.9735107421875, "pos_frac": 0.703125, "sample": [-2.6183242797851562, 46.21350860595703, 135.91329956054688, -77.25834655761719, 218.55406188964844, 182.56195068359375, -77.49380493164062, 104.37455749511719, 287.0736999511719, 218.27981567382812, 95.07290649414062, 407.75860595703125, 12.17823600769043, -106.08126831054688, -7.826543807983398, 209.30252075195312, -13.369333267211914, 40.94426727294922, 206.0090789794922, 101.46211242675781, -158.7220458984375, 39.550140380859375, -32.41633987426758, 262.21978759765625, 16.05889892578125, 261.7965393066406, 101.83607482910156, 241.92111206054688, 211.93057250976562, -3.6519432067871094, 37.569480895996094, 94.21336364746094, 137.2412567138672, -176.8955535888672, -63.321327209472656, 82.5128173828125, 282.27252197265625, 660.9735107421875, -15.857704162597656, -21.086997985839844, 220.51345825195312, 216.76710510253906, -278.459228515625, 252.47817993164062, -334.67755126953125, -55.739662170410156, 211.75408935546875, 15.554239273071289, -170.59957885742188, 211.30050659179688, 16.0325927734375, 54.1617431640625, -17.494796752929688, 43.876564025878906, 33.36854553222656, 120.7935562133789, 91.36810302734375, 46.18731689453125, 212.43057250976562, 33.981201171875, 121.91603088378906, 132.52706909179688, -202.01914978027344, 173.11598205566406], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000472.npy"}
|
||||
{"epoch": 0.7135298563869993, "step": 473, "batch_size": 64, "mean": 111.99749755859375, "std": 179.8339080810547, "min": -234.53945922851562, "p10": -107.26325378417968, "median": 73.41994094848633, "p90": 344.14010925292973, "max": 621.878173828125, "pos_frac": 0.765625, "sample": [336.5162658691406, 241.08334350585938, 71.02894592285156, -168.93746948242188, -147.33091735839844, 227.03004455566406, 89.88963317871094, 2.6513519287109375, 41.837013244628906, 267.36279296875, 197.0750732421875, 70.12542724609375, -8.175912857055664, 250.04014587402344, 10.947052001953125, -31.43065643310547, 425.8277587890625, 352.9228515625, 21.600914001464844, 253.37603759765625, -103.90982055664062, 14.083702087402344, 322.18951416015625, -92.32844543457031, -112.99411010742188, 193.1156768798828, 188.68626403808594, 482.8482360839844, 6.961137771606445, -75.74785614013672, 6.71051025390625, 11.807769775390625, -176.044921875, 5.425804138183594, -11.052412033081055, 102.26361846923828, 544.3004150390625, 332.9653625488281, 50.33543014526367, 36.857017517089844, 405.6905822753906, 221.8197021484375, 621.878173828125, 74.65585327148438, 144.16029357910156, 79.60055541992188, 72.18402862548828, 9.300201416015625, 83.95890045166016, 203.73390197753906, 202.50457763671875, -9.209739685058594, 143.533447265625, 155.72650146484375, -17.053810119628906, 202.02528381347656, 347.407470703125, -175.73191833496094, 18.771575927734375, -108.700439453125, 68.7083969116211, 162.0702667236328, 265.4332580566406, -234.53945922851562], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000473.npy"}
|
||||
{"epoch": 0.7150415721844293, "step": 474, "batch_size": 64, "mean": 87.50584411621094, "std": 151.22195434570312, "min": -223.0008544921875, "p10": -80.35122756958008, "median": 62.9516544342041, "p90": 270.478744506836, "max": 514.7073974609375, "pos_frac": 0.71875, "sample": [254.583984375, 195.26382446289062, 161.3708953857422, -78.64788818359375, -17.10831069946289, 35.75332260131836, 376.3478698730469, 255.83905029296875, 43.34675979614258, 71.8624267578125, 112.97720336914062, 66.22221374511719, -223.0008544921875, 180.92669677734375, 154.15884399414062, 246.4390411376953, -114.03475189208984, 0.3293266296386719, -54.047950744628906, 31.304443359375, -81.08123016357422, -56.418495178222656, 203.93019104003906, 39.515045166015625, 183.46527099609375, 47.9151611328125, 0.1880168914794922, -3.7553558349609375, -28.884315490722656, 77.49596405029297, 222.61227416992188, 59.681095123291016, -185.57037353515625, 21.091064453125, -56.55917739868164, 236.8496856689453, 72.4678955078125, 293.0556640625, 276.7528991699219, -3.9469528198242188, 20.530736923217773, 84.47457885742188, 338.37310791015625, 182.87571716308594, -95.76658630371094, 340.79583740234375, 514.7073974609375, 74.35566711425781, 123.76607513427734, 6.126935958862305, 200.69740295410156, 463.3028564453125, 31.975173950195312, 21.275291442871094, 66.98436737060547, -48.494266510009766, 227.391845703125, -130.39898681640625, -18.223857879638672, 137.3968048095703, 192.29795837402344, -184.8080596923828, -11.850692749023438, 43.89854049682617], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000474.npy"}
|
||||
{"epoch": 0.7165532879818595, "step": 475, "batch_size": 64, "mean": 119.0408706665039, "std": 154.34825134277344, "min": -210.5455322265625, "p10": -40.95334167480468, "median": 75.17076110839844, "p90": 336.3102935791016, "max": 551.7037353515625, "pos_frac": 0.75, "sample": [-1.9505863189697266, 20.91671371459961, -12.154518127441406, -43.64369201660156, 199.0934295654297, 241.38803100585938, 205.5575714111328, 249.90170288085938, -27.970155715942383, -85.48975372314453, 271.1095275878906, -210.5455322265625, 375.4056091308594, 551.7037353515625, 62.092018127441406, -69.94081115722656, 186.6664276123047, 236.2384796142578, 8.00967788696289, 175.23834228515625, 111.5843734741211, 38.08439636230469, -34.67585754394531, 9.342658996582031, 196.06707763671875, 2.0684242248535156, 22.01726531982422, 433.9302978515625, 194.15652465820312, -14.314506530761719, 249.12127685546875, 215.88894653320312, 259.5577392578125, 240.33316040039062, 40.53369140625, 2.375011444091797, 213.44281005859375, 17.241249084472656, 172.05274963378906, 43.626155853271484, 342.63330078125, 197.80206298828125, 20.13034439086914, 74.01411437988281, 195.87294006347656, -22.708431243896484, 28.60280990600586, 76.32740783691406, 424.63592529296875, 61.967201232910156, 211.7663116455078, 356.5523986816406, 367.42108154296875, 133.41775512695312, 201.32391357421875, -125.79662322998047, -45.771888732910156, 321.5566101074219, -25.295928955078125, 198.98794555664062, -1.3321723937988281, -18.758995056152344, -111.88276672363281, 13.090938568115234], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000475.npy"}
|
||||
{"epoch": 0.7180650037792895, "step": 476, "batch_size": 64, "mean": 70.36421203613281, "std": 152.1599884033203, "min": -267.0314636230469, "p10": -101.55504989624023, "median": 67.61914825439453, "p90": 243.99602050781252, "max": 494.9729919433594, "pos_frac": 0.609375, "sample": [270.2314453125, 67.93565368652344, -20.07123374938965, 67.30264282226562, -47.46022033691406, 42.198272705078125, -72.26847839355469, 204.27676391601562, 494.9729919433594, 84.17359924316406, 419.3973083496094, -5.183986663818359, 174.79513549804688, 199.22406005859375, 164.88026428222656, 84.98485565185547, 423.0945739746094, -20.42443084716797, 18.77233123779297, -33.5260009765625, -121.06041717529297, 4.308345794677734, 220.42405700683594, 190.8349151611328, 42.38386535644531, -156.77328491210938, 84.05957794189453, -95.2643051147461, -90.52694702148438, -0.9528083801269531, 239.3784942626953, -71.32826232910156, 185.35011291503906, -28.29641342163086, 77.4390640258789, -94.32066345214844, -157.2041778564453, -181.58221435546875, 122.49531555175781, -5.8182830810546875, -104.25108337402344, 245.97496032714844, 13.802764892578125, -113.12893676757812, 176.28781127929688, 106.14527893066406, 177.9142608642578, -33.35327911376953, -88.87063598632812, 137.80416870117188, 139.18917846679688, -267.0314636230469, 250.18243408203125, -50.089874267578125, 111.860595703125, 331.1501159667969, 16.176836013793945, 87.1093521118164, 203.51025390625, 112.25875854492188, -19.782655715942383, 222.03550720214844, 197.9380340576172, -30.374330520629883], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000476.npy"}
|
||||
{"epoch": 0.7195767195767195, "step": 477, "batch_size": 64, "mean": 88.23861694335938, "std": 188.51358032226562, "min": -411.6215515136719, "p10": -145.45727691650387, "median": 57.43521499633789, "p90": 275.6151123046875, "max": 608.974365234375, "pos_frac": 0.734375, "sample": [-189.8316192626953, -101.55899047851562, 133.971923828125, 38.01325988769531, -66.9873046875, 208.10809326171875, 31.980323791503906, 41.448707580566406, 123.8746337890625, 10.0059814453125, 276.10546875, 51.42339324951172, -164.2708282470703, -7.904014587402344, 22.752504348754883, 14.416301727294922, -41.68623352050781, 379.52447509765625, 195.7368927001953, -100.12548828125, 405.9752197265625, 198.9484405517578, -4.906890869140625, 48.43022918701172, 98.48959350585938, -95.40068817138672, -225.36627197265625, 7.158683776855469, 233.31854248046875, 10.146675109863281, -276.89263916015625, 7.299705505371094, 87.98590087890625, 72.75653076171875, -250.3367919921875, -19.280193328857422, -100.46177673339844, 56.994483947753906, 246.84596252441406, 35.269317626953125, 262.06072998046875, -2.7672576904296875, 274.470947265625, 258.44171142578125, 75.62667846679688, 257.3768005371094, -220.60885620117188, 474.1655578613281, 184.36911010742188, 57.875946044921875, 457.4334716796875, 171.6148223876953, 52.10096740722656, 245.32887268066406, 307.78131103515625, 198.98365783691406, -411.6215515136719, 608.974365234375, 249.79481506347656, 246.22512817382812, 121.6043930053711, 200.862060546875, 3.6781005859375, 181.5285186767578], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000477.npy"}
|
||||
{"epoch": 0.7210884353741497, "step": 478, "batch_size": 64, "mean": 62.91624450683594, "std": 137.67335510253906, "min": -247.74734497070312, "p10": -146.16981430053707, "median": 52.0715446472168, "p90": 223.64567718505862, "max": 338.65960693359375, "pos_frac": 0.6875, "sample": [-17.03498649597168, 189.17555236816406, 253.89599609375, 184.82095336914062, 170.01065063476562, -18.28331756591797, 231.31887817382812, 171.68927001953125, -41.171226501464844, 197.74710083007812, 19.58399200439453, 18.478485107421875, 132.53683471679688, 297.8707275390625, -38.24694061279297, 26.259124755859375, -206.4591827392578, 79.91818237304688, -37.61407470703125, 134.224365234375, -49.93787384033203, -11.563539505004883, 200.7503662109375, 217.2191619873047, 207.10623168945312, -111.26847076416016, 198.944091796875, 176.50067138671875, 110.96511840820312, -177.21685791015625, 8.904922485351562, 8.051410675048828, -247.74734497070312, -188.82635498046875, 32.8139533996582, 47.66673278808594, 226.0555877685547, 54.08953857421875, 186.29336547851562, 62.498329162597656, 218.02255249023438, -14.926218032836914, -161.12753295898438, 50.053550720214844, 214.76553344726562, 31.577285766601562, -15.133697509765625, 338.65960693359375, 65.05072021484375, -19.008525848388672, 281.64007568359375, 84.38642120361328, -76.1544189453125, 189.52569580078125, 19.165420532226562, 274.0237121582031, 125.91938781738281, -93.79907989501953, 10.422340393066406, 74.53782653808594, -175.93484497070312, -186.50440979003906, 71.64476776123047, 19.813753128051758], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000478.npy"}
|
||||
{"epoch": 0.7226001511715797, "step": 479, "batch_size": 64, "mean": 57.79898452758789, "std": 144.48826599121094, "min": -348.2002258300781, "p10": -101.53748092651367, "median": 43.909372329711914, "p90": 233.07729949951172, "max": 550.1000366210938, "pos_frac": 0.65625, "sample": [7.963283538818359, 96.63554382324219, 234.68408203125, -215.6887969970703, -120.51809692382812, 136.00125122070312, 282.08502197265625, 149.18692016601562, 144.3421630859375, -98.82013702392578, 42.80227279663086, -102.70205688476562, 7.70770263671875, -28.78864860534668, -9.631462097167969, 98.72616577148438, 337.6241455078125, 250.59097290039062, -18.604673385620117, -6.719707489013672, -73.23670959472656, 26.342243194580078, 550.1000366210938, 237.42872619628906, 30.58403968811035, -174.6499786376953, -22.621185302734375, 157.94720458984375, -173.13809204101562, -21.65062713623047, 15.226425170898438, 111.7845230102539, -10.5006103515625, 148.40943908691406, -16.937286376953125, 86.74200439453125, -194.26214599609375, -6.690032958984375, 40.5206298828125, 190.78379821777344, 161.9243621826172, -348.2002258300781, -72.92803192138672, 102.48828125, 229.32814025878906, 183.65463256835938, 88.8822250366211, 55.08221435546875, 1.1241912841796875, 82.67212677001953, 51.220489501953125, 52.69922637939453, -95.95376586914062, 306.2224426269531, -20.888124465942383, 121.40638732910156, 152.69798278808594, 14.22673225402832, 45.01647186279297, 163.82196044921875, 6.054716110229492, -11.879852294921875, 154.27113342285156, 187.13311767578125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000479.npy"}
|
||||
{"epoch": 0.7241118669690099, "step": 480, "batch_size": 64, "mean": 88.87548828125, "std": 145.72544860839844, "min": -253.57598876953125, "p10": -107.44057617187498, "median": 93.53521728515625, "p90": 239.76333312988282, "max": 551.48046875, "pos_frac": 0.734375, "sample": [113.1575698852539, 4.72705078125, 213.55722045898438, -177.54833984375, 139.50743103027344, 58.094703674316406, 175.7178497314453, -253.57598876953125, 9.6693115234375, 551.48046875, -25.840492248535156, 130.04495239257812, 238.21014404296875, -27.65447235107422, 176.97195434570312, -81.68522644042969, -202.37557983398438, 236.63218688964844, -89.7469482421875, -9.31149673461914, -134.15243530273438, 223.37741088867188, -5.526847839355469, -137.91098022460938, 240.42898559570312, 179.62892150878906, 470.92559814453125, 102.755126953125, 241.8488311767578, 56.48265075683594, 134.297119140625, 2.9429244995117188, 86.77497100830078, 185.1573944091797, 50.74040222167969, 113.90792846679688, 255.1299591064453, 88.0689697265625, -154.22244262695312, 164.850341796875, 184.51876831054688, 141.07421875, 72.26787567138672, 201.88978576660156, 136.27537536621094, 179.9740753173828, -16.620468139648438, 84.27783203125, 26.037899017333984, 36.7368049621582, -16.269262313842773, -115.0235595703125, 256.865234375, 42.37907409667969, 200.7049560546875, 132.35809326171875, 212.20834350585938, 184.6807861328125, -21.200061798095703, -31.26958465576172, 55.782257080078125, 264.9723815917969, 30.872161865234375, 99.00146484375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000480.npy"}
|
||||
{"epoch": 0.7256235827664399, "step": 481, "batch_size": 64, "mean": 80.409912109375, "std": 143.77056884765625, "min": -213.76608276367188, "p10": -78.39293899536132, "median": 68.02350234985352, "p90": 223.57691345214843, "max": 622.6358032226562, "pos_frac": 0.71875, "sample": [96.9719467163086, -35.06598663330078, 223.83837890625, 24.32225799560547, -139.73219299316406, 148.5874481201172, 201.57891845703125, 55.16177749633789, -53.048133850097656, -163.44346618652344, 94.024658203125, 125.86927795410156, 150.3091583251953, 69.80400848388672, 41.551979064941406, 76.23294067382812, 195.50091552734375, 79.12519836425781, 173.69207763671875, 110.69136810302734, -206.2415008544922, -0.8595809936523438, 327.77337646484375, 66.24299621582031, -12.511547088623047, -52.237815856933594, 22.202468872070312, 196.83065795898438, 0.2982597351074219, 75.67977905273438, 4.651460647583008, 219.59976196289062, -33.85548400878906, 150.69927978515625, 33.05104064941406, -111.68254089355469, 234.81204223632812, 19.938892364501953, 13.12155532836914, 19.525238037109375, 195.78683471679688, 15.977928161621094, 180.13526916503906, 222.96682739257812, -0.562347412109375, 246.51943969726562, -14.632293701171875, -80.87799835205078, 238.43930053710938, 204.6790771484375, 205.3997802734375, 33.62699890136719, 137.12203979492188, -18.045143127441406, -72.59446716308594, 622.6358032226562, -213.76608276367188, 158.33407592773438, 148.91799926757812, 459.7128601074219, 114.80867767333984, -4.502777099609375, -98.202880859375, 21.345104217529297], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000481.npy"}
|
||||
{"epoch": 0.72713529856387, "step": 482, "batch_size": 64, "mean": 97.53868103027344, "std": 185.36927795410156, "min": -413.64349365234375, "p10": -92.94067916870115, "median": 81.9638442993164, "p90": 247.30429992675784, "max": 706.9136962890625, "pos_frac": 0.71875, "sample": [45.51661682128906, -413.64349365234375, -30.31444549560547, 167.57077026367188, 6.416313171386719, 138.89324951171875, 106.59361267089844, -49.69762420654297, 19.736019134521484, 301.9505615234375, 149.2724609375, 221.43106079101562, 204.22113037109375, 77.37512969970703, 179.73757934570312, 203.63992309570312, 509.31597900390625, -24.759010314941406, -205.95741271972656, 86.55255889892578, 248.51690673828125, 73.36721801757812, 638.8231201171875, -43.01252746582031, -11.830711364746094, 146.76513671875, 224.9781494140625, 118.02040100097656, -40.93379211425781, 29.130542755126953, 49.731605529785156, 192.2476806640625, -105.66741943359375, 131.81793212890625, 244.47488403320312, 186.3402099609375, 87.79882049560547, 47.30344009399414, -17.993122100830078, -168.60574340820312, 706.9136962890625, 192.00942993164062, 47.224403381347656, 55.750274658203125, -0.4389190673828125, 213.04714965820312, 37.09754180908203, 6.624689102172852, 496.472900390625, -115.21025085449219, 134.78573608398438, 130.1525115966797, 146.6334228515625, 66.36439514160156, 193.7235107421875, 1.5687274932861328, -75.52820587158203, 405.835693359375, -79.4419937133789, -177.0515899658203, 165.8483428955078, -60.451560974121094, -98.725830078125, 124.14774322509766], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000482.npy"}
|
||||
{"epoch": 0.7286470143613001, "step": 483, "batch_size": 64, "mean": 107.86482238769531, "std": 145.6510009765625, "min": -229.44918823242188, "p10": -35.291317749023435, "median": 82.00151062011719, "p90": 262.24139404296875, "max": 498.9541320800781, "pos_frac": 0.796875, "sample": [50.433982849121094, 249.16346740722656, 8.414966583251953, 156.0414276123047, 156.95022583007812, 162.8666534423828, 196.0140838623047, -181.0732421875, 4.530426025390625, -26.152591705322266, 53.40856170654297, 207.24551391601562, 245.43777465820312, 263.37518310546875, -104.43719482421875, 191.34359741210938, 103.13859558105469, -129.14437866210938, 4.06048583984375, 189.03736877441406, 162.59715270996094, 218.39840698242188, 331.6548156738281, 160.2738800048828, 259.59588623046875, 45.031455993652344, 344.95611572265625, 0.5395126342773438, 60.86442565917969, 232.73214721679688, 207.11727905273438, -229.44918823242188, 240.50692749023438, 12.823881149291992, 31.510772705078125, 498.9541320800781, -37.55992889404297, -211.24880981445312, 55.025665283203125, -24.646156311035156, 29.12220001220703, 22.288841247558594, 358.302001953125, 111.30939483642578, -0.8647079467773438, 111.87712097167969, 365.3333740234375, 330.2724609375, 26.102134704589844, 193.0198974609375, 56.483802795410156, 150.1162872314453, 21.381324768066406, 249.0020751953125, -36.292762756347656, -12.235017776489258, 58.141849517822266, 10.170831680297852, 256.5406799316406, 187.80624389648438, 40.82780456542969, -32.954612731933594, -3.4314231872558594, 250.69488525390625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000483.npy"}
|
||||
{"epoch": 0.7301587301587301, "step": 484, "batch_size": 64, "mean": 66.19149780273438, "std": 109.76258087158203, "min": -232.96664428710938, "p10": -80.51574172973632, "median": 52.496726989746094, "p90": 222.02778167724617, "max": 311.88787841796875, "pos_frac": 0.796875, "sample": [45.96862030029297, 2.574859619140625, 6.0934295654296875, -18.74127197265625, 243.10476684570312, 5.202768325805664, 135.52796936035156, 34.50209045410156, 101.1925048828125, 34.95647430419922, 59.74016571044922, -86.88941192626953, 73.4112548828125, 16.92977523803711, 105.17376708984375, 119.15437316894531, -110.03300476074219, -43.60909652709961, 117.97761535644531, 176.71600341796875, 11.289602279663086, 270.0240478515625, 147.9839324951172, 4.541053771972656, 172.5199432373047, 150.3211669921875, 228.88233947753906, 38.9833984375, 131.13975524902344, 29.662019729614258, 51.400142669677734, 17.132102966308594, -11.695388793945312, -65.64384460449219, 287.1050720214844, -14.293842315673828, 46.077606201171875, 5.0648193359375, 108.68574523925781, -11.773513793945312, 158.00091552734375, -115.55553436279297, 288.3878479003906, -232.96664428710938, 178.15931701660156, 89.42223358154297, 206.0338134765625, -90.95411682128906, 76.09822082519531, 6.8302001953125, 311.88787841796875, 14.583904266357422, 121.67718505859375, 107.65242004394531, -103.17133331298828, -149.26840209960938, 69.205810546875, 241.10610961914062, 102.84002685546875, 143.601318359375, 39.550048828125, 78.42001342773438, 53.59331130981445, 24.761245727539062], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000484.npy"}
|
||||
{"epoch": 0.7316704459561603, "step": 485, "batch_size": 64, "mean": 74.10181427001953, "std": 128.699951171875, "min": -235.00900268554688, "p10": -46.13805313110351, "median": 51.892024993896484, "p90": 221.59051818847658, "max": 482.9898681640625, "pos_frac": 0.734375, "sample": [3.3431015014648438, 155.43472290039062, 86.52729797363281, 183.49012756347656, 121.21989440917969, -24.00354766845703, -2.1905040740966797, 4.123403549194336, 150.67576599121094, -61.073036193847656, -165.5449981689453, 46.91004943847656, 211.42788696289062, 262.1902770996094, 15.396484375, 328.44207763671875, 23.092098236083984, 187.6037139892578, 280.3576354980469, -0.6221942901611328, 26.589269638061523, 288.1612548828125, -31.251327514648438, 393.957763671875, -6.661186218261719, 3.1284141540527344, -235.00900268554688, 87.57499694824219, 74.58185577392578, 66.45770263671875, -38.97962951660156, -6.422145843505859, 112.52913665771484, 1.8141498565673828, 167.9592742919922, -128.5992889404297, 482.9898681640625, 223.24673461914062, 42.37211608886719, 208.93125915527344, 27.365314483642578, -39.83142852783203, 62.214202880859375, 13.347618103027344, 107.31465911865234, 2.1263561248779297, -11.249853134155273, 212.78355407714844, 119.88932037353516, -46.49784851074219, 23.007747650146484, 10.936243057250977, 118.33980560302734, 204.2775421142578, 105.5880355834961, 107.29925537109375, 56.874000549316406, 217.72601318359375, -45.29853057861328, -70.48172760009766, 67.38124084472656, 15.227886199951172, 60.53858947753906, -116.533447265625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000485.npy"}
|
||||
{"epoch": 0.7331821617535903, "step": 486, "batch_size": 64, "mean": 77.09710693359375, "std": 172.5797882080078, "min": -290.024658203125, "p10": -136.31027221679685, "median": 69.98495101928711, "p90": 306.750814819336, "max": 601.156005859375, "pos_frac": 0.6875, "sample": [3.5585250854492188, -42.45805358886719, 486.09326171875, 4.108636856079102, -85.81533813476562, -5.866847991943359, 77.95246887207031, 601.156005859375, 23.063194274902344, 15.673690795898438, -8.150957107543945, 294.80029296875, 151.59457397460938, 60.59175109863281, 73.59110260009766, 82.85911560058594, -56.937286376953125, 234.067138671875, -1.7842941284179688, 319.1787414550781, 155.13893127441406, 86.21061706542969, -91.66696166992188, -194.53872680664062, 270.8671569824219, -207.91371154785156, 75.0638427734375, -62.76462936401367, 36.498085021972656, -290.024658203125, 311.8724670410156, 86.12997436523438, 148.71389770507812, -148.22760009765625, 0.21549415588378906, 183.7144775390625, 82.84139251708984, 176.90599060058594, -163.0569610595703, 91.45677185058594, 3.788928985595703, 195.97500610351562, 149.83221435546875, 209.0491943359375, 213.7334747314453, -77.49207305908203, 339.04583740234375, 201.03741455078125, 105.55560302734375, 203.249755859375, 287.3573913574219, -108.503173828125, 49.59501647949219, 313.834228515625, 0.7359085083007812, -226.4736785888672, -4.525798797607422, 132.4497528076172, 393.412841796875, -40.16888427734375, 35.176513671875, -217.73904418945312, 66.37879943847656, -65.8016128540039], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000486.npy"}
|
||||
{"epoch": 0.7346938775510204, "step": 487, "batch_size": 64, "mean": 88.57109069824219, "std": 168.46060180664062, "min": -252.26882934570312, "p10": -82.06388778686522, "median": 66.44516563415527, "p90": 254.34713439941413, "max": 791.07470703125, "pos_frac": 0.71875, "sample": [217.6291046142578, -154.15724182128906, 236.06063842773438, 141.94186401367188, 42.215919494628906, 11.058111190795898, -88.34547424316406, 29.623329162597656, 78.68340301513672, 191.8487091064453, -252.26882934570312, -39.29545593261719, 76.0769271850586, -102.5267333984375, -33.083030700683594, 196.5455322265625, 73.76264953613281, 26.835453033447266, 613.4612426757812, 87.83863830566406, -3.2298812866210938, 356.16375732421875, 112.27024841308594, 228.51214599609375, 173.29708862304688, -31.63192367553711, 0.5056228637695312, -168.1190185546875, 28.544227600097656, 157.25521850585938, 142.14053344726562, 101.05225372314453, 142.78335571289062, -1.5914440155029297, 223.39501953125, -67.40685272216797, 152.3091278076172, -51.77679443359375, 51.13063049316406, 101.797119140625, 307.55810546875, 11.457828521728516, 146.43710327148438, 268.56512451171875, -156.6473388671875, 192.66127014160156, 5.956443786621094, -46.696807861328125, 96.3722915649414, -191.9136199951172, -4.378854751586914, 205.6563720703125, 56.82917022705078, -3.388336181640625, 12.985458374023438, 262.1842041015625, 59.127681732177734, 16.8272705078125, -36.1046142578125, 285.07098388671875, 153.59188842773438, 46.95203399658203, 791.07470703125, 187.06594848632812], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000487.npy"}
|
||||
{"epoch": 0.7362055933484505, "step": 488, "batch_size": 64, "mean": 100.19713592529297, "std": 146.33941650390625, "min": -186.4898681640625, "p10": -41.40212860107422, "median": 90.94294738769531, "p90": 280.1284057617189, "max": 612.7838134765625, "pos_frac": 0.71875, "sample": [40.285736083984375, 27.80925750732422, -21.6909122467041, -17.858007431030273, 95.05367279052734, -121.85687255859375, -36.97217559814453, 227.93365478515625, 387.7958984375, 126.84022521972656, 2.2062816619873047, 15.194854736328125, 223.00401306152344, 36.26373291015625, 30.56780242919922, 297.5065002441406, -60.28895568847656, 193.8881378173828, 2.7751083374023438, 79.66812133789062, 182.84027099609375, 94.43441009521484, 22.07306671142578, -53.05552673339844, -13.262001037597656, 146.65234375, -138.8377685546875, 215.68890380859375, -38.057464599609375, 223.54473876953125, -29.783615112304688, -42.83555603027344, -36.693199157714844, 184.37098693847656, 295.764404296875, 153.073974609375, 199.5617218017578, -10.714149475097656, 314.1763916015625, 448.23724365234375, 352.943603515625, 207.84173583984375, 210.2521514892578, -25.700408935546875, 199.53610229492188, 130.72061157226562, -13.626724243164062, 144.7362518310547, -117.97087097167969, 133.48284912109375, 89.3751220703125, 130.76815795898438, -186.4898681640625, 92.51077270507812, -7.453369140625, 94.29480743408203, 612.7838134765625, 47.431575775146484, 230.38304138183594, 57.97097396850586, 33.55479431152344, 101.81427001953125, 4.507606506347656, 243.6444091796875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000488.npy"}
|
||||
{"epoch": 0.7377173091458806, "step": 489, "batch_size": 64, "mean": 57.72705841064453, "std": 151.690185546875, "min": -291.80059814453125, "p10": -159.72070922851563, "median": 43.3091926574707, "p90": 239.98035430908203, "max": 392.01922607421875, "pos_frac": 0.671875, "sample": [-192.4093780517578, 173.15524291992188, 58.82484436035156, 240.48016357421875, 295.946533203125, 192.12020874023438, -116.06158447265625, 205.99658203125, 65.75898742675781, 348.3724365234375, 98.05651092529297, 18.034812927246094, 208.63711547851562, -88.77410888671875, -12.56201171875, -248.90476989746094, -49.06391906738281, -42.2181282043457, 88.1998291015625, 7.883119583129883, 238.8141326904297, 75.13818359375, 182.8535614013672, 50.0324821472168, 172.75946044921875, 188.81480407714844, 2.8923873901367188, 2.7178726196289062, -158.72616577148438, 52.98908615112305, -172.76979064941406, -40.199153900146484, 244.2626953125, 26.77220916748047, 36.718170166015625, -45.22630310058594, -58.305259704589844, -7.616724014282227, 207.46041870117188, 163.52127075195312, 72.09538269042969, 265.7515563964844, 8.693794250488281, 392.01922607421875, 146.52151489257812, -190.7400360107422, 230.19869995117188, 42.312217712402344, 137.14901733398438, 238.4662322998047, -239.23410034179688, 43.47002410888672, -39.124385833740234, 18.362119674682617, -18.25995635986328, -291.80059814453125, 285.1171569824219, 1.2908248901367188, -160.14694213867188, 188.57037353515625, -89.25824737548828, 196.8260498046875, -1.272409439086914, 43.14836120605469], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000489.npy"}
|
||||
{"epoch": 0.7392290249433107, "step": 490, "batch_size": 64, "mean": 73.6213150024414, "std": 171.2005157470703, "min": -392.6087646484375, "p10": -128.10580291748045, "median": 67.9537467956543, "p90": 244.939306640625, "max": 678.474609375, "pos_frac": 0.640625, "sample": [-37.61590576171875, -21.547836303710938, -392.6087646484375, -12.532920837402344, -5.425142288208008, 95.12075805664062, 97.35734558105469, 180.8194122314453, 279.3854064941406, 404.6257629394531, -184.77297973632812, 244.4651336669922, -223.4102783203125, 11.301877975463867, 21.237518310546875, 167.55252075195312, 244.59246826171875, 161.51019287109375, 415.2603759765625, 7.322242736816406, 64.60704803466797, 211.79403686523438, 162.08445739746094, 269.72991943359375, -110.46067810058594, -94.125, 214.50865173339844, -5.415889739990234, 88.82121276855469, -8.201286315917969, -186.21963500976562, 151.524658203125, 71.30044555664062, 140.45230102539062, 11.641778945922852, -50.51881408691406, -48.00183868408203, 245.08795166015625, -160.3056640625, -39.03998565673828, 164.35879516601562, -45.074462890625, 137.52011108398438, 183.21630859375, 11.894498825073242, 188.83860778808594, 103.697265625, 6.469972610473633, -40.50653839111328, -50.43194580078125, 678.474609375, 207.89752197265625, -42.89165115356445, 111.62870788574219, 330.75775146484375, 117.73243713378906, 215.45921325683594, 124.8593521118164, 24.124832153320312, 211.43572998046875, -210.1147918701172, -135.66799926757812, -2.4748992919921875, 38.65975570678711], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000490.npy"}
|
||||
{"epoch": 0.7407407407407407, "step": 491, "batch_size": 64, "mean": 71.86080932617188, "std": 133.89337158203125, "min": -289.39453125, "p10": -64.18415298461912, "median": 46.46835708618164, "p90": 221.17979583740234, "max": 570.2376708984375, "pos_frac": 0.6875, "sample": [198.82498168945312, -72.5230484008789, 152.18002319335938, 30.57958984375, 142.9031982421875, 104.1092300415039, -13.832216262817383, 175.40066528320312, 251.83602905273438, 76.54423522949219, -93.35107421875, -162.7076416015625, 214.65101623535156, 58.910362243652344, 171.64230346679688, -29.25336456298828, 30.209270477294922, 188.67633056640625, 27.199665069580078, 100.39772033691406, -103.68603515625, 353.4250183105469, 219.7742462158203, -20.869586944580078, 15.746236801147461, 194.98719787597656, 16.578330993652344, -44.72673034667969, -28.767478942871094, -9.329280853271484, 222.73789978027344, 85.65901184082031, -85.34109497070312, 47.11817169189453, 81.4974594116211, 20.450321197509766, 64.80060577392578, 570.2376708984375, 101.49636840820312, -82.40343475341797, 221.27589416503906, 45.81854248046875, 10.999847412109375, -40.49639892578125, -6.791110992431641, 74.15177917480469, -25.676513671875, 8.283828735351562, 108.5638198852539, 220.95556640625, -33.58758544921875, 178.77723693847656, 316.66375732421875, 190.81576538085938, 14.537714004516602, 73.1468505859375, -1.0249252319335938, 121.60501861572266, 9.575111389160156, -289.39453125, -12.194746017456055, 9.85362434387207, -40.088993072509766, 271.54022216796875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000491.npy"}
|
||||
{"epoch": 0.7422524565381708, "step": 492, "batch_size": 64, "mean": 69.74394226074219, "std": 167.87977600097656, "min": -405.59490966796875, "p10": -98.74490280151367, "median": 41.165245056152344, "p90": 285.0304321289063, "max": 546.9560546875, "pos_frac": 0.671875, "sample": [415.56976318359375, 546.9560546875, -94.86904907226562, 4.466423034667969, 255.98483276367188, 21.18062400817871, 132.15960693359375, 128.62429809570312, 90.46665954589844, -15.184122085571289, 287.397216796875, 224.207763671875, 204.66278076171875, -77.79390716552734, 38.2764892578125, 351.2480773925781, -206.49032592773438, 131.47267150878906, 10.597953796386719, 330.55584716796875, 154.365234375, -100.4059829711914, -29.37781524658203, -13.841178894042969, -23.432777404785156, 49.8517951965332, 134.35223388671875, 335.3010559082031, -80.7325439453125, -39.197105407714844, 197.24044799804688, 8.815996170043945, 30.46734619140625, 44.302276611328125, -5.814035415649414, 44.05400085449219, -405.59490966796875, 291.58380126953125, -256.9501953125, 19.341896057128906, 220.51882934570312, 5.955009460449219, 97.31978607177734, 17.310169219970703, -83.16950988769531, 46.466468811035156, 11.026424407958984, 54.09978485107422, -70.70914459228516, 181.0508270263672, -218.91294860839844, -68.97188568115234, -28.7103271484375, 224.3489227294922, 279.5079345703125, 47.9727897644043, 230.60671997070312, 231.22634887695312, 147.88890075683594, -15.777542114257812, 5.052301406860352, -128.88931274414062, -109.46682739257812, 254.04965209960938], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000492.npy"}
|
||||
{"epoch": 0.7437641723356009, "step": 493, "batch_size": 64, "mean": 97.55970764160156, "std": 160.24188232421875, "min": -260.3067626953125, "p10": -80.33562011718749, "median": 94.6252670288086, "p90": 272.4878265380859, "max": 801.5130615234375, "pos_frac": 0.6875, "sample": [-32.45024108886719, 10.760082244873047, -4.84002685546875, 184.4907989501953, -63.714027404785156, 113.46392822265625, -116.24868774414062, 14.90536117553711, -8.142263412475586, -92.35518646240234, 60.96490478515625, 349.4028015136719, 270.2759094238281, 236.30487060546875, 60.08561325073242, -86.50314331054688, 219.34860229492188, -2.4139633178710938, 145.8317413330078, 141.87509155273438, -7.238332748413086, 85.47188568115234, -4.210481643676758, -25.627410888671875, 42.325294494628906, -107.54454040527344, 161.79624938964844, -21.60118865966797, 238.80223083496094, 185.042724609375, 179.9443817138672, 273.435791015625, -260.3067626953125, 58.33222579956055, -123.89540100097656, 156.83050537109375, 214.26412963867188, -65.94473266601562, 300.8786315917969, 4.975885391235352, 337.2502136230469, 141.36695861816406, 101.31068420410156, 801.5130615234375, 161.81777954101562, 261.2730712890625, -47.30079650878906, 276.41510009765625, 188.98367309570312, -43.707664489746094, -29.69164276123047, 193.7819366455078, 114.16536712646484, 42.24354553222656, 278.8631591796875, 217.29739379882812, 111.91473388671875, 87.93984985351562, 117.73779296875, 211.93658447265625, 44.490272521972656, 148.55386352539062, 47.747215270996094, -208.85418701171875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000493.npy"}
|
||||
{"epoch": 0.745275888133031, "step": 494, "batch_size": 64, "mean": 90.83916473388672, "std": 182.07086181640625, "min": -382.48028564453125, "p10": -99.88109588623047, "median": 77.70990371704102, "p90": 227.86546630859377, "max": 700.4820556640625, "pos_frac": 0.671875, "sample": [180.12757873535156, 96.85970306396484, 73.52111053466797, 126.60884094238281, 230.27374267578125, 29.588058471679688, -15.76812744140625, 167.8025665283203, 165.65777587890625, -99.51937866210938, 10.778526306152344, -54.46729278564453, 2.4035491943359375, 126.43033599853516, 222.24615478515625, 148.93045043945312, 200.7354736328125, 9.688499450683594, 67.57341003417969, -236.49630737304688, 46.59939956665039, 700.4820556640625, 196.1609344482422, -45.37165832519531, 76.40210723876953, 212.58436584472656, 217.69447326660156, -30.361778259277344, 194.8870849609375, 206.68910217285156, 200.46466064453125, -241.2236328125, -250.64175415039062, 209.245849609375, -13.067316055297852, -41.77769470214844, -43.26891326904297, 194.4061279296875, 63.22156524658203, 139.21438598632812, 155.65052795410156, -6.874780654907227, -382.48028564453125, -18.597652435302734, 30.584224700927734, 49.263084411621094, 340.73455810546875, -16.28578758239746, -100.03611755371094, 127.58743286132812, 220.7298583984375, 79.0177001953125, -4.6727752685546875, 403.4814453125, 192.822509765625, 87.28766632080078, -130.47537231445312, -23.563610076904297, 312.1353759765625, 617.2596435546875, 390.9190979003906, -133.72970581054688, -33.4068717956543, 211.04238891601562], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000494.npy"}
|
||||
{"epoch": 0.7467876039304611, "step": 495, "batch_size": 64, "mean": 70.3248291015625, "std": 176.8968505859375, "min": -343.99078369140625, "p10": -107.84489364624024, "median": 21.46638774871826, "p90": 225.2054656982422, "max": 763.9939575195312, "pos_frac": 0.671875, "sample": [-77.93224334716797, -135.43258666992188, -343.99078369140625, -176.83238220214844, 227.42799377441406, -8.951881408691406, -168.7106475830078, 21.308395385742188, 109.57577514648438, -3.5157928466796875, 6.373466491699219, -162.9237060546875, -44.54386901855469, 130.5378875732422, 137.48611450195312, 225.0208740234375, -82.07744598388672, 42.6392822265625, 173.81512451171875, -21.0658016204834, 50.23868942260742, 189.61412048339844, 153.64308166503906, 4.275461196899414, 218.44308471679688, 225.28457641601562, -5.866519927978516, 108.84227752685547, 1.3659553527832031, 19.153026580810547, 763.9939575195312, 6.418603897094727, 11.875434875488281, -106.39857482910156, 21.624380111694336, 6.954994201660156, -5.139739990234375, 156.5807647705078, 448.2453308105469, 170.747314453125, -168.70553588867188, -12.501502990722656, 59.73445510864258, 78.23727416992188, 256.2417907714844, 39.94390106201172, 102.82776641845703, -11.696243286132812, 32.82659912109375, 221.16615295410156, 14.599126815795898, 198.27059936523438, 16.883262634277344, -19.510574340820312, -108.4647445678711, 645.4884643554688, 122.2764892578125, -23.719955444335938, -15.529228210449219, 164.42178344726562, 169.88414001464844, 36.29762268066406, 408.78900146484375, 4.924518585205078], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000495.npy"}
|
||||
