Do we have plan to make this model work with latest transformers?

#90
by Kaixuanliu - opened

When I use latest transformers to load this model, it returns error: "ImportError: cannot import name 'SlidingWindowCache' from 'transformers.cache_utils'", it seems the custom code has conflicts with latest transformers. After I modify related code, there are still a series of error. So can we update the remote code to make the model work?
Here is my env:

  • transformers 5.3.0
  • peft 0.18.1

And paste my test code:

from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("microsoft/Phi-4-multimodal-instruct", trust_remote_code=True, dtype="auto")
Kaixuanliu changed discussion status to closed

Sign up or log in to comment