OSError: no file named pytorch_model.bin
#3
by ymoslem - opened
Hello!
I tried this code:
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("amazon/FalconLite", trust_remote_code=True)
It results in this error:
OSError: amazon/FalconLite does not appear to have a file named pytorch_model.bin, tf_model.h5, model.ckpt or flax_model.msgpack.
I am not using SageMaker. Just a regular GPU.
Thanks!
You'll need to add the model_basename as well -> gptq_model-4bit-128g.safetensors
However, it looks like generation_config.json is missing, so it won't work locally
Error:
OSError: amazon/FalconLite does not appear to have a file named generation_config.json. Checkout 'https://huggingface.co/amazon/FalconLite/main' for available files.
I guess they only made change in TGI codebase, so it won't run locally with scaledRoPE because there is no code change from the original Falcon repo. Please correct me if I am wrong and eager to learn.
Does it means there is no way to finetune it (again)?