{"epoch": 0.7482993197278912, "step": 496, "batch_size": 64, "mean": 101.1602554321289, "std": 203.36314392089844, "min": -342.6988220214844, "p10": -84.41132965087888, "median": 50.51033020019531, "p90": 281.8760955810547, "max": 1076.6444091796875, "pos_frac": 0.640625, "sample": [159.3654327392578, 278.9403381347656, 260.1926574707031, 207.8180694580078, -4.206550598144531, 154.59471130371094, 94.48358154296875, 584.2945556640625, 45.62749481201172, 160.67771911621094, -51.95063781738281, 41.252159118652344, 67.1522445678711, 48.5848388671875, -0.019578933715820312, -8.161750793457031, -9.542499542236328, 11.076112747192383, 199.95640563964844, 188.84222412109375, -68.35956573486328, 1076.6444091796875, -5.397754669189453, 230.09030151367188, 44.17045593261719, -342.6988220214844, 542.6927490234375, 283.13427734375, -2.2828102111816406, 399.66925048828125, -91.29065704345703, 18.878986358642578, 232.32237243652344, 357.0281677246094, -1.1203765869140625, 4.11833381652832, -158.90496826171875, -18.183839797973633, 180.65371704101562, 104.20661163330078, 52.435821533203125, 100.77479553222656, 343.1595153808594, 130.63055419921875, 175.5633544921875, 63.741188049316406, -30.762176513671875, 147.1818084716797, -40.15886688232422, -18.362232208251953, 271.18389892578125, -230.10275268554688, -93.05122375488281, 4.310750961303711, 188.26812744140625, 209.79629516601562, -6.887580871582031, 125.2400894165039, -100.57537841796875, -27.490997314453125, -92.36839294433594, 93.8984375, 33.527122497558594, -40.044090270996094], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000496.npy"}
|
||||
{"epoch": 0.7498110355253212, "step": 497, "batch_size": 64, "mean": 89.47763061523438, "std": 149.60430908203125, "min": -378.9009704589844, "p10": -55.72209663391113, "median": 83.36662292480469, "p90": 276.95056762695316, "max": 515.2132568359375, "pos_frac": 0.6875, "sample": [203.2236328125, 4.570274353027344, 369.9419860839844, 143.19827270507812, 515.2132568359375, 54.570770263671875, -27.068618774414062, -8.702804565429688, 34.175506591796875, 110.19477844238281, 173.85556030273438, 187.62661743164062, 152.52748107910156, -57.35680389404297, 83.74337768554688, -2.8925399780273438, 256.0692138671875, -209.93328857421875, 193.3084716796875, 270.2369079589844, -6.974250793457031, -15.203182220458984, -13.371803283691406, -5.6643218994140625, 91.06083679199219, -91.69178771972656, -29.829986572265625, 245.77230834960938, 228.64744567871094, 227.53713989257812, -7.707780838012695, 158.9105682373047, 56.750823974609375, 280.1827392578125, -253.72219848632812, 183.5244140625, 316.5181579589844, 76.18115997314453, 99.79450988769531, -45.83900451660156, 184.92007446289062, 136.36465454101562, -41.712467193603516, 69.97866821289062, 173.52354431152344, -70.4058837890625, 293.05810546875, 93.77161407470703, -19.42742919921875, 52.36735534667969, 38.63731002807617, 82.9898681640625, 35.05479431152344, 177.4842987060547, 100.92815399169922, 309.8841247558594, -378.9009704589844, 9.405685424804688, -51.907779693603516, 225.4231414794922, 47.034088134765625, 120.6478042602539, -83.75651550292969, 279.8278503417969], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000497.npy"}
|
||||
{"epoch": 0.7513227513227513, "step": 498, "batch_size": 64, "mean": 94.83919525146484, "std": 143.8771514892578, "min": -484.4048156738281, "p10": -52.44064579010008, "median": 88.3508529663086, "p90": 265.60140991210943, "max": 416.36566162109375, "pos_frac": 0.796875, "sample": [-80.84284973144531, 159.97265625, 220.5923309326172, 152.6542205810547, 163.27139282226562, -13.939849853515625, 226.94171142578125, 252.598388671875, -11.742130279541016, -179.9712371826172, 271.17413330078125, 55.40346908569336, 76.39761352539062, -61.518882751464844, 183.05288696289062, 174.5118408203125, 22.121414184570312, -31.258092880249023, 157.98785400390625, 167.5679931640625, 416.36566162109375, 26.704017639160156, 216.63784790039062, 102.94454956054688, 60.183250427246094, 123.05027770996094, -78.87438201904297, 209.25225830078125, 217.89451599121094, 93.40454864501953, 285.49639892578125, 189.94290161132812, 10.243671417236328, 32.85472106933594, -4.175376892089844, -145.71331787109375, -20.050987243652344, 89.60317993164062, 5.289312362670898, -2.020742416381836, 218.4915313720703, 215.51292419433594, -484.4048156738281, 188.9181671142578, 139.18673706054688, 15.597007751464844, 12.532909393310547, 135.57229614257812, 22.914180755615234, 87.09852600097656, 305.2370300292969, 224.3641815185547, 139.34014892578125, 18.76305389404297, 310.3365478515625, 10.598451614379883, 347.70361328125, 51.209136962890625, 17.063112258911133, 20.848358154296875, 34.37548828125, -72.8224868774414, 47.96739959716797, 329.29779052734375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000498.npy"}
|
||||
{"epoch": 0.7528344671201814, "step": 499, "batch_size": 64, "mean": 108.51744842529297, "std": 192.99427795410156, "min": -341.41650390625, "p10": -148.06957244873047, "median": 115.08434295654297, "p90": 338.5129425048829, "max": 689.6978149414062, "pos_frac": 0.765625, "sample": [160.31729125976562, 297.7610778808594, -213.31222534179688, 446.242919921875, 58.16322326660156, 51.693721771240234, 73.93952941894531, -90.84395599365234, 49.138187408447266, 192.13221740722656, 55.80419158935547, 205.53558349609375, 101.03437805175781, -163.5740509033203, 347.10693359375, 29.34038543701172, 689.6978149414062, -116.04261016845703, 232.67103576660156, 210.08358764648438, 188.79722595214844, -197.04644775390625, -84.08186340332031, -148.6147003173828, -146.797607421875, 209.10382080078125, 209.92767333984375, 298.38531494140625, -150.74893188476562, 130.87625122070312, -65.12368774414062, 49.998416900634766, 158.56065368652344, 178.86061096191406, 136.76348876953125, -70.90133666992188, 31.432418823242188, 214.13040161132812, 16.171171188354492, 95.78474426269531, 318.4602966308594, 19.60637664794922, 252.49554443359375, 18.498096466064453, -341.41650390625, 148.86166381835938, 129.13430786132812, 141.90576171875, 534.9906005859375, 377.91717529296875, 265.5057373046875, 97.51663970947266, 34.155670166015625, -102.90335083007812, -234.2074737548828, 390.3224792480469, 219.52621459960938, 222.57522583007812, -111.93011474609375, 423.08221435546875, 44.074501037597656, 201.647705078125, 201.97650146484375, 20.954330444335938], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000499.npy"}
|
||||
{"epoch": 0.7543461829176115, "step": 500, "batch_size": 64, "mean": 78.49209594726562, "std": 159.56773376464844, "min": -480.70513916015625, "p10": -120.16419677734369, "median": 112.2297134399414, "p90": 251.21710357666024, "max": 406.70880126953125, "pos_frac": 0.765625, "sample": [-38.01017761230469, 119.98688507080078, 10.320732116699219, 16.052701950073242, 140.78582763671875, 120.16093444824219, 129.17428588867188, 220.72982788085938, 117.76852416992188, 233.47059631347656, 106.69090270996094, 192.81729125976562, 228.40353393554688, -45.075050354003906, 234.41209411621094, -51.47190856933594, 197.02491760253906, 136.8612518310547, 56.317604064941406, -29.27899169921875, 258.838623046875, 200.9837188720703, 43.338966369628906, 157.2183380126953, -21.575149536132812, 258.41925048828125, 72.02407836914062, 31.966434478759766, -39.870079040527344, -184.25820922851562, -192.40066528320312, 156.55738830566406, 39.012699127197266, 126.95893859863281, -179.37515258789062, 17.18593406677246, 85.85443115234375, 406.70880126953125, 273.53631591796875, 20.794044494628906, 186.51937866210938, 27.0427303314209, 271.04571533203125, 82.00654602050781, -17.075733184814453, 184.21461486816406, 97.95582580566406, -480.70513916015625, 188.4627685546875, -257.2784118652344, 290.1146240234375, -361.1406555175781, 3.3619384765625, 190.44476318359375, 188.3592071533203, 1.2430458068847656, 24.08007049560547, 141.25123596191406, 144.43643188476562, -64.36581420898438, -144.07778930664062, 190.6124725341797, 205.38722229003906, 302.5387268066406], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000500.npy"}
|
||||
{"epoch": 0.7558578987150416, "step": 501, "batch_size": 64, "mean": 70.76509094238281, "std": 105.613037109375, "min": -210.43594360351562, "p10": -38.22996215820312, "median": 42.14980697631836, "p90": 210.05485076904301, "max": 293.78936767578125, "pos_frac": 0.703125, "sample": [-30.263771057128906, 215.0736083984375, -34.30555725097656, 123.20651245117188, 169.24839782714844, 190.22853088378906, -69.05149841308594, 87.2608871459961, 39.37925338745117, 166.7581787109375, 127.09187316894531, 30.553510665893555, 81.69648742675781, -4.9597625732421875, -22.670072555541992, -28.74562644958496, 75.0849380493164, 184.97305297851562, -27.761878967285156, -19.95993995666504, 175.84628295898438, -29.511856079101562, 2.3791160583496094, 293.78936767578125, 160.89901733398438, 185.19857788085938, 238.367919921875, 285.8642578125, 8.000690460205078, 20.221893310546875, 53.776832580566406, -210.43594360351562, -9.127159118652344, 21.008075714111328, 131.8211669921875, 214.2281494140625, 159.18382263183594, -5.328071594238281, -13.608104705810547, 103.79478454589844, 126.84573364257812, 143.38873291015625, 36.85872268676758, 195.74026489257812, -52.93675994873047, -132.25267028808594, 99.8537826538086, 270.59698486328125, -53.57181930541992, 42.85356903076172, 41.446044921875, 119.7996597290039, 31.27228546142578, -39.91184997558594, 78.40647888183594, -3.5180301666259766, 26.858428955078125, 4.015283584594727, 223.4972686767578, 200.31715393066406, 16.606796264648438, 0.3047294616699219, 172.1348114013672, -58.8458251953125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000501.npy"}
|
||||
{"epoch": 0.7573696145124716, "step": 502, "batch_size": 64, "mean": 79.5633544921875, "std": 138.3057861328125, "min": -254.92080688476562, "p10": -63.47384567260742, "median": 57.18826103210449, "p90": 247.62675476074222, "max": 423.8639221191406, "pos_frac": 0.75, "sample": [-52.830177307128906, 22.316837310791016, 21.718841552734375, -6.4966278076171875, 233.6456298828125, 115.21511840820312, 55.51409149169922, 114.97007751464844, -116.56206512451172, -57.9571533203125, 57.186336517333984, 233.72467041015625, -24.75347900390625, -132.17160034179688, 423.8639221191406, 394.66845703125, 144.15972900390625, -180.4558868408203, -65.83814239501953, 81.27488708496094, 55.749969482421875, -43.33150863647461, 23.410146713256836, 199.2177734375, 267.9285888671875, 80.70574188232422, 118.94416809082031, 8.372417449951172, 3.9867725372314453, 93.546630859375, 57.190185546875, 16.70851707458496, 52.057579040527344, 20.976171493530273, 164.16485595703125, 120.19708251953125, 213.3165283203125, 267.5816650390625, -167.23159790039062, -8.38357162475586, 48.079017639160156, 10.354637145996094, 108.95284271240234, 19.842742919921875, 84.16656494140625, -39.47747802734375, 54.350372314453125, 251.50296020507812, 89.27080535888672, -167.59695434570312, 33.614952087402344, 184.6563720703125, 197.95062255859375, 238.582275390625, 310.6766662597656, -254.92080688476562, 129.4920654296875, 227.22097778320312, -49.05738830566406, 213.17269897460938, -8.907222747802734, 131.28704833984375, 386.899169921875, 85.64002227783203], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000502.npy"}
|
||||
{"epoch": 0.7588813303099018, "step": 503, "batch_size": 64, "mean": 78.66637420654297, "std": 133.47462463378906, "min": -348.05322265625, "p10": -47.90173416137695, "median": 68.43353462219238, "p90": 231.69041748046877, "max": 357.2164306640625, "pos_frac": 0.734375, "sample": [235.89263916015625, 19.700672149658203, -35.421051025390625, 94.26178741455078, 155.42514038085938, 112.8647232055664, 357.2164306640625, -348.05322265625, -14.56976318359375, 179.23223876953125, 233.7960205078125, -63.39183044433594, 21.73358154296875, 204.05531311035156, -50.41264343261719, 63.35348892211914, -5.246198654174805, -2.63189697265625, 321.7554931640625, -34.489891052246094, 33.69530487060547, 171.92213439941406, 204.25318908691406, 37.11314392089844, -10.230964660644531, 103.79348754882812, 163.26461791992188, 236.61624145507812, -33.22455978393555, 183.26507568359375, -36.20486068725586, 114.75098419189453, 233.38819885253906, 206.1662139892578, 224.78359985351562, -42.042945861816406, 176.23870849609375, 29.404136657714844, -193.85496520996094, 15.514961242675781, 3.745250701904297, 47.32867431640625, 12.220476150512695, 151.2830352783203, -1.8876533508300781, 32.04511260986328, 73.51358032226562, 246.94766235351562, 176.53175354003906, 16.365814208984375, 213.69610595703125, 172.83071899414062, -194.63809204101562, 14.4749755859375, 172.4392547607422, 152.09463500976562, 7.995401382446289, -134.5726318359375, 184.6477508544922, 227.7289276123047, 16.621402740478516, 200.70086669921875, 175.1172332763672, -196.26504516601562], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000503.npy"}
|
||||
{"epoch": 0.7603930461073318, "step": 504, "batch_size": 64, "mean": 102.62492370605469, "std": 129.82772827148438, "min": -211.69711303710938, "p10": -23.109402465820313, "median": 84.09253692626953, "p90": 253.9478744506836, "max": 403.4865417480469, "pos_frac": 0.796875, "sample": [-14.7847900390625, 199.70037841796875, 33.73130798339844, 174.98959350585938, 328.56097412109375, -22.12939453125, 80.36050415039062, 29.986547470092773, 242.20504760742188, 33.72016906738281, 130.33425903320312, 49.82806396484375, -22.296096801757812, 276.65264892578125, 37.4017219543457, 206.6378936767578, 188.3004913330078, -143.7540283203125, 130.65328979492188, 142.58590698242188, -32.49252700805664, 86.41514587402344, 3.8031768798828125, 221.3302764892578, 254.40516662597656, -42.541107177734375, 42.933570861816406, 233.00833129882812, 159.6974639892578, 403.4865417480469, 398.7424621582031, 356.8395080566406, -16.278629302978516, 71.98753356933594, 89.42764282226562, -23.457962036132812, 11.578498840332031, -211.69711303710938, 188.35415649414062, 81.45904541015625, 180.685791015625, 14.218597412109375, -5.427295684814453, 81.76992797851562, 124.928955078125, 163.94570922851562, 159.05577087402344, 42.79393768310547, -131.72183227539062, 178.5834503173828, 75.96795654296875, 139.3222198486328, -197.69647216796875, 345.7142028808594, 168.99945068359375, 53.146812438964844, 17.02601432800293, 138.6846160888672, 128.28201293945312, 67.16984558105469, -3.987020492553711, 252.880859375, 191.74252319335938, 22.223257064819336], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000504.npy"}
|
||||
{"epoch": 0.7619047619047619, "step": 505, "batch_size": 64, "mean": 73.75059509277344, "std": 164.14707946777344, "min": -385.5273132324219, "p10": -116.47633590698241, "median": 58.357200622558594, "p90": 228.2502182006836, "max": 668.5745239257812, "pos_frac": 0.71875, "sample": [174.67587280273438, 15.178722381591797, 85.78123474121094, -4.110507965087891, 5.299221038818359, 668.5745239257812, -172.14761352539062, 409.11260986328125, 193.7420196533203, 212.1834716796875, 218.96568298339844, -8.057640075683594, 119.09680938720703, 1.4298858642578125, 41.18225860595703, -30.00311279296875, -112.523193359375, -35.83217239379883, 192.7777862548828, 15.529552459716797, -100.55293273925781, 198.5887908935547, 184.26773071289062, 65.6513671875, 98.49295806884766, 379.13592529296875, -50.098304748535156, -146.66709899902344, 51.06303405761719, 0.8725776672363281, -125.80807495117188, 18.42278289794922, 222.67333984375, -218.07998657226562, 244.3558349609375, 216.61477661132812, 197.23171997070312, 227.9028778076172, -25.771055221557617, 8.510940551757812, 104.54275512695312, 23.079551696777344, -20.57115936279297, -118.17053985595703, 154.26605224609375, 70.81198120117188, 131.65847778320312, -30.06266975402832, 298.84393310546875, -211.34461975097656, 7.164329528808594, 200.25576782226562, 206.36622619628906, 86.14215087890625, 189.68832397460938, 83.41828918457031, 228.39907836914062, 70.38554382324219, 29.28307342529297, 242.17654418945312, 0.24162864685058594, 5.156761169433594, -83.82855987548828, -385.5273132324219], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000505.npy"}
|
||||
{"epoch": 0.763416477702192, "step": 506, "batch_size": 64, "mean": 78.02780151367188, "std": 162.52511596679688, "min": -425.1669921875, "p10": -127.20222930908196, "median": 83.9316635131836, "p90": 239.21649322509768, "max": 440.4542541503906, "pos_frac": 0.734375, "sample": [19.445409774780273, 152.57412719726562, 16.598773956298828, -53.183746337890625, 165.75572204589844, 236.5144500732422, 157.2459716796875, -63.00690841674805, 149.77392578125, -15.86175537109375, 8.871084213256836, 141.2120361328125, 122.8949203491211, -34.835594177246094, 204.11659240722656, 89.77975463867188, -234.83248901367188, 389.4990234375, 370.4106750488281, 50.54547882080078, 227.18954467773438, -230.0833740234375, -55.34974670410156, -425.1669921875, -186.67147827148438, -220.9952850341797, 79.31132507324219, 241.2073211669922, 440.4542541503906, 217.34298706054688, 186.98585510253906, 30.913253784179688, 433.0430908203125, -0.236785888671875, 206.4680938720703, 12.644157409667969, 195.53831481933594, 156.0570068359375, 232.75933837890625, 88.552001953125, -253.44622802734375, 32.68499755859375, 31.06805419921875, 23.853424072265625, 104.54989624023438, 94.11463928222656, -25.092418670654297, 123.3473892211914, 178.6136016845703, 59.81800079345703, 42.60198974609375, 240.37451171875, -13.621070861816406, 176.09512329101562, 168.66952514648438, -20.92884063720703, 8.374534606933594, 183.75216674804688, -154.71450805664062, -11.92612075805664, 34.97480010986328, 289.0771789550781, 112.927978515625, 65.1304931640625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000506.npy"}
|
||||
{"epoch": 0.764928193499622, "step": 507, "batch_size": 64, "mean": 94.48275756835938, "std": 182.8799591064453, "min": -386.67864990234375, "p10": -76.81861267089843, "median": 63.01972961425781, "p90": 289.37185974121104, "max": 679.6915283203125, "pos_frac": 0.6875, "sample": [223.31759643554688, 67.9610595703125, 13.987434387207031, 97.77484130859375, 594.5585327148438, -36.486793518066406, 495.05804443359375, -386.67864990234375, -50.47216796875, 159.26722717285156, 141.98306274414062, 29.40301513671875, -78.80914306640625, 50.463653564453125, -16.607322692871094, -210.6636962890625, 117.41618347167969, -60.00050354003906, 214.870361328125, 34.52303695678711, 137.3024444580078, 93.67767333984375, 203.36788940429688, -26.054933547973633, 264.0245666503906, -16.362497329711914, 58.924400329589844, 447.19775390625, 146.7296600341797, 20.56011962890625, 120.36837768554688, 3.19677734375, 521.8619384765625, -50.873619079589844, 242.1141357421875, 161.93406677246094, -132.06381225585938, 256.6751708984375, -107.06185913085938, 8.767066955566406, -72.17404174804688, -0.03436279296875, -12.658210754394531, 199.93531799316406, 181.07420349121094, 15.658950805664062, 89.09990692138672, 50.75950622558594, 679.6915283203125, 300.2349853515625, -65.417724609375, 131.75262451171875, -31.957008361816406, -2.6467208862304688, 150.47216796875, 67.11505889892578, 137.3831329345703, -98.08073425292969, 5.649574279785156, -127.17088317871094, 28.573760986328125, 315.6767578125, 178.10101318359375, 170.70648193359375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000507.npy"}
|
||||
{"epoch": 0.7664399092970522, "step": 508, "batch_size": 64, "mean": 114.18850708007812, "std": 182.78692626953125, "min": -264.4554443359375, "p10": -72.32296371459961, "median": 94.98739624023438, "p90": 351.11108398437506, "max": 551.5968017578125, "pos_frac": 0.65625, "sample": [161.99049377441406, 58.09141540527344, 273.370361328125, -63.86997985839844, 63.95228576660156, -72.70545959472656, 354.7901306152344, -1.3451919555664062, 184.47872924804688, 277.3477478027344, 419.2445068359375, 165.55038452148438, -3.0908660888671875, -4.5870361328125, 199.86080932617188, 175.85479736328125, -196.98558044433594, 121.6707763671875, 23.751388549804688, -131.64756774902344, 38.971099853515625, 34.99183654785156, 188.03280639648438, -71.43047332763672, -69.90409088134766, -264.4554443359375, 102.84515380859375, -4.333860397338867, 551.5968017578125, 142.60337829589844, -210.09365844726562, 476.739990234375, 191.4654083251953, 5.311943054199219, 223.30938720703125, 342.5266418457031, 140.4332733154297, 389.16851806640625, 226.26177978515625, -7.237495422363281, -0.7078628540039062, -68.42742156982422, 174.59539794921875, 264.1253356933594, -27.65717315673828, -124.50352478027344, 158.49276733398438, 526.2512817382812, 334.17047119140625, 16.62236213684082, 231.83477783203125, -6.3698883056640625, 85.31494140625, -8.191122055053711, -0.7371368408203125, -190.21810913085938, -30.876632690429688, 232.75149536132812, 271.0185546875, 512.880859375, 118.82352447509766, 87.129638671875, 48.12469482421875, 271.0924987792969], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000508.npy"}
|
||||
{"epoch": 0.7679516250944822, "step": 509, "batch_size": 64, "mean": 75.99424743652344, "std": 161.17434692382812, "min": -348.1330871582031, "p10": -126.24890747070309, "median": 80.59734344482422, "p90": 267.7131622314453, "max": 438.7183837890625, "pos_frac": 0.6875, "sample": [23.099746704101562, -95.81134033203125, 141.1747589111328, 218.70077514648438, 142.1304473876953, -6.909910202026367, 18.83838653564453, -28.197969436645508, -139.2935791015625, 144.7968292236328, 52.76018524169922, 74.16522216796875, -348.1330871582031, 147.546142578125, 262.5858459472656, -176.8982391357422, 438.7183837890625, 219.42874145507812, -60.5472412109375, 33.151153564453125, 214.5668182373047, -238.66854858398438, 43.21148681640625, -21.802139282226562, 308.2198791503906, 48.04685974121094, -72.65265655517578, 18.898799896240234, 178.18475341796875, -14.446014404296875, 330.7921447753906, -4.734569549560547, 257.28515625, 204.89642333984375, 34.714385986328125, -2.7818355560302734, 192.16683959960938, 119.786865234375, 280.6746826171875, 191.2794952392578, 91.0681381225586, -29.348983764648438, 273.5124816894531, -87.86027526855469, 194.44577026367188, 285.92926025390625, 90.67784118652344, 166.2599334716797, 254.06344604492188, 172.84556579589844, 255.95184326171875, -58.591773986816406, -84.48265838623047, -339.1400146484375, 17.593841552734375, 182.34043884277344, 79.24652862548828, -195.27825927734375, -194.15994262695312, 81.94815826416016, 18.275554656982422, 134.92678833007812, 269.91058349609375, 154.5535125732422], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000509.npy"}
|
||||
{"epoch": 0.7694633408919124, "step": 510, "batch_size": 64, "mean": 106.90889739990234, "std": 205.7588348388672, "min": -301.0037536621094, "p10": -162.65083160400388, "median": 64.25247764587402, "p90": 325.56013793945317, "max": 831.5870361328125, "pos_frac": 0.765625, "sample": [35.20158386230469, 232.8303985595703, 250.62313842773438, 208.92848205566406, 52.10552978515625, 40.910369873046875, 37.07090377807617, 279.6095886230469, 11.664146423339844, 47.350250244140625, -100.11768341064453, 79.61824798583984, -9.047977447509766, 49.11451721191406, -194.49932861328125, -21.99591064453125, -19.135040283203125, 69.55070495605469, -189.14346313476562, 24.647294998168945, 117.656982421875, 220.1050567626953, 283.44024658203125, -143.69235229492188, 11.249465942382812, 234.2215576171875, 36.995384216308594, -55.02571105957031, 0.20819473266601562, 152.60464477539062, 67.01019287109375, 144.83172607421875, 454.57977294921875, 17.776586532592773, 10.051044464111328, 268.70196533203125, 108.73967742919922, 386.9372863769531, 131.34765625, 76.4046630859375, 4.068967819213867, -187.4307861328125, 210.9309539794922, 61.4947624206543, 502.6146240234375, 24.468162536621094, 240.3062744140625, 327.3678283691406, 831.5870361328125, -170.77589416503906, 24.889366149902344, 85.78085327148438, 211.38650512695312, 198.09620666503906, -184.70321655273438, -2.802959442138672, 477.9329528808594, 321.3421936035156, -32.98517608642578, 178.1371307373047, 707.7987060546875, 109.88987731933594, -205.6509552001953, -301.0037536621094], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000510.npy"}
|
||||
{"epoch": 0.7709750566893424, "step": 511, "batch_size": 64, "mean": 112.98185729980469, "std": 138.3638916015625, "min": -161.4359893798828, "p10": -57.99022254943847, "median": 127.97677612304688, "p90": 294.06658630371095, "max": 398.5448303222656, "pos_frac": 0.765625, "sample": [301.25689697265625, 261.3717041015625, 109.224853515625, 217.8087921142578, 64.75021362304688, -46.20295715332031, 113.34707641601562, -18.80113983154297, 125.24180603027344, 241.17648315429688, 130.7117462158203, -137.33721923828125, 94.23985290527344, 68.7462387084961, 198.02435302734375, 374.62054443359375, 207.57862854003906, -15.552352905273438, 186.04537963867188, 181.75454711914062, 233.20237731933594, 139.0235595703125, 56.404945373535156, -120.16387176513672, -161.4359893798828, -128.20504760742188, 295.8950500488281, 151.6503143310547, 119.62610626220703, -49.957550048828125, 39.52375030517578, 307.7059020996094, -43.290771484375, 158.6767578125, 200.05612182617188, 230.6037139892578, 176.09042358398438, 41.633480072021484, -20.57758331298828, 27.295804977416992, 140.94728088378906, 21.14394760131836, 264.54571533203125, 289.8001708984375, 398.5448303222656, 177.49203491210938, -151.9643096923828, 74.23336791992188, 6.199796676635742, 209.4259796142578, 40.667991638183594, -0.5787353515625, 198.28379821777344, 230.627197265625, 8.530059814453125, 143.88751220703125, -153.03623962402344, 316.2927551269531, 309.7567138671875, 216.93804931640625, 237.2589111328125, -6.391529083251953, 7.903564453125, -61.432796478271484], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000511.npy"}
|
||||
{"epoch": 0.7724867724867724, "step": 512, "batch_size": 64, "mean": 66.19861602783203, "std": 158.31259155273438, "min": -492.46441650390625, "p10": -140.4457260131836, "median": 63.92988586425781, "p90": 235.2077728271485, "max": 465.322021484375, "pos_frac": 0.6875, "sample": [250.1834716796875, 183.65945434570312, 32.704261779785156, -192.67364501953125, -63.65017318725586, 207.40928649902344, 128.63514709472656, -69.76973724365234, 118.08057403564453, 153.52330017089844, 157.15892028808594, 326.50860595703125, -492.46441650390625, -76.17350769042969, 20.473052978515625, -129.18540954589844, -5.654212951660156, 29.305965423583984, 194.39224243164062, 13.088495254516602, 240.23867797851562, 212.19139099121094, -48.58051300048828, 251.56298828125, -80.97825622558594, 119.4271240234375, -145.27157592773438, 216.96484375, -211.39059448242188, 161.9336700439453, -3.8716468811035156, 294.9934387207031, 290.7855224609375, 6.742132186889648, 107.576416015625, 111.4366455078125, 10.6597900390625, 165.2834014892578, 15.970010757446289, 218.18023681640625, 53.00511169433594, -168.82081604003906, -17.630067825317383, -4.2010040283203125, 105.60919189453125, 465.322021484375, 26.504371643066406, 36.52778244018555, 135.60281372070312, 13.003070831298828, 189.68087768554688, -19.140777587890625, -15.320159912109375, -180.82977294921875, 74.85466003417969, 219.7578887939453, 148.94012451171875, 164.75733947753906, 188.66090393066406, 223.468994140625, 52.296287536621094, -60.82946014404297, 113.79057312011719, -227.70394897460938], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000512.npy"}
|
||||
{"epoch": 0.7739984882842026, "step": 513, "batch_size": 64, "mean": 86.80203247070312, "std": 133.7071075439453, "min": -238.54946899414062, "p10": -47.447229003906244, "median": 72.9126091003418, "p90": 242.9453842163086, "max": 563.3179321289062, "pos_frac": 0.765625, "sample": [405.9224853515625, 193.3011474609375, 205.52232360839844, 91.80174255371094, 152.8232421875, 44.68571853637695, 184.69863891601562, -7.567501068115234, -4.331821441650391, -29.718002319335938, 109.91350555419922, -84.28579711914062, 167.46896362304688, 180.6153564453125, 147.37411499023438, 149.40077209472656, -35.0567512512207, 30.12596893310547, 232.9142303466797, -38.683631896972656, 17.598617553710938, 276.7142333984375, 126.44921112060547, 109.63543701171875, 260.56634521484375, 10.8570556640625, -106.86112976074219, 115.36465454101562, 54.596221923828125, 145.1697998046875, 19.99042510986328, 144.39451599121094, 91.09551239013672, -238.54946899414062, 188.25274658203125, 31.601808547973633, -51.20305633544922, 235.15084838867188, 563.3179321289062, 70.30079650878906, 3.486553192138672, -3.062253952026367, 294.42413330078125, 11.3243408203125, 235.97015380859375, 245.9347686767578, 270.4814147949219, 75.52442169189453, 184.7120361328125, 103.80787658691406, 54.89928436279297, 5.6934356689453125, 17.96935272216797, -11.268386840820312, -163.3326416015625, -103.16091918945312, 83.34073638916016, 43.04988479614258, -30.512195587158203, 37.15099334716797, 133.052734375, -101.0271224975586, 2.3191375732421875, 3.1848411560058594], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000513.npy"}
|
||||
{"epoch": 0.7755102040816326, "step": 514, "batch_size": 64, "mean": 101.3864517211914, "std": 158.6385040283203, "min": -235.83700561523438, "p10": -95.10187072753904, "median": 115.34835052490234, "p90": 261.3062896728516, "max": 666.2753295898438, "pos_frac": 0.75, "sample": [8.063566207885742, 198.71710205078125, 165.5493927001953, 131.97959899902344, 76.37739562988281, 85.95178985595703, 230.32467651367188, -19.9921875, 141.78195190429688, -14.39169692993164, 9.63040542602539, 267.04974365234375, 191.86917114257812, 95.20919799804688, 5.435644149780273, 133.50985717773438, -233.10623168945312, 190.15249633789062, 20.651321411132812, 46.444664001464844, -232.86962890625, -3.9576034545898438, 185.3404998779297, 186.32333374023438, 268.0671081542969, 30.731216430664062, 75.88589477539062, 185.48648071289062, 172.20469665527344, 283.38970947265625, -128.476806640625, -158.5765380859375, 146.00389099121094, 264.8740539550781, -9.330238342285156, 120.09062957763672, 252.98150634765625, 183.43588256835938, -3.386566162109375, 156.8629608154297, 73.69332122802734, -105.87815856933594, -69.95719909667969, 18.37969207763672, 243.71559143066406, 113.04889678955078, 200.58799743652344, -23.867111206054688, 59.118995666503906, 117.6478042602539, -23.514503479003906, 93.17245483398438, -37.724708557128906, 339.16265869140625, -235.83700561523438, 190.58694458007812, 250.2147674560547, 236.90109252929688, 427.38330078125, 9.283498764038086, 666.2753295898438, 210.29364013671875, -200.6381378173828, 230.39535522460938], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000514.npy"}
|
||||
{"epoch": 0.7770219198790628, "step": 515, "batch_size": 64, "mean": 83.19206237792969, "std": 144.9114227294922, "min": -387.04888916015625, "p10": -94.38264923095701, "median": 97.7415542602539, "p90": 260.0276733398438, "max": 330.68853759765625, "pos_frac": 0.671875, "sample": [11.795764923095703, 179.78321838378906, -47.308555603027344, 26.862281799316406, 330.68853759765625, 195.77845764160156, -6.607791900634766, 204.09219360351562, 16.442138671875, 259.056884765625, 80.12967681884766, 37.28089141845703, 119.76629638671875, -10.896150588989258, 164.6772003173828, 199.7440185546875, -103.82740783691406, 275.98583984375, -142.899169921875, -117.66802978515625, 152.90182495117188, 17.894737243652344, -6.902099609375, 116.49171447753906, -387.04888916015625, 186.59120178222656, 219.48373413085938, 39.117645263671875, -61.089698791503906, -133.53982543945312, 125.0059585571289, 181.99105834960938, -1.3466072082519531, -2.490041732788086, 67.10269165039062, 191.971435546875, -155.49252319335938, 260.4437255859375, -40.14313507080078, 251.9895477294922, -72.34487915039062, 146.30625915527344, 311.9896240234375, -182.12652587890625, 187.05101013183594, 291.27813720703125, 30.608247756958008, -31.083242416381836, 115.35343170166016, 223.27586364746094, 27.179710388183594, 242.8741455078125, 152.55227661132812, 214.7019805908203, 297.940673828125, -72.34465026855469, 188.78887939453125, 3.8674354553222656, 298.01513671875, -55.249534606933594, 156.17062377929688, 163.30758666992188, -3.1503849029541016, -6.478279113769531], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000515.npy"}
|
||||
{"epoch": 0.7785336356764928, "step": 516, "batch_size": 64, "mean": 77.0119400024414, "std": 144.3265838623047, "min": -300.4265441894531, "p10": -117.50267181396482, "median": 91.50203323364258, "p90": 231.36369934082035, "max": 416.660888671875, "pos_frac": 0.75, "sample": [120.59616088867188, 238.7161407470703, 158.38584899902344, -135.03379821777344, -98.07316589355469, 71.62954711914062, -125.5816421508789, 161.5338134765625, 204.36676025390625, 202.8391571044922, 95.62860870361328, 48.782188415527344, 416.660888671875, 291.327880859375, -2.7168846130371094, 2.879220962524414, 49.90838623046875, 203.93760681152344, 315.59173583984375, 204.44358825683594, -98.65174102783203, -242.91319274902344, 106.68098449707031, 188.6361083984375, 140.57470703125, 79.70601654052734, 233.46475219726562, 10.291030883789062, 202.724853515625, 190.67669677734375, -300.4265441894531, 8.961990356445312, 1.2785968780517578, 15.476615905761719, -40.98523712158203, 171.67578125, 217.24301147460938, -17.295087814331055, 5.763233184814453, 24.717411041259766, 187.0281982421875, -259.7868957519531, 194.90487670898438, 226.46124267578125, 27.813720703125, 99.77328491210938, 200.4159393310547, 87.37545776367188, 118.08551025390625, 188.70970153808594, 8.16571044921875, 268.643310546875, 33.63018798828125, -61.99127960205078, -175.57009887695312, 150.3792266845703, -31.542278289794922, 104.48149108886719, -17.965438842773438, 2.4947242736816406, 173.9436798095703, -3.201763153076172, -161.74020385742188, 244.8337860107422], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000516.npy"}
|
||||
{"epoch": 0.780045351473923, "step": 517, "batch_size": 64, "mean": 59.117767333984375, "std": 139.33090209960938, "min": -347.8854064941406, "p10": -79.20635757446288, "median": 39.368669509887695, "p90": 222.91862945556645, "max": 526.8890380859375, "pos_frac": 0.625, "sample": [72.39218139648438, -72.27615356445312, 13.962200164794922, 114.64118194580078, -11.729988098144531, 19.766132354736328, -122.03538513183594, 148.6458740234375, 275.6861267089844, 88.71493530273438, -16.332534790039062, -8.534797668457031, -62.6707763671875, -26.19475555419922, -12.679901123046875, -68.23815155029297, -17.456682205200195, -40.36937713623047, 34.806640625, 44.388893127441406, 134.98765563964844, -161.5337677001953, 526.8890380859375, 206.67025756835938, -347.8854064941406, 45.049293518066406, -30.62994384765625, 102.69815826416016, 115.40646362304688, 174.41172790527344, 54.909889221191406, 194.94801330566406, 4.401430130004883, 149.33074951171875, -87.46701049804688, -46.727210998535156, 47.17869567871094, 370.2347412109375, -48.157806396484375, -82.17644500732422, -8.241203308105469, 233.05023193359375, 344.03265380859375, 26.47006607055664, 149.60800170898438, 194.8550567626953, 8.481304168701172, 102.94656372070312, -148.50247192382812, -43.682701110839844, -105.75483703613281, 151.9150390625, 43.389495849609375, 193.999755859375, 264.4933166503906, 228.130126953125, 57.92060852050781, -17.132648468017578, 30.309165954589844, 35.347843170166016, 210.7584686279297, -53.21446228027344, 141.1490020751953, 66.18441772460938], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000517.npy"}
|
||||
{"epoch": 0.781557067271353, "step": 518, "batch_size": 64, "mean": 88.2010498046875, "std": 151.6498260498047, "min": -274.04376220703125, "p10": -71.69699707031249, "median": 47.89365577697754, "p90": 251.14901123046874, "max": 485.91497802734375, "pos_frac": 0.703125, "sample": [164.51480102539062, 1.141387939453125, -3.8259353637695312, 251.9169921875, -57.62617492675781, 20.40943145751953, -12.572257995605469, 39.13615798950195, -56.92416000366211, 60.27625274658203, 141.63150024414062, -75.50994873046875, 161.06533813476562, 207.8383331298828, 127.7617416381836, -183.0423126220703, 98.48088073730469, 178.63319396972656, -121.96372985839844, 236.60598754882812, 485.91497802734375, -274.04376220703125, -2.1053695678710938, 169.9715118408203, 181.1771240234375, 231.59097290039062, -94.67442321777344, 181.42041015625, -90.40087890625, 168.56686401367188, 24.606014251708984, -41.878456115722656, 12.59974479675293, 170.8483123779297, 73.334228515625, 14.914276123046875, 2.657733917236328, -18.414749145507812, 114.03135681152344, 368.361328125, 29.240951538085938, 107.02074432373047, -0.20827293395996094, 249.3570556640625, 114.15077209472656, 456.3482666015625, -62.80010986328125, 218.08123779296875, 409.99652099609375, -81.0774154663086, -22.57547950744629, -32.63164520263672, 371.4596252441406, 0.4655342102050781, 20.61695098876953, 219.97979736328125, 214.38784790039062, 56.651153564453125, 21.48822593688965, -16.799179077148438, 9.982612609863281, 11.567167282104492, 380.96221923828125, 112.77752685546875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000518.npy"}
|
||||
{"epoch": 0.783068783068783, "step": 519, "batch_size": 64, "mean": 52.63594055175781, "std": 169.19480895996094, "min": -502.1024169921875, "p10": -165.61136474609373, "median": 46.181209564208984, "p90": 271.30258178710943, "max": 309.4244384765625, "pos_frac": 0.609375, "sample": [-194.02865600585938, 154.1319580078125, 231.74685668945312, -232.87501525878906, -79.61334228515625, 228.38653564453125, -17.223400115966797, 294.318603515625, 3.9913177490234375, 149.99337768554688, 165.89044189453125, 35.053382873535156, 262.9420166015625, -130.44223022460938, 149.61083984375, 158.38522338867188, 224.45529174804688, 24.790292739868164, -204.8794708251953, -103.70751953125, 218.4813995361328, 294.41656494140625, -172.4564666748047, 309.4244384765625, 246.26231384277344, 213.93215942382812, -102.63958740234375, -27.748504638671875, 40.91511535644531, 177.34585571289062, -33.4755859375, 204.10763549804688, 86.15122985839844, -82.85397338867188, 114.56680297851562, -64.95285034179688, -101.77018737792969, 16.966289520263672, 293.458251953125, -115.41329956054688, 99.85655212402344, 13.450065612792969, 284.0716552734375, 74.48831939697266, 29.18457794189453, -502.1024169921875, 117.85966491699219, -92.99554443359375, 70.05926513671875, -170.12973022460938, 207.5362548828125, -54.401283264160156, -155.06851196289062, 299.7340393066406, 177.0835418701172, 51.447303771972656, 274.88568115234375, 174.26800537109375, -14.434024810791016, -193.50262451171875, -74.2443618774414, -49.798240661621094, 234.28985595703125, -68.48187255859375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000519.npy"}
|
||||
