{
  "experiment_name": "qwen3gen-chemistry-SRPO_TR-Qwen-Qwen3-4B-mbs32-tr0.1-dwtrue-train32-rollout8-lr5e-6-vllm0.8",
  "base_model": "Qwen/Qwen3-4B",
  "dataset": "chemistry",
  "method": "SRPO-TR",
  "checkpoint_dir": "/workspace/L2T/checkpoints/datasets/sciknoweval/chemistry/qwen3gen-chemistry-SRPO_TR-Qwen-Qwen3-4B-mbs32-tr0.1-dwtrue-train32-rollout8-lr5e-6-vllm0.8",
  "best_checkpoint": "global_step_100",
  "last_checkpoint": "global_step_100",
  "train_batch_size": 32,
  "train_max_samples": 3200,
  "teacher_regularization": "trust-region",
  "teacher_update_rate": 0.1,
  "rollout_n": 8,
  "validation_n": 16,
  "learning_rate": "5e-6",
  "vllm_gpu_memory_utilization": 0.8,
  "command_args": ""
}