{
  "merged_utc": "2026-05-21T12:07:12.605220+00:00",
  "repo_id": "beita6969/SkillFlow-Model",
  "base_model": "Qwen/Qwen3.5-9B",
  "adapter_role": "supervisor_theta",
  "checkpoint": "checkpoint_step_0110",
  "lora_rank": 64,
  "lora_alpha": 128,
  "target_modules": [
    "q_proj",
    "k_proj",
    "v_proj",
    "o_proj"
  ],
  "dtype": "bfloat16",
  "notes": "Backward policy phi is training-time only and is not merged into this inference model."
}