{"epoch": 0.7845804988662132, "step": 520, "batch_size": 64, "mean": 80.34745788574219, "std": 141.8923797607422, "min": -233.2618865966797, "p10": -44.99956817626952, "median": 66.2486686706543, "p90": 218.44210662841797, "max": 677.7756958007812, "pos_frac": 0.703125, "sample": [216.23190307617188, -16.6995792388916, -5.386600494384766, -31.29092025756836, 207.12210083007812, 182.49990844726562, 107.80781555175781, 17.82701873779297, -14.952342987060547, 677.7756958007812, 126.53677368164062, -2.4113521575927734, -31.081756591796875, -37.27825927734375, 198.80813598632812, 134.15325927734375, 109.01997375488281, 22.277359008789062, 18.865306854248047, 214.8427734375, 399.8492431640625, 151.25355529785156, 15.666143417358398, 265.53302001953125, 73.12457275390625, 71.08458709716797, 40.60832214355469, -144.28001403808594, -139.5654296875, 218.72503662109375, 89.12385559082031, -21.73175811767578, -49.662681579589844, 14.314918518066406, 26.152969360351562, -1.5963325500488281, 222.15280151367188, 101.65181732177734, 54.01179504394531, -110.32584381103516, -200.0474090576172, 46.54814147949219, 87.98834228515625, -48.30870056152344, 98.62506103515625, 122.77528381347656, 10.724319458007812, 164.07630920410156, 181.17803955078125, 103.79347229003906, 55.82325744628906, 229.6121368408203, 187.9744415283203, 76.87601470947266, 362.07318115234375, 72.21741485595703, 61.412750244140625, 10.882415771484375, -24.90585708618164, -1.8021926879882812, -18.17670440673828, 207.619384765625, 217.7819366455078, -233.2618865966797], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000520.npy"}
|
||||
{"epoch": 0.7860922146636432, "step": 521, "batch_size": 64, "mean": 113.25245666503906, "std": 159.7565460205078, "min": -245.2105712890625, "p10": -34.70717239379883, "median": 106.55374908447266, "p90": 327.61200256347666, "max": 691.0196533203125, "pos_frac": 0.71875, "sample": [212.59674072265625, 144.19577026367188, 265.4985656738281, 109.08169555664062, 195.0875701904297, 129.03155517578125, 185.39666748046875, -14.272911071777344, -29.801284790039062, 0.819671630859375, 61.6588134765625, -29.316967010498047, 337.6582946777344, 35.709251403808594, 193.2709197998047, 75.7781753540039, 17.179826736450195, 24.21868896484375, -37.75395202636719, 199.1859130859375, -0.8510818481445312, 247.15692138671875, 134.08914184570312, 39.70086669921875, -21.113054275512695, 159.51803588867188, -146.18882751464844, 240.54397583007812, -1.32037353515625, 185.2405242919922, 304.170654296875, 691.0196533203125, -144.1435546875, 39.57863235473633, -17.89844512939453, -37.93389129638672, -34.940528869628906, 62.65119934082031, -12.709548950195312, 406.02044677734375, 357.28607177734375, 10.882442474365234, -29.271156311035156, 166.01690673828125, -3.2985897064208984, 33.94017028808594, 194.4834442138672, 104.02580261230469, 5.864734649658203, 172.7534637451172, 176.52993774414062, 135.55450439453125, -245.2105712890625, 426.5392761230469, 404.45916748046875, 184.79017639160156, 82.6096420288086, -111.42361450195312, 126.44685363769531, -34.16267395019531, 136.43185424804688, 113.32048797607422, 274.1148986816406, 397.66064453125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000521.npy"}
|
||||
{"epoch": 0.7876039304610734, "step": 522, "batch_size": 64, "mean": 61.516441345214844, "std": 169.4438934326172, "min": -422.4541320800781, "p10": -139.2193634033203, "median": 58.27748107910156, "p90": 258.3660522460938, "max": 471.8248291015625, "pos_frac": 0.6875, "sample": [-124.65576171875, 71.61336517333984, 6.89543342590332, 111.42081451416016, 150.00979614257812, 225.51661682128906, 116.67838287353516, 296.9469909667969, 252.08416748046875, 21.780996322631836, -185.77198791503906, -170.4803466796875, 146.6832733154297, 276.4513244628906, 29.049789428710938, 174.57913208007812, -79.45870208740234, 102.5799560546875, 199.02540588378906, 11.092100143432617, 261.05828857421875, -9.736759185791016, 89.05477142333984, -352.5718688964844, 403.6493835449219, 4.658943176269531, 15.830692291259766, -145.46090698242188, -422.4541320800781, 83.59814453125, 0.7707138061523438, 224.55929565429688, -107.878173828125, 272.81719970703125, 202.84219360351562, 30.888198852539062, -20.81580352783203, 112.71635437011719, 29.382755279541016, 55.924659729003906, 60.63030242919922, -36.32563781738281, 7.896888732910156, -114.3381118774414, 163.9002227783203, 0.5166854858398438, 161.98550415039062, -228.6011199951172, 246.44288635253906, -19.228609085083008, -32.30620574951172, -219.65199279785156, 205.81509399414062, 471.8248291015625, -24.495132446289062, 70.1533432006836, -109.73130798339844, 382.34716796875, 180.74501037597656, 142.02606201171875, -40.734291076660156, -46.07033157348633, 104.93177795410156, 248.44442749023438], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000522.npy"}
|
||||
{"epoch": 0.7891156462585034, "step": 523, "batch_size": 64, "mean": 84.79135131835938, "std": 153.62168884277344, "min": -449.82244873046875, "p10": -90.48317794799804, "median": 91.52249908447266, "p90": 263.8786712646485, "max": 478.7437744140625, "pos_frac": 0.71875, "sample": [86.24575805664062, 279.2401428222656, -36.10268783569336, -44.30870819091797, 180.67984008789062, 114.9068832397461, 188.41639709472656, 109.39330291748047, 231.99267578125, 87.33888244628906, -0.46232032775878906, 128.8439483642578, 299.30419921875, -136.20726013183594, 148.4149627685547, 91.9375, 173.81170654296875, -23.720365524291992, 62.550750732421875, -158.60110473632812, 14.95001220703125, 269.6746826171875, -82.6881103515625, 47.09112548828125, 0.361785888671875, 183.97003173828125, -49.787269592285156, 63.9835319519043, 36.764625549316406, 111.23542785644531, 22.488454818725586, 181.37220764160156, 273.53125, 160.08419799804688, 176.7300567626953, 250.35464477539062, 208.02674865722656, 224.63438415527344, -24.77710723876953, -449.82244873046875, 162.64804077148438, -263.6552429199219, 287.374755859375, 131.80181884765625, 199.50070190429688, -93.82392120361328, -24.084331512451172, 217.8673095703125, 11.101943969726562, -52.704200744628906, 304.14111328125, 27.828792572021484, 50.41668701171875, -155.1912841796875, 224.05496215820312, 243.48825073242188, 140.69459533691406, 478.7437744140625, -13.8436279296875, -174.2171630859375, -29.76837921142578, 91.10749816894531, 87.72503662109375, 173.58627319335938], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000523.npy"}
|
||||
{"epoch": 0.7906273620559335, "step": 524, "batch_size": 64, "mean": 80.39669799804688, "std": 133.52499389648438, "min": -238.68023681640625, "p10": -65.09970397949218, "median": 66.99419784545898, "p90": 234.73946228027344, "max": 489.1827392578125, "pos_frac": 0.78125, "sample": [224.57839965820312, 68.27857208251953, 398.8699035644531, 59.6392936706543, -47.10186767578125, 13.184989929199219, -68.45726776123047, -28.614501953125, 76.40422058105469, 125.13177490234375, 265.5704650878906, 2.307281494140625, -112.91635131835938, 118.76272583007812, 62.205291748046875, 287.1920166015625, -219.25790405273438, -119.14663696289062, 73.95587158203125, 212.24911499023438, 197.7175750732422, -183.95785522460938, 256.8529052734375, -1.6165084838867188, 36.450897216796875, 7.178276062011719, 160.38565063476562, 103.83329010009766, 139.44485473632812, 45.97735595703125, -22.091140747070312, 138.81588745117188, 65.70982360839844, 108.58869171142578, 188.66094970703125, 41.97567367553711, 19.17304039001465, 170.58880615234375, -10.581253051757812, -99.61576843261719, -32.519622802734375, 4.424591064453125, 12.052589416503906, 489.1827392578125, 38.64563751220703, 236.14837646484375, 206.04576110839844, 9.949138641357422, 61.03483581542969, 68.29812622070312, 156.97756958007812, 93.20526885986328, 21.12786102294922, 243.64529418945312, 102.87184143066406, 231.45199584960938, 209.82794189453125, 129.583740234375, -238.68023681640625, -57.26538848876953, 201.53912353515625, 4.193122863769531, 178.6220245361328, 18.699790954589844], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000524.npy"}
|
||||
{"epoch": 0.7921390778533636, "step": 525, "batch_size": 64, "mean": 66.34283447265625, "std": 158.98562622070312, "min": -327.5507507324219, "p10": -66.1259033203125, "median": 32.920820236206055, "p90": 241.10190429687506, "max": 587.28369140625, "pos_frac": 0.625, "sample": [-57.6802978515625, -226.57077026367188, 31.637237548828125, -36.35300827026367, -64.94359588623047, 51.77464294433594, 20.48517608642578, 93.6895751953125, 357.0581359863281, -5.074806213378906, 181.5428009033203, -39.47540283203125, 187.94790649414062, 35.294097900390625, 162.0462646484375, 205.55715942382812, 142.4248809814453, 1.1802139282226562, 357.90069580078125, -164.57254028320312, 72.94718170166016, -3.002490997314453, 275.0101318359375, 203.7913360595703, -62.33924102783203, 123.41920471191406, -39.44734191894531, 82.99247741699219, 5.692543029785156, 149.14552307128906, -6.9562835693359375, 345.6572265625, 15.939857482910156, 166.40518188476562, 126.01400756835938, -13.390396118164062, 34.204402923583984, -34.316131591796875, 587.28369140625, -2.325593948364258, -105.72694396972656, 71.87425231933594, 0.9082298278808594, -327.5507507324219, -66.63260650634766, 248.6770782470703, 2.641437530517578, 25.64203643798828, 90.72140502929688, 136.3172607421875, -54.51293182373047, 203.8526611328125, -12.471366882324219, -31.132522583007812, 156.92193603515625, 460.8693542480469, -250.09085083007812, 57.714202880859375, 140.76165771484375, 223.42649841308594, -36.23321533203125, 206.86251831054688, -13.865636825561523, -143.62611389160156], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000525.npy"}
|
||||
{"epoch": 0.7936507936507936, "step": 526, "batch_size": 64, "mean": 66.49160766601562, "std": 147.5263671875, "min": -281.66387939453125, "p10": -114.79169235229489, "median": 48.337392807006836, "p90": 242.45170135498051, "max": 478.2291564941406, "pos_frac": 0.734375, "sample": [289.51611328125, 5.643962860107422, -32.57263946533203, 11.345739364624023, -80.3119125366211, 11.782707214355469, -2.965839385986328, 454.93572998046875, -281.66387939453125, -129.56874084472656, -57.8232421875, 92.81527709960938, -178.27764892578125, 32.411712646484375, 247.12400817871094, -42.67211151123047, 128.77175903320312, 218.56700134277344, 22.66016387939453, 139.38401794433594, 140.97601318359375, 271.6394958496094, 83.51567077636719, 65.73776245117188, 254.96966552734375, 2.9499244689941406, 231.54965209960938, 226.74114990234375, -142.0094757080078, 418.5029602050781, -7.607904434204102, 23.58184814453125, 43.972496032714844, -36.227500915527344, 207.409423828125, 3.456144332885742, 121.1675033569336, 54.139625549316406, 38.090057373046875, 83.34210968017578, 76.70052337646484, 42.979373931884766, -136.11964416503906, 154.49659729003906, 94.39668273925781, -205.33111572265625, 2.099903106689453, 142.7735595703125, 137.4150848388672, -56.603240966796875, 182.4144287109375, 15.408817291259766, 478.2291564941406, 23.299091339111328, 209.7579345703125, 10.902633666992188, -47.52934646606445, 52.70228958129883, 109.71439361572266, 83.49520874023438, 73.58601379394531, -75.13937377929688, 147.25448608398438, -196.43984985351562], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000526.npy"}
|
||||
{"epoch": 0.7951625094482238, "step": 527, "batch_size": 64, "mean": 58.9218635559082, "std": 152.26026916503906, "min": -282.6627502441406, "p10": -140.00935363769528, "median": 29.739057540893555, "p90": 251.53726501464848, "max": 537.46728515625, "pos_frac": 0.6875, "sample": [0.05664634704589844, 17.615951538085938, 31.267112731933594, 143.2632293701172, 55.46388244628906, 5.168815612792969, 280.952880859375, 130.45806884765625, 212.46058654785156, 537.46728515625, 199.22506713867188, -20.013214111328125, 255.20416259765625, 84.48990631103516, 103.55226135253906, -109.99201202392578, 295.34747314453125, -1.6436138153076172, 119.54104614257812, -153.5087432861328, -168.86651611328125, -148.92929077148438, 23.126724243164062, -88.15010070800781, 114.3568115234375, 4.63104248046875, 363.6541442871094, 170.6006317138672, 21.271800994873047, 24.140960693359375, 176.06491088867188, 18.260848999023438, -119.1961669921875, 35.66657638549805, 73.35993957519531, 239.214599609375, 19.31829071044922, -83.05682373046875, -15.882352828979492, 219.33778381347656, 91.702880859375, 28.211002349853516, 290.6213073730469, 45.9618034362793, -46.33258056640625, -31.47296905517578, -201.65451049804688, 204.8390655517578, 20.189651489257812, -3.8070545196533203, -10.8551025390625, 0.8884716033935547, 367.18890380859375, 165.7956085205078, -282.6627502441406, -36.64917755126953, 87.79177856445312, -39.56285858154297, 85.93812561035156, 242.98117065429688, -155.16627502441406, -239.4131317138672, 84.71833801269531, 36.44721221923828], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000527.npy"}
|
||||
{"epoch": 0.7966742252456538, "step": 528, "batch_size": 64, "mean": 65.15164184570312, "std": 169.98558044433594, "min": -425.91534423828125, "p10": -99.55973434448241, "median": 31.65285873413086, "p90": 258.1341674804688, "max": 543.232666015625, "pos_frac": 0.6875, "sample": [28.823989868164062, -425.91534423828125, 218.61520385742188, -26.101594924926758, 29.11510467529297, 159.05825805664062, 46.75557327270508, -2.9649200439453125, 22.943565368652344, -14.189651489257812, -16.519302368164062, -56.06211853027344, 274.2020263671875, -213.43487548828125, 12.048440933227539, -9.502052307128906, 211.16432189941406, 14.562063217163086, 12.752775192260742, -85.00674438476562, -105.7967300415039, 476.9979248046875, 2.107391357421875, -331.1366271972656, -12.145286560058594, 73.06893157958984, -61.839500427246094, -3.8369979858398438, 34.19061279296875, 241.42837524414062, 84.83399963378906, 103.16615295410156, 50.93526840209961, 308.7175598144531, 167.7604217529297, 25.478910446166992, 306.13861083984375, -23.559242248535156, 207.44264221191406, 146.35623168945312, 247.37799072265625, 262.74395751953125, 15.436857223510742, 36.24907684326172, -206.95347595214844, 130.5326385498047, 12.847888946533203, 42.682098388671875, 201.34060668945312, -7.9683837890625, 75.2581787109375, -177.09130859375, -213.1100311279297, 451.06365966796875, 543.232666015625, 97.30766296386719, 2.2805213928222656, -18.733055114746094, 192.1229248046875, 11.945398330688477, 173.5525665283203, 76.91888427734375, 223.77041625976562, 126.24446105957031], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000528.npy"}
|
||||
{"epoch": 0.7981859410430839, "step": 529, "batch_size": 64, "mean": 115.57971954345703, "std": 132.11009216308594, "min": -173.8137664794922, "p10": -33.17953376770019, "median": 104.78921890258789, "p90": 266.1681030273438, "max": 438.5747985839844, "pos_frac": 0.78125, "sample": [166.56687927246094, -144.31129455566406, 14.215164184570312, -9.362098693847656, 53.56698989868164, -33.85057067871094, 438.5747985839844, 227.99642944335938, -31.613780975341797, 35.716148376464844, 18.244871139526367, -2.499094009399414, -30.13665771484375, 161.3687286376953, 194.7581024169922, 180.7537841796875, -119.84600067138672, 221.9712371826172, -4.605621337890625, 158.20086669921875, 252.38360595703125, -39.26029968261719, -45.1690673828125, 272.07574462890625, 143.49551391601562, 215.6514892578125, 22.03174591064453, 30.073341369628906, 217.0592803955078, 38.135589599609375, 342.0606994628906, 209.59226989746094, 105.87614440917969, 203.89356994628906, 197.3941650390625, 143.49423217773438, 211.31927490234375, 250.1326446533203, 67.92179870605469, 288.63922119140625, 319.4190979003906, 205.0670928955078, 322.50836181640625, 59.59889221191406, 217.7906494140625, 212.10354614257812, 81.94004821777344, -3.800243377685547, 103.7022933959961, -22.48712158203125, 61.026878356933594, 61.93816375732422, 53.37340545654297, 236.1757354736328, -173.8137664794922, 247.0341796875, 94.39130401611328, 242.20445251464844, 145.4703369140625, 75.72154235839844, -156.666015625, 25.854835510253906, 20.282493591308594, 345.756103515625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000529.npy"}
|
||||
{"epoch": 0.799697656840514, "step": 530, "batch_size": 64, "mean": 81.57747650146484, "std": 152.14047241210938, "min": -278.2035827636719, "p10": -78.18483505249023, "median": 58.487083435058594, "p90": 223.5988754272461, "max": 561.4807739257812, "pos_frac": 0.6875, "sample": [561.4807739257812, 99.91920471191406, 8.202150344848633, -66.79508972167969, 1.9453353881835938, -27.90240478515625, 415.36859130859375, -2.4614791870117188, 169.69139099121094, 159.4793701171875, 26.161640167236328, 385.6976013183594, 24.203527450561523, 15.005149841308594, 168.2759246826172, 56.098716735839844, 87.51422119140625, 126.37860107421875, 134.44960021972656, 71.65518951416016, 3.6134605407714844, 69.0212631225586, -278.2035827636719, 175.60592651367188, 222.38916015625, -2.1901378631591797, 170.48902893066406, 36.12942123413086, 22.33159637451172, 217.96453857421875, -6.217414855957031, 39.30706024169922, 317.57293701171875, 546.720703125, 211.96815490722656, 60.875450134277344, 71.47744750976562, 133.3082275390625, -37.63651657104492, 26.143375396728516, -86.07554626464844, 188.44020080566406, -83.06615447998047, 111.53347778320312, -118.11505889892578, -41.328819274902344, 195.6589813232422, -39.14320373535156, -219.3350372314453, 5.069854736328125, -13.901941299438477, -12.868911743164062, 210.63931274414062, -93.36799621582031, 232.9456024169922, 130.9870147705078, 191.55294799804688, -40.750701904296875, -32.13682556152344, 67.51585388183594, 224.11732482910156, 116.23721313476562, -3.9700050354003906, -84.71737670898438], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000530.npy"}
|
||||
{"epoch": 0.8012093726379441, "step": 531, "batch_size": 64, "mean": 82.14053344726562, "std": 147.268310546875, "min": -310.8333435058594, "p10": -61.184253692626946, "median": 57.97163009643555, "p90": 249.65704498291015, "max": 531.9940185546875, "pos_frac": 0.65625, "sample": [-21.523330688476562, 172.3505401611328, 345.1661682128906, 230.5282440185547, 174.852783203125, 120.68295288085938, -5.720085144042969, 531.9940185546875, 107.33100891113281, -18.386653900146484, 36.946449279785156, -6.125154495239258, 367.28564453125, 11.949783325195312, 180.32241821289062, 179.3216552734375, -155.701904296875, -54.555397033691406, 28.233848571777344, 225.00831604003906, 230.98988342285156, 311.82318115234375, -94.36705780029297, -23.60308074951172, -44.132843017578125, 13.938943862915039, 58.361289978027344, 71.92941284179688, 46.36882019042969, -0.423187255859375, 65.33325958251953, -23.323762893676758, -13.105245590209961, 57.58197021484375, -278.0107421875, -14.482463836669922, 248.4959716796875, 189.3668212890625, 0.4616546630859375, -21.586074829101562, -64.02519226074219, 112.28401947021484, 217.89674377441406, 213.78887939453125, -93.43373107910156, -310.8333435058594, 189.20962524414062, 194.36917114257812, 100.60319519042969, 45.7947998046875, -33.07989501953125, 75.78056335449219, 59.523956298828125, 250.15464782714844, 203.3597412109375, 9.236309051513672, 128.60336303710938, -77.9482421875, 232.56680297851562, 277.6896667480469, -4.631462097167969, -3.6156482696533203, 264.7726745605469, 37.3497314453125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000531.npy"}
|
||||
{"epoch": 0.8027210884353742, "step": 532, "batch_size": 64, "mean": 85.1446304321289, "std": 151.08372497558594, "min": -226.30816650390625, "p10": -109.55307693481444, "median": 68.93500900268555, "p90": 298.68677978515626, "max": 418.7492980957031, "pos_frac": 0.6875, "sample": [167.47607421875, 71.2327651977539, 405.9220886230469, 418.7492980957031, -0.257659912109375, -64.5311508178711, 182.2059326171875, 207.19427490234375, -3.248218536376953, 328.19268798828125, -97.71778106689453, 175.30557250976562, 139.8728790283203, 50.369140625, 80.54026794433594, -18.1671142578125, 56.42337417602539, 120.586181640625, -17.045366287231445, -146.6689910888672, -36.16551208496094, 22.321290969848633, 215.97610473632812, 19.53843879699707, -34.14306640625, 105.51881408691406, 80.34815979003906, 311.6707763671875, 156.88433837890625, 322.16278076171875, 326.822509765625, 26.56362533569336, -226.30816650390625, 250.85272216796875, 60.74352264404297, 158.13331604003906, 200.6110076904297, -111.50730895996094, 67.2740478515625, 36.43721008300781, -176.20962524414062, -24.009292602539062, 43.64168930053711, -41.08730697631836, 243.30184936523438, 94.17818450927734, 198.82876586914062, -120.5582046508789, 70.29411315917969, 266.86041259765625, 191.9822540283203, 67.5759048461914, -33.93927764892578, 252.0493621826172, -208.55126953125, 74.26304626464844, -175.80221557617188, 297.5502624511719, 208.81610107421875, -104.99320220947266, 5.68614387512207, -11.933786392211914, 21.969772338867188, 299.1738586425781], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000532.npy"}
|
||||
{"epoch": 0.8042328042328042, "step": 533, "batch_size": 64, "mean": 98.59541320800781, "std": 156.9241485595703, "min": -209.1740264892578, "p10": -69.77065200805663, "median": 81.49782943725586, "p90": 241.3049667358399, "max": 741.7922973632812, "pos_frac": 0.703125, "sample": [212.79766845703125, 202.2449951171875, 74.64826202392578, 59.858009338378906, 183.0166778564453, 67.5300521850586, 37.84367370605469, -76.29301452636719, 204.60409545898438, 183.43467712402344, -189.35861206054688, -10.486186981201172, -9.499214172363281, -4.000494003295898, 153.0735626220703, 741.7922973632812, 43.816627502441406, 199.18942260742188, -184.95889282226562, 5.264961242675781, 24.30254364013672, 0.7962818145751953, 178.08355712890625, -8.114578247070312, -62.372047424316406, 157.34463500976562, 180.87318420410156, 151.26458740234375, -10.787918090820312, -193.4844512939453, -55.27613830566406, 36.7985954284668, 206.13333129882812, 134.23495483398438, -5.902009963989258, 301.406494140625, 170.43487548828125, 406.0721130371094, 39.66149139404297, -48.478763580322266, -10.970199584960938, 183.2744140625, 211.08409118652344, 246.97463989257812, 302.2300109863281, 217.16802978515625, 115.78520202636719, -110.86872863769531, 17.8675537109375, -3.2230377197265625, 23.13728904724121, 223.01467895507812, 228.0757293701172, -72.94148254394531, -2.393430709838867, 116.85066986083984, 216.22850036621094, 161.4757080078125, 40.85485076904297, 209.26004028320312, 286.5545959472656, 333.9847412109375, -209.1740264892578, 88.34739685058594], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000533.npy"}
|
||||
{"epoch": 0.8057445200302343, "step": 534, "batch_size": 64, "mean": 75.15812683105469, "std": 154.51028442382812, "min": -359.9790344238281, "p10": -83.7206527709961, "median": 71.43360137939453, "p90": 244.72378540039065, "max": 504.4884033203125, "pos_frac": 0.625, "sample": [115.11087799072266, 21.61281394958496, 422.099365234375, 41.390926361083984, -42.85101318359375, 91.00916290283203, -13.270553588867188, -269.9767761230469, 169.35768127441406, -90.27668762207031, -14.300836563110352, 108.334716796875, 247.09573364257812, 193.63824462890625, -158.12545776367188, 504.4884033203125, 134.94805908203125, -9.147335052490234, 108.90406799316406, -61.993408203125, 130.19821166992188, 380.4791259765625, 170.42518615722656, 124.81112670898438, -23.679893493652344, -359.9790344238281, -156.1025390625, -14.693288803100586, 263.8573913574219, 10.099655151367188, 162.7101593017578, -88.97665405273438, 257.4557189941406, 218.6229248046875, 221.39976501464844, 3.73126220703125, -82.595703125, 37.32715606689453, 400.52276611328125, 206.7364959716797, 121.50941467285156, 222.22482299804688, -27.68134307861328, -18.180999755859375, 167.7774200439453, 94.61447143554688, -3.525604248046875, 102.80555725097656, 35.253929138183594, -58.08512878417969, -12.574600219726562, 239.18923950195312, 116.0972900390625, -11.671680450439453, -56.53765869140625, 40.844818115234375, 183.916748046875, -67.55279541015625, 135.65948486328125, -84.20277404785156, 74.85688018798828, 68.01032257080078, 210.00128173828125, -23.026775360107422], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000534.npy"}
|
||||
{"epoch": 0.8072562358276644, "step": 535, "batch_size": 64, "mean": 72.41929626464844, "std": 152.9510955810547, "min": -242.90611267089844, "p10": -135.11899108886718, "median": 58.810142517089844, "p90": 281.14203186035155, "max": 404.3035888671875, "pos_frac": 0.65625, "sample": [58.318275451660156, -89.48344421386719, 168.9431610107422, 69.43794250488281, 29.922691345214844, 37.801883697509766, -212.4828338623047, -135.93576049804688, -164.00448608398438, -27.807373046875, 404.3035888671875, -115.611328125, 44.95378875732422, 185.99148559570312, -2.1917495727539062, -16.22362518310547, -19.015838623046875, 71.94830322265625, -1.1856517791748047, -31.177871704101562, 64.45884704589844, 121.56434631347656, 201.57827758789062, -161.7191162109375, 158.9727783203125, 59.30200958251953, 279.0042419433594, -160.56069946289062, -133.21319580078125, 3.920564651489258, 282.0582275390625, 310.99713134765625, 129.5826873779297, 5.722694396972656, 106.68666076660156, 197.1287078857422, 65.007080078125, 390.012939453125, 14.47412109375, 185.34014892578125, 26.216468811035156, -70.87608337402344, 376.9334411621094, 224.00347900390625, -18.918014526367188, -242.90611267089844, 122.25640869140625, 197.83013916015625, -24.377506256103516, 61.688758850097656, 64.81246185302734, -21.99322509765625, 34.583614349365234, 259.525146484375, 29.348709106445312, 250.9362335205078, -19.138221740722656, 213.46820068359375, 181.44180297851562, -145.47592163085938, 143.52633666992188, -58.339805603027344, 351.6477966308594, 321.8211669921875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000535.npy"}
|
||||
{"epoch": 0.8087679516250945, "step": 536, "batch_size": 64, "mean": 113.66795349121094, "std": 140.55970764160156, "min": -149.42503356933594, "p10": -21.208507728576656, "median": 76.54175567626953, "p90": 293.73800659179693, "max": 495.8575134277344, "pos_frac": 0.78125, "sample": [75.72964477539062, 49.44477081298828, -23.250770568847656, 217.5004119873047, 359.4565734863281, 58.769287109375, -8.01470947265625, -45.81620788574219, 212.45172119140625, 19.220415115356445, -135.68516540527344, 211.725830078125, 139.1341552734375, 286.069580078125, 191.8357696533203, 167.79278564453125, 236.68109130859375, 224.6845703125, 423.4678955078125, 79.20365905761719, 2.6831188201904297, 8.222963333129883, 238.79257202148438, 340.56982421875, 77.35386657714844, 275.55816650390625, 215.8098907470703, 10.042192459106445, 203.75360107421875, 180.90338134765625, 281.0006103515625, 4.660152435302734, -0.2585105895996094, 91.57150268554688, -47.173583984375, -59.923065185546875, 229.48313903808594, 9.242660522460938, 188.65589904785156, 28.751951217651367, 297.02447509765625, 245.52420043945312, 66.77410125732422, -6.500497817993164, 23.017230987548828, 13.17108154296875, 13.257575988769531, 10.884393692016602, 33.57986831665039, 495.8575134277344, -8.456090927124023, 196.9670867919922, 120.78854370117188, 382.8133850097656, 12.856155395507812, 47.71405029296875, -2.6857223510742188, 338.11688232421875, -16.443227767944336, -149.42503356933594, 80.31926727294922, 146.63140869140625, -83.03091430664062, -4.1086578369140625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000536.npy"}
|
||||
{"epoch": 0.8102796674225246, "step": 537, "batch_size": 64, "mean": 102.42313385009766, "std": 165.35552978515625, "min": -198.0934600830078, "p10": -98.94177017211913, "median": 93.59720611572266, "p90": 340.737582397461, "max": 462.00115966796875, "pos_frac": 0.640625, "sample": [171.70718383789062, 108.38042449951172, -52.358123779296875, 348.5052185058594, 394.8172607421875, 212.2024383544922, 81.95892333984375, 1.0615253448486328, 271.68914794921875, 95.41291809082031, 171.8083953857422, -12.821653366088867, 225.55091857910156, 276.5423583984375, 140.06492614746094, 303.96954345703125, 222.32240295410156, 4.947484970092773, -0.1349334716796875, -7.395078659057617, -2.277507781982422, 175.98968505859375, -7.0627899169921875, 5.930290222167969, -10.058977127075195, 413.7008056640625, 237.2181396484375, 50.56452178955078, -189.13153076171875, -19.712995529174805, 120.36235809326172, 91.781494140625, 181.3957977294922, -97.38920593261719, -174.8440399169922, 53.87888717651367, 440.0650634765625, 114.74874877929688, -44.520721435546875, -48.3512077331543, 68.38346862792969, 235.22528076171875, -198.0934600830078, 150.31044006347656, -37.388153076171875, -6.792028427124023, 366.5625, -63.78125762939453, 148.53794860839844, 216.1924285888672, -28.136077880859375, 114.51861572265625, -38.824302673339844, 189.39366149902344, 252.59231567382812, 462.00115966796875, -172.4130401611328, 322.61309814453125, -116.40380859375, 80.05888366699219, 245.43978881835938, 365.5859680175781, -151.41415405273438, -99.6071548461914], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000537.npy"}
|
||||
{"epoch": 0.8117913832199547, "step": 538, "batch_size": 64, "mean": 95.32467651367188, "std": 153.0237274169922, "min": -366.5926513671875, "p10": -57.088642883300764, "median": 81.44656753540039, "p90": 241.5744140625, "max": 563.7374877929688, "pos_frac": 0.796875, "sample": [24.159828186035156, 204.57211303710938, -27.561256408691406, 23.43014144897461, 146.56077575683594, -211.09429931640625, 65.64967346191406, -66.06977844238281, 145.1136016845703, 185.1940155029297, 58.913658142089844, 216.95513916015625, 6.66558837890625, 166.28150939941406, 3.607168197631836, 62.4197998046875, 202.46063232421875, 161.46531677246094, 48.143028259277344, 25.400043487548828, 96.04986572265625, 3.6180343627929688, 239.9920654296875, 402.5462951660156, 90.25055694580078, -113.05732727050781, 187.35513305664062, -118.94922637939453, 242.2525634765625, -19.066200256347656, -29.293731689453125, 228.525634765625, -36.132659912109375, 115.1549072265625, 15.011238098144531, 35.297882080078125, 20.337188720703125, 38.1318359375, 3.9206695556640625, 24.78631591796875, 441.0082092285156, 233.7694091796875, 345.6387634277344, 72.642578125, -155.9654541015625, 183.2281036376953, -132.54208374023438, 197.1878662109375, 222.0379180908203, 563.7374877929688, 203.6292724609375, 133.8626251220703, 280.35870361328125, -366.5926513671875, 130.26535034179688, 135.8209991455078, 336.1209411621094, -6.596080780029297, 124.91932678222656, 1.05712890625, 139.81436157226562, 98.24949645996094, 58.88269805908203, -8.753429412841797], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000538.npy"}
|
||||
{"epoch": 0.8133030990173847, "step": 539, "batch_size": 64, "mean": 74.76457977294922, "std": 146.95254516601562, "min": -216.62109375, "p10": -74.04322662353515, "median": 35.149179458618164, "p90": 236.1975524902344, "max": 603.7640380859375, "pos_frac": 0.71875, "sample": [64.19721984863281, 77.89248657226562, 18.600013732910156, 278.19537353515625, 15.785482406616211, -79.60498809814453, 231.31878662109375, 203.49691772460938, 163.70822143554688, 38.001243591308594, 77.97956848144531, -105.90338134765625, 127.8228759765625, 0.5682525634765625, -35.917022705078125, 6.0488739013671875, 16.113815307617188, -4.7996826171875, 258.02728271484375, 179.3497314453125, -7.053159713745117, -64.69960021972656, 78.54843139648438, -78.04763793945312, 18.265052795410156, -216.62109375, -12.597478866577148, 175.7103271484375, 379.9072265625, 42.330535888671875, 45.408287048339844, -36.945037841796875, -54.85627746582031, 18.346817016601562, 487.7364807128906, 68.22830963134766, 23.81922149658203, 124.49703979492188, 238.2884521484375, -10.585805892944336, -19.268173217773438, 36.891693115234375, 162.3504638671875, 33.57732391357422, 112.44140625, 145.32911682128906, -92.4639892578125, 134.2906494140625, 7.248847961425781, -31.429290771484375, 217.51068115234375, -10.987403869628906, 36.72103500366211, -139.70480346679688, 185.16331481933594, 155.41915893554688, 15.109909057617188, 19.422466278076172, -158.04649353027344, 159.20660400390625, 0.019472122192382812, 437.7107238769531, 603.7640380859375, 24.095306396484375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000539.npy"}
|
||||
{"epoch": 0.8148148148148148, "step": 540, "batch_size": 64, "mean": 120.58941650390625, "std": 158.08071899414062, "min": -189.978759765625, "p10": -53.77247161865233, "median": 101.23342895507812, "p90": 306.43393554687503, "max": 703.790283203125, "pos_frac": 0.78125, "sample": [1.59698486328125, -25.094131469726562, 703.790283203125, 213.5609130859375, 8.954572677612305, -189.978759765625, 199.94195556640625, 264.58685302734375, 111.24579620361328, -10.865001678466797, 25.303241729736328, 307.47412109375, 198.38465881347656, 37.22515106201172, -137.60655212402344, 78.24868774414062, 172.73776245117188, 150.50784301757812, 37.18204879760742, 246.0543670654297, 38.61601257324219, -87.43998718261719, 162.34719848632812, 46.135955810546875, 325.4588928222656, 149.13601684570312, -44.63916778564453, 260.3180847167969, -162.3575439453125, 26.435806274414062, 346.0362243652344, 209.31346130371094, 283.21331787109375, 203.6250762939453, 300.0713195800781, -15.099105834960938, -57.686744689941406, 223.73291015625, -122.84823608398438, 5.515260696411133, -3.147388458251953, 336.99713134765625, 182.0289306640625, 254.08209228515625, 235.09767150878906, 81.31828308105469, 304.0068359375, 13.406963348388672, 256.2524108886719, -110.7884292602539, 66.32186126708984, 341.2429504394531, 100.58367919921875, 23.476112365722656, -3.354949951171875, 57.286529541015625, -12.259300231933594, 327.181396484375, 23.104110717773438, 101.8831787109375, 272.7252502441406, 26.880247116088867, 105.35298156738281, 254.9088592529297], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000540.npy"}
|
||||
{"epoch": 0.8163265306122449, "step": 541, "batch_size": 64, "mean": 87.25973510742188, "std": 125.52012634277344, "min": -211.35342407226562, "p10": -87.983837890625, "median": 95.16609191894531, "p90": 218.5764343261719, "max": 334.4557189941406, "pos_frac": 0.703125, "sample": [212.41140747070312, 122.40895080566406, 158.43692016601562, 210.11151123046875, -48.25621032714844, 86.60772705078125, -90.61375427246094, 219.798828125, 193.8982696533203, 17.966278076171875, 170.8393096923828, -200.43533325195312, -34.039756774902344, 176.80838012695312, -11.581008911132812, -211.35342407226562, 194.9187469482422, 161.7174072265625, 76.30926513671875, 215.72418212890625, -104.0394058227539, 40.35869598388672, 213.2316131591797, 237.03366088867188, 182.296630859375, 147.68247985839844, 319.20025634765625, -31.238815307617188, -81.84736633300781, -9.967731475830078, 221.08572387695312, 45.39575958251953, 213.43528747558594, 334.4557189941406, 28.289024353027344, 170.0713348388672, 202.9298858642578, -22.592193603515625, 169.78810119628906, 11.957326889038086, -5.270299911499023, -105.936767578125, 168.47927856445312, 190.2594757080078, 32.63658905029297, 176.36839294433594, 200.53076171875, 238.58660888671875, -144.78318786621094, 86.71418762207031, -0.07004737854003906, 189.26832580566406, 90.31771087646484, 91.67601013183594, -47.22746276855469, -130.68310546875, 143.84536743164062, 98.65617370605469, -26.975839614868164, 52.309425354003906, 152.7727813720703, -9.370145797729492, 220.4281768798828, 12.886512756347656], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000541.npy"}
|
||||
{"epoch": 0.817838246409675, "step": 542, "batch_size": 64, "mean": 98.68695068359375, "std": 153.1207733154297, "min": -324.34576416015625, "p10": -88.95968170166016, "median": 115.77136611938477, "p90": 257.6123840332031, "max": 460.84307861328125, "pos_frac": 0.765625, "sample": [145.8972930908203, 256.5374755859375, 42.46569061279297, 425.97314453125, 128.3855743408203, 65.69143676757812, 148.61184692382812, 39.95512390136719, 241.52493286132812, 115.56890106201172, 8.925884246826172, -48.259521484375, -202.230224609375, 282.3493957519531, 25.082983016967773, 228.01611328125, 53.834800720214844, 460.84307861328125, -202.69529724121094, 210.06741333007812, 209.78948974609375, 328.4972839355469, 118.98735046386719, 10.128013610839844, -147.71218872070312, -324.34576416015625, 196.26629638671875, -67.86610412597656, 174.68858337402344, -89.31747436523438, 209.34107971191406, 226.4308319091797, 2.122404098510742, 257.616943359375, -129.75706481933594, -25.864215850830078, -17.5025634765625, 161.25674438476562, 8.689567565917969, 172.36285400390625, -22.16717529296875, 52.09004211425781, 231.79180908203125, -120.45829772949219, 116.88551330566406, 0.9574661254882812, 220.77517700195312, 63.101219177246094, 49.03553009033203, 313.1587829589844, 123.05309295654297, 257.60174560546875, -88.12483215332031, -14.124229431152344, -14.84721565246582, 44.21913528442383, 52.12351989746094, 231.81851196289062, 174.20184326171875, 43.900367736816406, 115.97383117675781, 356.7388000488281, 200.5181884765625, 227.38400268554688], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000542.npy"}
|
||||
{"epoch": 0.8193499622071051, "step": 543, "batch_size": 64, "mean": 105.5288314819336, "std": 154.14315795898438, "min": -262.4385070800781, "p10": -51.64688835144042, "median": 90.14621353149414, "p90": 281.66623840332034, "max": 487.385498046875, "pos_frac": 0.75, "sample": [-21.838668823242188, 28.32119369506836, 88.17567443847656, 2.4892635345458984, 170.4139862060547, -2.6235389709472656, 257.8514709472656, 251.00302124023438, -224.54412841796875, 487.385498046875, 181.82644653320312, 209.759033203125, 197.5489959716797, 275.84588623046875, 284.1606750488281, 174.23265075683594, 197.9970703125, -120.43788146972656, 92.11675262451172, 259.8405456542969, 10.247854232788086, -33.78663635253906, 27.543537139892578, 34.87835693359375, 12.920269012451172, 207.36224365234375, -262.4385070800781, 2.1989669799804688, -148.84339904785156, -87.58708953857422, -3.5832366943359375, 155.39202880859375, 332.42230224609375, -19.360034942626953, 200.63584899902344, 221.5774688720703, 216.00086975097656, 3.3403377532958984, -20.838096618652344, -28.143104553222656, 86.67582702636719, 107.03301239013672, 472.5794982910156, 47.43609619140625, 207.3179473876953, 72.20438385009766, 35.163963317871094, 423.6064147949219, -12.456058502197266, 149.44895935058594, -39.46165084838867, 139.55154418945312, 193.00515747070312, -56.86913299560547, 344.826171875, 116.02621459960938, 151.932861328125, 50.72862243652344, 13.826812744140625, -100.09028625488281, 250.80213928222656, 346.4673767089844, 7.587116241455078, 137.03854370117188], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000543.npy"}
