info.json

{
  "config": {
    "model_name": "meta-llama/Llama-3.1-8B-Instruct",
    "train_path": "/kaggle/input/bengali-empathy-data/train_enhanced.jsonl",
    "val_path": "/kaggle/input/bengali-empathy-data/val_enhanced.jsonl",
    "data_sample": 0.4,
    "max_length": 256,
    "lora_r": 16,
    "lora_alpha": 32,
    "lora_dropout": 0.05,
    "target_modules": [
      "q_proj",
      "k_proj",
      "v_proj",
      "o_proj",
      "gate_proj",
      "up_proj",
      "down_proj"
    ],
    "output_dir": "/kaggle/working/bengali-llama-3h",
    "num_epochs": 2,
    "batch_size": 4,
    "grad_accum": 1,
    "learning_rate": 0.0003,
    "warmup_ratio": 0.05,
    "logging_steps": 50,
    "eval_steps": 300,
    "save_steps": 300
  },
  "time": "2025-12-04T11:29:26.333371"
}