Problem initializing weights using from_pretrained() method

#1
by Kelvin99 - opened

When I use this line of code:
model = AutoModelForSeq2SeqLM.from_pretrained('hkunlp/T5_base_finetune_all_tasks_2upsample2')
It returns this error:
Some weights of T5ForConditionalGeneration were not initialized from the model checkpoint
and then goes on to say all of T5's layers are randomly initialized.

Is this model not meant for T5ForConditionalGeneration?

NLP Group of The University of Hong Kong org

Hi, could you check about whether the version of huggingface transformers is the same with on in UnifiedSKG?

Sign up or log in to comment