|
||||
{"epoch": 0.8208616780045351, "step": 544, "batch_size": 64, "mean": 119.11622619628906, "std": 179.4794464111328, "min": -222.04469299316406, "p10": -108.8839385986328, "median": 120.4982681274414, "p90": 367.65511474609383, "max": 657.2489624023438, "pos_frac": 0.734375, "sample": [61.09100341796875, -215.19505310058594, 103.21795654296875, 32.96173095703125, 138.00726318359375, 88.87640380859375, -149.64393615722656, 41.593605041503906, -89.64596557617188, -72.72301483154297, -12.444671630859375, 116.32862854003906, -117.1287841796875, 407.23681640625, 172.78265380859375, 27.82008934020996, -205.57931518554688, 187.86978149414062, 162.20285034179688, 183.07125854492188, 128.24429321289062, 190.81893920898438, 266.83074951171875, 191.5094451904297, 319.0668640136719, 5.886850357055664, 97.98829650878906, -30.946853637695312, -3.035036087036133, -222.04469299316406, 452.71380615234375, -150.08570861816406, 208.43470764160156, -15.859275817871094, 222.38685607910156, 287.32196044921875, 657.2489624023438, -16.01165771484375, 104.87323760986328, 165.6514892578125, 174.5803680419922, 44.51152801513672, 377.35186767578125, 97.73419952392578, 68.12010955810547, 124.66790771484375, 172.61146545410156, 189.89938354492188, 280.6945495605469, -135.41702270507812, 34.2486572265625, 196.02593994140625, 470.0914306640625, 417.18878173828125, -62.34449005126953, -71.93995666503906, 345.02935791015625, -29.80255126953125, 266.1713562011719, 145.37445068359375, 3.175506591796875, 146.45343017578125, 197.52862548828125, 449.7906494140625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000544.npy"}
|
||||
{"epoch": 0.8223733938019653, "step": 545, "batch_size": 64, "mean": 61.362571716308594, "std": 150.39944458007812, "min": -337.72576904296875, "p10": -112.09016494750975, "median": 35.633358001708984, "p90": 253.15842132568363, "max": 356.3799133300781, "pos_frac": 0.6875, "sample": [152.3137664794922, 52.04065704345703, -132.08164978027344, 256.5658264160156, 21.806865692138672, -19.890655517578125, 2.8426742553710938, -156.16586303710938, -62.735069274902344, 0.10265731811523438, 350.6115417480469, 13.112335205078125, 218.06614685058594, 49.01670837402344, 100.05730438232422, -13.572898864746094, 100.835693359375, 245.2078094482422, -121.62545776367188, 84.1663818359375, -77.10547637939453, 216.31146240234375, -43.706581115722656, 294.84454345703125, 256.6576232910156, -77.66963958740234, 128.13638305664062, 41.028724670410156, -48.63383483886719, 15.236509323120117, -11.389873504638672, -85.59811401367188, -337.72576904296875, 167.16934204101562, -29.61199951171875, 23.10935401916504, 204.2351837158203, 236.42483520507812, 168.1370391845703, 8.265907287597656, -20.678462982177734, 356.3799133300781, 219.83250427246094, 30.237991333007812, 45.94921875, 126.9443130493164, -89.84114837646484, 156.52667236328125, 195.12112426757812, -259.9130859375, 61.32281494140625, 2.0994873046875, -4.509765625, 353.41302490234375, 111.65510559082031, -128.68690490722656, 8.28226089477539, 181.5975341796875, -290.9407043457031, 2.0777835845947266, 180.36770629882812, 185.58114624023438, 296.6687316894531, 18.927043914794922], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000545.npy"}
|
||||
{"epoch": 0.8238851095993953, "step": 546, "batch_size": 64, "mean": 68.55207061767578, "std": 158.21592712402344, "min": -382.0150146484375, "p10": -144.1125030517578, "median": 72.95977401733398, "p90": 217.63687133789062, "max": 415.2786865234375, "pos_frac": 0.703125, "sample": [230.70819091796875, -128.31219482421875, 206.9408416748047, 178.94119262695312, 70.54750061035156, 181.18093872070312, -64.6263427734375, -103.85520935058594, 143.69131469726562, -142.5353240966797, -382.0150146484375, 202.05435180664062, 14.461490631103516, 59.62782287597656, -211.32247924804688, 383.8797302246094, 20.514570236206055, -25.086692810058594, -80.95492553710938, -1.6578140258789062, -205.0169677734375, 42.253509521484375, 147.27615356445312, 192.41024780273438, 18.522621154785156, 85.82855224609375, 203.06292724609375, -22.996337890625, 200.61996459960938, 263.6234130859375, 199.87716674804688, -92.06511688232422, 185.9742431640625, 114.21653747558594, 63.24470138549805, -216.83642578125, 75.3720474243164, 415.2786865234375, 167.45457458496094, 218.57180786132812, 5.505046844482422, 67.5848617553711, -144.78843688964844, 215.45535278320312, -6.505865097045898, 204.96847534179688, -22.704017639160156, 41.23527908325195, 104.0518569946289, 156.2460479736328, 143.19395446777344, 66.61997985839844, 302.9425048828125, -307.3260498046875, -1.3750076293945312, 99.24240112304688, 202.5963134765625, 123.57421875, 208.94210815429688, 53.21033477783203, 246.94845581054688, -191.09518432617188, 196.56675720214844, 13.388975143432617], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000546.npy"}
|
||||
{"epoch": 0.8253968253968254, "step": 547, "batch_size": 64, "mean": 108.49510955810547, "std": 155.18724060058594, "min": -320.8662414550781, "p10": -41.578947448730446, "median": 84.71110916137695, "p90": 306.60527038574224, "max": 554.3244018554688, "pos_frac": 0.828125, "sample": [5.559356689453125, 57.11979675292969, 14.492055892944336, 33.00517272949219, 394.66046142578125, 125.43470001220703, -114.72986602783203, 354.6850891113281, 175.78549194335938, 166.25164794921875, 310.36492919921875, 169.77899169921875, 33.40099334716797, 132.9458770751953, 37.07515335083008, -19.292274475097656, 188.77830505371094, 213.5429229736328, 57.7192497253418, 49.535491943359375, 350.6500244140625, 233.4611358642578, 207.8773193359375, 62.70623779296875, 313.26715087890625, 192.72003173828125, -76.94146728515625, 179.51800537109375, -8.427734375, 30.839447021484375, 198.8981475830078, -5.872629165649414, -320.8662414550781, 219.93600463867188, 85.22132873535156, 73.76799011230469, 233.02120971679688, 16.858379364013672, 277.54791259765625, 40.27618408203125, 107.7021713256836, 224.22238159179688, 151.5081329345703, 185.3162078857422, -205.71768188476562, 97.19303894042969, 60.399559020996094, -308.716552734375, 83.24542236328125, 31.846664428710938, 293.0015869140625, 229.2572479248047, 554.3244018554688, 0.02033233642578125, 131.73419189453125, -60.25, -5.750118255615234, 297.8327331542969, 2.4011974334716797, 84.20088958740234, 15.95924186706543, 4.0064239501953125, 330.508056640625, -51.13037872314453], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000547.npy"}
|
||||
{"epoch": 0.8269085411942555, "step": 548, "batch_size": 64, "mean": 67.02081298828125, "std": 148.81980895996094, "min": -339.645751953125, "p10": -114.88126068115233, "median": 37.010562896728516, "p90": 262.28125305175786, "max": 481.1037902832031, "pos_frac": 0.65625, "sample": [166.15919494628906, 198.36627197265625, -87.75607299804688, -147.14991760253906, -161.03341674804688, 153.09059143066406, -9.992881774902344, 275.6430969238281, 210.96900939941406, 57.911964416503906, -78.30706787109375, 221.32174682617188, 0.7612800598144531, 34.54192352294922, 131.20196533203125, -15.82496452331543, -135.28067016601562, 40.30811309814453, -22.786148071289062, 264.5217590332031, 312.9732971191406, 62.20417785644531, -95.27003479003906, 39.47920227050781, 189.52392578125, 69.06764221191406, 304.7125549316406, 244.06854248046875, 8.571685791015625, 174.17239379882812, -123.28607177734375, 162.68621826171875, 10.066314697265625, 254.03993225097656, 68.47410583496094, -30.028274536132812, 275.16436767578125, 18.039260864257812, -55.849700927734375, 150.790283203125, -7.009864807128906, -25.20403480529785, 7.785911560058594, -339.645751953125, 481.1037902832031, 31.899169921875, -40.914772033691406, 64.56141662597656, -151.88136291503906, 121.80304718017578, 89.75137329101562, -163.95713806152344, 31.501441955566406, 179.64996337890625, 286.37384033203125, -7.804527282714844, 190.78489685058594, -12.859710693359375, 34.29078674316406, -72.74037170410156, 257.05340576171875, 253.5713348388672, -71.11761474609375, 16.071372985839844], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000548.npy"}
|
||||
{"epoch": 0.8284202569916855, "step": 549, "batch_size": 64, "mean": 77.1980209350586, "std": 185.53387451171875, "min": -418.5877990722656, "p10": -139.66795196533204, "median": 62.20751762390137, "p90": 256.82032775878906, "max": 689.4319458007812, "pos_frac": 0.6875, "sample": [-152.61721801757812, 44.96625518798828, 8.809822082519531, 82.7837905883789, 37.973121643066406, -320.2158203125, 109.6732406616211, 6.337808609008789, 12.739051818847656, 259.5361022949219, 263.3389892578125, -0.30184173583984375, 89.53112030029297, 263.3625183105469, 55.883270263671875, -116.12317657470703, -138.5740203857422, 94.64862060546875, 187.24612426757812, 221.37095642089844, 232.24954223632812, 166.46633911132812, 557.701171875, -156.07713317871094, 104.80502319335938, 21.808074951171875, 62.80668258666992, 239.02081298828125, 171.3835906982422, -60.02001190185547, 227.21871948242188, 250.4835205078125, 689.4319458007812, -53.92884826660156, 175.61727905273438, 397.5126647949219, -26.249975204467773, -322.8897399902344, -36.91871643066406, 197.85025024414062, -418.5877990722656, 61.60835266113281, -19.319541931152344, 86.10861206054688, 54.53132247924805, 180.68116760253906, -13.777374267578125, 147.05526733398438, 357.0657043457031, 140.59365844726562, -0.09328460693359375, 190.5155792236328, 233.83450317382812, 177.0323944091797, 78.56576538085938, -17.840011596679688, 9.325109481811523, -10.098865509033203, -61.823184967041016, 36.89635467529297, 210.61293029785156, -229.115478515625, -140.13677978515625, 38.39906311035156], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000549.npy"}
|
||||
{"epoch": 0.8299319727891157, "step": 550, "batch_size": 64, "mean": 59.847496032714844, "std": 144.7289581298828, "min": -266.098388671875, "p10": -126.85975952148438, "median": 30.269164085388184, "p90": 237.14650726318368, "max": 480.2826232910156, "pos_frac": 0.6875, "sample": [119.39776611328125, 268.193115234375, 215.08314514160156, 0.637359619140625, -127.08541870117188, 39.564125061035156, 1.5762882232666016, -35.44432067871094, -2.0174331665039062, 44.923736572265625, 201.71218872070312, 101.05518341064453, 3.088897705078125, -4.954048156738281, 26.435808181762695, 296.1812744140625, 246.60223388671875, 131.1834716796875, 76.49673461914062, 10.83401107788086, 185.44180297851562, -42.972774505615234, 195.66891479492188, 124.47097778320312, 280.3006591796875, -68.87086486816406, 421.4330139160156, -19.215007781982422, 18.287506103515625, -175.74449157714844, -174.35055541992188, 65.21345520019531, -126.33322143554688, 144.2683563232422, 2.13031005859375, 108.3826904296875, 128.31314086914062, 86.21197509765625, -39.15306854248047, 23.08331298828125, -78.634765625, 126.74250793457031, 29.873950958251953, 170.4908905029297, -49.81105041503906, 170.5138702392578, 9.716156005859375, -2.4833812713623047, 93.88446044921875, 198.47911071777344, -156.58584594726562, 111.71107482910156, 480.2826232910156, 30.664377212524414, 172.41241455078125, 13.055337905883789, -266.098388671875, 26.0673885345459, 312.03228759765625, -202.52728271484375, 175.58164978027344, -132.55142211914062, -55.50856399536133, -97.09809875488281], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000550.npy"}
|
||||
{"epoch": 0.8314436885865457, "step": 551, "batch_size": 64, "mean": 74.19692993164062, "std": 152.1179656982422, "min": -327.74005126953125, "p10": -38.862120819091786, "median": 47.22571563720703, "p90": 227.8930847167969, "max": 776.7337036132812, "pos_frac": 0.6875, "sample": [-28.335594177246094, 61.8682861328125, 112.28730773925781, 57.00213623046875, 71.38694763183594, 44.011741638183594, 776.7337036132812, 24.695674896240234, 57.85877990722656, 115.92113494873047, -15.63052749633789, 57.562713623046875, 194.87408447265625, -12.596746444702148, 74.92434692382812, -3.227558135986328, 396.2861633300781, 58.608848571777344, -327.74005126953125, -29.94567108154297, -42.68345642089844, 4.178962707519531, 118.2072525024414, 76.44833374023438, 98.8897705078125, 202.20697021484375, -66.35160827636719, 43.26264190673828, 72.23356628417969, 190.77696228027344, -18.843297958374023, -106.86913299560547, 2.3613662719726562, -114.94219970703125, 41.40350341796875, 482.6270751953125, 8.345489501953125, 167.51171875, -43.168304443359375, -13.620254516601562, 19.98956298828125, 68.97832489013672, -26.77539825439453, -8.30764389038086, 186.635009765625, 317.078857421875, 225.80667114257812, 8.820026397705078, 41.305633544921875, 7.076137542724609, 99.99781036376953, 228.78726196289062, 85.392578125, -18.50424575805664, 50.43968963623047, -11.639511108398438, -16.876453399658203, 277.51751708984375, 54.714691162109375, 21.80306625366211, -95.2200927734375, -4.512104034423828, 191.3291473388672, 256.2460632324219], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000551.npy"}
|
||||
{"epoch": 0.8329554043839759, "step": 552, "batch_size": 64, "mean": 85.6976089477539, "std": 143.1346893310547, "min": -263.33807373046875, "p10": -57.955185699462874, "median": 75.66669082641602, "p90": 259.4958145141602, "max": 472.1492004394531, "pos_frac": 0.796875, "sample": [40.50349426269531, 84.83255004882812, 31.26850128173828, -65.2815170288086, 172.4639129638672, 37.829463958740234, 5.419973373413086, 294.55596923828125, 22.247726440429688, 169.27383422851562, 472.1492004394531, 402.742431640625, 119.34175109863281, 74.18673706054688, -27.046409606933594, -94.3228530883789, 31.169891357421875, 171.7427520751953, 87.47016143798828, -33.44068908691406, 92.17195129394531, 76.06297302246094, 81.89640808105469, 172.8400115966797, 195.6071319580078, -40.86041259765625, 80.74080657958984, 179.1031494140625, 316.604736328125, -259.3583679199219, 0.8815212249755859, -8.955245971679688, 19.800384521484375, -225.43658447265625, 79.9169692993164, -2.843414306640625, 399.9443054199219, 59.08055114746094, 12.068000793457031, -38.19871520996094, 228.84347534179688, 43.12347412109375, -116.76721954345703, 154.7307586669922, 349.2934875488281, 35.440765380859375, 48.061622619628906, 88.5611801147461, -263.33807373046875, 267.0792236328125, 239.5767364501953, 96.4814224243164, 25.766202926635742, -130.39816284179688, 162.70428466796875, 75.2704086303711, 125.73955535888672, 139.63528442382812, 241.8011932373047, 162.27008056640625, 55.282798767089844, 32.48174285888672, 196.3312225341797, 38.50230407714844], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000552.npy"}
|
||||
{"epoch": 0.8344671201814059, "step": 553, "batch_size": 64, "mean": 89.30570983886719, "std": 173.42564392089844, "min": -349.7249755859375, "p10": -72.01437454223633, "median": 59.11882019042969, "p90": 335.31622924804697, "max": 563.9495239257812, "pos_frac": 0.6875, "sample": [361.44940185546875, -44.61176300048828, -16.454513549804688, 143.03189086914062, 8.852806091308594, 11.107488632202148, 184.03359985351562, 563.9495239257812, -1.6209182739257812, -28.868133544921875, 27.990196228027344, 478.7879638671875, 281.3104553222656, 375.31146240234375, -5.635169982910156, 189.762451171875, -27.15138053894043, 26.317604064941406, 83.54833984375, -246.70004272460938, -7.353582382202148, 140.91246032714844, -20.73700714111328, 193.39755249023438, 214.20941162109375, 40.757530212402344, 59.058998107910156, 83.03976440429688, 21.2755126953125, -73.68020629882812, 404.22271728515625, 224.57022094726562, 75.1439437866211, 524.4735717773438, 60.84869384765625, 150.82736206054688, 210.45591735839844, 246.7491912841797, 245.43130493164062, 59.17864227294922, 16.605445861816406, 89.17915344238281, 14.536283493041992, -5.663875579833984, -98.86614990234375, 101.27286529541016, -0.9965476989746094, 314.7454833984375, -215.7913818359375, -181.83087158203125, 66.30450439453125, 202.62062072753906, 113.62255859375, -53.50554656982422, -38.58765411376953, 25.307693481445312, 39.55081558227539, -349.7249755859375, 164.59124755859375, 68.7156753540039, -68.12743377685547, -101.27836608886719, 344.13226318359375, 51.56065368652344], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000553.npy"}
|
||||
{"epoch": 0.8359788359788359, "step": 554, "batch_size": 64, "mean": 99.23629760742188, "std": 141.9836883544922, "min": -437.962890625, "p10": -58.98865165710448, "median": 116.68282699584961, "p90": 236.74808654785156, "max": 364.41534423828125, "pos_frac": 0.78125, "sample": [109.37503051757812, 295.5787353515625, 247.15386962890625, 364.41534423828125, -33.51625061035156, 136.76577758789062, 7.321321487426758, 7.681735992431641, 144.08348083496094, 223.57949829101562, -41.896827697753906, 123.9906234741211, -19.644454956054688, 170.55355834960938, -220.6668243408203, 53.92620086669922, 127.01878356933594, -16.97332763671875, 224.46368408203125, 66.1868896484375, 7.900611877441406, 185.29437255859375, 78.09536743164062, 5.367767333984375, 315.96966552734375, 203.68203735351562, 234.880615234375, 101.27482604980469, -1.8793907165527344, 58.44072723388672, 230.62405395507812, -77.85450744628906, 196.66372680664062, 173.6099090576172, 10.28350830078125, -106.3361587524414, 220.411865234375, 51.12611389160156, 234.5562744140625, -3.62017822265625, -437.962890625, 267.70257568359375, 96.15449523925781, 150.18128967285156, -65.42094421386719, 165.28884887695312, 39.40393829345703, 159.73434448242188, 90.92750549316406, 4.25006103515625, 222.1436767578125, -69.0585708618164, -43.9799690246582, 178.04385375976562, 234.3831024169922, 49.07184600830078, 237.54843139648438, 196.5145263671875, 126.69811248779297, 233.79656982421875, 363.69384765625, -162.636962890625, 194.78390502929688, 31.973220825195312], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000554.npy"}
|
||||
{"epoch": 0.8374905517762661, "step": 555, "batch_size": 64, "mean": 68.56673431396484, "std": 157.5604705810547, "min": -425.5113525390625, "p10": -157.15854034423828, "median": 80.97270965576172, "p90": 250.74394683837892, "max": 333.72308349609375, "pos_frac": 0.703125, "sample": [-0.5048847198486328, -270.0225524902344, 82.40512084960938, 98.63218688964844, 300.7500305175781, 9.651298522949219, -425.5113525390625, -3.406463623046875, -183.58682250976562, 168.4614715576172, 258.9960021972656, -155.39173889160156, 218.5849609375, 28.690448760986328, 186.99595642089844, 297.4682922363281, 23.20555877685547, -7.183372497558594, -6.1438446044921875, 288.7276306152344, 206.78369140625, 19.658370971679688, 262.49932861328125, 79.29277038574219, 218.6082763671875, 79.54029846191406, 126.65806579589844, 221.08486938476562, 141.43128967285156, 247.55606079101562, 216.24754333496094, 204.02716064453125, 76.61726379394531, 156.3494873046875, -157.91574096679688, 252.1101837158203, -246.56954956054688, 195.9068603515625, -60.893531799316406, 3.8052406311035156, 9.1448974609375, 40.351951599121094, -54.66027069091797, 83.5578384399414, -159.18580627441406, -28.35928726196289, 85.76071166992188, 125.6546401977539, 191.65985107421875, -40.63041687011719, 188.71925354003906, 68.28638458251953, -305.9901123046875, -4.463172912597656, 194.801513671875, 126.49708557128906, 43.68666458129883, 104.44638061523438, 160.84054565429688, -137.518310546875, 333.72308349609375, 26.673147201538086, 183.37112426757812, -1.712564468383789], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000555.npy"}
|
||||
{"epoch": 0.8390022675736961, "step": 556, "batch_size": 64, "mean": 78.46257019042969, "std": 145.9959259033203, "min": -269.3580017089844, "p10": -63.561038970947266, "median": 53.96899223327637, "p90": 261.66493225097656, "max": 501.1920166015625, "pos_frac": 0.703125, "sample": [234.53939819335938, 40.68939208984375, 110.21240234375, -24.665470123291016, 25.813522338867188, 216.17391967773438, 259.5388488769531, 111.96957397460938, -27.48497200012207, -30.957763671875, 18.508317947387695, -16.03106689453125, 63.332908630371094, 312.3273620605469, 9.417049407958984, 29.903656005859375, 122.60348510742188, -4.767173767089844, 104.94419860839844, 157.1075439453125, -61.437889099121094, 192.6876678466797, 236.868408203125, 93.0930404663086, -28.05869483947754, 52.693702697753906, -269.3580017089844, 16.162378311157227, 147.2481689453125, -152.40277099609375, 7.356973648071289, -213.06369018554688, 119.12724304199219, 145.20614624023438, 133.14015197753906, 2.809202194213867, 198.2232208251953, 197.15560913085938, 21.78649139404297, 12.783231735229492, 191.3673858642578, 501.1920166015625, 16.34072494506836, -41.33625030517578, 389.1451416015625, 334.1650390625, -48.8839111328125, 18.5084228515625, -180.9024200439453, -0.1533203125, 262.57611083984375, -60.33685302734375, -61.84294891357422, 299.58453369140625, 55.24428176879883, -111.38186645507812, 139.2066650390625, -123.50302124023438, 120.19288635253906, 212.66839599609375, 216.4185791015625, 264.6336975097656, 127.8025894165039, -64.29736328125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000556.npy"}
|
||||
{"epoch": 0.8405139833711263, "step": 557, "batch_size": 64, "mean": 111.04303741455078, "std": 147.02940368652344, "min": -158.74041748046875, "p10": -46.74002571105956, "median": 78.83616256713867, "p90": 278.8652679443359, "max": 626.138916015625, "pos_frac": 0.765625, "sample": [250.0145263671875, 114.66401672363281, 184.593505859375, 18.602798461914062, -32.7511100769043, 12.4566650390625, 226.84669494628906, 82.11669158935547, 227.06082153320312, 199.99420166015625, -52.73527526855469, -0.5916900634765625, -11.081611633300781, 55.113136291503906, -2.856700897216797, 142.4867706298828, 29.80685806274414, -12.872764587402344, 212.0064697265625, 134.32870483398438, 39.5498161315918, 30.0816650390625, 277.0805969238281, 212.5218505859375, -16.937911987304688, -130.0186767578125, 186.87481689453125, 305.0277404785156, 39.86240768432617, -127.05741119384766, 485.8996276855469, 146.7635955810547, 171.99880981445312, 3.223430633544922, -55.30906677246094, 223.79629516601562, 184.19500732421875, 162.87176513671875, 51.89707946777344, 359.292724609375, -64.44331359863281, 34.514678955078125, -20.650108337402344, 15.0374755859375, -4.513277053833008, -158.74041748046875, 626.138916015625, 390.4249572753906, 208.14593505859375, 43.129051208496094, 26.780418395996094, 194.81053161621094, 35.23518371582031, 197.71401977539062, 191.02847290039062, 155.1436767578125, 284.31634521484375, 60.82125473022461, 75.55563354492188, 121.89239501953125, -122.63262939453125, 2.1211090087890625, 279.630126953125, 206.47683715820312], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000557.npy"}
|
||||
{"epoch": 0.8420256991685563, "step": 558, "batch_size": 64, "mean": 116.85623931884766, "std": 163.41110229492188, "min": -211.45205688476562, "p10": -65.3848663330078, "median": 84.45694732666016, "p90": 307.9159912109376, "max": 592.963623046875, "pos_frac": 0.734375, "sample": [217.0765380859375, 284.76275634765625, 287.1994934082031, -37.98029327392578, 205.346435546875, 18.482650756835938, 214.16445922851562, -33.45126724243164, 5.442771911621094, 315.6907958984375, -51.35620880126953, 237.9827880859375, -143.37107849121094, 51.45172882080078, 5.44866943359375, 173.44239807128906, 215.18789672851562, -7.4237060546875, 174.65309143066406, 82.01800537109375, 592.963623046875, 224.1016082763672, -18.837657928466797, 208.8408966064453, 160.18804931640625, -140.8760223388672, 335.1927795410156, 58.468833923339844, 323.6943359375, 45.638938903808594, 188.58599853515625, 80.28375244140625, 224.15542602539062, 289.7747802734375, -1.7173004150390625, 7.166067123413086, 156.10311889648438, 38.040863037109375, 46.33732604980469, 403.982177734375, 170.56423950195312, -125.96255493164062, 530.270263671875, 86.89588928222656, 212.60845947265625, 205.32550048828125, -33.882080078125, -68.63262176513672, 380.1343994140625, -10.5762939453125, 260.5604248046875, -211.45205688476562, 49.95367431640625, -167.0252685546875, 206.23880004882812, -26.0302734375, 148.37557983398438, 16.238975524902344, -57.80677032470703, 15.5614013671875, 269.1302795410156, 73.63420867919922, 192.46002197265625, -74.64056396484375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000558.npy"}
|
||||
{"epoch": 0.8435374149659864, "step": 559, "batch_size": 64, "mean": 107.38188171386719, "std": 185.38734436035156, "min": -220.50115966796875, "p10": -81.95881423950195, "median": 56.9778938293457, "p90": 310.4173980712891, "max": 815.6401977539062, "pos_frac": 0.671875, "sample": [55.87782287597656, -106.1428451538086, -22.995929718017578, -31.196434020996094, 126.4049301147461, 815.6401977539062, 349.71002197265625, 485.56597900390625, 41.091331481933594, -22.196758270263672, -220.50115966796875, 440.9667663574219, 58.077964782714844, 265.415771484375, 15.321439743041992, 237.0689697265625, 277.6762390136719, 32.688026428222656, 215.5270233154297, 93.91651916503906, 241.2918701171875, -3.1481170654296875, -85.64244079589844, 218.10955810546875, -11.482154846191406, -33.059974670410156, -177.7464141845703, 196.61367797851562, 106.28466796875, 216.54229736328125, 97.21072387695312, 159.04136657714844, -59.183998107910156, 481.57086181640625, 260.36151123046875, -73.36368560791016, 146.6373748779297, 19.7729549407959, -107.27713775634766, -38.713890075683594, -30.41571807861328, 24.162635803222656, -16.85274887084961, 16.392059326171875, 302.59100341796875, 261.3377380371094, 195.0959930419922, -8.346443176269531, 167.1026611328125, 35.07383728027344, 21.45380973815918, 299.8056640625, -172.1410369873047, 86.82072448730469, 171.2996063232422, 313.0442199707031, 40.585811614990234, 325.2864685058594, -3.1997604370117188, 18.679841995239258, 138.49908447265625, 304.28814697265625, -70.81224822998047, -209.0460662841797], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000559.npy"}
|
||||
{"epoch": 0.8450491307634165, "step": 560, "batch_size": 64, "mean": 98.25459289550781, "std": 131.8840789794922, "min": -118.3954086303711, "p10": -72.28191909790038, "median": 101.9729232788086, "p90": 238.02785797119142, "max": 475.00885009765625, "pos_frac": 0.75, "sample": [102.26239776611328, 50.525482177734375, 216.10597229003906, 69.6117935180664, -107.70877075195312, -46.20692443847656, 226.51861572265625, 30.691574096679688, 401.6328430175781, -0.11679649353027344, 139.32435607910156, 14.522773742675781, 159.67721557617188, 28.063140869140625, -17.781158447265625, 55.348663330078125, -6.5647735595703125, 9.750452041625977, 51.06319046020508, 148.5320281982422, -98.22027587890625, 212.42825317382812, 107.71398162841797, 5.7391815185546875, 214.0982666015625, 156.18780517578125, 38.16631317138672, 224.12216186523438, 232.2196502685547, -21.847742080688477, -32.798675537109375, -26.338172912597656, -81.31336212158203, 280.8778076171875, 226.05885314941406, -72.84332275390625, 86.64727020263672, -102.67805480957031, 210.1887664794922, -70.97197723388672, 240.51708984375, 224.54005432128906, 113.3257064819336, 153.772216796875, 47.50928497314453, 129.94781494140625, 9.469734191894531, 217.6047821044922, 59.44991683959961, -30.995101928710938, 105.04946899414062, 475.00885009765625, -118.3954086303711, 14.284530639648438, 417.7669677734375, 282.04034423828125, -114.27803039550781, 101.752197265625, 201.8585205078125, 127.56314086914062, 251.47210693359375, 102.19364929199219, 103.7974853515625, 160.34986877441406], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000560.npy"}
|
||||
{"epoch": 0.8465608465608465, "step": 561, "batch_size": 64, "mean": 80.63005065917969, "std": 147.78367614746094, "min": -227.67408752441406, "p10": -110.32618713378905, "median": 38.38784408569336, "p90": 273.74470214843757, "max": 450.6033020019531, "pos_frac": 0.671875, "sample": [6.347480773925781, 27.00128173828125, 31.65704345703125, -12.173669815063477, 32.66737365722656, 255.4561767578125, 119.28701782226562, -22.081260681152344, 200.51019287109375, 70.454833984375, 30.41888427734375, 199.64564514160156, -10.788280487060547, 85.02885437011719, 36.477020263671875, -130.74609375, 65.94050598144531, 196.89028930664062, 250.76443481445312, -0.9481048583984375, -38.98027038574219, 362.6855773925781, 163.4717254638672, -90.93113708496094, 121.31082153320312, -179.7990264892578, 181.59210205078125, -31.035282135009766, -227.67408752441406, 404.88134765625, -4.336215972900391, 35.28923034667969, 334.022705078125, 281.5826416015625, 53.79765319824219, 73.07394409179688, 69.63532257080078, 39.87580871582031, 3.1441478729248047, 450.6033020019531, 351.7911376953125, 342.452880859375, 210.23568725585938, 242.97645568847656, 197.52163696289062, -131.66180419921875, -172.20123291015625, -1.5571823120117188, 118.26240539550781, -5.877153396606445, -116.52371215820312, 168.51138305664062, 36.899879455566406, -12.294683456420898, 32.328399658203125, 227.05079650878906, 2.598722457885742, 156.264892578125, 108.23333740234375, -122.44076538085938, -0.9042186737060547, -3.897336959838867, 194.39892578125, -95.86529541015625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000561.npy"}
|
||||
{"epoch": 0.8480725623582767, "step": 562, "batch_size": 64, "mean": 115.04664611816406, "std": 175.64068603515625, "min": -259.80438232421875, "p10": -79.14426422119139, "median": 115.16543197631836, "p90": 306.4789520263672, "max": 734.4572143554688, "pos_frac": 0.703125, "sample": [33.012481689453125, 232.96388244628906, -85.33856201171875, 475.8678894042969, 281.07568359375, 248.57928466796875, 137.85330200195312, 734.4572143554688, 175.67037963867188, 1.5954818725585938, 121.46897888183594, 135.5509033203125, -172.4110565185547, -4.536304473876953, 170.82830810546875, -6.431423187255859, 137.69691467285156, 0.463226318359375, 127.42864990234375, -123.84950256347656, 226.14254760742188, 211.33270263671875, -16.615005493164062, -7.857734680175781, 42.6881103515625, 211.02320861816406, -19.19222068786621, 113.40068817138672, 48.33612060546875, -4.436420440673828, 206.58729553222656, -110.0931396484375, 228.36111450195312, 351.4189453125, 104.02694702148438, 602.0421752929688, -7.738254547119141, -64.69090270996094, 1.7926750183105469, 318.7427062988281, -7.320213317871094, 40.54277801513672, -0.44357872009277344, 119.62910461425781, 309.9013671875, 229.22219848632812, 298.4933166503906, 5.1263427734375, 143.54989624023438, -101.5426025390625, 175.62255859375, 168.54690551757812, -7.940439224243164, 115.17890167236328, 172.49549865722656, 115.15196228027344, 209.7021942138672, 473.89312744140625, 181.40567016601562, -133.38134765625, 67.99069213867188, -259.80438232421875, -18.433815002441406, 8.181869506835938], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000562.npy"}
|
||||
{"epoch": 0.8495842781557067, "step": 563, "batch_size": 64, "mean": 93.92084503173828, "std": 148.3907470703125, "min": -277.3353271484375, "p10": -73.8696563720703, "median": 86.27223587036133, "p90": 268.47344970703125, "max": 362.31964111328125, "pos_frac": 0.75, "sample": [155.6494140625, 201.60142517089844, 35.17495346069336, 0.085296630859375, 245.54127502441406, 177.91702270507812, -252.04705810546875, 274.6119689941406, 289.83868408203125, 84.68242645263672, 35.034122467041016, 271.31414794921875, 57.595802307128906, -52.23939514160156, 207.57371520996094, 43.58489990234375, -199.16461181640625, 184.55831909179688, -52.29229736328125, 325.88519287109375, 44.180419921875, 362.31964111328125, -35.83140563964844, 288.12176513671875, 178.17694091796875, 129.81143188476562, 155.13352966308594, 6.664466857910156, 42.0718994140625, -205.7481689453125, 25.711959838867188, -41.760250091552734, 35.50432586669922, -6.0464935302734375, -3.711200714111328, 245.17935180664062, 71.09942626953125, -109.52291870117188, 181.62686157226562, 64.72184753417969, 234.70877075195312, 261.5815124511719, 213.0114288330078, 64.787841796875, -253.38595581054688, 128.94142150878906, 87.86204528808594, -40.30925750732422, 238.79205322265625, 147.87179565429688, -277.3353271484375, 78.41368103027344, -3.401538848876953, 175.5869140625, 184.7665252685547, 65.55873107910156, 261.84515380859375, 345.4156494140625, 124.78346252441406, 229.1542205810547, 178.587158203125, -83.11709594726562, 188.30010986328125, -4.0936737060546875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000563.npy"}
|
||||
{"epoch": 0.8510959939531368, "step": 564, "batch_size": 64, "mean": 116.18414306640625, "std": 160.9722442626953, "min": -282.5938720703125, "p10": -66.10667114257811, "median": 114.71464538574219, "p90": 313.30309753417976, "max": 533.4224243164062, "pos_frac": 0.8125, "sample": [208.5447998046875, 40.639564514160156, 276.01617431640625, 303.5741271972656, 88.0804443359375, 348.9535217285156, 76.01710510253906, -239.3611297607422, -71.09136962890625, 32.821502685546875, 257.2593994140625, 193.88534545898438, 110.11216735839844, -34.12724304199219, 20.660751342773438, 35.70701599121094, -11.183509826660156, 206.5543212890625, 317.47265625, 99.92237091064453, 168.80577087402344, 0.0014438629150390625, 159.28280639648438, 290.6590576171875, 253.42652893066406, 221.30917358398438, 329.61328125, -54.4757080078125, -7.056791305541992, 162.88687133789062, 343.2491149902344, 339.05584716796875, 276.0851745605469, 122.76188659667969, 19.39093017578125, -281.9429931640625, 202.30694580078125, 14.138179779052734, 119.31712341308594, 14.331169128417969, 212.8628387451172, 101.27046203613281, 156.97557067871094, -18.881135940551758, -112.61163330078125, 58.32952880859375, 54.507469177246094, 533.4224243164062, 32.959327697753906, -112.82798767089844, 89.53593444824219, 166.84547424316406, 425.554931640625, 28.854028701782227, 184.45867919921875, 94.80716705322266, 182.51123046875, 271.9821472167969, -282.5938720703125, 228.28054809570312, -175.7008056640625, 25.774860382080078, 132.38516235351562, 203.4791717529297], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000564.npy"}
|
||||
{"epoch": 0.8526077097505669, "step": 565, "batch_size": 64, "mean": 96.39464569091797, "std": 141.9251251220703, "min": -283.5038757324219, "p10": -59.02729492187499, "median": 98.47580337524414, "p90": 243.26978454589846, "max": 470.6758117675781, "pos_frac": 0.796875, "sample": [-2.0031204223632812, 144.47735595703125, 28.68682098388672, 80.07464599609375, 288.3252258300781, 97.4466552734375, 328.5148620605469, 135.2608642578125, 58.896697998046875, 81.68500518798828, 152.78958129882812, 180.63882446289062, 470.6758117675781, -202.10235595703125, -46.906280517578125, 217.1865997314453, 88.40240478515625, 119.09672546386719, 445.7144470214844, 281.3017578125, 247.16452026367188, -106.27757263183594, 19.166603088378906, 63.709861755371094, 39.12799072265625, 168.17343139648438, 118.29014587402344, 31.929725646972656, 87.01367950439453, 129.33526611328125, 213.18435668945312, 23.852373123168945, -180.42758178710938, -34.3818244934082, 163.49867248535156, -45.68159866333008, 14.85667610168457, 193.81613159179688, -205.64141845703125, 146.311767578125, 234.18206787109375, 3.5110130310058594, -107.46894836425781, 156.68283081054688, -283.5038757324219, 169.0167236328125, 100.44856262207031, 170.906005859375, 308.23699951171875, 3.9270706176757812, 223.1759033203125, -32.62663269042969, -15.722686767578125, 1.1785621643066406, 69.58289337158203, 84.39065551757812, -64.22201538085938, 221.17184448242188, 124.03014373779297, 59.895111083984375, 216.1172637939453, 178.59902954101562, 99.50495147705078, 213.05975341796875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000565.npy"}
|
||||
{"epoch": 0.854119425547997, "step": 566, "batch_size": 64, "mean": 84.08949279785156, "std": 191.19776916503906, "min": -441.35150146484375, "p10": -142.40941314697264, "median": 81.67677688598633, "p90": 283.26748046875, "max": 623.64453125, "pos_frac": 0.671875, "sample": [97.37274169921875, 52.56011199951172, -72.81172180175781, 203.01275634765625, -64.3983154296875, 74.72712707519531, 223.0426025390625, 298.48828125, 0.7931880950927734, 196.9013671875, 110.12556457519531, 313.1246337890625, 38.93547058105469, 47.26620101928711, 27.413963317871094, 623.64453125, -441.35150146484375, 162.45068359375, 256.1720886230469, 188.09732055664062, 286.8689270019531, -35.04119873046875, -12.830814361572266, 101.63996124267578, 4.236665725708008, -266.2919921875, 118.14466857910156, 178.72561645507812, 299.1102600097656, -8.602157592773438, 274.8641052246094, 230.24928283691406, 199.6941375732422, -146.13400268554688, -94.2447509765625, -214.43133544921875, -55.68242645263672, 15.96487045288086, 598.6888427734375, -61.55443572998047, 95.07234191894531, -133.7187042236328, 493.374267578125, -52.97412872314453, 214.5555419921875, 16.460670471191406, -169.48094177246094, -207.33709716796875, 175.61843872070312, -33.917694091796875, -84.74089050292969, 31.799339294433594, -109.33177185058594, 214.51287841796875, 3.3542747497558594, 115.35115051269531, -25.662818908691406, 242.90304565429688, 88.62642669677734, -156.50331115722656, 213.46788024902344, 173.47525024414062, 261.30841064453125, 266.5738220214844], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000566.npy"}
|
||||
{"epoch": 0.8556311413454271, "step": 567, "batch_size": 64, "mean": 111.06340026855469, "std": 163.1463623046875, "min": -251.9830322265625, "p10": -117.17632598876949, "median": 102.29769515991211, "p90": 279.646875, "max": 724.8946533203125, "pos_frac": 0.796875, "sample": [233.9875946044922, -7.098049163818359, 4.098381042480469, 195.13343811035156, -234.2620086669922, 293.052978515625, 248.1751708984375, 180.969482421875, 198.81292724609375, 112.90239715576172, 232.28416442871094, 57.244842529296875, 178.17376708984375, -4.8497467041015625, 176.42758178710938, -55.186988830566406, -12.293540954589844, 77.07196044921875, 80.04737091064453, 53.0916862487793, 363.23529052734375, 23.39025115966797, 3.4879894256591797, 259.6777038574219, -251.9830322265625, -2.738473892211914, -171.51458740234375, -137.31394958496094, 178.13656616210938, -164.02615356445312, 202.06637573242188, 65.67259216308594, 274.94537353515625, 724.8946533203125, 282.3642578125, 2.847057342529297, 159.20375061035156, 62.8759765625, 308.662841796875, 236.5395965576172, -176.46078491210938, 183.1363525390625, 267.35797119140625, 186.5084686279297, 204.89706420898438, 160.4482879638672, 16.537841796875, 0.0821075439453125, 281.66180419921875, 249.09750366210938, 62.432533264160156, 80.35568237304688, 247.42254638671875, 6.7965087890625, 52.781524658203125, -145.63406372070312, -70.18853759765625, 91.6929931640625, 304.1661682128906, 191.59417724609375, 181.68772888183594, 88.18141174316406, 21.39495277404785, 193.8997802734375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000567.npy"}
