training_params.json

{
    "model": "google/paligemma-3b-pt-224",
    "project_name": "ryukijano-paligemma-finetuned",
    "data_path": "abhishek/vqa_small",
    "train_split": "train",
    "valid_split": "validation",
    "trainer": "vqa",
    "log": "tensorboard",
    "disable_gradient_checkpointing": false,
    "logging_steps": -1,
    "eval_strategy": "epoch",
    "save_total_limit": 1,
    "auto_find_batch_size": false,
    "mixed_precision": "fp16",
    "lr": 2e-05,
    "epochs": 100,
    "batch_size": 2,
    "warmup_ratio": 0.1,
    "gradient_accumulation": 16,
    "optimizer": "adamw_torch",
    "scheduler": "linear",
    "weight_decay": 0.0,
    "max_grad_norm": 1.0,
    "seed": 42,
    "quantization": "int4",
    "target_modules": "all-linear",
    "merge_adapter": false,
    "peft": true,
    "lora_r": 16,
    "lora_alpha": 32,
    "lora_dropout": 0.05,
    "image_column": "image",
    "text_column": "multiple_choice_answer",
    "prompt_text_column": "question",
    "push_to_hub": true,
    "username": "Ryukijano"
}