| { | |
| "run": { | |
| "task": "image_text_pretrain", | |
| "lr_sched": "linear_warmup_cosine_lr", | |
| "init_lr": 3e-05, | |
| "min_lr": 1e-05, | |
| "warmup_lr": 1e-06, | |
| "weight_decay": 0.05, | |
| "max_epoch": 50, | |
| "iters_per_epoch": 2100, | |
| "batch_size_train": 7, | |
| "batch_size_eval": 1, | |
| "num_workers": 4, | |
| "warmup_steps": 200, | |
| "seed": 42, | |
| "output_dir": "output/xraygpt_vqa_finetune", | |
| "amp": true, | |
| "resume_ckpt_path": null, | |
| "evaluate": false, | |
| "train_splits": [ | |
| "train" | |
| ], | |
| "device": "cuda", | |
| "world_size": 8, | |
| "dist_url": "env://", | |
| "distributed": true, | |
| "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": "Joycean0301/Llama-3.2-3B-Instruct-Medical-Conversational", | |
| "prompt": "", | |
| "lora_r": 16, | |
| "lora_alpha": 16, | |
| "lora_target_modules": [ | |
| "q_proj", | |
| "k_proj", | |
| "v_proj", | |
| "o_proj", | |
| "gate_proj", | |
| "up_proj", | |
| "down_proj" | |
| ], | |
| "lora_dropout": 0.1, | |
| "lora_bias": "none", | |
| "model_type": "pretrain_vicuna", | |
| "max_txt_len": 160, | |
| "end_sym": "<|eot_id|>", | |
| "ckpt": "24_10_29_03_29_41_epoch160.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": { | |
| "vqa": { | |
| "data_type": "images", | |
| "build_info": { | |
| "storage": "dataset/Mix_llavamed_mimic_fientune" | |
| }, | |
| "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": 3e-05, | |
| "min_lr": 1e-05, | |
| "warmup_lr": 1e-06, | |
| "weight_decay": 0.05, | |
| "max_epoch": 50, | |
| "iters_per_epoch": 2100, | |
| "batch_size_train": 6, | |
| "batch_size_eval": 1, | |
| "num_workers": 4, | |
| "warmup_steps": 200, | |
| "seed": 42, | |
| "output_dir": "output/xraygpt_vqa_finetune", | |
| "amp": true, | |
| "resume_ckpt_path": null, | |
| "evaluate": false, | |
| "train_splits": [ | |
| "train" | |
| ], | |
| "device": "cuda", | |
| "world_size": 8, | |
| "dist_url": "env://", | |
| "distributed": true, | |
| "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": "Joycean0301/Llama-3.2-3B-Instruct-Medical-Conversational", | |
| "prompt": "", | |
| "lora_r": 16, | |
| "lora_alpha": 16, | |
| "lora_target_modules": [ | |
| "q_proj", | |
| "k_proj", | |
| "v_proj", | |
| "o_proj", | |
| "gate_proj", | |
| "up_proj", | |
| "down_proj" | |
| ], | |
| "lora_dropout": 0.1, | |
| "lora_bias": "none", | |
| "model_type": "pretrain_vicuna", | |
| "max_txt_len": 160, | |
| "end_sym": "<|eot_id|>", | |
| "ckpt": "24_10_29_03_29_41_epoch160.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": { | |
| "vqa": { | |
| "data_type": "images", | |
| "build_info": { | |
| "storage": "dataset/Mix_llavamed_mimic_fientune" | |
| }, | |
| "vis_processor": { | |
| "train": { | |
| "name": "blip2_image_train", | |
| "image_size": 224 | |
| } | |
| }, | |
| "text_processor": { | |
| "train": { | |
| "name": "blip_caption" | |
| } | |
| } | |
| } | |
| } | |
| } | |
| {"train_lr": "0.000029", "train_loss": "1.119681"} | |
| {"train_lr": "0.000030", "train_loss": "0.967288"} | |
| {"train_lr": "0.000030", "train_loss": "0.919890"} | |
| {"train_lr": "0.000030", "train_loss": "0.885236"} | |
| {"train_lr": "0.000030", "train_loss": "0.854326"} | |
| {"train_lr": "0.000029", "train_loss": "0.825763"} | |
| {"train_lr": "0.000029", "train_loss": "0.797093"} | |
| {"train_lr": "0.000029", "train_loss": "0.771083"} | |
| {"train_lr": "0.000029", "train_loss": "0.744359"} | |
| {"train_lr": "0.000028", "train_loss": "0.716633"} | |
| {"train_lr": "0.000028", "train_loss": "0.688450"} | |
| {"train_lr": "0.000027", "train_loss": "0.659950"} | |
| {"train_lr": "0.000027", "train_loss": "0.633209"} | |
| {"train_lr": "0.000027", "train_loss": "0.604965"} | |
| {"train_lr": "0.000026", "train_loss": "0.575196"} | |
| {"train_lr": "0.000026", "train_loss": "0.549392"} | |
| {"train_lr": "0.000025", "train_loss": "0.522796"} | |
| {"train_lr": "0.000025", "train_loss": "0.493527"} | |
| {"train_lr": "0.000024", "train_loss": "0.467043"} | |
| {"train_lr": "0.000023", "train_loss": "0.443777"} | |
| {"train_lr": "0.000023", "train_loss": "0.417607"} | |
| {"train_lr": "0.000022", "train_loss": "0.395030"} | |
| {"train_lr": "0.000022", "train_loss": "0.373894"} | |
| {"train_lr": "0.000021", "train_loss": "0.351892"} | |
| {"train_lr": "0.000020", "train_loss": "0.332837"} | |
| {"train_lr": "0.000020", "train_loss": "0.312894"} | |
| {"train_lr": "0.000019", "train_loss": "0.295802"} | |
| {"train_lr": "0.000018", "train_loss": "0.280947"} | |
| {"train_lr": "0.000018", "train_loss": "0.265249"} | |
| {"train_lr": "0.000017", "train_loss": "0.250452"} | |
| {"train_lr": "0.000017", "train_loss": "0.238245"} | |
| {"train_lr": "0.000016", "train_loss": "0.225438"} | |
| {"train_lr": "0.000015", "train_loss": "0.214808"} | |
| {"train_lr": "0.000015", "train_loss": "0.203942"} | |
| {"train_lr": "0.000014", "train_loss": "0.194547"} | |
| {"train_lr": "0.000014", "train_loss": "0.185850"} | |
| {"train_lr": "0.000013", "train_loss": "0.178339"} | |
| {"train_lr": "0.000013", "train_loss": "0.169570"} | |
| {"train_lr": "0.000013", "train_loss": "0.163180"} | |
| {"train_lr": "0.000012", "train_loss": "0.156634"} | |
| {"train_lr": "0.000012", "train_loss": "0.150682"} | |
| {"train_lr": "0.000011", "train_loss": "0.144779"} | |
| {"train_lr": "0.000011", "train_loss": "0.140201"} | |
| {"train_lr": "0.000011", "train_loss": "0.135027"} | |
| {"train_lr": "0.000011", "train_loss": "0.130042"} | |
| {"train_lr": "0.000010", "train_loss": "0.126120"} | |