how to solve this error

[its-athira]

[j-min]

Hi @its-athira, thanks for your interest. But there's not enough information from the screenshot, as I don't know your environment, package version, etc. Could you please share more information about your environment, your edits and where the error comes from?

[its-athira]

Hi, This is the code I used https://colab.research.google.com/drive/1lQREId2RY9n02YllV_95no_V8bahVkME?authuser=1#scrollTo=Rdl7I779z3wm

…

On Sun, 30 Jul 2023, 1:41 am Jaemin Cho, ***@***.***> wrote: Hi @its-athira <https://github.com/its-athira>, thanks for your interest. But there's not enough information from the screenshot, as I don't know your environment, package version, etc. Could you please share more information about your environment, your edits and where the error comes from? — Reply to this email directly, view it on GitHub <#34 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BBAYU6KX63X24NGWQWXXIR3XSVU7FANCNFSM6AAAAAA24T5H6I> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[j-min]

I don't have access to the linked colab. Would you please make it public?

[its-athira]

Sorry, I changed it . https://colab.research.google.com/drive/1lQREId2RY9n02YllV_95no_V8bahVkME#scrollTo=Rdl7I779z3wm&uniqifier=1

…

On Sun, Jul 30, 2023 at 3:52 PM Jaemin Cho ***@***.***> wrote: I don't have access to the linked colab. Would you please make it public? — Reply to this email directly, view it on GitHub <#34 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BBAYU6PWE7A7HXCYFDDXLC3XSYYWZANCNFSM6AAAAAA24T5H6I> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[j-min]

You're using the latest transformers version==4.31.0, but the current codebase is based on 4.2.1, as mentioned in requirements.txt

Here (https://github.com/j-min/VL-T5/blob/HF-4.31.0/VL-T5/src/modeling_t5.py), I'm working on a version compatible to 4.31.0, which you might wanna check. For tokenization, I suggest to use VLT5Tokenizer instead of VLT5TokenizerFast for now (check #21)

[its-athira]

Thanks, I'll check it out.

…

On Mon, Jul 31, 2023 at 8:43 PM Jaemin Cho ***@***.***> wrote: You're using the latest transformers version==4.31.0, but the current codebase is based on 4.2.1, as mentioned in requirements.txt Here ( https://github.com/j-min/VL-T5/blob/HF-4.31.0/VL-T5/src/modeling_t5.py), I'm working on a version compatible to 4.31.0, which you might wanna check. For tokenization, I suggest to use VLT5Tokenizer instead of VLT5TokenizerFast for now (check #21 <#21>) — Reply to this email directly, view it on GitHub <#34 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BBAYU6PUFIV3FYOGKAW5HZ3XS7DQLANCNFSM6AAAAAA24T5H6I> . You are receiving this because you were mentioned.Message ID: ***@***.***>