{
    "run": {
        "task": "image_text_pretrain",
        "lr_sched": "linear_warmup_cosine_lr",
        "init_lr": 1e-05,
        "min_lr": 1e-06,
        "warmup_lr": 1e-06,
        "weight_decay": 0.05,
        "max_epoch": 10,
        "batch_size_train": 1,
        "batch_size_eval": 1,
        "num_workers": 12,
        "warmup_steps": 500,
        "seed": 42,
        "output_dir": "ft/Meta-Llama-3-8B-Instruct-hf/",
        "amp": true,
        "resume_ckpt_path": null,
        "evaluate": false,
        "train_splits": [
            "train"
        ],
        "device": "cuda",
        "world_size": 1,
        "dist_url": "env://",
        "distributed": true,
        "stage": 2,
        "rank": 0,
        "gpu": 0,
        "dist_backend": "nccl"
    },
    "model": {
        "arch": "mini_gpt4",
        "image_size": 224,
        "drop_path_rate": 0,
        "use_grad_checkpoint": false,
        "vit_precision": "fp16",
        "freeze_vit": true,
        "freeze_qformer": true,
        "num_query_token": 32,
        "llama_model": "/home/ubuntu/ckpt/hf/Meta-Llama-3-8B-Instruct-hf/",
        "prompt": "",
        "model_type": "pretrain_vicuna",
        "max_txt_len": 256,
        "end_sym": "###",
        "prompt_template": "###Human: {} ###Assistant: ",
        "ckpt": "/home/ubuntu/proteinchat/minigpt4/output/Meta-Llama-3-8B-Instruct-hf/20240606190/checkpoint_2.pth"
    },
    "preprocess": {
        "vis_processor": {
            "train": {
                "name": "blip2_image_train",
                "image_size": 224
            },
            "eval": {
                "name": "blip2_image_eval",
                "image_size": 224
            }
        },
        "text_processor": {
            "train": {
                "name": "blip_caption"
            },
            "eval": {
                "name": "blip_caption"
            }
        }
    },
    "datasets": {
        "cc_sbu_align": {
            "data_type": "images",
            "build_info": {
                "storage": "PATH_TO_$cc_sbu_align"
            },
            "vis_processor": {
                "train": {
                    "name": "blip2_image_train",
                    "image_size": 224
                }
            },
            "text_processor": {
                "train": {
                    "name": "blip_caption"
                }
            }
        }
    }
}
{
    "run": {
        "task": "image_text_pretrain",
        "lr_sched": "linear_warmup_cosine_lr",
        "init_lr": 1e-05,
        "min_lr": 1e-06,
        "warmup_lr": 1e-06,
        "weight_decay": 0.05,
        "max_epoch": 10,
        "batch_size_train": 1,
        "batch_size_eval": 1,
        "num_workers": 12,
        "warmup_steps": 5000,
        "seed": 42,
        "output_dir": "ft/Meta-Llama-3-8B-Instruct-hf/",
        "amp": true,
        "resume_ckpt_path": null,
        "evaluate": false,
        "train_splits": [
            "train"
        ],
        "device": "cuda",
        "world_size": 1,
        "dist_url": "env://",
        "distributed": true,
        "stage": 2,
        "rank": 0,
        "gpu": 0,
        "dist_backend": "nccl"
    },
    "model": {
        "arch": "mini_gpt4",
        "image_size": 224,
        "drop_path_rate": 0,
        "use_grad_checkpoint": false,
        "vit_precision": "fp16",
        "freeze_vit": true,
        "freeze_qformer": true,
        "num_query_token": 32,
        "llama_model": "/home/ubuntu/ckpt/hf/Meta-Llama-3-8B-Instruct-hf/",
        "prompt": "",
        "model_type": "pretrain_vicuna",
        "max_txt_len": 256,
        "end_sym": "###",
        "prompt_template": "###Human: {} ###Assistant: ",
        "ckpt": "/home/ubuntu/proteinchat/minigpt4/output/Meta-Llama-3-8B-Instruct-hf/20240606190/checkpoint_2.pth"
    },
    "preprocess": {
        "vis_processor": {
            "train": {
                "name": "blip2_image_train",
                "image_size": 224
            },
            "eval": {
                "name": "blip2_image_eval",
                "image_size": 224
            }
        },
        "text_processor": {
            "train": {
                "name": "blip_caption"
            },
            "eval": {
                "name": "blip_caption"
            }
        }
    },
    "datasets": {
        "cc_sbu_align": {
            "data_type": "images",
            "build_info": {
                "storage": "PATH_TO_$cc_sbu_align"
            },
            "vis_processor": {
                "train": {
                    "name": "blip2_image_train",
                    "image_size": 224
                }
            },
            "text_processor": {
                "train": {
                    "name": "blip_caption"
                }
            }
        }
    }
}
{"train_lr": "0.000", "train_loss": "1.119"}
{"train_lr": "0.000", "train_loss": "0.898"}
{"train_lr": "0.000", "train_loss": "0.796"}
{"train_lr": "0.000", "train_loss": "0.697"}
{"train_lr": "0.000", "train_loss": "0.606"}
{"train_lr": "0.000", "train_loss": "0.524"}