Hi, I'm trying your new model to reproduce the result, but your released model's shape is different from the released code.
I downloaded all of your checkpoints, but shell script shows these error messages as follows.
RuntimeError: Error(s) in loading state_dict for AMRBartForConditionalGeneration:
size mismatch for final_logits_bias: copying a param with shape torch.Size([1, 53631]) from checkpoint, the shape in current model is torch.Size([1, 53587]).
size mismatch for model.shared.weight: copying a param with shape torch.Size([53631, 1024]) from checkpoint, the shape in current model is torch.Size([53587, 1024]).
size mismatch for model.encoder.embed_tokens.weight: copying a param with shape torch.Size([53631, 1024]) from checkpoint, the shape in current model is torch.Size([53587, 1024]).
size mismatch for model.decoder.embed_tokens.weight: copying a param with shape torch.Size([53631, 1024]) from checkpoint, the shape in current model is torch.Size([53587, 1024]).