{
  "stage": "final",
  "base": "google/gemma-4-E4B-it",
  "base_revision": "fee6332c1abaafb77f6f9624236c63aa2f1d0187",
  "dataset_repo": "guifav/caramelo-dataset",
  "dataset_file": "caramelo_style_pairs.jsonl",
  "dataset_revision": "05188bac9287f993aa84a206ceac98d7dcbd88f1",
  "dataset_sha256": "85a20744e4c45b5f366338fb5ca421d0b97aeb83afbc44a9b7d4084a0d8e6486",
  "adapter_repo": "guifav/caramelo-gemma4-e4b-lora-adapter",
  "epochs": 3,
  "batch_size": 1,
  "grad_accum": 16,
  "learning_rate": 0.0002,
  "lora_r": 16,
  "lora_alpha": 32,
  "lora_dropout": 0.05,
  "max_length": 2048,
  "warmup_ratio": 0.03,
  "eval_ratio": 0.1,
  "resume_from_checkpoint": null,
  "train_examples": 372,
  "eval_examples": 42,
  "target_modules": ".*language_model\\..*?layers\\.\\d+\\..*\\.(q_proj|k_proj|v_proj|o_proj|gate_proj|up_proj|down_proj)$",
  "seed": 42,
  "bnb_4bit_quant_type": "nf4",
  "bnb_4bit_compute_dtype": "bfloat16",
  "bnb_4bit_use_double_quant": true,
  "load_in_4bit": true,
  "bf16": true,
  "gradient_checkpointing": true,
  "gradient_checkpointing_use_reentrant": false,
  "optim": "adamw_torch",
  "lr_scheduler_type": "cosine",
  "lora_bias": "none",
  "task_type": "CAUSAL_LM",
  "trust_remote_code": false,
  "skip_upload": false,
  "versions": {
    "transformers": "5.12.1",
    "torch": "2.12.1",
    "peft": "0.19.1",
    "bitsandbytes": "0.49.2",
    "accelerate": "1.14.0",
    "datasets": "5.0.0",
    "huggingface_hub": "1.21.0"
  }
}