mae · Issues· 133 open

TypeError: __init__() got an unexpected keyword argument 'qk_scale'

Use your own pre-trained model for downstream segmentation

Confusion over the naming in pos_embed

MAE is available in HuggingFace Transformers (both PyTorch and TF)

patchify and unpatchify

Layer-wise learning rate decay (lrd) not applied correctly due to missing lr parameter in param groups

Colab execution does not work

mask token is shared

Customize masking

Is the visualization result normal?

Linear probing weights

Reconstructed image darkening

Dataset for "Fine-Tuning Pre-trained MAE for classification"

Cannot load the checkpoints

Doesn't get the result of visual demo