|
||||
{"epoch": 0.8571428571428571, "step": 568, "batch_size": 64, "mean": 129.8306884765625, "std": 167.57688903808594, "min": -179.5448760986328, "p10": -62.53005561828613, "median": 111.65616226196289, "p90": 309.9176635742188, "max": 689.7515869140625, "pos_frac": 0.765625, "sample": [234.7001953125, 135.7733917236328, 73.71162414550781, 90.17294311523438, 149.20700073242188, 198.07699584960938, 207.03811645507812, -103.44763946533203, 51.35770797729492, 57.36528015136719, -70.15958404541016, 223.74952697753906, 134.23944091796875, 357.5592956542969, -0.5397109985351562, -48.35218811035156, -9.450843811035156, -28.42076873779297, 62.14891815185547, 124.49589538574219, -74.7359619140625, -63.81011962890625, 93.98077392578125, 184.77902221679688, 296.69952392578125, 275.48419189453125, 200.217529296875, 418.05828857421875, -59.54323959350586, 90.94143676757812, 266.7767333984375, 159.14971923828125, 60.841148376464844, -30.465065002441406, 191.05938720703125, 2.9947376251220703, -29.867141723632812, 271.1231689453125, 3.076436996459961, 181.26754760742188, 13.091659545898438, 315.58258056640625, 131.70570373535156, 104.47824096679688, 193.9956817626953, 118.8340835571289, 73.00718688964844, 27.825225830078125, -130.07778930664062, 96.84234619140625, 241.96975708007812, 235.4083251953125, -121.4305419921875, 185.9310302734375, 289.282470703125, 59.8726806640625, -5.624092102050781, 642.55859375, 441.7371520996094, 329.36639404296875, -179.5448760986328, 14.878456115722656, 262.4688720703125, 689.7515869140625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000568.npy"}
|
||||
{"epoch": 0.8586545729402872, "step": 569, "batch_size": 64, "mean": 58.385337829589844, "std": 165.0227813720703, "min": -387.81976318359375, "p10": -140.70086135864253, "median": 42.363210678100586, "p90": 221.5376724243164, "max": 669.8720092773438, "pos_frac": 0.703125, "sample": [174.149169921875, -387.81976318359375, 1.4301586151123047, 177.58450317382812, 35.766448974609375, 222.31732177734375, 56.30683898925781, 669.8720092773438, 53.953033447265625, 206.3994140625, 62.024688720703125, -46.974266052246094, 46.31606674194336, 38.41035461425781, -255.646240234375, 74.61676025390625, 16.213909149169922, 382.089111328125, -28.030439376831055, 68.86798095703125, 200.09730529785156, 376.25335693359375, -11.927265167236328, 305.0362243652344, -46.73956298828125, 211.22085571289062, 155.60813903808594, 12.96194076538086, 93.27349853515625, -193.16079711914062, -161.4988250732422, 129.12149047851562, 18.042068481445312, 29.304271697998047, -46.929893493652344, -38.61955261230469, 223.01168823242188, -10.92909049987793, 95.74982452392578, 57.42839813232422, 28.310819625854492, -73.51517486572266, -303.810791015625, -92.17227935791016, 209.87986755371094, -221.88436889648438, 164.89532470703125, 23.249923706054688, 19.956432342529297, 118.50511932373047, 89.76017761230469, 143.7368927001953, 219.71849060058594, -33.695594787597656, -21.142629623413086, 27.285165786743164, 100.03050231933594, 134.3809356689453, 239.71630859375, -36.78648376464844, -162.02316284179688, 6.979393005371094, 176.6392822265625, 13.496513366699219], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000569.npy"}
|
||||
{"epoch": 0.8601662887377173, "step": 570, "batch_size": 64, "mean": 113.80067443847656, "std": 170.29562377929688, "min": -277.248046875, "p10": -77.7626640319824, "median": 126.1423225402832, "p90": 275.63339233398443, "max": 895.7664794921875, "pos_frac": 0.796875, "sample": [240.09878540039062, -188.45635986328125, 204.2711181640625, 171.5616455078125, 180.6251220703125, 63.383750915527344, -114.6114501953125, -1.7296295166015625, 55.66741943359375, -48.92265319824219, 108.5491714477539, 53.56233215332031, -2.482156753540039, 202.6487274169922, -26.376243591308594, 212.81161499023438, 1.6820430755615234, 17.427265167236328, 151.83047485351562, 154.22323608398438, 134.7531280517578, -186.9976348876953, 59.449302673339844, 202.76625061035156, 250.31112670898438, 2.271116256713867, 23.525768280029297, 249.23951721191406, 282.6468811035156, -88.9740219116211, 8.999732971191406, 109.23441314697266, 206.12911987304688, -104.01921081542969, 6.434123992919922, 167.3006591796875, 259.2685852050781, 29.798933029174805, 216.2923583984375, 19.146509170532227, 314.50848388671875, 190.61203002929688, 369.8417663574219, -51.60282897949219, 13.242835998535156, 199.18572998046875, 298.885498046875, 282.6971435546875, 247.09625244140625, -176.63003540039062, 236.46771240234375, 0.16747283935546875, 122.59796905517578, 188.63648986816406, 295.28851318359375, 111.95187377929688, 129.68667602539062, -5.31550407409668, 134.39199829101562, 195.22500610351562, 895.7664794921875, 37.69943618774414, 246.74932861328125, -277.248046875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000570.npy"}
|
||||
{"epoch": 0.8616780045351474, "step": 571, "batch_size": 64, "mean": 74.37464904785156, "std": 167.22982788085938, "min": -259.6907958984375, "p10": -143.1432342529297, "median": 74.77637481689453, "p90": 208.95118103027346, "max": 843.9486083984375, "pos_frac": 0.6875, "sample": [157.61387634277344, 185.09722900390625, 126.95669555664062, 11.633977890014648, -6.6702423095703125, 182.34518432617188, 278.7004089355469, 114.6086196899414, 166.6305694580078, 843.9486083984375, 136.28724670410156, -13.829429626464844, -162.74798583984375, -5.325588226318359, -7.660858154296875, 163.0167694091797, 114.6605224609375, 204.2450408935547, 77.50552368164062, 130.92001342773438, 211.15074157714844, 38.009498596191406, 72.04722595214844, 101.18360900878906, 281.6856384277344, 89.41896057128906, 121.16566467285156, 186.6051788330078, -259.6907958984375, 34.31139373779297, 1.1603336334228516, -138.96327209472656, -29.18920135498047, -201.28988647460938, 193.2101287841797, -6.030500411987305, 22.71173858642578, 6.473194122314453, -144.9346466064453, -26.8994140625, 41.411460876464844, -21.938018798828125, 194.3684539794922, 183.850830078125, 99.24967956542969, 210.30377197265625, 360.6524963378906, -216.81700134277344, 39.382835388183594, -84.16558837890625, -42.87351989746094, -211.91314697265625, 185.91690063476562, 58.423065185546875, 196.70242309570312, -87.5040512084961, 205.79513549804688, 172.1530303955078, 6.081321716308594, 250.50535583496094, -45.877750396728516, 36.587982177734375, -215.26556396484375, 194.87596130371094], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000571.npy"}
|
||||
{"epoch": 0.8631897203325775, "step": 572, "batch_size": 64, "mean": 62.669471740722656, "std": 150.4054412841797, "min": -311.283447265625, "p10": -74.09503479003907, "median": 54.124732971191406, "p90": 244.68882446289066, "max": 451.4916076660156, "pos_frac": 0.65625, "sample": [20.07018280029297, -66.01510620117188, -6.731651306152344, 104.72236633300781, 286.24664306640625, 49.79310607910156, 200.32070922851562, 58.45635986328125, -9.741634368896484, 233.60226440429688, -74.06986999511719, 185.87890625, 451.4916076660156, 236.65142822265625, -58.80439376831055, 248.1334228515625, -145.13333129882812, 4.84514045715332, 30.318838119506836, 157.53172302246094, 63.430850982666016, -235.61624145507812, 149.3442840576172, 208.40249633789062, 257.4183654785156, 127.9415283203125, -60.379615783691406, 154.99957275390625, -29.350914001464844, 36.78782653808594, 250.76834106445312, 202.0072479248047, 184.82228088378906, -311.283447265625, 165.14892578125, 2.141712188720703, 112.42495727539062, -60.09181213378906, 13.98648452758789, 150.9465789794922, 193.94992065429688, -31.791563034057617, 165.98211669921875, 201.6219940185547, 16.703521728515625, 220.44808959960938, -15.751419067382812, -198.75013732910156, 249.68704223632812, -37.16331481933594, 96.17288208007812, 138.71485900878906, -74.10581970214844, -35.75084686279297, -249.44879150390625, -71.25028991699219, 77.1724624633789, 69.48971557617188, -43.108123779296875, -283.2366943359375, 308.2237854003906, -2.609445571899414, 6.829092025756836, 17.400957107543945], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000572.npy"}
|
||||
{"epoch": 0.8647014361300076, "step": 573, "batch_size": 64, "mean": 91.24996948242188, "std": 195.68453979492188, "min": -324.9803161621094, "p10": -126.38059310913086, "median": 74.7486801147461, "p90": 355.54168395996095, "max": 725.6461181640625, "pos_frac": 0.625, "sample": [51.46504211425781, 159.210693359375, -214.8910675048828, 219.99440002441406, -136.23690795898438, 101.75897216796875, -4.799432754516602, 40.259605407714844, 236.82089233398438, -127.44819641113281, -211.26943969726562, -260.08892822265625, -20.20369529724121, -146.99281311035156, -77.80191040039062, 376.6322937011719, -35.27360534667969, -12.634017944335938, 222.58071899414062, -324.9803161621094, 130.87234497070312, 96.64195251464844, 293.8108825683594, 120.34980010986328, -48.927459716796875, 33.71330642700195, 725.6461181640625, -5.949489593505859, 197.2857666015625, 76.45709991455078, -10.62469482421875, 55.63189697265625, 112.48538208007812, -20.631301879882812, 186.03515625, -58.87672805786133, 189.39093017578125, 106.61456298828125, 28.923194885253906, 440.0945739746094, 166.21475219726562, 224.2714385986328, 213.46571350097656, -11.304948806762695, -116.92399597167969, 134.1517333984375, 358.24932861328125, -40.18693542480469, 37.07238006591797, 170.84384155273438, 349.2238464355469, 430.1658935546875, 2.326812744140625, 157.54702758789062, -7.686286926269531, -108.84465026855469, 597.1060180664062, -123.88951873779297, 123.76228332519531, 73.0402603149414, 166.7764129638672, 157.5032958984375, -59.331817626953125, 461.3998107910156], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000573.npy"}
|
||||
{"epoch": 0.8662131519274376, "step": 574, "batch_size": 64, "mean": 119.26148986816406, "std": 157.3621368408203, "min": -119.2750015258789, "p10": -58.41600036621094, "median": 130.14168548583984, "p90": 255.81843414306644, "max": 903.341796875, "pos_frac": 0.75, "sample": [111.80341339111328, -76.3783187866211, -35.680213928222656, -116.24010467529297, 183.85475158691406, 161.06332397460938, 33.02752685546875, 8.834259033203125, 14.664018630981445, -58.879234313964844, -18.286849975585938, 197.66563415527344, -83.44740295410156, 109.31859588623047, -2.406036376953125, 9.863197326660156, -4.511157989501953, 126.73347473144531, 49.420448303222656, 149.1881866455078, 133.54989624023438, 79.85933685302734, -115.85940551757812, 196.72543334960938, -29.94617462158203, 259.74346923828125, 172.47616577148438, 275.64208984375, -1.0392513275146484, 242.7777099609375, 276.33026123046875, -2.7783203125, 68.17253112792969, 94.4364013671875, 412.93682861328125, 214.63314819335938, -119.2750015258789, 164.89535522460938, 230.3626708984375, 6.941774368286133, 185.166259765625, 137.71438598632812, 197.310546875, 346.1217041015625, 184.0175018310547, 224.46316528320312, 136.57861328125, 229.32391357421875, 141.8954620361328, 31.034996032714844, -63.43243408203125, 226.6674041748047, -44.649810791015625, 197.59225463867188, 166.08810424804688, 52.13709259033203, 50.47805404663086, 184.03851318359375, 227.47357177734375, 305.3209228515625, -57.335121154785156, 246.66001892089844, 903.341796875, 104.53663635253906], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000574.npy"}
|
||||
{"epoch": 0.8677248677248677, "step": 575, "batch_size": 64, "mean": 86.34864807128906, "std": 190.67453002929688, "min": -262.8981628417969, "p10": -137.64875793457028, "median": 60.211870193481445, "p90": 294.4310302734375, "max": 810.3237915039062, "pos_frac": 0.671875, "sample": [127.48352813720703, -189.05072021484375, -262.8981628417969, -12.010377883911133, 810.3237915039062, 330.93768310546875, -71.55506134033203, 164.0184326171875, 84.05632019042969, -36.80617904663086, 151.45294189453125, -203.28939819335938, 278.056396484375, 2.947620391845703, -194.91943359375, 103.99494934082031, -152.069091796875, 201.14126586914062, -9.581703186035156, 16.94733428955078, 61.55048370361328, 292.76434326171875, 199.43032836914062, 1.5894317626953125, -11.945777893066406, 26.891860961914062, -208.23988342285156, 198.2772674560547, 445.3360900878906, -0.7611923217773438, 37.11100769042969, 101.54438018798828, 223.1554718017578, 420.40411376953125, 209.54776000976562, -74.16509246826172, 295.14532470703125, -15.37506103515625, 327.936767578125, 192.71803283691406, -103.08985900878906, 61.28546142578125, 98.7699966430664, 78.81698608398438, 7.6175689697265625, -46.561561584472656, 99.37161254882812, -241.09913635253906, 152.13404846191406, 580.5020751953125, 289.6375732421875, 27.729782104492188, 25.194198608398438, 59.13827896118164, -104.00131225585938, 55.299110412597656, 141.96548461914062, -2.755046844482422, 1.3821048736572266, 116.71018981933594, -38.872169494628906, 269.1771240234375, 140.9412841796875, -5.076023101806641], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000575.npy"}
|
||||
{"epoch": 0.8692365835222978, "step": 576, "batch_size": 64, "mean": 129.92605590820312, "std": 171.40249633789062, "min": -369.90875244140625, "p10": -57.297668075561525, "median": 123.12753295898438, "p90": 358.8748291015626, "max": 573.5824584960938, "pos_frac": 0.75, "sample": [-60.236907958984375, 190.83563232421875, -36.125404357910156, 117.24755096435547, 573.5824584960938, 22.246170043945312, 224.28036499023438, 10.346223831176758, 315.7947082519531, 138.24562072753906, -57.593528747558594, -78.20501708984375, 243.47012329101562, -66.5814208984375, 176.74984741210938, -183.917236328125, -31.61779022216797, 197.59133911132812, 472.9560546875, 483.59521484375, 228.40162658691406, 305.9646301269531, 127.02021789550781, 19.242971420288086, 154.6100616455078, 111.84237670898438, 22.76664161682129, -107.42744445800781, 380.61407470703125, -46.21410369873047, 95.80216979980469, 110.66449737548828, 167.50894165039062, 210.42196655273438, 27.896942138671875, 119.23484802246094, 381.43829345703125, -33.79692077636719, -1.928955078125, 33.66975402832031, 261.82733154296875, -56.60732650756836, 200.78402709960938, 53.576568603515625, 295.86981201171875, 8.687164306640625, 244.49342346191406, 265.5653076171875, -13.712682723999023, 256.6744384765625, 324.9755859375, -369.90875244140625, 373.403076171875, 438.30682373046875, 145.90408325195312, 252.45883178710938, 14.712881088256836, -3.5786285400390625, 63.32801055908203, 111.61518096923828, -1.795969009399414, 130.7354736328125, 156.10324096679688, 201.4533233642578], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000576.npy"}
|
||||
{"epoch": 0.8707482993197279, "step": 577, "batch_size": 64, "mean": 76.81715393066406, "std": 159.3997802734375, "min": -409.4928894042969, "p10": -107.11437606811522, "median": 60.272491455078125, "p90": 238.900895690918, "max": 516.5628662109375, "pos_frac": 0.765625, "sample": [12.005935668945312, 10.703750610351562, -88.40575408935547, 34.42877197265625, 20.06494903564453, 140.590087890625, 0.3759803771972656, -22.984384536743164, 305.8855285644531, 0.5860671997070312, -90.89240264892578, 230.0784454345703, 181.0201873779297, -55.396514892578125, -126.99688720703125, 242.68194580078125, 516.5628662109375, 191.8416748046875, -157.16761779785156, 163.4213104248047, 271.1060791015625, 117.06610870361328, 19.616687774658203, 157.9781494140625, 51.99652099609375, -115.2121810913086, -165.35853576660156, 193.38986206054688, 176.6085662841797, 432.0058898925781, -57.46369934082031, 188.1420440673828, -114.066650390625, 40.50212097167969, 169.57040405273438, 55.0380859375, 65.50689697265625, 11.439605712890625, -13.809791564941406, 16.56557846069336, 66.17101287841797, 25.34765625, 284.26043701171875, 8.59699821472168, 227.79364013671875, 400.2928161621094, -320.4171447753906, 207.07994079589844, 45.72428894042969, 121.24203491210938, 93.69148254394531, 207.50038146972656, 168.7032470703125, 204.47821044921875, -7.6477203369140625, 73.5023193359375, -409.4928894042969, 1.9024181365966797, -79.52597045898438, 167.96197509765625, 2.7689895629882812, 189.64962768554688, 113.01344299316406, 114.67488098144531], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000577.npy"}
|
||||
{"epoch": 0.872260015117158, "step": 578, "batch_size": 64, "mean": 72.7741470336914, "std": 179.59555053710938, "min": -407.2259216308594, "p10": -114.85526580810546, "median": 48.59760284423828, "p90": 242.90944976806645, "max": 655.504638671875, "pos_frac": 0.6875, "sample": [156.75672912597656, -7.295021057128906, 177.13442993164062, 52.476409912109375, 211.6556396484375, -58.74430847167969, 148.95347595214844, 38.79084777832031, 230.99856567382812, 7.099397659301758, 48.22991943359375, 655.504638671875, 23.763763427734375, 189.5348358154297, 221.19204711914062, -333.2230224609375, 170.29588317871094, 214.69830322265625, 193.24037170410156, 105.18844604492188, 179.94064331054688, 135.8706512451172, 1.7625370025634766, -8.279804229736328, 19.327116012573242, 33.606834411621094, 289.5048828125, 13.61407470703125, 103.85260772705078, -9.011360168457031, 447.7576904296875, 216.58837890625, 229.27719116210938, -407.2259216308594, 9.248695373535156, 45.02442932128906, -15.666608810424805, 116.02055358886719, 59.79369354248047, 129.00892639160156, 248.0141143798828, -292.2146911621094, 74.34284973144531, -117.26445770263672, -8.485427856445312, 22.7254638671875, -141.1522216796875, 191.65773010253906, -152.49624633789062, 252.74703979492188, 167.43490600585938, -109.23381805419922, 564.2031860351562, -0.23166847229003906, -108.30014038085938, -90.34867858886719, 48.96528625488281, 80.8565444946289, -36.74414825439453, 2.785886764526367, -51.187129974365234, -8.242195129394531, -195.8114013671875, 279.2576904296875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000578.npy"}
|
||||
{"epoch": 0.873771730914588, "step": 579, "batch_size": 64, "mean": 109.25408172607422, "std": 147.12567138671875, "min": -300.94757080078125, "p10": -40.618591308593736, "median": 117.84310150146484, "p90": 251.29219665527344, "max": 485.87860107421875, "pos_frac": 0.796875, "sample": [159.22470092773438, 112.33380126953125, 73.42183685302734, 49.05402374267578, 62.291664123535156, -217.41522216796875, 266.8182373046875, 117.06062316894531, 183.85043334960938, 485.87860107421875, 118.62557983398438, 188.95639038085938, 151.74057006835938, 232.2576904296875, 103.69960021972656, 27.288375854492188, 181.53884887695312, -122.67056274414062, 6.658302307128906, 459.9718017578125, 306.07611083984375, -109.77732849121094, 204.5941162109375, -10.803136825561523, 120.27799987792969, 198.95701599121094, 246.54205322265625, 31.834426879882812, 223.69468688964844, 218.6154327392578, -27.107032775878906, 9.30865478515625, 166.27938842773438, -2.3929805755615234, 198.01815795898438, 157.08523559570312, 49.406005859375, -26.294981002807617, -152.53016662597656, 211.73086547851562, 62.405975341796875, 0.9596195220947266, -27.565048217773438, 52.64879608154297, 111.0460205078125, -0.5767498016357422, -46.21296691894531, -300.94757080078125, 224.27761840820312, 235.46768188476562, 216.89556884765625, 421.5472717285156, 208.51895141601562, 286.7255554199219, 142.57269287109375, 11.340200424194336, 165.39683532714844, 133.83609008789062, 63.84801483154297, 111.19976806640625, -185.02430725097656, 42.49333572387695, 253.32797241210938, 153.97999572753906], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000579.npy"}
|
||||
{"epoch": 0.8752834467120182, "step": 580, "batch_size": 64, "mean": 105.17242431640625, "std": 191.0565185546875, "min": -237.51171875, "p10": -79.80401916503905, "median": 94.96662902832031, "p90": 337.27085571289064, "max": 676.508056640625, "pos_frac": 0.671875, "sample": [-208.86453247070312, 8.7845458984375, -63.989837646484375, 16.464704513549805, -58.883148193359375, 33.874473571777344, 212.67715454101562, 113.42300415039062, 202.88967895507812, 66.6288833618164, 304.2939453125, 110.61519622802734, 191.54168701171875, -53.836788177490234, -38.342872619628906, 89.26710510253906, 208.85581970214844, 148.18734741210938, 9.522344589233398, 56.27366638183594, -11.45157241821289, 162.31405639648438, 340.115966796875, 53.592185974121094, 103.4491195678711, -69.55165100097656, 629.6636352539062, -16.914398193359375, -27.00434684753418, 330.63226318359375, 292.3405456542969, 232.70716857910156, -10.250007629394531, 187.99649047851562, -89.70755004882812, 190.9366455078125, -15.333141326904297, -237.51171875, 23.370990753173828, 580.9476928710938, -207.23446655273438, 98.23556518554688, -84.19789123535156, 411.28460693359375, 448.56683349609375, -8.805421829223633, 103.3205337524414, 106.2752914428711, 448.2225646972656, 171.12106323242188, 131.44976806640625, 172.75205993652344, -59.81908416748047, -152.5004119873047, 18.966922760009766, -52.820068359375, -28.337295532226562, 142.6136932373047, 143.78497314453125, 676.508056640625, 91.69769287109375, -214.3379364013672, 188.42237854003906, 186.1414794921875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000580.npy"}
|
||||
{"epoch": 0.8767951625094482, "step": 581, "batch_size": 64, "mean": 91.3229751586914, "std": 178.31736755371094, "min": -365.3561096191406, "p10": -106.4706298828125, "median": 98.70301818847656, "p90": 271.3562286376954, "max": 782.673828125, "pos_frac": 0.71875, "sample": [171.244873046875, 292.2120666503906, 76.5535659790039, 2.3241653442382812, 36.37425231933594, 28.401443481445312, 47.46318054199219, -106.13818359375, 171.93136596679688, 338.2587890625, -155.780517578125, 25.925617218017578, 17.873159408569336, 26.3505916595459, 197.53643798828125, -106.5970458984375, 225.81979370117188, 122.93059539794922, 209.23573303222656, 95.38108825683594, 210.33863830566406, 56.8680419921875, 233.45114135742188, -13.91461181640625, 323.35260009765625, 280.2532043457031, 241.7133331298828, -95.390380859375, 80.78983306884766, 73.48895263671875, -42.702545166015625, 241.7685546875, 116.74848937988281, 237.09674072265625, 167.08262634277344, 102.02494812011719, 117.15568542480469, -273.8365478515625, 350.96783447265625, 367.72332763671875, -87.2471923828125, -365.3561096191406, 230.77706909179688, 107.40220642089844, 165.17355346679688, -106.1756591796875, 782.673828125, -10.587432861328125, -30.808223724365234, -195.64996337890625, 250.59661865234375, -95.1837387084961, 108.01933288574219, 203.03013610839844, 38.33659744262695, 171.15098571777344, -7.0335845947265625, -174.95802307128906, -145.17686462402344, 7.560764312744141, 102.40255737304688, 208.3057403564453, 232.43423461914062, -37.29707336425781], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000581.npy"}
|
||||
{"epoch": 0.8783068783068783, "step": 582, "batch_size": 64, "mean": 79.22227478027344, "std": 157.45005798339844, "min": -235.160400390625, "p10": -115.75077362060547, "median": 45.837154388427734, "p90": 258.5525268554688, "max": 561.073974609375, "pos_frac": 0.71875, "sample": [86.6991958618164, 204.56399536132812, 42.710044860839844, 193.52017211914062, 59.48497009277344, 80.197998046875, 334.2147216796875, 35.91188430786133, 259.74835205078125, 146.87234497070312, 114.96929931640625, 83.80062103271484, 490.41864013671875, 561.073974609375, -73.37063598632812, 233.46353149414062, -157.80010986328125, 19.033889770507812, 118.98471069335938, 154.85104370117188, -5.880073547363281, 171.25119018554688, 15.411548614501953, 287.9804992675781, 317.2972717285156, 358.62945556640625, -41.92266845703125, -10.473165512084961, -167.47927856445312, -206.78976440429688, -7.504232406616211, -20.038330078125, 151.29385375976562, -47.266693115234375, -41.62293243408203, 204.64015197753906, 136.70428466796875, 34.18803024291992, 255.76226806640625, 21.242931365966797, 9.13336181640625, 136.0966033935547, 100.4996566772461, 15.171541213989258, 15.696760177612305, 4.7858734130859375, -116.20655059814453, 224.40505981445312, 20.84392547607422, -235.160400390625, 3.968595504760742, 48.964263916015625, 0.9734382629394531, -7.076271057128906, -120.73705291748047, 76.32081604003906, 187.96714782714844, -114.68729400634766, 198.19276428222656, -71.96360778808594, -202.53668212890625, 26.209556579589844, 251.99050903320312, 222.6003875732422], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000582.npy"}
|
||||
{"epoch": 0.8798185941043084, "step": 583, "batch_size": 64, "mean": 125.09844970703125, "std": 195.6817626953125, "min": -286.802734375, "p10": -77.70235214233398, "median": 105.4932632446289, "p90": 322.8122528076172, "max": 713.1751708984375, "pos_frac": 0.75, "sample": [102.42431640625, -44.51634216308594, 50.60553741455078, 0.6208534240722656, 194.05874633789062, 108.56221008300781, 31.294654846191406, -7.5838165283203125, -175.08859252929688, -27.334365844726562, 277.53961181640625, -16.50338363647461, -92.82138061523438, 190.52877807617188, 117.93751525878906, -54.725502014160156, 70.08155822753906, 7.003854751586914, 235.09832763671875, 218.6356201171875, 26.28498649597168, -191.76980590820312, 244.06947326660156, 28.37206268310547, 472.9896240234375, 146.46957397460938, 713.1751708984375, -286.802734375, 37.91325759887695, 282.5760803222656, 73.62245178222656, 61.5206298828125, -68.91773986816406, 49.99668884277344, 574.6685791015625, 301.34844970703125, 6.1002655029296875, -81.4671859741211, 325.1775817871094, 189.5755157470703, 136.93704223632812, 281.3659973144531, -226.61367797851562, 528.2819213867188, 9.713388442993164, 213.50308227539062, 193.29135131835938, 193.30929565429688, 181.15948486328125, -25.24517059326172, 427.9320068359375, 178.0125732421875, 237.3435821533203, 24.04197883605957, 152.76947021484375, 589.8291625976562, -187.3082275390625, 127.22960662841797, -4.638755798339844, 209.66015625, 277.4945068359375, 317.29315185546875, 93.89189147949219, -13.674545288085938], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000583.npy"}
|
||||
{"epoch": 0.8813303099017384, "step": 584, "batch_size": 64, "mean": 97.82256317138672, "std": 173.66452026367188, "min": -218.65733337402344, "p10": -72.8850715637207, "median": 65.21877098083496, "p90": 351.8833251953128, "max": 601.956787109375, "pos_frac": 0.671875, "sample": [261.7776794433594, 426.8930969238281, 44.05415725708008, -30.44482421875, 136.86631774902344, -19.97624397277832, 387.29962158203125, 281.36798095703125, 196.57513427734375, 212.21669006347656, 138.30130004882812, 0.6486797332763672, -36.819435119628906, -66.2747573852539, -177.61985778808594, 124.79676055908203, 174.3144989013672, 243.55995178222656, 214.26553344726562, -13.15365982055664, 219.25704956054688, 382.10418701171875, -35.049766540527344, 240.70899963378906, 601.956787109375, 188.01956176757812, 110.80382537841797, 69.51441955566406, 96.52558898925781, 34.735626220703125, 218.43080139160156, -6.8967132568359375, 158.7321319580078, 5.1016845703125, -180.5941619873047, 271.1880798339844, -12.991790771484375, 2.02435302734375, -30.40123748779297, -51.762298583984375, 11.778575897216797, 174.63848876953125, -75.71806335449219, 17.507959365844727, 108.84427642822266, 492.80902099609375, -0.44948387145996094, 84.08798217773438, 130.34674072265625, -164.87835693359375, -218.65733337402344, 60.92312240600586, -20.82610321044922, 196.04443359375, -5.306617736816406, 425.0990905761719, 424.52703857421875, 20.146638870239258, 45.277984619140625, 21.594650268554688, -50.59949493408203, 155.89328002929688, -145.91094970703125, -206.58428955078125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000584.npy"}
|
||||
{"epoch": 0.8828420256991686, "step": 585, "batch_size": 64, "mean": 64.44938659667969, "std": 154.39505004882812, "min": -340.92462158203125, "p10": -69.29181289672852, "median": 45.61509323120117, "p90": 240.50338134765633, "max": 639.6097412109375, "pos_frac": 0.65625, "sample": [33.58806228637695, 320.3389892578125, 104.033447265625, -9.983322143554688, 1.9684677124023438, 46.7371826171875, 135.89227294921875, 356.94732666015625, -43.82958984375, 12.544776916503906, -194.49176025390625, 51.961544036865234, 306.8540344238281, -35.85182189941406, -5.840045928955078, 639.6097412109375, -172.896728515625, 11.921363830566406, 133.80039978027344, 136.8892364501953, 96.77154541015625, 2.2431392669677734, -14.959943771362305, -158.55421447753906, -33.39134216308594, 2.3016490936279297, -74.16740417480469, 31.18256378173828, 300.4488525390625, 144.2109375, 6.008899688720703, 57.783050537109375, -26.419445037841797, -20.834423065185547, -287.18902587890625, -14.150833129882812, -20.000457763671875, 191.26876831054688, 248.24346923828125, 123.22845458984375, 123.62556457519531, 195.40386962890625, 111.07290649414062, -32.800926208496094, -340.92462158203125, 222.44317626953125, 261.7682800292969, 212.40267944335938, -64.26850128173828, -69.55693817138672, 86.34278869628906, 44.493003845214844, -68.67318725585938, 7.572275161743164, 60.97309112548828, 197.58248901367188, 144.62994384765625, 219.79432678222656, 214.71820068359375, 83.62274169921875, 130.15585327148438, -17.966232299804688, 74.56444549560547, -56.432472229003906], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000585.npy"}
|
||||
{"epoch": 0.8843537414965986, "step": 586, "batch_size": 64, "mean": 83.542724609375, "std": 113.68426513671875, "min": -131.2705841064453, "p10": -28.38106689453124, "median": 60.08673286437988, "p90": 236.00364990234382, "max": 416.316162109375, "pos_frac": 0.703125, "sample": [-63.430908203125, 33.555030822753906, 189.9752197265625, 117.58931732177734, 14.900787353515625, 107.21745300292969, 47.32159423828125, 6.8019256591796875, -18.6910400390625, -2.0041332244873047, 171.196533203125, -32.533935546875, 0.2507476806640625, 296.7456359863281, 263.66827392578125, -12.907228469848633, 147.37979125976562, 68.60713195800781, 85.29817962646484, 162.4613800048828, 10.382442474365234, 79.09092712402344, 51.56633377075195, 214.60618591308594, 105.74581146240234, -7.53602409362793, -8.855186462402344, 18.14630126953125, 166.3623046875, 48.77840805053711, 266.8136901855469, 213.6976318359375, 85.85005950927734, 123.44581604003906, 319.081787109375, -8.356307983398438, -131.2705841064453, 310.97271728515625, -4.474208831787109, 11.045034408569336, -10.818464279174805, -59.1331787109375, -86.70616149902344, 195.87620544433594, 103.99203491210938, -0.7976646423339844, 4.307853698730469, 84.90791320800781, -11.375236511230469, 140.8716583251953, 416.316162109375, -12.537429809570312, 34.89134216308594, 18.24787712097168, 220.4188232421875, 96.24573516845703, -48.607383728027344, 242.682861328125, 169.2978973388672, -87.63023376464844, 189.21661376953125, -10.625141143798828, 188.34072875976562, 120.8568115234375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000586.npy"}
|
||||
{"epoch": 0.8858654572940288, "step": 587, "batch_size": 64, "mean": 80.63211822509766, "std": 210.62132263183594, "min": -982.007568359375, "p10": -119.28530120849607, "median": 80.9754524230957, "p90": 252.66788482666018, "max": 595.9761962890625, "pos_frac": 0.78125, "sample": [41.44423294067383, -168.3280792236328, -9.606056213378906, 157.1269989013672, 246.63330078125, -187.59373474121094, 60.857791900634766, -145.32644653320312, 228.15115356445312, 227.00869750976562, 34.78702926635742, 76.66223907470703, -132.65000915527344, 119.17338562011719, 7.7800140380859375, 85.28866577148438, 192.35379028320312, -7.943416595458984, -69.60711669921875, 255.25413513183594, 285.0772705078125, 160.67477416992188, -0.17110443115234375, 213.2967071533203, 116.0955810546875, 54.80261993408203, 283.6973571777344, -14.113662719726562, -982.007568359375, 51.58658981323242, 71.1927261352539, 189.7489776611328, 105.59718322753906, 158.9336700439453, 24.74508285522461, 50.335693359375, 163.32081604003906, -443.309814453125, 3.01416015625, 543.4451904296875, 180.54774475097656, 29.94620132446289, 227.10076904296875, 165.21218872070312, 8.574827194213867, 169.28045654296875, 33.53594207763672, 196.49002075195312, 202.348876953125, 117.22364807128906, 273.2247314453125, -88.10098266601562, 135.80825805664062, 32.37322998046875, 62.29985046386719, 440.49560546875, 36.69805145263672, -33.15638732910156, -177.9674530029297, 119.13408660888672, 155.48123168945312, 595.9761962890625, 207.95980834960938, 22.53954315185547], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000587.npy"}
|
||||
{"epoch": 0.8873771730914588, "step": 588, "batch_size": 64, "mean": 95.76359558105469, "std": 197.413330078125, "min": -410.7747497558594, "p10": -96.3503303527832, "median": 85.98844528198242, "p90": 296.64302673339853, "max": 648.6570434570312, "pos_frac": 0.765625, "sample": [36.85096740722656, -98.73469543457031, -90.78681182861328, 319.6194763183594, 44.07310485839844, 108.63607788085938, 197.1768798828125, 76.51054382324219, 278.25262451171875, 175.3272705078125, 595.859619140625, 88.38339233398438, 17.41590690612793, 195.583984375, 86.26110076904297, 174.35975646972656, 257.5398864746094, 553.22802734375, 34.958953857421875, 648.6570434570312, 190.57550048828125, 186.0208740234375, -35.52274703979492, 208.6271209716797, 0.42763710021972656, 29.116455078125, -6.458747863769531, -14.178535461425781, 9.913070678710938, -322.3017883300781, 9.122270584106445, 156.3650665283203, -212.02066040039062, 249.6900177001953, 167.43060302734375, 315.40313720703125, 48.60027313232422, 141.43716430664062, 127.06535339355469, -90.2836685180664, -45.943016052246094, -33.093475341796875, 93.55109405517578, 304.5246276855469, 189.8065948486328, 41.82777404785156, 203.5660858154297, -17.867244720458984, 76.74879455566406, 205.6977081298828, 75.45652770996094, -410.7747497558594, 85.71578979492188, 69.0907974243164, 179.5994110107422, 53.80699157714844, -137.38258361816406, 33.220848083496094, 100.44288635253906, -384.8098449707031, 196.2732391357422, 456.8856506347656, 241.82308959960938, -307.49786376953125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000588.npy"}
|
||||
{"epoch": 0.8888888888888888, "step": 589, "batch_size": 64, "mean": 92.2740707397461, "std": 139.5054473876953, "min": -290.5388488769531, "p10": -64.48282546997066, "median": 90.35986328125, "p90": 237.8907470703125, "max": 457.35015869140625, "pos_frac": 0.796875, "sample": [52.218597412109375, 19.85749053955078, -184.976318359375, 223.237060546875, 114.89368438720703, -2.620941162109375, -175.34271240234375, -32.987060546875, 75.34268951416016, -13.794769287109375, -88.46885681152344, 17.059587478637695, 457.35015869140625, 5.416919708251953, 244.8260955810547, 77.08560943603516, 118.14344024658203, 161.8588409423828, 213.4518280029297, 249.48800659179688, 127.57437133789062, 161.9998779296875, 204.87667846679688, -248.695068359375, 403.68182373046875, 143.74337768554688, -77.98101043701172, 117.49144744873047, 174.76466369628906, 201.4817657470703, 53.47136688232422, 15.165225982666016, 207.15054321289062, 141.28302001953125, -16.21551513671875, 77.97235107421875, 135.4034423828125, -21.847267150878906, 91.46541595458984, 215.16644287109375, 79.49629211425781, 176.56814575195312, 239.01473999023438, 156.57525634765625, 60.296173095703125, -193.39974975585938, 183.53970336914062, -27.067615509033203, 11.030010223388672, 66.91301727294922, -290.5388488769531, 235.26809692382812, 184.9806671142578, 71.0313491821289, 89.25431060791016, 200.0852813720703, 154.56678771972656, 176.81036376953125, 41.71186065673828, 53.94769287109375, 274.2168884277344, 15.79043197631836, 304.965576171875, 0.49134254455566406], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000589.npy"}
|
||||
{"epoch": 0.890400604686319, "step": 590, "batch_size": 64, "mean": 82.88819122314453, "std": 137.9273681640625, "min": -244.4742431640625, "p10": -92.0363868713379, "median": 82.4420280456543, "p90": 249.1248962402344, "max": 362.8565979003906, "pos_frac": 0.6875, "sample": [82.97406768798828, 33.876197814941406, 172.85247802734375, 3.0167160034179688, 254.15274047851562, 180.10833740234375, 238.4034423828125, -244.4742431640625, 9.265300750732422, -20.941978454589844, 81.90998840332031, -4.863689422607422, 34.01042175292969, 94.16551971435547, -189.54405212402344, 201.6881561279297, 199.7307586669922, -110.45037841796875, -11.518505096435547, -0.3210601806640625, -91.62709045410156, 239.24822998046875, 32.677955627441406, 245.44464111328125, 231.3431396484375, 9.387748718261719, -45.69462585449219, 141.3904571533203, 340.4801330566406, 186.83273315429688, -92.21179962158203, 362.8565979003906, 20.269685745239258, -62.2359619140625, 184.032958984375, 91.99168395996094, 86.86493682861328, 28.477066040039062, 119.68440246582031, 48.20018005371094, 262.7940979003906, 154.2530517578125, 331.0702209472656, 250.7021484375, 207.718994140625, 107.52742004394531, -50.08343505859375, 286.32403564453125, -143.1341552734375, 17.094772338867188, -113.45558166503906, 229.76817321777344, 141.89215087890625, -66.46998596191406, -13.945755004882812, 226.32846069335938, -13.779603958129883, 115.7857437133789, 211.54095458984375, -5.566783905029297, 74.81771850585938, -50.30877685546875, 202.9181671142578, -140.40081787109375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000590.npy"}
|
||||
{"epoch": 0.891912320483749, "step": 591, "batch_size": 64, "mean": 119.92483520507812, "std": 203.6518096923828, "min": -288.23590087890625, "p10": -93.1168830871582, "median": 107.95354461669922, "p90": 363.65546875000007, "max": 903.0629272460938, "pos_frac": 0.703125, "sample": [3.3643741607666016, 252.10247802734375, 193.2225799560547, 220.1344757080078, 261.3690490722656, 134.3341064453125, 220.16224670410156, 191.0183563232422, 79.61026000976562, 99.0633544921875, 116.84373474121094, -264.6233825683594, 279.76214599609375, -97.12451934814453, 118.36636352539062, -17.771865844726562, 11.712448120117188, -5.401214599609375, 162.50189208984375, 13.76346206665039, 308.5402526855469, 24.935190200805664, 516.47705078125, -271.43023681640625, 197.2418670654297, -186.5776824951172, 344.362060546875, 214.64346313476562, -11.79193115234375, -32.238677978515625, -288.23590087890625, -202.22068786621094, 524.616455078125, 304.7286376953125, -6.283317565917969, 248.39027404785156, -40.309146881103516, 56.013427734375, 222.6229705810547, 13.053691864013672, 15.671669006347656, -14.987655639648438, 63.468109130859375, 169.8981170654297, 2.159158706665039, 371.924072265625, -6.696222305297852, -83.76573181152344, -3.6927261352539062, 903.0629272460938, -109.32801818847656, -2.8474655151367188, 375.8058166503906, 47.256797790527344, -20.62405776977539, 254.87814331054688, 380.4650573730469, 183.1648712158203, 172.16542053222656, 146.23480224609375, 289.36199951171875, 400.71649169921875, 2.4642887115478516, 229.48538208007812], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000591.npy"}
