The shape of checkpoint and the shape of model do not match

[yulongwanxue]

Thank you for your wonderful research！
but,when I was doing a zero-shot test, I got this error.

just as：
size mismatch for vision_encoder.layers.2.blocks.2.attn.relative_position_bias_table: copying a param with shape torch.Size([169, 16]) from checkpoint, the shape in current model is torch.Size([529, 16])
size mismatch for vision_encoder.layers.2.blocks.9.attn_mask: copying a param with shape torch.Size([4, 49, 49]) from checkpoint, the shape in current model is torch.Size([4, 144, 144])

I followed your code，but get this problem。Do you know the answer to the question？