初始化项目，由ModelHub XC社区提供模型

Model: SeongryongJung/Qwen3-8B-Material-RLSD-TR
Source: Original Platform

results/hyperparameters.csv

@@ -0,0 +1,38 @@
+section,parameter,value,source
+Run identity,Base model,Qwen/Qwen3-8B,queue/script override
+Run identity,Dataset,Material / SciKnowEval material,run_qwen3_generalization.sh
+Run identity,Method,RLSD_TR,run_qwen3_generalization.sh
+Run identity,Config,rlsd,run_qwen3_generalization.sh
+Run identity,Experiment,qwen3gen-material-RLSD_TR-Qwen-Qwen3-8B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8,run_qwen3_generalization.sh
+Run identity,W&B run,run-20260703_115716-skza3cpu,wandb
+Data,Train file,datasets/sciknoweval/material/train.parquet,script override
+Data,Validation file,datasets/sciknoweval/material/test.parquet,script override
+Data,Train batch size,32,queue/script override
+Data,Train max samples,3200,queue/script override
+Schedule,Total training steps,100,queue/script override
+Schedule,Validation before train,False,queue/script override
+Schedule,Save frequency,10,queue/script override
+Schedule,Validation frequency,10,queue/script override
+Sequence,Max prompt length,2048,queue/script override
+Sequence,Max response length,8192,queue/script override
+Sequence,Max model length,10240,queue/script override
+Rollout,Train rollout n,8,queue/script override
+Rollout,Validation rollout n,16,queue/script override
+Rollout,vLLM GPU memory utilization,0.8,queue/script override
+Optimization,Learning rate,1e-6,RLSD_TR method override
+Optimization,Weight decay,0.01,script override
+PPO/GRPO,PPO mini batch size,8,queue/script override
+PPO/GRPO,Normalize GRPO advantages by std,False,baseline_grpo.yaml / script override
+Rollout correction,Importance sampling mode,token,script override
+Rollout correction,IS threshold,2.0,script override
+Checkpoint/Logging,Checkpoint root,checkpoints/datasets/sciknoweval/material/qwen3gen-material-RLSD_TR-Qwen-Qwen3-8B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8,script override
+Checkpoint/Logging,Latest checkpointed iteration,100,latest_checkpointed_iteration.txt
+Checkpoint/Logging,External actor archive,checkpoints/datasets/sciknoweval/material/qwen3gen-material-RLSD_TR-Qwen-Qwen3-8B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/_actor_archive,preserve_actor_checkpoints.py
+Checkpoint/Logging,Logger,"console, wandb",ppo_trainer.yaml
+RLSD_TR,Policy loss mode,rlsd,method override
+RLSD_TR,Teacher regularization,trust-region,method override
+RLSD_TR,Trust-region mix / teacher update rate,0.1,queue/script override
+RLSD_TR,Token reweight lambda,0.5,queue/script override
+RLSD_TR,Token reweight eps_w,0.2,queue/script override
+RLSD_TR,Token reweight decay steps,0,queue/script override
+RLSD_TR,Fused kernels,False,method override

results/summary.json

@@ -0,0 +1,29 @@
+{
+  "repo_id": "SeongryongJung/Qwen3-8B-Material-RLSD-TR",
+  "output_dir": "/mnt/mole/SDPO/L2T/hf_upload_tr/Qwen3-8B-Material-RLSD-TR",
+  "experiment": "qwen3gen-material-RLSD_TR-Qwen-Qwen3-8B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8",
+  "best_step": 100,
+  "best_val_mean16": 0.7646276595744681,
+  "best_actor_dir": "/mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/material/qwen3gen-material-RLSD_TR-Qwen-Qwen3-8B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_100/actor",
+  "final_step": 100,
+  "final_val_mean16": 0.7646276595744681,
+  "final_actor_dir": "/mnt/mole/SDPO/L2T/checkpoints/datasets/sciknoweval/material/qwen3gen-material-RLSD_TR-Qwen-Qwen3-8B-mbs8-decay0-tr0.1-train32-rollout8-lr1e-6-vllm0.8/global_step_100/actor",
+  "train_rows": 90,
+  "val_rows": 10,
+  "hf_model_files": [
+    "added_tokens.json",
+    "chat_template.jinja",
+    "config.json",
+    "generation_config.json",
+    "merges.txt",
+    "model.safetensors.index.json",
+    "special_tokens_map.json",
+    "tokenizer.json",
+    "tokenizer_config.json",
+    "vocab.json",
+    "model-00001-of-00004.safetensors",
+    "model-00002-of-00004.safetensors",
+    "model-00003-of-00004.safetensors",
+    "model-00004-of-00004.safetensors"
+  ]
+}

results/training_score.png

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8b033785d929211e5f4e519e7c4a4bd294a8424ceceb6a979c4471637e0e5c16
+size 128191

results/training_scores.csv

@@ -0,0 +1,91 @@
+step,critic_score_mean
+1,0.53125
+2,0.38671875
+3,0.49609375
+4,0.51953125
+5,0.55078125
+6,0.46484375
+7,0.5703125
+8,0.54296875
+9,0.57421875
+11,0.5390625
+12,0.62890625
+13,0.5546875
+14,0.71875
+15,0.61328125
+16,0.66015625
+17,0.55078125
+18,0.62890625
+19,0.60546875
+21,0.54296875
+22,0.69140625
+23,0.73828125
+24,0.8515625
+25,0.625
+26,0.66796875
+27,0.81640625
+28,0.5859375
+29,0.6171875
+31,0.65234375
+32,0.65625
+33,0.7109375
+34,0.80859375
+35,0.81640625
+36,0.61328125
+37,0.8515625
+38,0.60546875
+39,0.7578125
+41,0.59765625
+42,0.59375
+43,0.515625
+44,0.72265625
+45,0.7734375
+46,0.703125
+47,0.6640625
+48,0.84375
+49,0.66796875
+51,0.66796875
+52,0.78515625
+53,0.64453125
+54,0.57421875
+55,0.7421875
+56,0.765625
+57,0.765625
+58,0.8046875
+59,0.7265625
+61,0.80859375
+62,0.8515625
+63,0.6953125
+64,0.80078125
+65,0.6171875
+66,0.69921875
+67,0.80078125
+68,0.78125
+69,0.796875
+71,0.7421875
+72,0.765625
+73,0.6796875
+74,0.8359375
+75,0.70703125
+76,0.703125
+77,0.55078125
+78,0.6640625
+79,0.8046875
+81,0.71875
+82,0.734375
+83,0.68359375
+84,0.8203125
+85,0.69140625
+86,0.83984375
+87,0.609375
+88,0.671875
+89,0.71484375
+91,0.84765625
+92,0.71875
+93,0.8046875
+94,0.69921875
+95,0.7109375
+96,0.58984375
+97,0.84375
+98,0.89453125
+99,0.765625

results/validation_mean16.csv

@@ -0,0 +1,11 @@
+step,val_mean16
+10,0.6555851063829787
+20,0.6888297872340425
+30,0.6974734042553191
+40,0.7340425531914894
+50,0.7466755319148937
+60,0.7513297872340425
+70,0.7519946808510638
+80,0.7446808510638298
+90,0.7539893617021277
+100,0.7646276595744681