{
  "run_name": "simnpo_gemma-3-12b-pt_20260427_203900",
  "run_dir": "/runs/simnpo_gemma-3-12b-pt_20260427_203900",
  "config": {
    "model_name": "google/gemma-3-12b-pt",
    "forget_path": "unlearning-cleanslate/fsid-curated-gemma-12b:forget:baseline",
    "retain_path": "unlearning-cleanslate/fsid-curated-gemma-12b:retain:baseline",
    "output_dir": "/runs",
    "objective": "simnpo",
    "num_epochs": 2,
    "batch_size": 8,
    "gradient_accum_steps": 4,
    "lr": 5e-05,
    "weight_decay": 0.0,
    "warmup_steps": 10,
    "max_grad_norm": 1.0,
    "forget_weight": 5.0,
    "retain_weight": 1.0,
    "retain_loss_type": "nll",
    "npo_beta": 2.0,
    "simnpo_delta": 0.25,
    "rmu_steering_coeff": 2.0,
    "rmu_module_regex": "model\\.layers\\.7",
    "rmu_control_vec_seed": 0,
    "undial_beta": 10.0,
    "save_every": 200,
    "fsdp_sharding_strategy": "full_shard",
    "dtype": "bfloat16",
    "seed": 42,
    "num_workers": 0,
    "log_every": 10,
    "gradient_checkpointing": false,
    "torch_compile": false,
    "push_to_hub": true,
    "hub_repo_id": "unlearning-cleanslate/gemma-3-12b-simnpo-baseline",
    "hub_private": false,
    "optim": "adamw_torch"
  },
  "num_forget_rows": 27936,
  "num_forget_pairs": 27936,
  "num_retain_rows": 1646,
  "num_retain_sequences": 1646,
  "num_trainable_params": 12187325040,
  "num_total_params": 12187325040,
  "run_seconds": 5790.328476428986,
  "checkpoint_dir": "/runs/simnpo_gemma-3-12b-pt_20260427_203900",
  "is_main_process": true
}