IndexError: The shape of the mask [425] at index 0 does not match the shape of the indexed tensor [413, 768] at index 0

[lenijwp]

When I enter a long code snippet, I get an error like this: “IndexError: The shape of the mask [425] at index 0 does not match the shape of the indexed tensor [413, 768] at index 0”. Does this mean that the maximum input length supported by the model configuration is 413 tokens? I was wondering if the API could be further refined to support arbitrary length or automatic truncation.

[urialon]

Hi @lenijwp ,
Thank you for your interest in our work!

Our code supports unlimited inputs, by chunking long inputs and encoding each chunks separately.
However, 413 tokens do not even require this chunking, because the models' input size is 512.

Can you check whether there's a "newline" character at the end of the input?
If so, can you remove it and check again?
If not, can you provide us with the problematic inputs?

Thanks,
Uri

[lenijwp]

I checked as you suggested and removed the trailing whitespaces and newline characters from the input. The problem has been solved. Thank you for your timely and helpful assistance!

[urialon]

Thank you for reporting this! I'll make sure to strip all inputs on our side.

…

On Mon, Mar 6, 2023 at 21:36 Lenic Jiang ***@***.***> wrote: I checked as you suggested and removed the trailing whitespaces and newline characters from the input. The problem has been solved. Thank you for your timely and helpful assistance! — Reply to this email directly, view it on GitHub <#3 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADSOXMFM25KOPJOEGCP2BZTW22NLXANCNFSM6AAAAAAVR4DRTE> . You are receiving this because you commented.Message ID: ***@***.***>