search

amazon-science / semimtr-text-recognition

Multimodal Semi-Supervised Learning for Text Recognition (SemiMTR)
Apache License 2.0
81 stars 12 forks source link

NaN in input tensor #4

Closed dikubab closed 2 years ago

dikubab commented 2 years ago

The language model gives NaN or Inf found in input tensor. train.txt Can help why it is failing to train on non English character?

aaberdam commented 2 years ago

Hi,

Thank you for the interest in our work. Please refer in this issue, to the changes that are required for training on non-English language. In particular, make sure that you created AmharicText_train.csv and AmharicText_eval.csv with the right character set.

Let me know if it works for you. Aviad

dikubab commented 2 years ago

Thank you for your prompt response. As per your suggestion, I have cleaned my train and valid text datasets and now it is working for me. I am closing the issue.