Transformers 4.57.1 not working with torch < 2.3 #41960
Description
System Info
In the Readme it is said
Transformers works with Python 3.9+, and PyTorch 2.1+.
I use torch 2.2.2 on Mac Sonoma 14.7 (Intel) and I get the following error:
inputs = processor.apply_chat_template(
File "/home/venv310/lib/python3.10/site-packages/transformers/utils/deprecation.py", line 172, in wrapped_func
return func(*args, **kwargs)
File "/home/venv310/lib/python3.10/site-packages/transformers/utils/deprecation.py", line 172, in wrapped_func
return func(*args, **kwargs)
File "/home/venv310/lib/python3.10/site-packages/transformers/processing_utils.py", line 1705, in apply_chat_template
out = self(
File "/home/venv310/lib/python3.10/site-packages/transformers/models/internvl/processing_internvl.py", line 229, in __call__
video_inputs = self.video_processor(videos=videos, **video_kwargs)
File "/home/venv310/lib/python3.10/site-packages/transformers/video_processing_utils.py", line 206, in __call__
return self.preprocess(videos, **kwargs)
File "/home/venv310/lib/python3.10/site-packages/transformers/video_processing_utils.py", line 387, in preprocess
preprocessed_videos = self._preprocess(videos=videos, **kwargs)
File "/home/venv310/lib/python3.10/site-packages/transformers/models/internvl/video_processing_internvl.py", line 126, in _preprocess
stacked_videos = self.resize(stacked_videos, size=size, interpolation=interpolation)
File "/home/venv310/lib/python3.10/site-packages/transformers/image_processing_utils_fast.py", line 361, in resize
if torch.compiler.is_compiling() and is_rocm_platform():
AttributeError: module 'torch.compiler' has no attribute 'is_compiling''is_compiling' NOT exist in <= 2.2.2
https://github.com/pytorch/pytorch/blob/v2.2.2/torch/compiler/__init__.py
Who can help?
@zucchini-nlp @gante @stevhliu
Information
- The official example scripts
- My own modified scripts
Tasks
- An officially supported task in the
examplesfolder (such as GLUE/SQuAD, ...) - My own task or dataset (give details below)
Reproduction
Steps to reproduce:
- Install latest transformers version and torch 2.2.2
- Run video/image inference with InternVL3-1B model
Expected behavior
Run w/o errors with Torch 2.2.2