|
||||
{"epoch": 0.8934240362811792, "step": 592, "batch_size": 64, "mean": 91.32894897460938, "std": 151.55987548828125, "min": -164.49093627929688, "p10": -63.73913955688476, "median": 56.22427177429199, "p90": 262.58880004882815, "max": 682.4593505859375, "pos_frac": 0.734375, "sample": [143.6046600341797, 271.8008728027344, 263.75640869140625, 12.838937759399414, 213.1262969970703, 218.8485870361328, 319.2003173828125, 30.73678207397461, 25.015769958496094, -24.038978576660156, -49.78166961669922, 169.83639526367188, 73.71247863769531, 46.582984924316406, -128.16201782226562, 122.81404876708984, 221.82235717773438, 152.20693969726562, 0.40523529052734375, 23.733440399169922, -108.80274963378906, 221.72265625, 74.09185791015625, 345.8877868652344, 16.671192169189453, 213.70106506347656, 119.0013198852539, 205.64236450195312, 259.8643798828125, -35.49809265136719, -110.3598403930664, 158.98153686523438, -54.740882873535156, 191.8134307861328, 43.80474090576172, 159.38040161132812, -112.30464172363281, 215.95684814453125, -164.49093627929688, -53.65174865722656, -11.481563568115234, 59.20744323730469, 58.94709014892578, -67.59553527832031, -141.19761657714844, 38.69464111328125, 179.69200134277344, 500.529052734375, 307.8371887207031, 35.62249755859375, 682.4593505859375, -0.283203125, 21.342876434326172, 2.0701446533203125, -3.52960205078125, 9.802457809448242, 71.71817016601562, -5.589591979980469, -51.970882415771484, 11.834972381591797, 53.5014533996582, 72.04256439208984, 173.32540893554688, 153.34255981445312], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000592.npy"}
|
||||
{"epoch": 0.8949357520786092, "step": 593, "batch_size": 64, "mean": 100.82471466064453, "std": 144.75909423828125, "min": -416.8070983886719, "p10": -62.81305160522461, "median": 123.9395866394043, "p90": 267.5563171386719, "max": 363.66729736328125, "pos_frac": 0.78125, "sample": [270.01544189453125, 58.277156829833984, 241.3587646484375, 64.6737060546875, 202.11566162109375, -83.0459976196289, 184.05148315429688, -416.8070983886719, 265.6439208984375, 60.74383544921875, 225.63088989257812, 55.58626174926758, 363.66729736328125, 225.58616638183594, 100.98225402832031, 37.235965728759766, 193.5858154296875, 198.86746215820312, 190.16375732421875, 171.78060913085938, 63.40856170654297, 171.87892150878906, 311.1459655761719, 327.25262451171875, -64.81704711914062, -15.010948181152344, 16.855981826782227, 210.2064208984375, 175.4918212890625, -136.13619995117188, 153.7071533203125, -56.010887145996094, 135.87867736816406, 93.75425720214844, 162.3515167236328, 203.84307861328125, -1.313812255859375, -25.802337646484375, 123.64585876464844, -206.3049774169922, -167.03070068359375, 5.604400634765625, 285.10443115234375, -143.22329711914062, 181.89712524414062, 194.9496307373047, 212.0629119873047, 21.62827491760254, -36.67076873779297, 323.8945617675781, 152.71954345703125, -26.346765518188477, 244.42971801757812, 175.85110473632812, 94.95511627197266, -58.137062072753906, 25.392343521118164, 268.37591552734375, 3.7270774841308594, 28.254356384277344, 84.3070068359375, 124.23331451416016, 2.1672515869140625, 200.49876403808594], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000593.npy"}
|
||||
{"epoch": 0.8964474678760394, "step": 594, "batch_size": 64, "mean": 77.0777587890625, "std": 144.57806396484375, "min": -348.21356201171875, "p10": -80.08570098876953, "median": 31.980688095092773, "p90": 253.98745422363285, "max": 409.2379455566406, "pos_frac": 0.734375, "sample": [73.60699462890625, 301.67974853515625, -121.70840454101562, -27.698570251464844, 84.42471313476562, 179.49014282226562, 0.8136672973632812, 46.25909423828125, 98.14143371582031, 25.63166046142578, -134.81924438476562, 33.323421478271484, -13.739639282226562, 258.0875244140625, 29.3504581451416, 19.61550521850586, -79.90951538085938, 3.704448699951172, 20.690673828125, 30.637954711914062, 271.5013122558594, 134.4839630126953, 4.012434005737305, 224.3485870361328, 234.5846405029297, 95.74476623535156, 216.75222778320312, -54.01251983642578, 194.5227508544922, -3.231447219848633, 244.42062377929688, 374.8039245605469, 349.1651306152344, 188.26657104492188, 166.13339233398438, -112.59812927246094, -90.15792083740234, 146.43289184570312, 188.72772216796875, -60.00123596191406, 201.63758850097656, 24.35997200012207, 20.038209915161133, 190.32611083984375, 8.233222961425781, 8.816869735717773, 23.537229537963867, 217.63458251953125, -6.19416618347168, 226.75341796875, 409.2379455566406, -348.21356201171875, 27.1143798828125, 316.16717529296875, -210.12371826171875, 106.76780700683594, 0.02654266357421875, 38.822425842285156, 195.1537628173828, -50.74317169189453, 154.92306518554688, -40.12750244140625, -42.492515563964844, -80.16120910644531], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000594.npy"}
|
||||
{"epoch": 0.8979591836734694, "step": 595, "batch_size": 64, "mean": 111.36105346679688, "std": 174.62818908691406, "min": -292.3653259277344, "p10": -68.68059921264648, "median": 85.18917846679688, "p90": 303.21954040527345, "max": 660.4705810546875, "pos_frac": 0.78125, "sample": [-0.18704986572265625, 77.8775405883789, 660.4705810546875, 163.98626708984375, -67.3885269165039, 15.65294075012207, 138.49160766601562, 43.54318618774414, 38.34019470214844, 230.2789306640625, 336.4300231933594, 24.87541961669922, 147.9434356689453, -185.43643188476562, 234.2613067626953, -15.708572387695312, -0.30025672912597656, 16.911468505859375, 246.3630828857422, 36.673988342285156, 28.717308044433594, 147.06753540039062, 298.13543701171875, 23.922523498535156, 592.3953857421875, 92.50081634521484, 149.40756225585938, 140.96522521972656, 304.83734130859375, 120.33078002929688, 16.86896514892578, -138.96957397460938, 54.62825012207031, -56.991058349609375, 218.39398193359375, 211.16539001464844, 358.5352783203125, -292.3653259277344, -249.84815979003906, 285.14886474609375, 240.59942626953125, 233.06959533691406, 317.61566162109375, 42.685821533203125, -89.0036849975586, 249.7754669189453, 74.90255737304688, 151.17974853515625, -69.23434448242188, 105.95783996582031, -0.7280483245849609, 365.3897399902344, 185.1555633544922, 67.41548919677734, -47.85601806640625, 55.54816436767578, 239.97850036621094, 212.43960571289062, 67.08311462402344, 1.7717132568359375, 0.014661788940429688, -243.12025451660156, 219.09678649902344, 299.4446716308594], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000595.npy"}
|
||||
{"epoch": 0.8994708994708994, "step": 596, "batch_size": 64, "mean": 39.14437484741211, "std": 130.04591369628906, "min": -259.80950927734375, "p10": -111.53581848144532, "median": 18.281320571899414, "p90": 200.41349945068362, "max": 377.65447998046875, "pos_frac": 0.5625, "sample": [-103.74746704101562, -34.05040740966797, 269.79931640625, 15.856834411621094, 51.09209442138672, -120.04999542236328, -10.819042205810547, 1.0856704711914062, -45.810707092285156, 215.54946899414062, -151.8948516845703, 43.87617874145508, -28.52960205078125, -111.55368041992188, 377.65447998046875, 197.52816772460938, 173.93936157226562, 126.5372314453125, -154.72085571289062, 50.05152130126953, -83.938720703125, 201.6500701904297, 64.414306640625, 137.49131774902344, 215.37966918945312, 170.21380615234375, 163.34716796875, -17.407278060913086, 120.6560287475586, -259.80950927734375, -111.494140625, 108.03073120117188, 56.79017639160156, 105.64591217041016, 145.52191162109375, 2.293659210205078, 126.94718170166016, 309.58074951171875, 20.705806732177734, -24.53209686279297, -58.791168212890625, -28.397666931152344, 12.880016326904297, -218.02561950683594, -37.78630065917969, -24.853649139404297, 55.75982666015625, -33.58902359008789, 285.6582946777344, 171.9390869140625, 163.19741821289062, -7.3563232421875, -60.33873748779297, -133.91897583007812, -49.797218322753906, 168.9297332763672, 51.79219055175781, 196.36187744140625, -15.273653030395508, 24.588808059692383, 43.5660514831543, -42.12358093261719, -89.56700134277344, -82.89484405517578], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000596.npy"}
|
||||
{"epoch": 0.9009826152683296, "step": 597, "batch_size": 64, "mean": 79.73832702636719, "std": 124.60186004638672, "min": -194.4435272216797, "p10": -71.95443496704101, "median": 58.19932174682617, "p90": 233.323649597168, "max": 350.7869567871094, "pos_frac": 0.75, "sample": [-16.148395538330078, -36.38904571533203, -105.65635681152344, 20.551681518554688, 222.42550659179688, -60.70001983642578, 25.99423599243164, -103.81922149658203, -22.935720443725586, 229.30223083496094, 8.65884017944336, 133.59571838378906, 0.7387561798095703, 268.5372619628906, -50.32243347167969, 14.839506149291992, 191.96017456054688, 265.8076171875, 208.80422973632812, 11.936101913452148, 116.76412963867188, 33.536895751953125, -194.4435272216797, 84.05098724365234, 226.66790771484375, 202.61021423339844, 3.033109664916992, -94.29707336425781, 253.66571044921875, 15.056985855102539, 195.55868530273438, 79.861328125, 234.40145874023438, 101.69702911376953, 71.94194793701172, 175.2116241455078, -76.77775573730469, 103.21566772460938, 10.992189407348633, 83.6561279296875, 2.739490509033203, -100.56964111328125, 26.246925354003906, -42.20098114013672, 32.105262756347656, -13.084051132202148, 4.104188919067383, 44.456695556640625, 174.26715087890625, 149.15658569335938, 250.97702026367188, 9.895076751708984, 72.62055206298828, -4.8689422607421875, -164.06410217285156, 129.65240478515625, -0.047084808349609375, 230.8087615966797, 349.57220458984375, 350.7869567871094, 139.42364501953125, 214.10467529296875, 186.61993408203125, 226.9658203125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000597.npy"}
|
||||
{"epoch": 0.9024943310657596, "step": 598, "batch_size": 64, "mean": 60.90877151489258, "std": 149.34071350097656, "min": -226.65977478027344, "p10": -108.06357345581054, "median": 33.181575775146484, "p90": 238.9902297973633, "max": 526.3843994140625, "pos_frac": 0.65625, "sample": [-6.070024490356445, -142.54649353027344, 42.932395935058594, 69.64826202392578, 144.46234130859375, 5.584415435791016, 10.55078125, 211.92498779296875, 232.8363800048828, -213.7335205078125, 55.78142547607422, 17.530550003051758, 69.53627014160156, -11.58726692199707, 179.74989318847656, -34.71118927001953, 351.9014892578125, 3.753204345703125, 207.4625701904297, 43.97523880004883, -13.220666885375977, -226.65977478027344, -55.81321716308594, 142.4603271484375, 85.07203674316406, 23.430755615234375, 75.9278564453125, 229.09738159179688, 20.949607849121094, -113.08961486816406, -61.60786437988281, -174.8663330078125, 15.811731338500977, 157.99546813964844, 326.2773742675781, -9.067951202392578, -13.947891235351562, -2.138416290283203, 43.79972839355469, 378.1874084472656, 300.1674499511719, 48.987457275390625, -3.549196243286133, 221.63592529296875, -50.65510559082031, 178.89479064941406, 83.97596740722656, 8.528602600097656, 184.15133666992188, 91.8558349609375, -163.23419189453125, -74.75577545166016, 13.888664245605469, 46.4617919921875, 17.220064163208008, -65.77182006835938, 151.23135375976562, 153.2173309326172, -86.79302978515625, -96.33614349365234, 526.3843994140625, 241.62759399414062, -200.11965942382812, 303.568115234375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000598.npy"}
|
||||
{"epoch": 0.9040060468631897, "step": 599, "batch_size": 64, "mean": 65.76770782470703, "std": 158.4142608642578, "min": -349.5719909667969, "p10": -144.21954650878905, "median": 57.95877456665039, "p90": 244.12462615966803, "max": 480.4495849609375, "pos_frac": 0.625, "sample": [-163.4145050048828, -43.876708984375, 37.191261291503906, 90.22039031982422, -108.14698028564453, 116.9806137084961, 21.508228302001953, 394.54486083984375, -149.91603088378906, -349.5719909667969, -2.8728790283203125, -68.27660369873047, 52.373619079589844, 71.73011779785156, 324.9598388671875, 30.476768493652344, -199.7891082763672, -0.8689231872558594, 2.218639373779297, -16.41204833984375, 217.4760284423828, 168.20530700683594, 145.18942260742188, -171.16751098632812, -7.149803161621094, -130.92774963378906, 63.54393005371094, 213.07766723632812, -82.60050964355469, 145.66262817382812, 330.5404052734375, 213.86068725585938, -176.63824462890625, -63.77867889404297, 71.84318542480469, 1.0001220703125, 96.65882110595703, -72.25106811523438, 154.84812927246094, 63.90411376953125, 167.53170776367188, -1.6399917602539062, 41.602020263671875, -58.32977294921875, 187.89813232421875, 90.30953979492188, 290.3163146972656, -12.086685180664062, -207.42721557617188, -12.863544464111328, 208.09860229492188, 373.6001281738281, 117.7275390625, 213.32325744628906, 226.5344696044922, 23.6043643951416, 480.4495849609375, 248.81082153320312, 179.21372985839844, -11.128473281860352, -2.6243743896484375, 95.8510971069336, 233.19017028808594, 116.81674194335938], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000599.npy"}
|
||||
{"epoch": 0.9055177626606198, "step": 600, "batch_size": 64, "mean": 79.5253677368164, "std": 165.0675811767578, "min": -278.912109375, "p10": -102.87413330078124, "median": 40.567237854003906, "p90": 294.7888641357422, "max": 626.19140625, "pos_frac": 0.6875, "sample": [113.40234375, 70.71497344970703, 14.83024787902832, 39.40447998046875, 279.2734680175781, -84.88594055175781, 10.502262115478516, -172.3636932373047, 181.7198486328125, 295.76641845703125, 18.403831481933594, -6.676555633544922, 97.82978057861328, 141.2936553955078, 1.992431640625, 30.75926399230957, -110.58335876464844, -121.5490951538086, -278.912109375, 379.5797119140625, 44.51430130004883, 92.60617065429688, 93.97023010253906, 136.04302978515625, -35.03923797607422, 517.6828002929688, 164.995849609375, 222.46014404296875, -51.70563507080078, 277.8865966796875, 319.72564697265625, -14.222221374511719, -227.0655975341797, -4.541496276855469, 115.84269714355469, -73.28396606445312, 6.73193359375, 178.78717041015625, 2.874286651611328, 626.19140625, -13.084144592285156, 292.5079040527344, -1.3914642333984375, 346.3911437988281, -9.026533126831055, -0.2547111511230469, -112.16194152832031, -50.64326477050781, -157.27581787109375, 58.459224700927734, 8.937210083007812, 144.33763122558594, 121.83812713623047, 41.72999572753906, 216.2149200439453, 13.34344482421875, 155.73875427246094, 68.36785888671875, 74.2972412109375, 272.75128173828125, 3.7648849487304688, 0.8477706909179688, -13.38897705078125, 332.36688232421875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000600.npy"}
|
||||
{"epoch": 0.9070294784580499, "step": 601, "batch_size": 64, "mean": 106.2969741821289, "std": 177.9225311279297, "min": -312.05841064453125, "p10": -81.43291778564452, "median": 101.60733032226562, "p90": 300.19039001464853, "max": 579.2440795898438, "pos_frac": 0.734375, "sample": [199.31312561035156, 44.14916229248047, 258.7698669433594, -194.3009490966797, 265.4038391113281, 149.950927734375, 554.3253173828125, -250.71441650390625, 110.61994934082031, -17.800140380859375, 278.14105224609375, -226.232177734375, 223.30372619628906, 24.925806045532227, -21.835535049438477, 201.95639038085938, 207.0472412109375, 423.02728271484375, -70.72515869140625, 213.93771362304688, 268.3020324707031, 21.728546142578125, 1.5980072021484375, 309.6401062011719, 240.1510009765625, 175.04318237304688, -8.194639205932617, -86.02195739746094, 214.35446166992188, 9.695449829101562, 18.144325256347656, 157.39710998535156, 387.36993408203125, -45.35163116455078, -7.109918594360352, 186.1770782470703, -18.783485412597656, 161.78915405273438, 70.44149780273438, 15.278051376342773, -156.5695037841797, 197.04437255859375, 12.629512786865234, 254.81893920898438, 316.704345703125, 191.7555694580078, 44.169368743896484, -31.747413635253906, -17.878503799438477, 12.916793823242188, -11.524589538574219, 399.19793701171875, 2.957487106323242, 6.716167449951172, 212.24337768554688, 248.80801391601562, 579.2440795898438, 124.43327331542969, 175.98806762695312, 4.5322265625, -163.24412536621094, 174.36306762695312, 92.59471130371094, -312.05841064453125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000601.npy"}
|
||||
{"epoch": 0.90854119425548, "step": 602, "batch_size": 64, "mean": 89.7902603149414, "std": 143.67820739746094, "min": -407.5876159667969, "p10": -42.57334098815918, "median": 70.34272766113281, "p90": 263.80186157226564, "max": 542.1339111328125, "pos_frac": 0.796875, "sample": [9.213088989257812, 184.62179565429688, 255.7828369140625, 542.1339111328125, -21.999862670898438, 64.97261047363281, 128.94570922851562, 3.0442638397216797, -63.056854248046875, -15.36404037475586, 291.17169189453125, 14.846744537353516, 267.23858642578125, 197.5556640625, -129.49472045898438, 273.30633544921875, -407.5876159667969, 326.4207458496094, 364.17022705078125, -34.35764694213867, 77.141845703125, -8.245002746582031, 31.054611206054688, -17.004451751708984, 75.71284484863281, 229.58001708984375, 128.47921752929688, 221.1292724609375, 26.03664779663086, 102.20303344726562, 53.985965728759766, 188.9081573486328, 111.94844818115234, 62.93812942504883, 34.81376266479492, 11.488880157470703, 20.715869903564453, 108.2610092163086, 191.23793029785156, 90.62601470947266, 45.351497650146484, 142.91668701171875, 194.26939392089844, 182.193115234375, 30.694000244140625, -190.3021240234375, 58.063194274902344, 103.64753723144531, 4.7698211669921875, 8.150947570800781, -40.82807540893555, 8.448989868164062, 194.7315673828125, -104.19937896728516, -43.321311950683594, 224.34730529785156, 108.4112777709961, 93.92546081542969, 283.7477722167969, 252.7208709716797, -106.8907699584961, 198.8124237060547, 40.95326232910156, 63.387725830078125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000602.npy"}
|
||||
{"epoch": 0.91005291005291, "step": 603, "batch_size": 64, "mean": 99.57186889648438, "std": 169.5417022705078, "min": -328.4748840332031, "p10": -68.10823974609374, "median": 64.57506561279297, "p90": 230.8127868652344, "max": 714.7083740234375, "pos_frac": 0.734375, "sample": [212.306640625, 256.31866455078125, 327.9334716796875, 30.023155212402344, -97.74638366699219, 168.3256378173828, 49.14051055908203, 177.41567993164062, -4.7820587158203125, 41.15715026855469, 210.58778381347656, 154.87034606933594, -8.511665344238281, -70.9687271118164, 11.942054748535156, 714.7083740234375, -155.47816467285156, -2.9445114135742188, 49.45915985107422, 561.789306640625, 45.89775848388672, -15.593109130859375, 211.53794860839844, 121.17536926269531, -22.250797271728516, 110.45156860351562, 224.5332794189453, -114.04705810546875, 183.83570861816406, -241.25608825683594, -10.408935546875, 28.3450927734375, 214.9482879638672, 162.47537231445312, -61.43376922607422, 198.99195861816406, 4.169086456298828, 47.15755844116211, -19.36391258239746, 212.70901489257812, 62.05189514160156, 233.37051391601562, 175.5906524658203, 48.929649353027344, 93.2407455444336, 264.72662353515625, 83.42472839355469, -73.39501953125, 113.55915832519531, 24.849763870239258, 62.68449401855469, 66.46563720703125, 195.59616088867188, 222.15184020996094, 76.6598129272461, -53.1201171875, 8.804210662841797, -328.4748840332031, 204.0858154296875, 220.64715576171875, 224.84475708007812, 8.264986038208008, 556.5408325195312, -26.32074737548828], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000603.npy"}
|
||||
{"epoch": 0.9115646258503401, "step": 604, "batch_size": 64, "mean": 74.13782501220703, "std": 166.96762084960938, "min": -332.99664306640625, "p10": -155.76795349121093, "median": 61.704505920410156, "p90": 237.52424926757814, "max": 646.475341796875, "pos_frac": 0.71875, "sample": [315.9189453125, 174.52639770507812, -25.393451690673828, 9.679931640625, -4.755207061767578, 204.65878295898438, 31.146533966064453, 0.8720645904541016, 218.6154327392578, -31.292251586914062, 138.81961059570312, -3.8159866333007812, 4.645803451538086, 238.14340209960938, -160.496826171875, -95.46014404296875, 198.60072326660156, 94.07823181152344, 10.088386535644531, 320.28533935546875, -332.99664306640625, -144.73391723632812, 13.749130249023438, -166.9647216796875, 170.47415161132812, 26.42075538635254, 15.858003616333008, 137.52020263671875, 75.000732421875, -319.3656005859375, 22.93948745727539, -195.45074462890625, 62.74237823486328, -209.4149169921875, 203.3257293701172, -66.3648910522461, 188.7647705078125, -170.5787353515625, 218.42242431640625, 159.712890625, 104.4674072265625, 231.96548461914062, 646.475341796875, 53.20026397705078, 171.02365112304688, 44.985626220703125, 60.66663360595703, 236.07955932617188, 269.48968505859375, -76.41812133789062, 296.4734802246094, -10.567886352539062, 119.8825912475586, 112.56832122802734, 200.37606811523438, 75.58909606933594, 229.35964965820312, 53.85913848876953, 7.055883407592773, 292.2698059082031, 193.48609924316406, -0.2459259033203125, -128.31968688964844, 233.1724090576172], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000604.npy"}
|
||||
{"epoch": 0.9130763416477702, "step": 605, "batch_size": 64, "mean": 89.56930541992188, "std": 155.55946350097656, "min": -272.0816650390625, "p10": -97.66140136718747, "median": 90.79202270507812, "p90": 264.5126007080078, "max": 463.16552734375, "pos_frac": 0.71875, "sample": [91.5851058959961, 154.01951599121094, -181.98721313476562, 463.16552734375, -42.41718292236328, 257.7284240722656, 250.68528747558594, 320.4806213378906, 123.5223159790039, -9.07928466796875, 388.79620361328125, -40.35877990722656, 65.35846710205078, -19.733436584472656, -28.600994110107422, -71.61778259277344, 112.59474182128906, 189.97125244140625, 163.9587860107422, 66.32682800292969, 3.6087646484375, 25.366790771484375, 188.46405029296875, -272.0816650390625, 10.088768005371094, 158.70533752441406, 214.8023681640625, 211.96478271484375, 34.84514617919922, 112.32284545898438, 223.75192260742188, 89.99893951416016, -259.717041015625, 203.8097381591797, -9.362640380859375, 297.0880432128906, 250.4442901611328, 209.68368530273438, 46.85603332519531, 101.2845458984375, -108.82295227050781, -26.089252471923828, 238.73092651367188, 29.887739181518555, 240.57681274414062, 52.40454864501953, 267.42010498046875, 203.7578582763672, 214.70468139648438, -60.9747314453125, 359.4910888671875, 34.51564025878906, 188.5421142578125, -164.18325805664062, -10.031295776367188, 8.844690322875977, 15.601045608520508, -216.59921264648438, 92.33934020996094, -148.80764770507812, 17.322237014770508, 305.36334228515625, -23.900482177734375, 126.0191650390625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000605.npy"}
|
||||
{"epoch": 0.9145880574452003, "step": 606, "batch_size": 64, "mean": 114.13306427001953, "std": 158.2932891845703, "min": -196.22341918945312, "p10": -79.77193527221678, "median": 118.0014419555664, "p90": 290.67737426757816, "max": 555.718017578125, "pos_frac": 0.734375, "sample": [-4.291847229003906, 294.93170166015625, 116.37884521484375, 357.9329833984375, -86.9211196899414, 51.82080841064453, 167.45404052734375, 263.0567626953125, -2.634153366088867, 195.40402221679688, 225.63284301757812, -58.62931823730469, 142.32504272460938, 280.7506103515625, 235.43862915039062, -187.83517456054688, 171.8034210205078, 98.56936645507812, 18.463041305541992, 196.26339721679688, 555.718017578125, -6.33135986328125, 64.14649963378906, 193.71209716796875, -56.834327697753906, -149.50491333007812, 474.79736328125, -63.508689880371094, 175.1116943359375, 236.06686401367188, 194.3961639404297, 8.007720947265625, -196.22341918945312, 192.09393310546875, 80.02088928222656, 2.7469043731689453, 130.7635040283203, 27.98372459411621, 257.6813049316406, 16.01561737060547, 262.1523742675781, 317.2392883300781, 39.46306610107422, 119.62403869628906, -0.11457061767578125, 435.3214111328125, -12.260421752929688, 164.04898071289062, -119.55982208251953, -86.74189758300781, 215.00924682617188, 101.43377685546875, -148.3348388671875, 17.63702392578125, -44.36968994140625, 173.17379760742188, 138.6007080078125, 11.74545669555664, 173.5612335205078, 383.6275634765625, 215.27015686035156, 238.09963989257812, 102.45501708984375, -5.339178085327148], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000606.npy"}
|
||||
{"epoch": 0.9160997732426304, "step": 607, "batch_size": 64, "mean": 95.0063247680664, "std": 168.209716796875, "min": -413.24053955078125, "p10": -94.06137084960938, "median": 107.03781509399414, "p90": 264.38142242431655, "max": 606.036865234375, "pos_frac": 0.6875, "sample": [-5.801765441894531, 50.32415771484375, -164.3391571044922, 198.29583740234375, 178.09097290039062, 6.453315734863281, 201.05706787109375, 331.693359375, -413.24053955078125, -9.226966857910156, -12.535938262939453, -174.30233764648438, -30.898406982421875, 214.9354705810547, 60.82173156738281, -162.334228515625, -24.34772491455078, 6.620733261108398, -22.597190856933594, 289.68536376953125, 70.3485107421875, 280.66204833984375, 206.89114379882812, 119.37687683105469, 192.32376098632812, 31.63568878173828, -47.76293182373047, 190.71409606933594, -13.905914306640625, -43.54436492919922, 297.6806640625, 117.99254608154297, 176.5736083984375, 173.8714599609375, -43.9246826171875, 211.49249267578125, 457.546875, -23.06247329711914, 225.7814178466797, 140.35438537597656, -96.95935821533203, 20.154502868652344, 49.786415100097656, 226.39329528808594, 137.24473571777344, 28.365127563476562, -114.9124526977539, 160.96304321289062, 471.6963195800781, 208.06478881835938, 10.235614776611328, 96.08308410644531, 178.29083251953125, 606.036865234375, 126.31208801269531, 221.36358642578125, -3.381877899169922, 156.15481567382812, 209.79444885253906, 220.52932739257812, 50.48942565917969, -209.35919189453125, -87.29940032958984, 174.95989990234375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000607.npy"}
|
||||
{"epoch": 0.9176114890400605, "step": 608, "batch_size": 64, "mean": 104.33794403076172, "std": 163.63670349121094, "min": -198.49148559570312, "p10": -96.06854934692383, "median": 79.10475158691406, "p90": 296.6643432617188, "max": 585.7202758789062, "pos_frac": 0.765625, "sample": [229.89675903320312, -124.44200134277344, -0.8670158386230469, 225.3278045654297, 0.8408412933349609, -198.49148559570312, 178.98458862304688, 12.924530029296875, 26.126644134521484, -183.88819885253906, 164.36709594726562, 155.66629028320312, 229.65199279785156, 224.6400604248047, 65.95620727539062, 172.38961791992188, 342.91461181640625, -191.44512939453125, -6.255891799926758, 45.697383880615234, 159.56455993652344, -33.03255081176758, -0.9820423126220703, 294.99517822265625, 214.10067749023438, 37.34322738647461, 7.550788879394531, 133.01710510253906, 12.007888793945312, 240.0410614013672, 575.0307006835938, -97.13867950439453, 337.7590637207031, 182.6817626953125, 73.17703247070312, 123.98757934570312, 365.9925842285156, -53.423561096191406, -170.98069763183594, -30.7294921875, -21.700889587402344, 14.735496520996094, 138.60853576660156, 40.025978088378906, -93.57157897949219, 184.8315887451172, 35.02252197265625, 202.28382873535156, 585.7202758789062, 201.8568878173828, 104.10015869140625, 58.1199951171875, 95.91742706298828, 223.45858764648438, 6.294010162353516, 150.19955444335938, -112.52632141113281, 213.3023681640625, 85.032470703125, 21.69445037841797, 40.67882537841797, 297.37969970703125, 34.76225280761719, 430.44561767578125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000608.npy"}
|
||||
{"epoch": 0.9191232048374905, "step": 609, "batch_size": 64, "mean": 100.04335021972656, "std": 124.40462493896484, "min": -284.7781066894531, "p10": -62.887982177734365, "median": 113.8303108215332, "p90": 234.01604614257815, "max": 485.5521240234375, "pos_frac": 0.8125, "sample": [290.5751953125, 241.0824432373047, 189.9344482421875, 17.60388946533203, 197.4659423828125, 36.277313232421875, 208.80242919921875, 211.62744140625, -14.679275512695312, 7.357597351074219, -66.05842590332031, 118.88923645019531, -55.49028015136719, 21.432750701904297, 485.5521240234375, 70.40087890625, 219.05160522460938, 159.57989501953125, 88.06196594238281, 266.06646728515625, -93.3274154663086, 40.8065185546875, 200.9632568359375, 141.07443237304688, -114.40957641601562, 250.9716796875, 231.24366760253906, 57.47583770751953, -141.7244873046875, 193.69329833984375, 114.9850845336914, -3.016641616821289, 201.4970703125, 56.498558044433594, -82.9415283203125, 160.34063720703125, 89.94136047363281, 2.117525100708008, 241.03961181640625, 188.0746612548828, 124.29293060302734, -28.355052947998047, 137.95993041992188, 208.06686401367188, 204.0372772216797, -284.7781066894531, 21.579402923583984, 81.94595336914062, 138.28964233398438, 29.587186813354492, 89.88539123535156, 126.90513610839844, 160.97080993652344, 105.1612777709961, 5.657522201538086, 36.185340881347656, 235.20420837402344, 139.97274780273438, 196.9619598388672, -87.58008575439453, 159.11184692382812, 74.34590148925781, 112.675537109375, -14.146133422851562], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000609.npy"}
|
||||
{"epoch": 0.9206349206349206, "step": 610, "batch_size": 64, "mean": 85.75836181640625, "std": 191.68252563476562, "min": -633.699462890625, "p10": -85.11733474731444, "median": 61.2169303894043, "p90": 306.2153625488282, "max": 575.20556640625, "pos_frac": 0.75, "sample": [28.544273376464844, 396.75665283203125, 384.4696350097656, 346.72894287109375, 61.93573760986328, 253.9931182861328, -8.053131103515625, 11.409097671508789, -535.6602783203125, 38.828765869140625, 104.7461166381836, -129.1898193359375, 138.04969787597656, 1.6540069580078125, -27.64691162109375, 53.60160827636719, 212.79244995117188, -89.53495788574219, 3.100250244140625, -74.8095474243164, 25.73175811767578, -133.1321563720703, -68.27202606201172, 152.32374572753906, -46.89366149902344, -219.26393127441406, 60.49812316894531, 215.3592529296875, 89.55267333984375, 291.46990966796875, 44.082977294921875, 575.20556640625, 311.62567138671875, 224.00924682617188, 27.812118530273438, 205.9383544921875, 209.4204559326172, -4.46699333190918, 203.35296630859375, 31.9619140625, 21.96466064453125, 165.45555114746094, 23.441261291503906, -633.699462890625, 314.5194091796875, 3.649944305419922, 26.96068572998047, 19.69765853881836, 197.83834838867188, 288.3700256347656, 151.60386657714844, 74.95375061035156, 133.85362243652344, -9.959535598754883, -3.4406375885009766, -54.182640075683594, -115.43553161621094, 245.123046875, 293.59130859375, 154.05433654785156, 114.36344909667969, 179.6955108642578, 336.71331787109375, 191.3715057373047], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000610.npy"}
|
||||
{"epoch": 0.9221466364323507, "step": 611, "batch_size": 64, "mean": 82.84901428222656, "std": 129.34963989257812, "min": -204.83786010742188, "p10": -71.71212615966795, "median": 72.04349899291992, "p90": 257.90910644531255, "max": 419.0058898925781, "pos_frac": 0.703125, "sample": [126.43707275390625, 146.3646240234375, 143.02597045898438, 27.52133560180664, 227.5910186767578, -2.9268569946289062, -40.895751953125, 61.46492004394531, -6.367908477783203, 30.168319702148438, 6.288852691650391, 144.03843688964844, 110.03582000732422, 132.8626708984375, -8.520545959472656, 26.06610107421875, -132.99319458007812, 267.98504638671875, -26.454448699951172, 419.0058898925781, 219.297119140625, 49.087974548339844, 269.3004455566406, -126.9464111328125, -119.92062377929688, -6.879219055175781, 267.4212646484375, 89.1106948852539, 104.51885223388672, 345.6864929199219, -7.460296630859375, -39.742286682128906, 102.69285583496094, 245.63885498046875, -93.06942749023438, -18.862396240234375, -204.83786010742188, 319.175537109375, 142.59262084960938, 18.886825561523438, 263.16778564453125, 63.490386962890625, -174.82940673828125, 63.00447082519531, 80.59661102294922, 105.6777572631836, 113.07012939453125, 88.62625122070312, 192.97080993652344, 222.49136352539062, 61.65684509277344, 215.65609741210938, -3.9835758209228516, 133.33377075195312, 5.343902587890625, 31.658554077148438, 233.87948608398438, -83.9043960571289, 222.21713256835938, 94.44255828857422, -43.26349639892578, 10.969955444335938, -6.7693634033203125, 206.44525146484375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000611.npy"}
|
||||
{"epoch": 0.9236583522297808, "step": 612, "batch_size": 64, "mean": 121.72428894042969, "std": 148.88917541503906, "min": -312.6594543457031, "p10": -29.229713821411128, "median": 118.92846298217773, "p90": 322.13038940429686, "max": 423.44122314453125, "pos_frac": 0.765625, "sample": [-9.874183654785156, -15.270355224609375, 257.0502624511719, 98.28594207763672, 35.24955749511719, -10.795021057128906, 208.6580352783203, 196.84625244140625, -20.700668334960938, 74.23638153076172, 190.61495971679688, 139.25135803222656, 21.90465545654297, 84.67301940917969, 234.25369262695312, 130.690673828125, 71.55213165283203, -30.96611785888672, 82.61695861816406, 213.35501098632812, 174.542724609375, -36.93413543701172, 156.2061767578125, 37.78318786621094, 86.44056701660156, 72.05645751953125, 322.280517578125, -25.178104400634766, -73.961181640625, 32.38140106201172, 125.02069854736328, 255.5742950439453, 5.522670745849609, -17.277326583862305, -19.265785217285156, 276.1969909667969, 11.002021789550781, 311.8283386230469, 357.43060302734375, -130.6488494873047, 338.26055908203125, 112.83622741699219, 408.49066162109375, -181.05616760253906, 416.6175842285156, 232.9929962158203, 268.7837829589844, -37.04486083984375, 180.3643341064453, 321.78009033203125, 67.2103042602539, -10.086326599121094, 404.6787414550781, 145.247802734375, 8.282293319702148, 423.44122314453125, 179.94775390625, 77.5444107055664, 143.7130126953125, -312.6594543457031, 134.84542846679688, 199.2901611328125, 180.48727416992188, 213.75303649902344], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000612.npy"}
|
||||
{"epoch": 0.9251700680272109, "step": 613, "batch_size": 64, "mean": 96.05809783935547, "std": 152.7236328125, "min": -488.5910339355469, "p10": -47.007836151123044, "median": 101.33101654052734, "p90": 264.41122894287116, "max": 384.0367736816406, "pos_frac": 0.75, "sample": [-25.515243530273438, 51.66302490234375, 7.611968994140625, 114.14317321777344, 153.29058837890625, 221.5787353515625, 195.31768798828125, 130.80703735351562, -41.62932586669922, -12.12108039855957, 79.6584243774414, 75.16525268554688, -217.2901611328125, 223.51040649414062, 200.40228271484375, 176.97813415527344, -22.682764053344727, 4.092863082885742, 373.7748107910156, 239.21847534179688, 51.322105407714844, -28.243253707885742, -488.5910339355469, 19.811296463012695, 204.0439453125, 91.84115600585938, 194.9946746826172, -10.502962112426758, -109.65491485595703, 33.88575744628906, 317.5080871582031, 51.005680084228516, 180.82415771484375, -9.162158966064453, -27.421592712402344, 83.80596923828125, -99.02490997314453, 269.4715270996094, 163.43885803222656, 302.7962646484375, 181.30718994140625, 232.69137573242188, 86.3082046508789, 120.62188720703125, 349.22918701171875, 6.0159759521484375, 384.0367736816406, 0.7771663665771484, 14.121627807617188, 252.60386657714844, -49.31291198730469, -180.9383544921875, 194.34652709960938, 326.1409912109375, 110.82087707519531, 182.93783569335938, 26.254688262939453, 232.583984375, 169.6815643310547, 226.84591674804688, -30.579933166503906, 151.7520751953125, 166.3153076171875, -126.96627807617188], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000613.npy"}
|
||||
{"epoch": 0.926681783824641, "step": 614, "batch_size": 64, "mean": 93.28669738769531, "std": 153.3185577392578, "min": -337.5926513671875, "p10": -46.17200851440429, "median": 85.63825988769531, "p90": 260.0227722167969, "max": 517.08251953125, "pos_frac": 0.828125, "sample": [209.25003051757812, 149.50350952148438, 4.551395416259766, 141.240234375, 71.66661834716797, 75.96257781982422, 2.009349822998047, 0.36359405517578125, 148.8666534423828, 83.68328857421875, 325.5828857421875, 159.21197509765625, 468.03387451171875, 294.43878173828125, 291.2712097167969, 198.24888610839844, 129.2573699951172, 53.148780822753906, 144.38824462890625, 264.781005859375, -225.4783477783203, 36.74784851074219, -27.543846130371094, -337.5926513671875, 517.08251953125, 43.602027893066406, 39.491600036621094, 87.59323120117188, 202.87831115722656, -251.83056640625, -171.2027587890625, 27.676883697509766, 14.701976776123047, 213.00283813476562, 30.641979217529297, 11.041101455688477, 158.51683044433594, 137.18463134765625, -219.67469787597656, 180.83737182617188, -15.887779235839844, 180.61859130859375, 217.69107055664062, 169.98121643066406, 198.447021484375, 176.12515258789062, 23.57605743408203, 104.55599975585938, 166.463134765625, 31.342636108398438, 44.165435791015625, 74.92131805419922, -183.72532653808594, -48.68523406982422, 45.28708267211914, 314.9261474609375, 107.88142395019531, 46.04370880126953, 208.13815307617188, -2.59588623046875, 144.61341857910156, 74.71626281738281, -40.30781555175781, 248.92022705078125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000614.npy"}
|
||||
{"epoch": 0.9281934996220711, "step": 615, "batch_size": 64, "mean": 114.22479248046875, "std": 192.3758087158203, "min": -374.9309997558594, "p10": -98.81486968994139, "median": 91.44125366210938, "p90": 314.6914245605469, "max": 760.8919067382812, "pos_frac": 0.75, "sample": [160.4920654296875, 150.68751525878906, 208.8228302001953, 45.7757568359375, 145.35110473632812, 212.58767700195312, 232.6077880859375, -160.88136291503906, 187.01119995117188, 54.497413635253906, -144.46331787109375, 48.55785369873047, -154.4290771484375, 318.1484375, 29.116289138793945, -84.84275817871094, 233.40823364257812, 306.62506103515625, 281.1120300292969, 53.75621032714844, 209.633544921875, -37.07189178466797, 367.642578125, 1.0051040649414062, 219.86007690429688, 85.4251708984375, -70.69210815429688, 117.57643127441406, 11.138031005859375, 250.62620544433594, 234.4032440185547, 97.45733642578125, 5.927543640136719, 54.99599075317383, -50.578758239746094, 199.69424438476562, 657.2445068359375, -66.36465454101562, 70.29512023925781, 188.29367065429688, 319.8164367675781, 338.3616943359375, 193.16787719726562, -129.15684509277344, 284.09814453125, -6.964763641357422, 76.55075073242188, -53.7928466796875, 73.79104614257812, 238.81117248535156, -104.80291748046875, 760.8919067382812, 32.30305480957031, 121.52909851074219, -48.31000518798828, 532.776123046875, 106.45518493652344, 219.62228393554688, 213.83798217773438, -0.73486328125, 39.42744445800781, 66.54426574707031, -259.3573913574219, -374.9309997558594], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000615.npy"}
