size mismatch

I launch a training, stop it and want to continue it.

I set the source model as the model of the latest epoch and the resume from saved training state to the latest state folder. Everyelse are similar.

I got this issue:

RuntimeError: Error(s) in loading state_dict for UNet2DConditionModel:
size mismatch for down_blocks.0.attentions.0.proj_in.weight: copying a param with shape torch.Size([320, 320]) from checkpoint, the shape in current model is torch.Size([320, 320, 1, 1]).
size mismatch for down_blocks.0.attentions.0.transformer_blocks.0.attn2.to_k.weight: copying a param with shape torch.Size([320, 1024]) from checkpoint, the shape in current model is torch.Size([320, 768]).
size mismatch for down_blocks.0.attentions.0.transformer_blocks.0.attn2.to_v.weight: copying a param with shape torch.Size([320, 1024]) from checkpoint, the shape in current model is torch.Size([320, 768]).

Start by reproducing the training-resume flow described: use the latest-epoch model as the source and the latest saved training-state folder as the resume state. Compare the model and resume configurations against the reported UNet2DConditionModel state_dict shapes; done means resuming completes without the listed size-mismatch errors.

Written by the indexing model from the issue text.