How to use stefanj0/t5gemma-math-corrector with PEFT:
from peft import PeftModel from transformers import AutoModelForSeq2SeqLM base_model = AutoModelForSeq2SeqLM.from_pretrained("google/t5gemma-s-s-ul2-it") model = PeftModel.from_pretrained(base_model, "stefanj0/t5gemma-math-corrector")
How to use stefanj0/t5gemma-math-corrector with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("stefanj0/t5gemma-math-corrector", device_map="auto")
What is a pickle import?
How to fix it?
