About train.py
lincong8722 opened this issue · 1 comments
Every time I start training, there are always mistakes like this
usage: train.py [-h] --name NAME [--dataset {CUB_200_2011,car,dog,nabirds}] [--data_root DATA_ROOT] [--model_type {ViT-B_16,ViT-B_32,ViT-L_16,ViT-L_32,ViT-H_14}] [--pretrained_dir PRETRAINED_DIR] [--pretrained_model PRETRAINED_MODEL] [--output_dir OUTPUT_DIR] [--img_size IMG_SIZE] [--train_batch_size TRAIN_BATCH_SIZE] [--eval_batch_size EVAL_BATCH_SIZE] [--eval_every EVAL_EVERY] [--learning_rate LEARNING_RATE] [--weight_decay WEIGHT_DECAY] [--num_steps NUM_STEPS] [--decay_type {cosine,linear}] [--warmup_steps WARMUP_STEPS] [--max_grad_norm MAX_GRAD_NORM] [--local_rank LOCAL_RANK] [--seed SEED] [--gradient_accumulation_steps GRADIENT_ACCUMULATION_STEPS] [--fp16] [--fp16_opt_level FP16_OPT_LEVEL] [--loss_scale LOSS_SCALE] [--smoothing_value SMOOTHING_VALUE] [--split SPLIT] [--slide_step SLIDE_STEP] train.py: error: the following arguments are required: --name
I don't know how to solve it. If you know , please tell me. Thank you！
It may be that the name is not specified.When training, you need to specify enough parameters