Valid mask not found when train on mini image1k dataset

[oggyfaker]

Hi authors, i got some bug with this logging 'Valid mask not found, decreasing acceptable-regions' when i use the small part of image1k full. And when i start train command , the code is stuck with self._sample_block_mask in line 144 (multiblock.py)

My structure directory like this. And my config model is below. How can i fix this issue , please let know. Thanks very much !!!

` data: batch_size: 128 color_jitter_strength: 0.0 crop_scale:

• 0.3
• 1.0 crop_size: 32 image_folder: /home/oggy/Desktop/DINO_ISIC_2024/DINOv2_TryOn/ijepa/data/ num_workers: 12 pin_mem: true root_path: /home/oggy/Desktop/DINO_ISIC_2024/DINOv2_TryOn/ijepa use_color_distortion: false use_gaussian_blur: false use_horizontal_flip: false logging: folder: /home/oggy/Desktop/DINO_ISIC_2024/DINOv2_TryOn/ijepa/logs/vits14_224_bs_128_ep_100 write_tag: jepa mask: allow_overlap: false aspect_ratio:
• 0.75
• 1.5 enc_mask_scale:
• 0.85
• 1.0 min_keep: 10 num_enc_masks: 1 num_pred_masks: 4 patch_size: 14 pred_mask_scale:
• 0.15
• 0.2 meta: copy_data: false load_checkpoint: false model_name: vit_small pred_depth: 8 pred_emb_dim: 192 read_checkpoint: null use_bfloat16: false optimization: ema:
• 0.996
• 1.0 epochs: 100 final_lr: 1.0e-06 final_weight_decay: 0.4 ipe_scale: 1.0 lr: 0.001 start_lr: 0.0002 warmup: 40 weight_decay: 0.04 `

[oggyfaker]

sorry my fault , change crop_size to 224 can fix the problem.