|
||||
{"epoch": 0.9297052154195011, "step": 616, "batch_size": 64, "mean": 89.8245849609375, "std": 147.71636962890625, "min": -258.660400390625, "p10": -50.36716728210449, "median": 75.87528228759766, "p90": 257.32084960937505, "max": 529.1097412109375, "pos_frac": 0.6875, "sample": [276.6578063964844, 54.026939392089844, 51.3277587890625, -51.26201629638672, 243.79299926757812, -1.6883373260498047, 341.3289489746094, 242.2857208251953, 2.7770118713378906, 177.9071502685547, 90.5696029663086, 18.857330322265625, 95.3009262084961, -229.22817993164062, 76.7576904296875, -32.63959503173828, 147.41030883789062, -19.74289321899414, -40.37190246582031, 300.2715148925781, 103.40015411376953, 226.04312133789062, -15.968606948852539, -12.21574592590332, 238.09347534179688, -2.2225894927978516, -28.907865524291992, 31.101524353027344, -30.871110916137695, 212.9611358642578, 189.78836059570312, 56.72430419921875, 111.468994140625, 180.65953063964844, 318.6463623046875, 206.72032165527344, -35.293758392333984, 92.40145874023438, -172.06661987304688, 212.57861328125, 216.3944091796875, 51.40781784057617, 61.97772216796875, 74.99287414550781, 222.87188720703125, 529.1097412109375, -108.6246109008789, 41.86479568481445, -258.660400390625, 27.01172637939453, -48.2791862487793, -94.97151184082031, 111.4980697631836, 133.31298828125, 26.78728485107422, 263.1184997558594, 159.88516235351562, -15.975765228271484, 168.4708251953125, -46.66911315917969, 204.62783813476562, 421.86968994140625, 118.04815673828125, -138.6750946044922], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000616.npy"}
|
||||
{"epoch": 0.9312169312169312, "step": 617, "batch_size": 64, "mean": 76.78437805175781, "std": 160.73204040527344, "min": -259.21490478515625, "p10": -118.95574111938477, "median": 60.25336837768555, "p90": 250.31733093261718, "max": 625.0146484375, "pos_frac": 0.734375, "sample": [86.22794342041016, -43.36537551879883, -110.11305236816406, 50.883033752441406, 390.07568359375, 139.07147216796875, 474.2865905761719, 202.03109741210938, 18.242046356201172, -119.4899673461914, 141.06777954101562, 41.58318328857422, -259.0096435546875, 1.775747299194336, 8.498950958251953, 232.68734741210938, -154.27056884765625, 376.1637878417969, 48.936737060546875, 9.336662292480469, 625.0146484375, -26.31509780883789, -32.73857498168945, 251.13717651367188, 108.07159423828125, 1.5020599365234375, -1.7662086486816406, 119.64532470703125, 90.36811828613281, -15.632221221923828, 13.206981658935547, 34.99055862426758, 128.77886962890625, 223.80718994140625, 146.93621826171875, 300.09869384765625, 81.07530975341797, -154.77835083007812, -84.36772918701172, 218.52442932128906, 248.40435791015625, 70.82087707519531, -117.70921325683594, -16.030216217041016, 251.43020629882812, 217.36862182617188, -138.2369384765625, -259.21490478515625, -176.88206481933594, 62.17738342285156, 125.41545104980469, 205.19781494140625, 80.54096984863281, 26.353918075561523, 170.1522216796875, 104.45323181152344, 56.15892028808594, 36.244361877441406, 58.32935333251953, 195.60040283203125, 62.9952392578125, 45.14238739013672, 117.634521484375, -74.32537841796875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000617.npy"}
|
||||
{"epoch": 0.9327286470143613, "step": 618, "batch_size": 64, "mean": 102.11787414550781, "std": 170.08888244628906, "min": -255.33364868164062, "p10": -51.84481010437011, "median": 62.63600158691406, "p90": 259.2687927246094, "max": 691.38720703125, "pos_frac": 0.78125, "sample": [401.67816162109375, 292.0362548828125, 56.120296478271484, 156.88882446289062, -3.8951950073242188, 7.15852165222168, -52.74527359008789, 117.39974212646484, 691.38720703125, 2.463634490966797, 12.100700378417969, 619.8899536132812, 165.3476104736328, 112.24217224121094, 199.56423950195312, 12.161720275878906, 45.17290496826172, 4.0842437744140625, 556.8231201171875, 161.5587158203125, 87.2774429321289, 404.2330322265625, 2.7253971099853516, 62.61833953857422, 53.303932189941406, 245.342529296875, 33.76599884033203, 125.5351333618164, 180.92715454101562, 46.766456604003906, 122.22557067871094, -4.191307067871094, 260.13739013671875, -5.479747772216797, 91.16276550292969, -255.33364868164062, 91.07490539550781, -139.35006713867188, 184.68829345703125, 62.653663635253906, -172.23709106445312, -9.53121566772461, 19.970447540283203, -201.64244079589844, 200.3455810546875, 36.22383117675781, 53.509605407714844, -49.74372863769531, 154.854736328125, 257.2420654296875, 149.68832397460938, 198.07725524902344, 24.8095703125, -11.116493225097656, 185.43856811523438, 53.66511535644531, -18.057235717773438, 110.6415023803711, 210.90042114257812, 188.2988739013672, -64.58216857910156, 105.97470092773438, -136.34466552734375, 41.63697814941406], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000618.npy"}
|
||||
{"epoch": 0.9342403628117913, "step": 619, "batch_size": 64, "mean": 104.84233093261719, "std": 168.9849090576172, "min": -344.8590393066406, "p10": -62.82659225463866, "median": 85.48475646972656, "p90": 306.8263946533204, "max": 587.82177734375, "pos_frac": 0.78125, "sample": [158.50732421875, 69.51066589355469, 4.395147323608398, 86.68138122558594, 204.56753540039062, 317.9620361328125, 279.1400146484375, -66.33597564697266, -174.69525146484375, 389.64288330078125, 227.24603271484375, 136.00531005859375, 256.83294677734375, 16.20233726501465, -12.15252685546875, 3.0144805908203125, 112.5536117553711, 53.331512451171875, -14.828470230102539, 1.5428466796875, -173.1717529296875, 334.6470031738281, 46.02294921875, -115.42591857910156, -4.148653030395508, 219.12527465820312, 214.86886596679688, 488.40350341796875, 12.307775497436523, -54.638031005859375, 1.8800582885742188, -344.8590393066406, -14.057510375976562, 577.3970947265625, 161.5826416015625, 208.90838623046875, 38.86606216430664, 212.8653564453125, 280.8432312011719, -128.04017639160156, 136.46507263183594, 363.8695373535156, 84.28813171386719, 1.6966800689697266, 223.59474182128906, 173.44923400878906, 140.74696350097656, 160.30014038085938, 202.23094177246094, 9.629718780517578, 41.47368621826172, 101.33261108398438, 12.78424072265625, 106.67759704589844, 587.82177734375, 24.31122589111328, 98.82186889648438, 125.74747467041016, -16.86855125427246, 24.32921028137207, -14.150177001953125, 171.78927612304688, 77.03523254394531, -139.96881103515625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000619.npy"}
|
||||
{"epoch": 0.9357520786092215, "step": 620, "batch_size": 64, "mean": 73.36974334716797, "std": 152.43173217773438, "min": -456.9232177734375, "p10": -91.69808654785156, "median": 74.27894592285156, "p90": 276.1803405761719, "max": 375.16302490234375, "pos_frac": 0.6875, "sample": [41.524452209472656, -271.11083984375, -3.88995361328125, 118.51708984375, -163.42550659179688, -1.0212478637695312, 130.72654724121094, 245.48341369628906, -93.93704986572266, 92.0688705444336, 278.18865966796875, 206.9784698486328, 129.73171997070312, 284.9362487792969, -8.794853210449219, 375.16302490234375, 132.8427734375, 40.07096862792969, 24.6910400390625, -2.52838134765625, -261.25244140625, 51.07578659057617, 91.17141723632812, 315.64434814453125, -104.47177124023438, 158.2276611328125, -27.732154846191406, -29.37445068359375, 246.46844482421875, -86.47383880615234, 192.7191619873047, 305.1845703125, -24.36673355102539, -13.731260299682617, -17.220027923583984, 24.194091796875, -17.124866485595703, 84.45068359375, 199.2698974609375, 8.880149841308594, 194.18185424804688, 71.00550842285156, 134.48397827148438, 86.78870391845703, -15.438167572021484, -22.134872436523438, 330.69122314453125, 271.4942626953125, 14.927560806274414, 180.8938446044922, 1.5091381072998047, 310.861083984375, -456.9232177734375, 25.528657913208008, -175.19366455078125, 50.93004608154297, 194.65121459960938, 77.55238342285156, 141.8042755126953, 48.4796142578125, 175.76397705078125, 135.17425537109375, 169.7128143310547, 97.16503143310547], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000620.npy"}
|
||||
{"epoch": 0.9372637944066515, "step": 621, "batch_size": 64, "mean": 106.710205078125, "std": 200.2146759033203, "min": -319.0491027832031, "p10": -103.99588317871093, "median": 67.08253860473633, "p90": 345.12158203125006, "max": 845.0745849609375, "pos_frac": 0.75, "sample": [-12.241813659667969, 57.07684326171875, 53.98912811279297, -71.70960998535156, -1.48297119140625, -213.32708740234375, 10.727447509765625, -91.26420593261719, -121.5782241821289, 68.72608184814453, 196.89553833007812, 112.21586608886719, 153.7264404296875, -24.42375373840332, 201.8924560546875, 114.30673217773438, 106.37723541259766, 246.54855346679688, 56.104854583740234, 351.849609375, 136.21096801757812, 153.24862670898438, 491.0791015625, -169.5599365234375, 287.0085144042969, 195.67019653320312, 90.109619140625, -319.0491027832031, 400.54864501953125, -138.6338653564453, 845.0745849609375, 68.96329498291016, 199.76174926757812, 440.084228515625, 174.94883728027344, 16.75946044921875, 41.41243362426758, 20.8080997467041, 86.64511108398438, 18.814483642578125, -109.45231628417969, 65.43899536132812, 314.79144287109375, 124.53009033203125, 0.9427013397216797, -19.129535675048828, 5.664764404296875, -5.025848388671875, 25.634353637695312, -39.56380081176758, 9.083768844604492, 225.06808471679688, 138.86007690429688, 0.03756904602050781, 133.02569580078125, 2.542980194091797, -15.81597900390625, 628.6795654296875, -190.53372192382812, 129.0623321533203, 582.9708251953125, 46.53694152832031, 212.3672332763672, 329.4228515625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000621.npy"}
|
||||
{"epoch": 0.9387755102040817, "step": 622, "batch_size": 64, "mean": 93.49225616455078, "std": 170.3873291015625, "min": -281.4933166503906, "p10": -91.16560363769531, "median": 68.69210815429688, "p90": 269.74256286621096, "max": 808.3889770507812, "pos_frac": 0.71875, "sample": [179.61309814453125, 66.02279663085938, 87.45030975341797, -281.4933166503906, 31.570701599121094, 58.109413146972656, -93.00704956054688, 28.77642822265625, -86.868896484375, 40.15260314941406, 270.2677917480469, 137.53517150878906, -1.655160903930664, -53.86471939086914, 237.4146270751953, 94.0768814086914, 244.89271545410156, 214.27191162109375, 298.9273681640625, 247.37420654296875, 202.63645935058594, 213.51409912109375, -10.374456405639648, -33.508880615234375, 358.0486755371094, 21.555946350097656, 98.26829528808594, 265.7345886230469, 30.829944610595703, 75.70405578613281, 390.53057861328125, -220.50088500976562, 27.53236961364746, 43.37912368774414, 114.14948272705078, -17.963064193725586, 331.3404235839844, 183.1341552734375, 16.728851318359375, 77.14554595947266, 177.96676635742188, 120.46173858642578, 183.87680053710938, -28.224355697631836, 155.41079711914062, 230.02110290527344, 808.3889770507812, -221.38034057617188, 5.874561309814453, -52.05951690673828, 268.51702880859375, 40.35509490966797, -0.29224395751953125, -141.6986083984375, -19.9287052154541, -121.79373168945312, 316.6428527832031, -68.04639434814453, -112.67860412597656, 46.927207946777344, 175.0584259033203, 24.743650436401367, 236.54812622070312, 71.36141967773438], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000622.npy"}
|
||||
{"epoch": 0.9402872260015117, "step": 623, "batch_size": 64, "mean": 101.80033111572266, "std": 191.4267120361328, "min": -482.1900634765625, "p10": -119.38596115112304, "median": 91.62480545043945, "p90": 318.90174255371096, "max": 777.7373046875, "pos_frac": 0.734375, "sample": [170.82528686523438, 88.6832504272461, -41.96693801879883, -199.20558166503906, 104.85981750488281, 346.3577880859375, 168.25387573242188, 225.63250732421875, 289.6581115722656, 92.56267547607422, -123.6215591430664, 115.01075744628906, 304.8838195800781, 450.6433410644531, -48.95683288574219, 320.99298095703125, 12.650066375732422, 213.39309692382812, 26.124916076660156, 108.14444732666016, 53.137054443359375, 15.187410354614258, 285.5498046875, -11.7567138671875, -55.24059295654297, 189.93040466308594, 85.10893249511719, 257.21978759765625, -482.1900634765625, -109.50289916992188, 173.43914794921875, 225.7042999267578, 103.22723388671875, 420.9236755371094, -71.021728515625, -151.41085815429688, 194.307373046875, -11.389495849609375, 179.181640625, 236.50469970703125, 314.0221862792969, 242.84771728515625, 63.854976654052734, 6.07496452331543, -49.21129608154297, -24.802106857299805, -150.42947387695312, 17.723737716674805, 10.973278045654297, 207.28176879882812, 329.2626953125, 777.7373046875, 116.77589416503906, 419.5555419921875, 17.99651336669922, 12.997306823730469, -158.98268127441406, 17.991973876953125, 90.68693542480469, 2.5840511322021484, 195.55404663085938, -12.942726135253906, 137.89913940429688, -222.0655517578125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000623.npy"}
|
||||
{"epoch": 0.9417989417989417, "step": 624, "batch_size": 64, "mean": 100.86743927001953, "std": 159.08958435058594, "min": -228.21261596679688, "p10": -64.97950439453125, "median": 82.91977310180664, "p90": 304.1729644775393, "max": 522.9522705078125, "pos_frac": 0.703125, "sample": [-60.938995361328125, -34.02430725097656, 207.00497436523438, 112.03823852539062, 213.0680694580078, 208.29429626464844, -9.802711486816406, 100.45426940917969, -66.71115112304688, -31.845399856567383, 114.1122055053711, 19.792724609375, 121.08326721191406, 180.4303741455078, 188.47647094726562, 237.95773315429688, 12.86717414855957, 465.4990539550781, 373.87213134765625, 7.253028869628906, -75.01786804199219, -126.26094055175781, -36.712486267089844, 240.16403198242188, 19.676593780517578, 241.35476684570312, 90.83696746826172, -97.3865737915039, 29.721832275390625, 13.375930786132812, 435.7103576660156, 330.8204040527344, 149.69369506835938, 522.9522705078125, 17.996875762939453, -16.99405288696289, 234.70822143554688, 168.7560272216797, 481.6026611328125, 214.76190185546875, 75.00257873535156, 241.99560546875, 157.64718627929688, -35.586578369140625, -105.70684814453125, -56.40327453613281, 337.7420654296875, 12.238082885742188, -13.750757217407227, 179.75665283203125, 149.700927734375, 1.691171646118164, 25.681137084960938, 27.00884437561035, -228.21261596679688, 148.47360229492188, 94.20552062988281, -36.25635528564453, 210.04071044921875, -3.987041473388672, -184.65101623535156, -1.6056442260742188, 62.65931701660156, 199.19088745117188], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000624.npy"}
|
||||
{"epoch": 0.9433106575963719, "step": 625, "batch_size": 64, "mean": 80.48973083496094, "std": 168.67098999023438, "min": -236.43643188476562, "p10": -119.84320068359374, "median": 41.98299026489258, "p90": 298.8900848388672, "max": 532.5415649414062, "pos_frac": 0.671875, "sample": [97.34648132324219, 289.4774475097656, 40.70173645019531, -231.41839599609375, 0.4380626678466797, 186.32589721679688, 103.19966888427734, 140.77850341796875, 6.418081283569336, 418.9130859375, 160.91842651367188, -126.25839233398438, 233.66053771972656, -88.24452209472656, 392.7663269042969, 53.97044372558594, 355.5579833984375, 67.42344665527344, -53.59418869018555, -1.7965774536132812, 194.73133850097656, 184.14517211914062, 79.51036834716797, -104.87442016601562, -3.3116912841796875, 35.672607421875, 58.70777893066406, 43.13788604736328, 37.8320426940918, 63.3206787109375, -0.8007392883300781, 368.53082275390625, 532.5415649414062, -42.79869842529297, 231.507080078125, -8.343208312988281, -3.975893020629883, 476.59332275390625, -11.305686950683594, 5.302276611328125, 120.36822509765625, 260.8664855957031, 302.924072265625, 149.46702575683594, 11.455070495605469, 10.323776245117188, -202.28713989257812, -147.56393432617188, -189.9723358154297, 228.27259826660156, -60.40220642089844, -236.43643188476562, -216.7849884033203, 262.9965515136719, 155.35858154296875, -1.5053253173828125, 277.79437255859375, 40.828094482421875, 104.16059112548828, -34.87767791748047, 84.50009155273438, -6.058750152587891, 40.733558654785156, 14.475776672363281], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000625.npy"}
|
||||
{"epoch": 0.9448223733938019, "step": 626, "batch_size": 64, "mean": 113.05304718017578, "std": 189.9899139404297, "min": -208.90972900390625, "p10": -85.46521987915038, "median": 89.8435173034668, "p90": 266.71929321289065, "max": 874.2911376953125, "pos_frac": 0.734375, "sample": [251.43930053710938, 33.57463073730469, 223.12527465820312, -1.1431121826171875, 796.9636840820312, 223.37750244140625, 258.52001953125, -31.885095596313477, 101.7973861694336, 212.85597229003906, -84.00799560546875, 41.23833465576172, 54.27600860595703, 6.546764373779297, -110.97290802001953, 200.983642578125, 202.28945922851562, -151.8569793701172, 355.3055114746094, 76.89525604248047, -208.90972900390625, 13.9725341796875, 143.9053497314453, 178.7075653076172, 14.214950561523438, -10.503110885620117, 167.4976043701172, 267.1934814453125, -65.02604675292969, 874.2911376953125, 148.78176879882812, -86.0897445678711, 293.212890625, 237.60984802246094, 93.51221466064453, -136.57513427734375, -9.34234619140625, 60.391632080078125, -6.321958541870117, 142.95590209960938, 20.629981994628906, 5.831491470336914, 265.61285400390625, 39.49588394165039, -18.473175048828125, 5.616006851196289, 138.23606872558594, 198.82577514648438, 59.00581741333008, 232.98715209960938, 184.64601135253906, 86.17481994628906, -181.27444458007812, 223.50555419921875, -173.09149169921875, 214.41287231445312, 47.31075668334961, 311.8353271484375, 199.88565063476562, 185.11080932617188, 364.5230712890625, -82.27560424804688, 216.10757446289062, -82.04534149169922], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000626.npy"}
|
||||
{"epoch": 0.9463340891912321, "step": 627, "batch_size": 64, "mean": 123.1042251586914, "std": 156.0784912109375, "min": -246.9134521484375, "p10": -46.795706176757804, "median": 126.19868850708008, "p90": 312.3214965820313, "max": 574.1636962890625, "pos_frac": 0.8125, "sample": [154.77308654785156, 0.10585403442382812, 48.79652404785156, 275.3190612792969, 163.89813232421875, 154.08370971679688, 160.88690185546875, 383.31964111328125, 484.62908935546875, 328.67974853515625, 26.65106201171875, 299.8930358886719, 192.10708618164062, 228.81007385253906, 73.38914489746094, 213.82763671875, -38.6175537109375, 36.229270935058594, 267.7743835449219, 207.16348266601562, 150.70701599121094, 187.58702087402344, 73.63570404052734, -14.4229736328125, -11.760543823242188, -82.42056274414062, 224.432861328125, -99.31672668457031, 169.43472290039062, 9.495952606201172, 12.829912185668945, 253.92942810058594, 317.6479797363281, 25.10466766357422, 214.6856689453125, -50.300628662109375, 2.9035568237304688, 158.4663543701172, -59.764373779296875, 275.84991455078125, 330.5467529296875, 36.054237365722656, -194.97259521484375, -0.031951904296875, 114.7947769165039, 18.540115356445312, 20.586212158203125, 174.97030639648438, 70.89520263671875, 399.99285888671875, 66.27852630615234, -209.69772338867188, -14.171344757080078, 114.23554992675781, 236.14781188964844, 574.1636962890625, 83.08568572998047, -246.9134521484375, 38.908203125, 88.0673599243164, 137.60260009765625, 196.31619262695312, 167.85736083984375, 254.96954345703125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000627.npy"}
|
||||
{"epoch": 0.9478458049886621, "step": 628, "batch_size": 64, "mean": 104.03134155273438, "std": 184.8887481689453, "min": -505.2254333496094, "p10": -78.84615173339841, "median": 118.7235221862793, "p90": 269.0324981689453, "max": 496.61297607421875, "pos_frac": 0.796875, "sample": [259.4770202636719, 61.44989013671875, 150.0163116455078, 191.67323303222656, -44.216766357421875, -505.2254333496094, 320.129638671875, 50.4844970703125, 198.86965942382812, 242.95892333984375, 261.9165954589844, 397.5174255371094, 41.62879943847656, 60.924686431884766, 30.60257911682129, 238.33059692382812, -442.35711669921875, -216.10691833496094, 392.59246826171875, 151.14529418945312, 264.1616516113281, 139.6971893310547, 41.70074462890625, 7.72265625, 226.0023193359375, 230.78518676757812, -62.484527587890625, 196.91024780273438, 125.35575866699219, 265.55755615234375, 209.90794372558594, 394.70172119140625, 59.3651123046875, 239.93841552734375, -7.166568756103516, 496.61297607421875, -37.15386962890625, 17.044830322265625, 270.5217590332031, -267.80535888671875, 192.95050048828125, 112.0912857055664, 18.99073600769043, 186.1968231201172, 106.94231414794922, -43.04924011230469, 211.84512329101562, 73.52362060546875, 190.46383666992188, 233.6885223388672, 83.49108123779297, -117.23701477050781, 26.78753662109375, -219.9784393310547, 161.87693786621094, 444.0582275390625, 30.6976318359375, 170.1183319091797, 96.89854431152344, 27.121841430664062, -44.33422088623047, 140.153076171875, 7.380001068115234, -85.8582763671875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000628.npy"}
|
||||
{"epoch": 0.9493575207860923, "step": 629, "batch_size": 64, "mean": 76.27835083007812, "std": 151.2153778076172, "min": -191.99130249023438, "p10": -63.01355628967285, "median": 44.097511291503906, "p90": 281.17870178222665, "max": 543.14501953125, "pos_frac": 0.671875, "sample": [-133.58499145507812, 145.97079467773438, -191.99130249023438, 114.59844970703125, 41.94318389892578, -17.08771514892578, -52.43904113769531, 71.73275756835938, -182.30535888671875, 18.990737915039062, -63.60258483886719, 77.22769927978516, 13.716241836547852, 262.33050537109375, -53.483238220214844, -34.92531204223633, 4.951650619506836, -31.02606201171875, -156.327392578125, 75.20672607421875, 97.53689575195312, -0.6283187866210938, -17.353158950805664, 27.771808624267578, -61.639156341552734, 22.10955047607422, 471.09771728515625, 543.14501953125, -12.833517074584961, 186.00753784179688, 118.29061889648438, -41.37480926513672, 139.47825622558594, 93.39563751220703, -43.21622085571289, 163.58566284179688, 90.99746704101562, 27.868896484375, 8.26224136352539, -4.320329666137695, 48.155792236328125, 57.65717697143555, 338.01593017578125, 0.015094757080078125, -18.62310028076172, -120.73762512207031, 114.20404815673828, 134.8060760498047, 172.45358276367188, 206.0249481201172, -44.46928405761719, 84.92868041992188, 318.5547180175781, 16.094926834106445, 394.3193359375, 468.359130859375, 46.25183868408203, 289.2565002441406, 156.4312744140625, 219.08071899414062, 25.2716064453125, 162.64674377441406, -68.29734802246094, 163.33648681640625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000629.npy"}
|
||||
{"epoch": 0.9508692365835223, "step": 630, "batch_size": 64, "mean": 110.32024383544922, "std": 161.2056427001953, "min": -422.87127685546875, "p10": -43.021773147583005, "median": 94.43450546264648, "p90": 321.2943420410157, "max": 533.82861328125, "pos_frac": 0.78125, "sample": [-34.3624153137207, 239.29568481445312, -37.82955551147461, 39.39252471923828, 199.9541473388672, 379.4171142578125, 166.52249145507812, 7.905117034912109, 325.0871887207031, 210.47451782226562, -4.5433349609375, 224.53753662109375, -19.145301818847656, 221.8199462890625, 28.20348358154297, 232.041015625, -176.63926696777344, 15.975700378417969, 199.07308959960938, -422.87127685546875, 159.82699584960938, 37.87220001220703, 139.28073120117188, 218.4013671875, 384.2266540527344, 107.02068328857422, -28.781150817871094, 71.38432312011719, -232.18246459960938, 169.134033203125, -10.910743713378906, 99.39930725097656, 9.860736846923828, 338.62091064453125, 363.3913269042969, 50.40147399902344, 312.4443664550781, 64.01901245117188, 120.67375183105469, 151.68304443359375, 533.82861328125, 35.17570495605469, 237.84352111816406, -69.85246276855469, 226.17074584960938, 89.4697036743164, 309.812255859375, 80.24554443359375, 76.83148193359375, -172.74057006835938, 113.59513092041016, 37.57106018066406, 0.7968330383300781, 358.248046875, 241.89663696289062, 47.84397888183594, 159.89456176757812, 44.44015884399414, 212.3400421142578, 207.47811889648438, -49.15849304199219, -45.24700927734375, 74.01759338378906, -10.080455780029297], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000630.npy"}
|
||||
{"epoch": 0.9523809523809523, "step": 631, "batch_size": 64, "mean": 131.174072265625, "std": 182.98399353027344, "min": -323.634521484375, "p10": -67.11217956542968, "median": 125.26567077636719, "p90": 397.4954376220703, "max": 630.5668334960938, "pos_frac": 0.734375, "sample": [205.8296661376953, 98.58145141601562, -108.53497314453125, 281.9455871582031, 187.0148468017578, 179.78204345703125, 311.2935791015625, 158.71347045898438, 122.12936401367188, -69.6055908203125, 344.8695373535156, -11.865612030029297, 203.94834899902344, 101.99574279785156, 13.80643081665039, 39.880882263183594, 7.117950439453125, 66.93231201171875, -0.7457351684570312, 416.5882568359375, 420.5567626953125, -61.294219970703125, 398.6731262207031, 279.95867919921875, 269.6863708496094, 47.402008056640625, 42.69822692871094, 471.1749572753906, -30.356496810913086, 187.27166748046875, 21.008190155029297, 48.57093048095703, -211.28109741210938, 136.41531372070312, 159.86585998535156, 434.02789306640625, 206.89585876464844, -119.89212036132812, 630.5668334960938, 291.14141845703125, 133.80496215820312, -20.69049072265625, -323.634521484375, 216.33741760253906, -33.63055419921875, -150.54730224609375, -27.46040916442871, 1.0437870025634766, -23.576370239257812, 468.16455078125, -36.86750030517578, 128.4019775390625, -77.57159423828125, 272.5013732910156, -0.3688316345214844, 210.8291015625, 23.410282135009766, 308.54425048828125, 83.81024169921875, 394.74749755859375, 170.17523193359375, 210.532958984375, 12.168224334716797, 282.24920654296875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000631.npy"}
|
||||
{"epoch": 0.9538926681783825, "step": 632, "batch_size": 64, "mean": 111.63894653320312, "std": 151.67510986328125, "min": -290.4002990722656, "p10": -60.5258861541748, "median": 126.19716262817383, "p90": 285.9210998535156, "max": 481.4926452636719, "pos_frac": 0.78125, "sample": [161.89901733398438, 407.4613037109375, 107.370849609375, -4.137180328369141, 68.68881225585938, 217.73651123046875, 15.102245330810547, 152.0382080078125, -55.96357345581055, 234.45822143554688, 133.90057373046875, -270.2290954589844, 169.57290649414062, -13.261112213134766, -124.12638854980469, 268.5791015625, 283.89642333984375, 6.700675964355469, 193.66188049316406, -113.16106414794922, 92.13009643554688, 13.58578872680664, -194.96209716796875, 53.16889953613281, 59.291709899902344, 217.73428344726562, 170.79885864257812, 134.61172485351562, 62.79317092895508, -20.413494110107422, 238.13839721679688, 481.4926452636719, 128.33497619628906, -55.91735076904297, 353.6875305175781, 23.560731887817383, 218.59152221679688, 53.83477783203125, 124.0593490600586, 193.32073974609375, 267.8273620605469, 221.5985870361328, 31.811065673828125, -40.677940368652344, 14.604785919189453, 233.51309204101562, 286.788818359375, 161.71588134765625, 310.564208984375, 217.723876953125, 218.62265014648438, -290.4002990722656, 83.08135986328125, 197.81570434570312, -6.310462951660156, -62.481163024902344, 247.808837890625, 326.0227966308594, 297.8760986328125, 82.60867309570312, 131.14956665039062, 8.181537628173828, -67.63505554199219, 85.052490234375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000632.npy"}
|
||||
{"epoch": 0.9554043839758125, "step": 633, "batch_size": 64, "mean": 93.976318359375, "std": 153.21249389648438, "min": -175.98045349121094, "p10": -82.77049484252929, "median": 66.0151481628418, "p90": 304.0804809570313, "max": 465.04229736328125, "pos_frac": 0.71875, "sample": [-5.219442367553711, 20.960838317871094, 3.9560928344726562, 127.37300109863281, 306.29656982421875, -53.777923583984375, 199.68411254882812, 202.2496337890625, 35.58290100097656, 171.72418212890625, -16.326919555664062, 31.30046272277832, 447.87249755859375, 11.96213150024414, -25.397693634033203, 123.30403900146484, 83.2711181640625, 242.91795349121094, 74.97002410888672, 2.8577499389648438, 14.71697998046875, -90.65093231201172, 65.1466293334961, -3.9107894897460938, 215.90643310546875, 88.4497299194336, -18.53500747680664, -51.904563903808594, 78.26604461669922, 144.28054809570312, 298.90960693359375, 17.34632110595703, 227.0964813232422, 176.6045379638672, -83.71366882324219, 15.963871002197266, -175.98045349121094, -72.0192642211914, -84.43901062011719, -80.56975555419922, -167.48739624023438, 94.86299133300781, 2.409076690673828, 465.04229736328125, 19.48938751220703, 103.079345703125, 45.01976776123047, -93.86217498779297, 100.35661315917969, 212.2285919189453, 286.8843994140625, 205.86248779296875, 248.14492797851562, 240.08758544921875, 3.7277679443359375, -65.08344268798828, -8.576457977294922, 66.8836669921875, 320.418701171875, 277.421630859375, 434.43505859375, -151.62705993652344, 381.65740966796875, 326.5837707519531], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000633.npy"}
|
||||
{"epoch": 0.9569160997732427, "step": 634, "batch_size": 64, "mean": 31.083759307861328, "std": 166.67440795898438, "min": -477.64129638671875, "p10": -151.8791717529297, "median": 2.511465072631836, "p90": 222.70511016845708, "max": 660.10009765625, "pos_frac": 0.53125, "sample": [-25.514423370361328, -150.83474731445312, -12.330184936523438, 210.3892822265625, -90.67559814453125, 0.9705753326416016, -39.704402923583984, -14.489896774291992, 115.79683685302734, 62.75941467285156, 3.6689395904541016, -73.6784439086914, 3.2738876342773438, -103.53649139404297, -2.9139251708984375, 97.41969299316406, 174.30906677246094, 27.03823471069336, -105.48784637451172, 32.41291809082031, -42.92443084716797, -3.825977325439453, 230.3374786376953, -170.99258422851562, 227.9833221435547, 10.072257995605469, 271.72869873046875, 191.35472106933594, 660.10009765625, -185.5409698486328, -16.532684326171875, 19.52359962463379, -13.765911102294922, -24.393238067626953, 199.8139190673828, -16.453561782836914, -152.3267822265625, -157.83006286621094, 352.40301513671875, 201.05799865722656, 244.12840270996094, -134.49441528320312, -244.54571533203125, 34.46674728393555, -477.64129638671875, -264.5286560058594, -14.598560333251953, 1.7490425109863281, 202.29891967773438, 32.868289947509766, -23.19219970703125, -80.07008361816406, 179.84786987304688, 123.6949234008789, -5.158170700073242, -17.50656509399414, 208.39935302734375, 144.16741943359375, 51.11400604248047, 119.3025131225586, 116.1169204711914, 12.015295028686523, -143.01214599609375, 235.27688598632812], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000634.npy"}
|
||||
{"epoch": 0.9584278155706727, "step": 635, "batch_size": 64, "mean": 80.10987854003906, "std": 154.60284423828125, "min": -245.89212036132812, "p10": -118.43008956909179, "median": 79.79962539672852, "p90": 263.82424011230466, "max": 416.3206787109375, "pos_frac": 0.65625, "sample": [-152.27706909179688, 196.85009765625, -85.61270904541016, 27.007343292236328, 131.4405517578125, 68.57952117919922, -117.04623413085938, -198.08758544921875, -93.27511596679688, 207.14588928222656, -0.50775146484375, 184.009033203125, 48.184181213378906, 11.546989440917969, 327.6262512207031, 263.8026428222656, 172.38717651367188, 19.717056274414062, 256.9570007324219, 207.24026489257812, -151.41883850097656, 14.5968017578125, -119.0231704711914, 345.6596984863281, 144.97946166992188, -47.65968322753906, 245.15548706054688, -2.702932357788086, -12.34048843383789, 116.12783813476562, 396.7052001953125, -91.19182586669922, 250.3856201171875, 21.884559631347656, 140.78231811523438, -3.4906272888183594, 224.63461303710938, -26.43111801147461, 416.3206787109375, -243.41676330566406, 151.9014434814453, 313.6491394042969, -86.46047973632812, 263.83349609375, -10.00213623046875, -31.134933471679688, 149.31178283691406, 102.49117279052734, 91.01972961425781, 61.80424499511719, 96.87963104248047, 344.19219970703125, -30.3995361328125, 199.7120361328125, -245.89212036132812, 147.87969970703125, 103.86500549316406, 47.72998046875, -15.012176513671875, 165.8273162841797, -123.74783325195312, 143.83135986328125, 181.6055908203125, 8.903450012207031], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000635.npy"}
|
||||
{"epoch": 0.9599395313681028, "step": 636, "batch_size": 64, "mean": 105.12950897216797, "std": 142.60784912109375, "min": -246.77682495117188, "p10": -45.70656509399414, "median": 108.14578628540039, "p90": 276.48039855957035, "max": 485.021240234375, "pos_frac": 0.8125, "sample": [218.45594787597656, 28.03357696533203, -45.88710021972656, 9.008552551269531, 75.96222686767578, 57.556922912597656, 110.90147399902344, 9.120506286621094, -45.285316467285156, 222.11734008789062, 183.35601806640625, 1.5567455291748047, 198.11355590820312, 190.61878967285156, 76.61212158203125, 74.13664245605469, -96.62864685058594, 325.40631103515625, 117.70796966552734, 191.71597290039062, 266.1243591308594, 4.1700592041015625, 3.7210140228271484, 195.5175323486328, -0.8863906860351562, 99.7402114868164, 395.9178771972656, 7.77052116394043, -0.1396770477294922, 193.13864135742188, 108.31863403320312, 154.9749755859375, 343.6624755859375, 185.8895263671875, -44.527191162109375, 187.1156005859375, 134.51321411132812, 42.24681854248047, -99.30420684814453, 110.97891235351562, 107.97293853759766, 252.81561279296875, -4.502593994140625, 485.021240234375, 23.132183074951172, -216.07720947265625, 280.918701171875, 146.2188720703125, 6.368442535400391, 207.59576416015625, 337.98114013671875, 108.33557891845703, 240.93043518066406, 297.6405944824219, 255.0470428466797, -130.45193481445312, -246.77682495117188, 19.66636085510254, 114.74911499023438, 185.1798553466797, 78.8729248046875, -181.73968505859375, 76.96328735351562, 90.90428161621094], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000636.npy"}
|
||||
{"epoch": 0.9614512471655329, "step": 637, "batch_size": 64, "mean": 81.32229614257812, "std": 176.42218017578125, "min": -445.4052429199219, "p10": -104.31293792724608, "median": 49.30253601074219, "p90": 308.0421966552735, "max": 691.1748046875, "pos_frac": 0.703125, "sample": [-110.41319274902344, 351.8639831542969, 13.107978820800781, 691.1748046875, -45.56682586669922, -60.295257568359375, 8.883659362792969, 121.4719467163086, 313.3454895019531, 44.21731948852539, -14.901725769042969, 95.80794525146484, 183.25946044921875, -137.31610107421875, 206.965087890625, 54.06691360473633, 4.623359680175781, 191.78948974609375, 295.6678466796875, -32.10038757324219, 205.54676818847656, 37.98884582519531, 121.60903930664062, 51.55443572998047, 19.385793685913086, -9.5115966796875, -46.938682556152344, 133.27989196777344, 398.52020263671875, 244.68222045898438, 218.5075225830078, 13.220466613769531, -250.5067901611328, 139.94155883789062, -90.07901000976562, 109.43356323242188, -172.632080078125, -22.964500427246094, 31.877525329589844, 47.050636291503906, 212.33338928222656, 381.7402648925781, 3.7275943756103516, 13.162540435791016, 324.19091796875, 20.70470428466797, 20.147857666015625, 239.60037231445312, -167.16275024414062, -27.309547424316406, 75.99554443359375, 317.92156982421875, 143.41055297851562, 198.61328125, 161.4902801513672, -184.66195678710938, 281.4542236328125, 52.114662170410156, -44.10930633544922, -445.4052429199219, 114.24456024169922, 233.16818237304688, -60.167640686035156, -16.19451904296875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000637.npy"}
|
||||
{"epoch": 0.9629629629629629, "step": 638, "batch_size": 64, "mean": 96.26412963867188, "std": 166.7211151123047, "min": -355.5335998535156, "p10": -64.67456626892088, "median": 49.262210845947266, "p90": 280.16897888183604, "max": 712.074462890625, "pos_frac": 0.78125, "sample": [-355.5335998535156, 216.25331115722656, 247.99107360839844, -16.048702239990234, -120.42068481445312, 13.316452026367188, 257.0493469238281, 255.07528686523438, 51.39278793334961, -52.9494514465332, 20.46739959716797, -135.9856414794922, 6.2227935791015625, -232.38404846191406, -132.98095703125, 417.05084228515625, 16.503326416015625, -10.391464233398438, 294.2569580078125, 31.313949584960938, 288.2979736328125, 2.7058181762695312, 146.24258422851562, 342.034912109375, -109.3649673461914, 334.7196044921875, 139.23529052734375, 204.04307556152344, 2.6276931762695312, 70.7705078125, -11.533136367797852, 250.57806396484375, 299.20123291015625, 210.433349609375, 66.41192626953125, 40.17012023925781, 58.996177673339844, 194.4170684814453, 101.64427185058594, 254.8628692626953, 93.33733367919922, 253.92929077148438, 261.2013244628906, 43.664424896240234, 226.71591186523438, -69.69961547851562, -5.761402130126953, 198.41497802734375, 40.65052795410156, 26.056800842285156, 0.9391651153564453, 47.13163375854492, 712.074462890625, 2.3504638671875, -45.611690521240234, 19.61023712158203, 6.006919860839844, 232.87908935546875, 213.66111755371094, 23.368465423583984, -48.516380310058594, 5.3098297119140625, 207.56805419921875, 58.92979431152344], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000638.npy"}
|
||||
{"epoch": 0.9644746787603931, "step": 639, "batch_size": 64, "mean": 111.11836242675781, "std": 175.32342529296875, "min": -266.3203125, "p10": -84.87477188110351, "median": 82.70928955078125, "p90": 349.1716674804689, "max": 588.5953369140625, "pos_frac": 0.75, "sample": [232.96713256835938, 172.96701049804688, 199.71087646484375, -88.59646606445312, 45.190673828125, 245.42434692382812, 20.721054077148438, -53.726444244384766, -95.64154815673828, 179.3611602783203, 22.944290161132812, 103.95077514648438, 393.2664794921875, 77.81185913085938, 211.50550842285156, 53.920867919921875, 84.63389587402344, 12.43353271484375, -253.67214965820312, 86.20845794677734, 227.42965698242188, 286.3915710449219, -6.855813980102539, 383.39111328125, 26.94615936279297, 146.42723083496094, 93.77444458007812, 240.39427185058594, -149.61802673339844, 314.6759033203125, 363.95556640625, 7.410825729370117, 0.6361217498779297, 179.29364013671875, -6.611354827880859, 21.537002563476562, 157.72222900390625, -30.757797241210938, 484.8774108886719, 218.01272583007812, 203.7967529296875, 148.25335693359375, 588.5953369140625, 144.3560791015625, -266.3203125, -57.53605270385742, -76.1908187866211, -40.11080551147461, 80.78468322753906, 32.11330795288086, 448.6615295410156, 208.65493774414062, 33.20915985107422, 71.29019165039062, -89.37078857421875, -161.125, 62.462440490722656, -0.0426483154296875, 13.468223571777344, 563.7115478515625, -17.18584442138672, 151.6854248046875, 217.18177795410156, 240.81826782226562], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000639.npy"}
