{
  "model_name": "meta-llama/Llama-3.1-8B-Instruct",
  "run_name": "clinical-agent-transformer-mcp-full",
  "seq_len": 1024,
  "micro_batch_size": 1,
  "gradient_accumulation_steps": 8,
  "learning_rate": 2e-05,
  "train_steps": 1200,
  "warmup_steps": 100,
  "weight_decay": 0.01,
  "use_bf16": true,
  "gradient_checkpointing": true,
  "hf_cache_dir": "/cache/huggingface",
  "dataset_dir": "/root/generated_data",
  "checkpoint_dir": "/checkpoints",
  "output_dir": "/checkpoints/runs",
  "train_jsonl": "/root/generated_data/amr_mcp_traces_balanced_v1.jsonl",
  "text_field": "text_trace",
  "max_records": null,
  "validation_fraction": 0.1,
  "resume_trainable_state": "/checkpoints/runs/clinical-agent-transformer-clinical-full/trainable_state.pt",
  "log_every": 25,
  "eval_every": 100,
  "checkpoint_every": 0,
  "max_eval_batches": 75,
  "seed": 42,
  "optimizer_name": "adam8bit",
  "freeze_base_model": false,
  "adapter": "AdapterConfig(enabled=True, adapter_name='clinical_agent_adapter', r=8, lora_alpha=32, lora_dropout=0.05, target_modules=('q_proj', 'k_proj', 'v_proj', 'o_proj'))",
  "surgery": "SurgeryConfig(use_mla=True, use_qk_norm=True, use_post_norm=True, use_moe_layers=True, use_recurrent_layers=True, use_mtp_head=True, qk_norm_layers=(), qk_norm_eps=1e-05, mla_layers=(), moe_layers=(), recurrent_layers=())"
}