I am new to transformers. I have fine-tuned my MT5 model in Google Colab and saved it with save_pretrained(PATH)
. However, when I tried to load the model with MT5ForConditionalGeneration.from_pretrained(PATH)
, Google Colab keep crashing with log error:
terminate called after throwing an instance of 'std::__ios_failure'
what(): basic_filebuf::underflow error reading the file: iostream error
I suspect that it was out-of-memory problem, but I have no clue how to fix it.