|
||||
{"epoch": 0.9659863945578231, "step": 640, "batch_size": 64, "mean": 94.80416870117188, "std": 155.53555297851562, "min": -254.44786071777344, "p10": -62.92387619018554, "median": 66.39886856079102, "p90": 256.7825225830078, "max": 525.0386962890625, "pos_frac": 0.75, "sample": [211.9553680419922, 43.38747024536133, 525.0386962890625, -54.065574645996094, 41.444610595703125, 103.82192993164062, 180.89183044433594, 10.492630004882812, 406.3735046386719, 211.2151641845703, 108.94970703125, 110.3717269897461, 166.8714141845703, 193.56317138671875, -254.44786071777344, 182.01522827148438, 41.07140350341797, 42.3770637512207, -162.44723510742188, 255.89447021484375, -217.1291961669922, 257.1631164550781, 341.650634765625, 121.24992370605469, -38.848480224609375, -24.68191909790039, 177.84503173828125, 68.7699966430664, -25.414552688598633, 23.72518539428711, 189.41334533691406, 33.17851257324219, 117.86306762695312, 238.83682250976562, 31.415313720703125, -3.67669677734375, -47.433311462402344, -8.985549926757812, -165.53245544433594, 165.28472900390625, 64.02774047851562, 218.70565795898438, 13.395195007324219, -70.42697143554688, 242.88465881347656, 197.58029174804688, 132.4661865234375, -234.62396240234375, 9.556314468383789, 205.4692840576172, 181.84922790527344, 439.809326171875, 58.944000244140625, 415.361083984375, 34.202484130859375, -66.72029113769531, 195.5577392578125, -50.24198913574219, 262.654052734375, 21.643831253051758, -5.653759002685547, 37.429901123046875, 23.79958152770996, 140.32919311523438], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000640.npy"}
|
||||
{"epoch": 0.9674981103552532, "step": 641, "batch_size": 64, "mean": 81.43589782714844, "std": 131.7555694580078, "min": -254.4486541748047, "p10": -49.5817626953125, "median": 48.31868934631348, "p90": 224.68513793945314, "max": 633.864501953125, "pos_frac": 0.765625, "sample": [-19.48782730102539, -206.43392944335938, 28.407100677490234, 92.39729309082031, -15.486724853515625, -50.079612731933594, 138.8851318359375, -22.41088104248047, 24.37421417236328, -65.05229187011719, 28.766586303710938, 1.6229877471923828, 173.051513671875, 177.19468688964844, 228.73776245117188, 12.99476432800293, 35.293304443359375, 348.546875, 216.64126586914062, 223.14549255371094, 179.82745361328125, 31.71259307861328, 11.966377258300781, 17.800216674804688, 79.54456329345703, 225.47618103027344, 7.4130096435546875, 159.68504333496094, -0.0468902587890625, 161.67950439453125, -11.22930908203125, 149.3104248046875, -74.5313491821289, 103.6801528930664, 26.94409942626953, -86.41011810302734, 175.472900390625, 95.76246643066406, -48.42011260986328, 137.09088134765625, -254.4486541748047, 132.17349243164062, 113.6514892578125, 2.1816864013671875, 160.5770721435547, 288.4647521972656, 36.76104736328125, -6.949193954467773, 196.91351318359375, 54.32508850097656, 633.864501953125, 79.66606140136719, 286.69775390625, 225.34498596191406, 14.580780029296875, 141.6239471435547, 126.129638671875, 47.75288391113281, 48.88449478149414, -13.646636962890625, -63.01660919189453, 44.35870361328125, 37.19707489013672, 184.97357177734375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000641.npy"}
|
||||
{"epoch": 0.9690098261526833, "step": 642, "batch_size": 64, "mean": 59.711238861083984, "std": 162.07640075683594, "min": -239.47515869140625, "p10": -134.13451232910154, "median": 64.83204650878906, "p90": 228.13129272460938, "max": 781.34814453125, "pos_frac": 0.671875, "sample": [9.756004333496094, -79.85807037353516, 15.823371887207031, -183.76747131347656, 53.742515563964844, -2.6989498138427734, 179.21832275390625, 230.18299865722656, -138.55789184570312, -145.6336212158203, 781.34814453125, -5.366325378417969, 101.2789306640625, -75.83343505859375, 108.849365234375, 64.73519897460938, 150.2799835205078, 307.4199523925781, -111.5689468383789, -25.906497955322266, 205.14413452148438, 251.5921630859375, 258.2828369140625, -27.51186752319336, 132.40103149414062, 199.4139404296875, 216.78302001953125, 80.2911376953125, 194.08511352539062, 68.8360366821289, 185.21783447265625, 23.90008544921875, 64.92889404296875, 30.983829498291016, -13.608871459960938, 122.04107666015625, -84.70051574707031, 123.14258575439453, 7.249870300292969, -106.83430480957031, 56.594356536865234, -62.7125244140625, 11.050424575805664, -40.95165252685547, 186.93414306640625, 8.311395645141602, -123.81329345703125, 199.09629821777344, 120.719482421875, 282.4178161621094, 92.55746459960938, 68.70178985595703, 230.61056518554688, 88.17923736572266, 223.34397888183594, 78.93696594238281, -93.61620330810547, 124.86312103271484, -239.47515869140625, -222.55157470703125, 31.559341430664062, -201.73448181152344, -227.95091247558594, 65.36726379394531], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000642.npy"}
|
||||
{"epoch": 0.9705215419501134, "step": 643, "batch_size": 64, "mean": 107.99020385742188, "std": 175.41380310058594, "min": -254.8553466796875, "p10": -115.52609863281246, "median": 120.07634353637695, "p90": 305.56589355468753, "max": 706.63525390625, "pos_frac": 0.71875, "sample": [71.11831665039062, 63.44672775268555, 145.4864501953125, 256.80352783203125, 2.5866498947143555, 311.2173156738281, 10.876993179321289, 203.0631866455078, 107.9421615600586, 68.4533920288086, 536.1704711914062, 64.9282455444336, 188.79957580566406, -13.339675903320312, 120.92853546142578, 165.9229736328125, 3.0739898681640625, -14.853429794311523, 234.4976348876953, 264.667236328125, 203.52386474609375, -0.217559814453125, 232.81112670898438, 188.06761169433594, -53.580162048339844, 292.3792419433594, 238.22496032714844, -56.767059326171875, 233.81060791015625, 438.4279479980469, 149.3079833984375, 182.4048309326172, -60.772369384765625, 174.48141479492188, 133.16481018066406, -135.20518493652344, 23.031694412231445, -20.990509033203125, 340.64251708984375, 200.51710510253906, 184.37770080566406, -29.242698669433594, 217.00794982910156, -1.6920013427734375, 229.3741455078125, -180.1327667236328, -129.6806640625, -238.39627075195312, 11.826873779296875, 25.15224838256836, -161.78829956054688, 162.53309631347656, 317.1398010253906, 119.22415161132812, 122.8626708984375, -161.2919921875, -17.111557006835938, 322.47003173828125, 11.616630554199219, 67.0406723022461, -82.498779296875, 706.63525390625, 175.74905395507812, -254.8553466796875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000643.npy"}
|
||||
{"epoch": 0.9720332577475435, "step": 644, "batch_size": 64, "mean": 60.94654083251953, "std": 141.4824981689453, "min": -348.5711975097656, "p10": -99.88167724609373, "median": 40.59115982055664, "p90": 246.98864593505863, "max": 296.5062561035156, "pos_frac": 0.6875, "sample": [283.1700439453125, -157.60687255859375, -348.5711975097656, 173.85516357421875, -73.50862121582031, -105.60032653808594, 13.733665466308594, 34.6094970703125, 211.4776611328125, 289.5173645019531, 75.44111633300781, 7.043933868408203, 209.56683349609375, 201.46778869628906, 169.4019775390625, 88.87110900878906, 36.74616622924805, 71.75084686279297, 8.943138122558594, 208.92564392089844, -43.0079231262207, 256.9034423828125, -213.63775634765625, -74.21589660644531, -0.659912109375, 145.9034423828125, -1.7394180297851562, 161.9019775390625, -22.634689331054688, -86.53816223144531, 216.25352478027344, -59.28607177734375, -18.83795928955078, 225.7913360595703, 217.40728759765625, 0.3871307373046875, -165.08444213867188, 296.5062561035156, 256.04620361328125, -13.043052673339844, 165.0260009765625, 235.40325927734375, 251.9538116455078, -2.4329166412353516, 8.602554321289062, -244.20631408691406, -31.457778930664062, 94.59284973144531, 39.489540100097656, 161.89564514160156, 52.65553283691406, 80.84634399414062, 8.301565170288086, -155.98287963867188, 41.692779541015625, 3.61212158203125, -78.31808471679688, 169.0955810546875, 85.65985107421875, 5.019309997558594, 16.426193237304688, 193.40011596679688, 51.890663146972656, 269.7626953125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000644.npy"}
|
||||
{"epoch": 0.9735449735449735, "step": 645, "batch_size": 64, "mean": 75.61616516113281, "std": 143.52708435058594, "min": -296.30352783203125, "p10": -81.59333724975583, "median": 79.8933219909668, "p90": 213.41625213623047, "max": 425.54693603515625, "pos_frac": 0.765625, "sample": [63.95926284790039, 276.98797607421875, 171.10159301757812, -137.2186279296875, 97.78260803222656, 342.04638671875, 134.5736083984375, 1.8097515106201172, -33.907432556152344, 92.09720611572266, 12.12667465209961, -282.570556640625, 43.775367736816406, 13.136966705322266, 425.54693603515625, 119.91386413574219, 134.6553955078125, 9.233745574951172, -12.845703125, 38.4736328125, 213.5355224609375, 21.15166473388672, -3.0893630981445312, 109.32759857177734, 90.57349395751953, -55.343299865722656, 197.40029907226562, 127.15452575683594, 213.13795471191406, 148.50665283203125, 169.93609619140625, -3.960134506225586, 60.05658721923828, -48.67753601074219, 10.044204711914062, 126.64797973632812, -224.9523468017578, 48.41172790527344, -92.84335327148438, 286.71697998046875, 168.12767028808594, 180.43865966796875, 423.59295654296875, -47.201377868652344, 42.634822845458984, 155.40408325195312, 179.59307861328125, 141.0830535888672, 182.84600830078125, 125.22488403320312, 69.21315002441406, 150.88980102539062, 28.278583526611328, 219.82861328125, -106.880859375, 10.034431457519531, 45.981109619140625, -27.130138397216797, -296.30352783203125, -272.4544982910156, 52.05883026123047, 206.80526733398438, 189.25216674804688, 113.70366668701172], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000645.npy"}
|
||||
{"epoch": 0.9750566893424036, "step": 646, "batch_size": 64, "mean": 80.37705993652344, "std": 159.0143280029297, "min": -301.27984619140625, "p10": -100.13609771728515, "median": 62.56418800354004, "p90": 257.5453659057618, "max": 531.45166015625, "pos_frac": 0.65625, "sample": [-150.16909790039062, -0.5476703643798828, -39.813209533691406, 165.63658142089844, 215.318359375, 239.1805877685547, -118.35997009277344, 428.0570983886719, 131.6110382080078, 234.3032684326172, -17.459564208984375, 41.03343963623047, 220.03741455078125, 213.9959259033203, 193.10824584960938, -4.647247314453125, 110.0757827758789, 182.47726440429688, 225.88970947265625, 158.37733459472656, 278.614013671875, 362.0332336425781, -234.55516052246094, -83.47547149658203, -91.40213012695312, 171.15200805664062, 10.942180633544922, -31.764541625976562, 41.04496765136719, 76.75888061523438, 156.7239990234375, -12.613197326660156, -63.37012481689453, 265.4159851074219, 106.24737548828125, 10.829200744628906, 72.28236389160156, 35.87379455566406, 64.7456283569336, 183.02586364746094, -2.057535171508789, -87.10015869140625, 213.3583221435547, -301.27984619140625, 129.24655151367188, 531.45166015625, 198.80189514160156, 295.0626220703125, 56.93489456176758, 377.64678955078125, 12.856779098510742, -27.8719482421875, -78.1290283203125, -103.87922668457031, -213.0871124267578, 21.93468475341797, 178.37265014648438, -135.56004333496094, -13.241470336914062, 203.6802978515625, 60.382747650146484, -4.4356842041015625, 73.89149475097656, 10.538028717041016], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000646.npy"}
|
||||
{"epoch": 0.9765684051398337, "step": 647, "batch_size": 64, "mean": 126.61318969726562, "std": 170.3917236328125, "min": -330.0762939453125, "p10": -39.31314239501952, "median": 135.0353546142578, "p90": 291.98558044433605, "max": 614.8645629882812, "pos_frac": 0.8125, "sample": [-16.791637420654297, 213.32864379882812, 225.13790893554688, 29.534408569335938, 162.95187377929688, 61.69917297363281, 45.05322265625, 382.8697204589844, 246.11126708984375, 491.00439453125, 130.84059143066406, 216.235595703125, 242.65530395507812, 2.985980987548828, -28.781387329101562, 12.71044921875, 518.5962524414062, -23.424592971801758, -10.05063247680664, 614.8645629882812, 211.34889221191406, 110.203125, 163.33203125, 151.13226318359375, 265.1014404296875, 196.94732666015625, 248.73236083984375, 6.111444473266602, 248.7414093017578, 121.84521484375, 207.81369018554688, -27.11444091796875, 199.58782958984375, 17.63298797607422, 200.50982666015625, -43.826751708984375, 14.77083969116211, 232.09619140625, 61.40282440185547, -202.7271728515625, 160.76268005371094, 74.83399200439453, 27.764331817626953, 174.48733520507812, -330.0762939453125, 74.71137237548828, 100.79424285888672, 85.1602783203125, 62.6080322265625, 139.23011779785156, -59.344879150390625, 337.6195373535156, 207.9493408203125, -113.7862548828125, -280.3297119140625, 202.04254150390625, 95.25898742675781, 429.2575378417969, -172.3819580078125, 303.5073547363281, 245.39158630371094, 84.48711395263672, 178.58628845214844, 173.53802490234375], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000647.npy"}
|
||||
{"epoch": 0.9780801209372638, "step": 648, "batch_size": 64, "mean": 88.75203704833984, "std": 189.90379333496094, "min": -362.97784423828125, "p10": -147.4944625854492, "median": 99.87670516967773, "p90": 287.8381164550782, "max": 539.0840454101562, "pos_frac": 0.640625, "sample": [332.27783203125, 53.919090270996094, -12.730247497558594, -218.603515625, 261.8504638671875, -151.9750213623047, -246.82302856445312, 6.42413330078125, 275.82135009765625, 52.75287628173828, -42.6978645324707, 157.66635131835938, 255.19659423828125, -78.74441528320312, -137.03982543945312, -178.97589111328125, 28.57669448852539, -42.38756561279297, 539.0840454101562, 440.8410949707031, 228.76931762695312, 173.50787353515625, 19.025421142578125, 205.10516357421875, 160.86764526367188, -36.838287353515625, 138.12387084960938, 264.0326232910156, 15.966102600097656, 183.03738403320312, -181.74124145507812, -11.065437316894531, 495.90411376953125, 482.17596435546875, 2.3237533569335938, -127.41643524169922, -62.88710021972656, 112.20055389404297, 170.3663330078125, -28.34387969970703, 195.93807983398438, 237.21456909179688, 267.0330810546875, -89.05934143066406, -1.4286270141601562, 28.14971160888672, 145.79795837402344, 157.19741821289062, 295.412841796875, 292.9881591796875, 233.7674102783203, 197.63299560546875, 240.08538818359375, -362.97784423828125, -320.6034851074219, 237.88304138183594, 149.7161865234375, -91.92456817626953, 87.5528564453125, -16.866924285888672, -3.7589950561523438, -36.72657775878906, 178.02609252929688, 161.53408813476562], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000648.npy"}
|
||||
{"epoch": 0.9795918367346939, "step": 649, "batch_size": 64, "mean": 86.74510192871094, "std": 154.99478149414062, "min": -385.76519775390625, "p10": -94.98390541076657, "median": 103.10811996459961, "p90": 253.43425598144532, "max": 485.27655029296875, "pos_frac": 0.71875, "sample": [197.38888549804688, 71.42987060546875, -57.374080657958984, 189.55047607421875, 104.96109008789062, 183.21859741210938, -42.848388671875, -15.008234024047852, -28.517921447753906, 192.02197265625, -6.669610977172852, 485.27655029296875, 207.60606384277344, 1.8677330017089844, 258.17718505859375, 258.21600341796875, 215.24569702148438, 6.288145065307617, 195.518798828125, 252.71539306640625, 221.78564453125, -26.096071243286133, -130.61798095703125, 0.10169219970703125, -385.76519775390625, 177.70404052734375, 29.72985076904297, 57.009361267089844, -129.88174438476562, -188.95611572265625, 129.90223693847656, 144.690185546875, 157.158447265625, -5.141530990600586, 118.66085815429688, 253.74234008789062, 76.59307098388672, 207.7583770751953, 161.3026123046875, 207.77099609375, -14.082176208496094, 316.35797119140625, 93.23827362060547, 9.235836029052734, 65.75631713867188, 210.13153076171875, 6.482643127441406, 47.001441955566406, 15.232284545898438, 107.76652526855469, 101.2551498413086, 355.8470458984375, -139.78436279296875, -363.4856872558594, 306.45391845703125, -20.46139907836914, -111.10240173339844, -36.35472106933594, 118.9358139038086, 216.33555603027344, 229.11422729492188, 163.33023071289062, -11.164592742919922, 139.132080078125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000649.npy"}
|
||||
{"epoch": 0.981103552532124, "step": 650, "batch_size": 64, "mean": 97.00033569335938, "std": 171.285888671875, "min": -327.59417724609375, "p10": -115.3745506286621, "median": 57.95292854309082, "p90": 259.09851379394536, "max": 625.4169921875, "pos_frac": 0.734375, "sample": [-29.104541778564453, 412.42987060546875, 625.4169921875, 230.26382446289062, 45.136749267578125, 195.75701904296875, 224.07275390625, 183.60203552246094, 17.05028533935547, 24.340320587158203, 179.96609497070312, 209.00924682617188, -182.08648681640625, -9.014259338378906, 2.6348819732666016, 344.9752502441406, 188.3972930908203, 43.61689758300781, 337.20269775390625, 4.826356887817383, 242.42262268066406, 164.4312286376953, -116.77991485595703, 1.5214958190917969, 196.07366943359375, 241.235107421875, 220.87249755859375, 229.5670623779297, 138.17095947265625, -123.9305648803711, 6.839397430419922, -105.72096252441406, -23.68026351928711, 163.41680908203125, -327.59417724609375, 57.65440368652344, 231.4359893798828, 58.2514533996582, 19.11200714111328, 0.8225326538085938, -141.92343139648438, 96.61831665039062, -271.57891845703125, 198.59054565429688, -112.09536743164062, 47.12267303466797, 384.70391845703125, 262.7652282714844, 226.9442138671875, -27.021108627319336, 196.4770050048828, 12.07232666015625, 226.8035430908203, -15.519519805908203, -13.098722457885742, 31.795595169067383, -85.11555480957031, -117.353515625, 206.88925170898438, 392.7008056640625, 250.5428466796875, -36.45993423461914, 142.6780242919922, 28.86876678466797], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000650.npy"}
|
||||
{"epoch": 0.982615268329554, "step": 651, "batch_size": 64, "mean": 89.26908874511719, "std": 146.1291961669922, "min": -154.64816284179688, "p10": -57.68512878417968, "median": 42.88723945617676, "p90": 251.04876251220708, "max": 552.5018310546875, "pos_frac": 0.671875, "sample": [-7.1752777099609375, -12.1602783203125, 174.25779724121094, -78.27981567382812, 49.144248962402344, -37.28838348388672, -51.440467834472656, 98.68274688720703, 40.934967041015625, 216.31011962890625, -6.118408203125, -102.83322143554688, -17.447059631347656, 123.88072204589844, 170.94627380371094, -154.64816284179688, 27.36125946044922, 468.7348327636719, -40.047096252441406, 238.47853088378906, 238.19561767578125, -14.491065979003906, 70.04835510253906, 148.9203338623047, 49.45805358886719, 44.83951187133789, 274.4489440917969, 109.7100830078125, 21.703292846679688, 445.2533264160156, 8.452098846435547, 291.6866149902344, 24.378353118896484, -8.979560852050781, 106.40351867675781, -0.27221107482910156, 231.6686553955078, 12.480514526367188, -51.223670959472656, 111.35125732421875, 256.4360046386719, 1.9196319580078125, 228.03367614746094, -10.999446868896484, -18.308246612548828, 234.78280639648438, 364.1649169921875, 195.16537475585938, -154.36932373046875, 211.99005126953125, 16.675813674926758, 11.108642578125, 6.448371887207031, 145.93923950195312, 201.56692504882812, 118.17552947998047, 203.1276397705078, -6.073184967041016, 142.6446990966797, -110.63046264648438, 28.407958984375, -60.45061492919922, 552.5018310546875, -60.361412048339844], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000651.npy"}
|
||||
{"epoch": 0.9841269841269841, "step": 652, "batch_size": 64, "mean": 74.52107238769531, "std": 196.51026916503906, "min": -536.861328125, "p10": -105.25370864868164, "median": 34.3491096496582, "p90": 305.20503845214853, "max": 822.5033569335938, "pos_frac": 0.640625, "sample": [36.18537521362305, 214.88674926757812, 187.40994262695312, -15.816108703613281, -1.9195785522460938, 216.49478149414062, 1.2715167999267578, 141.9401092529297, 179.2429656982422, 12.962226867675781, 267.56494140625, -163.34732055664062, 4.285923004150391, -3.548154830932617, 78.06855773925781, 258.53997802734375, 58.7325439453125, 132.52227783203125, 9.335151672363281, 19.806228637695312, -75.95962524414062, 394.0738220214844, 277.05120849609375, 6.055305480957031, 38.29616165161133, -103.90279388427734, -536.861328125, -105.83267211914062, 313.0621337890625, 203.92404174804688, -28.7894287109375, -50.015220642089844, 32.51284408569336, -15.226905822753906, 117.32032012939453, -188.7208251953125, 312.5657043457031, 168.44442749023438, -20.637489318847656, -254.4441680908203, -61.4359130859375, 270.923583984375, -24.920211791992188, 198.12164306640625, -38.417236328125, 66.542236328125, 288.0301513671875, 382.414794921875, -2.3497257232666016, 405.4241027832031, 822.5033569335938, 95.4011001586914, 83.47433471679688, -29.42981719970703, 15.491382598876953, -74.70884704589844, -230.35755920410156, 3.662050247192383, -85.87826538085938, 215.05638122558594, 49.47210693359375, 127.6495361328125, -196.15194702148438, 371.2979431152344], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000652.npy"}
|
||||
{"epoch": 0.9856386999244142, "step": 653, "batch_size": 64, "mean": 89.91576385498047, "std": 153.09043884277344, "min": -305.1858825683594, "p10": -65.59363327026367, "median": 64.34658432006836, "p90": 263.51671142578124, "max": 486.3708190917969, "pos_frac": 0.75, "sample": [5.464662551879883, 146.50595092773438, 45.72418212890625, -3.5639190673828125, 125.38459777832031, 91.56375122070312, 27.903221130371094, -305.1858825683594, 231.2491455078125, 27.196563720703125, -33.43728256225586, -3.8914947509765625, 179.47177124023438, -108.70469665527344, 476.6332092285156, 259.1244201660156, 120.34356689453125, 326.861328125, 64.75819396972656, 260.921875, 127.0910415649414, 297.19866943359375, 158.37158203125, -27.39403533935547, 152.31716918945312, 58.165870666503906, 250.16473388671875, 17.66950225830078, -7.895359039306641, -15.0621337890625, 317.8568420410156, -217.1564483642578, -66.72733306884766, 30.993404388427734, 8.157564163208008, -119.37394714355469, 196.3850860595703, 26.501808166503906, 193.42498779296875, 77.984130859375, 63.934974670410156, -14.105361938476562, 486.3708190917969, 8.216278076171875, 145.74221801757812, -198.54434204101562, 218.17446899414062, -62.948333740234375, 227.3431854248047, 264.6287841796875, -20.152809143066406, 218.43453979492188, -238.63958740234375, 17.079391479492188, 6.349088668823242, 35.97112274169922, 205.24984741210938, 276.6542663574219, 17.289775848388672, 212.67578125, 52.57129669189453, 152.866455078125, 94.21208953857422, 192.23831176757812], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000653.npy"}
|
||||
{"epoch": 0.9871504157218443, "step": 654, "batch_size": 64, "mean": 94.11676025390625, "std": 158.4545135498047, "min": -232.99822998046875, "p10": -84.94240417480468, "median": 56.577247619628906, "p90": 293.99057312011723, "max": 593.6207275390625, "pos_frac": 0.71875, "sample": [31.37120819091797, 319.62786865234375, -232.99822998046875, 168.22137451171875, 38.266815185546875, -132.27337646484375, 258.04693603515625, 98.0198745727539, 4.818363189697266, -86.71986389160156, 18.493255615234375, -80.79499816894531, 248.347900390625, 12.385047912597656, -66.29080963134766, 45.309654235839844, 202.1728057861328, -140.38389587402344, 217.26792907714844, 58.9219970703125, 102.18806457519531, 45.15922164916992, 279.42437744140625, 399.9108581542969, 15.287858963012695, -35.697547912597656, 15.85220718383789, -16.07379913330078, -93.9372329711914, -18.327495574951172, -9.759262084960938, 258.2857971191406, 593.6207275390625, 99.28555297851562, 25.845985412597656, 298.8212890625, 323.9743957519531, -63.287994384765625, 48.62260437011719, 155.14437866210938, 196.52940368652344, 371.541748046875, 265.249267578125, -148.58180236816406, 54.23249816894531, 42.92546844482422, 227.2274627685547, 214.8815460205078, 70.162109375, 110.83341979980469, 226.77552795410156, 282.7189025878906, -0.24750900268554688, 199.50411987304688, 184.411865234375, -73.17987060546875, -25.79644012451172, 135.4969024658203, 307.0489501953125, 114.29759216308594, 45.09604263305664, 66.81631469726562, -27.321762084960938, -223.2994384765625], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000654.npy"}
|
||||
{"epoch": 0.9886621315192744, "step": 655, "batch_size": 64, "mean": 131.76361083984375, "std": 160.85032653808594, "min": -167.07362365722656, "p10": -52.09011611938477, "median": 110.84217071533203, "p90": 327.54430236816404, "max": 651.5057983398438, "pos_frac": 0.796875, "sample": [193.12783813476562, 195.80201721191406, -64.2905502319336, 11.32278060913086, 53.460723876953125, 250.17684936523438, 536.9923095703125, 114.96405029296875, 157.4107666015625, 42.79669189453125, -128.3817138671875, 215.30862426757812, -6.9768829345703125, -52.68999481201172, -2.425445556640625, -50.690399169921875, 81.12908935546875, 270.79656982421875, 240.05709838867188, 2.4530982971191406, 226.14035034179688, -34.86216735839844, 207.54257202148438, 651.5057983398438, 22.27971649169922, -47.37977600097656, 102.41426849365234, -69.44140625, 448.00714111328125, -104.63185119628906, 195.9600372314453, 122.72125244140625, 200.96942138671875, 78.1696548461914, 348.177734375, 48.812686920166016, 53.49372100830078, 177.15261840820312, 283.76580810546875, 327.697021484375, 18.523529052734375, 99.09100341796875, 21.934654235839844, 327.1879577636719, 190.0259246826172, -80.75213623046875, 60.221588134765625, 203.11351013183594, 193.8631591796875, 529.5803833007812, -17.76803970336914, 106.72029113769531, 330.5550231933594, 65.6797866821289, 27.370040893554688, 120.18110656738281, 126.75985717773438, 181.34060668945312, 252.49078369140625, 148.4512939453125, -167.07362365722656, 74.93385314941406, 95.00802612304688, 226.59474182128906], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000655.npy"}
|
||||
{"epoch": 0.9901738473167044, "step": 656, "batch_size": 64, "mean": 91.52388000488281, "std": 156.33270263671875, "min": -307.14166259765625, "p10": -63.45533485412597, "median": 55.16893768310547, "p90": 287.6999481201172, "max": 593.736572265625, "pos_frac": 0.71875, "sample": [192.46786499023438, 88.43634796142578, 213.34591674804688, 234.08709716796875, 68.70169067382812, -0.3303070068359375, -23.305049896240234, -34.043479919433594, 0.237335205078125, -65.33668518066406, -10.047321319580078, 328.21087646484375, -42.09751892089844, -8.641448974609375, -1.1281509399414062, 230.24293518066406, -307.14166259765625, -66.58675384521484, 184.46029663085938, 193.5359344482422, 366.7704772949219, 35.988990783691406, -198.26202392578125, 99.95301818847656, 285.67230224609375, -59.06551742553711, 135.52764892578125, 593.736572265625, 19.537307739257812, 173.74160766601562, 130.82778930664062, -6.419242858886719, 58.30265808105469, 52.03521728515625, 39.67253875732422, 221.47303771972656, 45.497344970703125, 17.174224853515625, 263.005859375, 152.05413818359375, 160.4569549560547, -165.82984924316406, 4.665275573730469, 321.84478759765625, 26.30926513671875, 288.5689392089844, 17.948122024536133, 27.628990173339844, 14.064617156982422, 214.9130401611328, 188.6885986328125, -37.579769134521484, -113.5936279296875, 272.25872802734375, 291.6507263183594, 38.30003356933594, 32.58442687988281, -207.63267517089844, -54.015769958496094, 95.10350036621094, 96.38278198242188, 147.1125946044922, 234.22772216796875, 361.1790466308594], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000656.npy"}
|
||||
{"epoch": 0.9916855631141346, "step": 657, "batch_size": 64, "mean": 117.31202697753906, "std": 179.16531372070312, "min": -272.09613037109375, "p10": -76.91509857177732, "median": 111.5990219116211, "p90": 324.042709350586, "max": 569.5018310546875, "pos_frac": 0.765625, "sample": [303.15087890625, -8.950164794921875, 31.794147491455078, 535.6343994140625, 169.24368286132812, 39.87921905517578, 537.9566650390625, 231.91525268554688, 61.979393005371094, -184.88185119628906, 20.372852325439453, 18.89045524597168, 168.09173583984375, 46.03667449951172, 519.5086669921875, 112.67176818847656, 62.92584991455078, 200.7145233154297, -221.72305297851562, 316.3835754394531, 110.5948486328125, 154.24075317382812, -200.0614013671875, 10.587640762329102, 44.790245056152344, 206.69032287597656, 209.6651611328125, -29.31268310546875, -28.36701774597168, 193.52386474609375, 438.9197998046875, 201.3577880859375, -180.70005798339844, 197.4606475830078, 139.07603454589844, 59.447509765625, 192.92575073242188, 197.41287231445312, 327.3251953125, 21.5390567779541, 569.5018310546875, 40.45372009277344, -4.813812255859375, 285.073974609375, 31.922119140625, 331.6313171386719, 281.2931823730469, -17.417436599731445, -47.641136169433594, 68.65298461914062, -89.4610824584961, -272.09613037109375, 130.87823486328125, -3.5753002166748047, 27.074918746948242, 135.01803588867188, -124.39920806884766, 190.9625244140625, 112.60319519042969, 241.15867614746094, 31.87881851196289, 165.90322875976562, -21.134864807128906, 215.79067993164062], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000657.npy"}
|
||||
{"epoch": 0.9931972789115646, "step": 658, "batch_size": 64, "mean": 92.16999816894531, "std": 197.2000732421875, "min": -630.2125244140625, "p10": -107.23861846923828, "median": 84.85910415649414, "p90": 321.61332092285164, "max": 642.5731201171875, "pos_frac": 0.734375, "sample": [204.6436767578125, 4.709506988525391, 301.22161865234375, -630.2125244140625, 263.6028747558594, -16.453638076782227, -209.7582550048828, 186.87977600097656, 217.01612854003906, 349.1197204589844, -112.56228637695312, 304.6211242675781, -59.56846618652344, 37.737789154052734, 252.76976013183594, 90.16256713867188, -94.81672668457031, 126.02799987792969, 642.5731201171875, 171.53195190429688, -27.52446937561035, 90.00033569335938, 47.878013610839844, 190.88168334960938, 229.0380859375, 219.9176788330078, -262.2395935058594, 203.60751342773438, 132.74749755859375, 50.4964599609375, 24.78900718688965, -14.537281036376953, 62.64009094238281, 112.31053161621094, 4.5759735107421875, 111.40894317626953, 328.89569091796875, -202.06103515625, 90.80530548095703, 15.022491455078125, 271.44732666015625, 7.362945556640625, 354.4422607421875, 286.432861328125, -5.971220016479492, 60.445091247558594, -89.60627746582031, 329.4412841796875, -14.264411926269531, 132.47335815429688, 76.00927734375, 4.815633773803711, -84.05199432373047, -197.3447723388672, 89.8970718383789, 147.8758544921875, 17.590984344482422, -207.93841552734375, 376.72796630859375, -6.729652404785156, 547.4652099609375, 79.82113647460938, 276.5599670410156, 8.079549789428711], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000658.npy"}
|
||||
{"epoch": 0.9947089947089947, "step": 659, "batch_size": 64, "mean": 68.84123229980469, "std": 171.0502471923828, "min": -425.72406005859375, "p10": -166.25197448730466, "median": 43.993900299072266, "p90": 242.6950759887696, "max": 554.2734375, "pos_frac": 0.625, "sample": [70.88534545898438, 140.7814178466797, 248.98106384277344, 174.36708068847656, -39.7591552734375, 411.2857971191406, 201.44461059570312, 143.04562377929688, 15.192026138305664, 270.7809753417969, 449.2996520996094, -21.280624389648438, 47.29710388183594, -3.6604042053222656, -1.1251792907714844, -20.512832641601562, 214.75552368164062, -199.778564453125, 21.80848503112793, -36.85517883300781, 228.02777099609375, -181.9244384765625, 118.62701416015625, 166.71771240234375, -0.6114292144775391, -23.792556762695312, 22.409265518188477, 22.125364303588867, -3.0028934478759766, 196.1779022216797, 93.83833312988281, 86.03182220458984, 204.45516967773438, 87.32872009277344, 138.87753295898438, 41.566627502441406, 227.84658813476562, 154.451904296875, 214.6826171875, -12.958988189697266, -47.948394775390625, -425.72406005859375, -129.68289184570312, 213.22239685058594, 14.913488388061523, -6.46258544921875, -20.817840576171875, 554.2734375, 165.82711791992188, -121.26825714111328, -17.037582397460938, 209.7111053466797, -218.16293334960938, -191.30361938476562, 287.66162109375, 31.100067138671875, 10.280067443847656, -207.0703887939453, 279.760498046875, 46.421173095703125, -40.35340118408203, 212.53273010253906, 197.61886596679688, -259.4783935546875], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000659.npy"}
|
||||
{"epoch": 0.9962207105064248, "step": 660, "batch_size": 64, "mean": 121.72052001953125, "std": 178.0220947265625, "min": -267.1798095703125, "p10": -58.93984069824218, "median": 114.73092651367188, "p90": 297.9386199951172, "max": 962.46435546875, "pos_frac": 0.8125, "sample": [238.37013244628906, 105.58783721923828, 136.5337371826172, -61.751983642578125, 88.48171997070312, -167.12255859375, -146.4996337890625, 222.2016143798828, -267.1798095703125, -23.40545654296875, 75.95927429199219, 292.6127624511719, 87.93260192871094, 175.41500854492188, 17.739791870117188, 436.3734436035156, 242.8258514404297, 173.9071807861328, 35.82447814941406, -127.07449340820312, 184.9089813232422, 331.83148193359375, 24.89875030517578, 62.089149475097656, 243.8142547607422, 219.539306640625, 2.1669158935546875, 300.22113037109375, 169.09426879882812, -110.640380859375, 16.244117736816406, 393.32940673828125, 962.46435546875, 161.75413513183594, -15.310291290283203, -52.378173828125, 37.53905487060547, -9.536590576171875, 206.6540069580078, 110.02803039550781, 55.886146545410156, 4.364103317260742, 284.1479187011719, 132.6523895263672, 204.89602661132812, 226.64813232421875, -146.32583618164062, 65.53633880615234, 210.56398010253906, 166.80380249023438, -3.222095489501953, 354.87054443359375, 1.8158798217773438, 6.409372329711914, 271.2255554199219, 131.07994079589844, 129.2850799560547, 119.43382263183594, 12.975555419921875, 4.745025634765625, 56.92991638183594, 196.48704528808594, 215.97837829589844, 311.4830017089844], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000660.npy"}
|
||||
{"epoch": 0.9977324263038548, "step": 661, "batch_size": 64, "mean": 75.77854919433594, "std": 181.62159729003906, "min": -439.49267578125, "p10": -165.0026107788086, "median": 62.80247497558594, "p90": 246.91444396972665, "max": 651.993408203125, "pos_frac": 0.703125, "sample": [-29.19598388671875, -0.481170654296875, -197.68505859375, -199.52139282226562, 101.23753356933594, 165.4066925048828, 218.4101104736328, 50.57879638671875, -439.49267578125, 227.34571838378906, -171.67051696777344, 25.68292236328125, -139.3166961669922, 210.04583740234375, 285.3553161621094, 167.38653564453125, 255.30104064941406, 76.27787780761719, 614.8023681640625, 88.30671691894531, -190.39390563964844, 221.6028594970703, 182.74319458007812, -115.56463623046875, -8.100597381591797, 56.49493408203125, 19.777732849121094, -32.39125442504883, 3.0251388549804688, 3.7391128540039062, 191.08010864257812, 97.75836944580078, -199.3589630126953, 13.407173156738281, 221.65963745117188, -2.1051063537597656, 130.6019744873047, 220.6643829345703, 33.4982795715332, 173.6680145263672, 28.850250244140625, 76.84001159667969, -170.91688537597656, 159.623779296875, 285.42315673828125, 149.55972290039062, 315.6033935546875, 16.82573890686035, 9.695821762084961, 181.97360229492188, -151.20263671875, -31.13782501220703, 9.010726928710938, -53.0272216796875, 69.11001586914062, 112.70438385009766, 211.87277221679688, 358.88433837890625, 187.41073608398438, 218.55880737304688, 32.515045166015625, -93.2958984375, -57.62828826904297, 651.993408203125], "npy": "outputs/llama3-8b-base-new-method-s_star0.4/margin_logs/step_0000661.npy"}
|
||||
3
model-00001-of-00007.safetensors
Normal file
3
model-00001-of-00007.safetensors
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:58aa6ea624d1e4337141e8246ce3f572101986e399d3655054d4f11a9119de50
|
||||
size 4886466168
|
||||
3
model-00002-of-00007.safetensors
Normal file
3
model-00002-of-00007.safetensors
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:8a3283db529373939816bd411938702ba07359e0170d251ec25bf57f4566566e
|
||||
size 4832007448
|
||||
3
model-00003-of-00007.safetensors
Normal file
3
model-00003-of-00007.safetensors
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:34384402ac82eaf7ea67469725b32d90ea00f07c337c52b77da4f3edcbadea73
|
||||
size 4999813112
|
||||
3
model-00004-of-00007.safetensors
Normal file
3
model-00004-of-00007.safetensors
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:74420786e21734ec9b8a8dc3e329802e557b2863b55f8809814e066a86c75f2e
|
||||
size 4999813128
|
||||
3
model-00005-of-00007.safetensors
Normal file
3
model-00005-of-00007.safetensors
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:1404952b5a2a24b5ebd86f620d6ba3d0ca8badd548a1ea75d9eb3d720a5d78be
|
||||
size 4832007496
|
||||
3
model-00006-of-00007.safetensors
Normal file
3
model-00006-of-00007.safetensors
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:cb5a966273f04d8a4ba9728bd20e785e1346441855f0b8afe322b9d2ae2ef78a
|
||||
size 4999813120
|
||||
3
model-00007-of-00007.safetensors
Normal file
3
model-00007-of-00007.safetensors
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:edbabe9b1b927fb020ffa6c965642ddd3470f5795347440ed1e1db7b98c08dc5
|
||||
size 2571158184
|
||||
298
model.safetensors.index.json
Normal file
298
model.safetensors.index.json
Normal file
@@ -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"
|
||||
}
|
||||
}
|
||||
23
special_tokens_map.json
Normal file
23
special_tokens_map.json
Normal file
@@ -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
|
||||
}
|
||||
}
|
||||
3
tokenizer.json
Normal file
3
tokenizer.json
Normal file
@@ -0,0 +1,3 @@
|
||||
version https://git-lfs.github.com/spec/v1
|
||||
oid sha256:3c5cf44023714fb39b05e71e425f8d7b92805ff73f7988b083b8c87f0bf87393
|
||||
size 17209961
|
||||
2064
tokenizer_config.json
Normal file
2064
tokenizer_config.json
Normal file
File diff suppressed because it is too large
Load Diff
9
train_results.json
Normal file
9
train_results.json
Normal file
@@ -0,0 +1,9 @@
|
||||
{
|
||||
"epoch": 0.999244142101285,
|
||||
"total_flos": 0.0,
|
||||
"train_loss": 1.1812975648311552,
|
||||
"train_runtime": 1809.2515,
|
||||
"train_samples": 42336,
|
||||
"train_samples_per_second": 23.4,
|
||||
"train_steps_per_second": 0.365
|
||||
}
|
||||
2621
trainer_state.json
Normal file
2621
trainer_state.json
Normal file
File diff suppressed because it is too large
Load Diff
Reference in New Issue
Block a user