Getting zero accuracy on the ImageNet dataset.

[shahidalihakro]

I don't know why accuracy is 0 even I have set the path of pre-trained weights parser.add_argument('--finetune', default='/mnt/d/My experimets on mamba/Vim-main/vim_t_midclstok_76p1acc.pth', help='finetune from checkpoint')
parser.add_argument('--resume', default='/mnt/d/My experimets on mamba/Vim-main/vim_t_midclstok_76p1acc.pth', help='resume from checkpoint')

Not using distributed mode
Namespace(gpu=0, batch_size=64, epochs=300, bce_loss=False, unscale_lr=False, model='vim_tiny_patch16_224_bimambav2_final_pool_mean_abs_pos_embed_with_midclstok_div2', input_size=224, drop=0.0, drop_path=0.1, model_ema=True, model_ema_decay=0.99996, model_ema_force_cpu=False, opt='adamw', opt_eps=1e-08, opt_betas=None, clip_grad=None, momentum=0.9, weight_decay=0.05, sched='cosine', lr=0.0005, lr_noise=None, lr_noise_pct=0.67, lr_noise_std=1.0, warmup_lr=1e-06, min_lr=1e-05, decay_epochs=30, warmup_epochs=5, cooldown_epochs=10, patience_epochs=10, decay_rate=0.1, color_jitter=0.3, aa='rand-m9-mstd0.5-inc1', smoothing=0.1, train_interpolation='bicubic', repeated_aug=True, train_mode=True, ThreeAugment=False, src=False, reprob=0.25, remode='pixel', recount=1, resplit=False, mixup=0.8, cutmix=1.0, cutmix_minmax=None, mixup_prob=1.0, mixup_switch_prob=0.5, mixup_mode='batch', teacher_model='regnety_160', teacher_path='', distillation_type='none', distillation_alpha=0.5, distillation_tau=1.0, cosub=False, finetune='/mnt/d/My experimets on mamba/Vim-main/vim_t_midclstok_76p1acc.pth', attn_only=False, data_path='/mnt/d/My experimets on mamba/Vim-main/imagenet', data_set='IMNET', inat_category='name', output_dir='', device='cuda', seed=0, resume='/mnt/d/My experimets on mamba/Vim-main/vim_t_midclstok_76p1acc.pth', start_epoch=0, eval='False', eval_crop_ratio=0.875, dist_eval=False, num_workers=1, pin_mem=True, distributed=False, world_size=1, dist_url='env://', if_amp=False, if_continue_inf=False, if_nan2num=False, if_random_cls_token_position=False, if_random_token_rank=False, local_rank=0)
Creating model: vim_tiny_patch16_224_bimambav2_final_pool_mean_abs_pos_embed_with_midclstok_div2
number of params: 7148008

[shahidalihakro]

I found that I was testing the code on the wrong dataset. After switching to the correct dataset (ImageNet1K validation set), everything is working as expected now.