| {"training_config": {"learning_rate": 1e-05, "warmup_steps": 500, "epochs": 2, "early_stopping": 5, "weight_decay": 0, "gradient_accumulation_steps": 1, "mix_precision": "fp16", "train_batch_size": 8, "validation_batch_size": 16}, "model_config": {"num_attn_heads": 16, "layer_connections": [[4, 6], [8, 8], [12, 11]], "anchor_model": "stabilityai/stablelm-2-zephyr-1_6b", "augmenting_model": "gpt2", "anchor_output_weight": 1, "augment_output_weight": 1, "freeze_anchor_params": true, "freeze_augment_params": true}, "logging_config": {"project_name": "text-math-v1", "ckpt_dir": "/home/vdhee/scratch/Nikhil/VA_LLM/checkpoints", "ckpt_freq": 1500, "keep_last_n_checkpoints": 20, "save_model_path": "best_model.pth", "save_model_conn_path": "best_model_conn.pth", "push_to_hub": true, "hf_username": "IamYash"}, "data_config": {"image_data_root": "/home/vdhee/scratch/Nikhil/VA_LLM/coco", "caption_data_path": "/home/vdhee/scratch/Nikhil/VA_LLM/CSV/llava_28k.csv", "test_size": 0.2}} |