{ "HYPER-PARAMETERS": { "MODEL_NAME": "unsloth/phi-4", "MAX_LEN": 3400, "LOAD_IN_4BIT": true, "BATCH_SIZE": 16, "GRAD_ACC": 4, "EPOCHS": 5, "LR": 0.000218, "LOG_STEPS": 100, "SAVE_STEPS": 100, "SAVE_LIMIT": 3, "EVAL_STEPS": 100, "WEIGHT_DECAY": 0.025, "WARMUP_RATIO": 0.047, "MAX_GRAD_NORM": 1.0, "LORA_R": 16, "LORA_ALPHA": 128, "LORA_DROPOUT": 0.1, "ES_THRESHOLD": 0.001, "ES_PATIENCE": 5, "DATA_FILE_PATH": "/dgx2/pawangcs/Dataset/v3_gitapress_final_3shot_prompts.csv", "OUTPUT_DIR": "/dgx2/pawangcs/Trained_Models/Phi4-3shot" }, "TRAIN_DATASET_LEN": 23316, "VAL_DATASET_LEN": 2918, "best_model": { "best_model_checkpoint": "/dgx2/pawangcs/Trained_Models/Phi4-3shot/checkpoint-700", "best_model_metric": 0.5273797512054443 } }