Foundations --> Transformers about madewithml

[knosing]

@shashankvasisht
Hi, I just got the same error and found the issue.
This error pops up if you are using the newer 4.0 ver of transformers.
So one solution is what @GokuMohandas recommended and the other would be to add the argument return_dict=False when you pass inputs to your model:
seq, pool = self.transformer(input_ids=ids, attention_mask=masks, return_dict=False)

Rest all works fine, cheers!

[GokuMohandas]

HI @shashankvasisht, based on the error messages, it looks like you're running this on a local notebook instead of running on Google colab directly? If not, make sure you have the exact version of the transformers (3.0.2) package (their future releases might be not be backwards compatible) and other packages as well (use pip freeze inside the notebook to check). I just reran the notebook and it ran successfully to completion.