accelerate 0.30.0

Accelerate

Stars: 7049, Watchers: 7049, Forks: 808, Open Issues: 136

The huggingface/accelerate repo was created 3 years ago and the last code push was 35 minutes ago.
The project is extremely popular with a mindblowing 7049 github stars!

How to Install accelerate

You can install accelerate using pip

pip install accelerate

or add it to a project with poetry

poetry add accelerate

Package Details

Author

The HuggingFace team

License

Apache

Homepage

https://github.com/huggingface/accelerate

PyPi:

https://pypi.org/project/accelerate/

GitHub Repo:

https://github.com/huggingface/accelerate

Classifiers

• Scientific/Engineering/Artificial Intelligence

Related Packages

Errors

A list of common accelerate errors.

Code Examples

Here are some accelerate code examples and snippets.

GitHub Issues

The accelerate package has 136 open issues on GitHub

• get_balanced_memory() ignores GPU:0 entirely if low_zero=True
• Accelerate test failed on 2 nodes with 8x A800
• Is CPU-offloading function in accelerate same with deepSpeed?
• example/cv_example.py fp8 error
• If save_state support automatic_checkpoint_naming, then load_state should be able to load latest checkpoint
• from_pretrained crash if device_map contains only "disk"
• device_map={"": "cuda"} raises Can't pickle <function Embedding.forward when using multiprocessing, while loading with PyTorch context with torch.device("cuda"): goes fine
• Issue loading tiiuae/falcon-40b with device_map="auto"
• "ImportError: Megatron is not installed. please build it from source." when launching accelerate
• Is there some way to get a FLOPS profile of a model trained with Accelerate?
• AttributeError: 'BitsAndBytesConfig' object has no attribute 'load_in_4bit'
• AttributeError: 'T5Attention' object has no attribute 'relative_attention_bias' when using load_checkpoint_and_dispatch to load T5-11b
• CUDA error: device-side assert triggered for LED Model using Accelerate, works outside of accelerate and setup works with other models
• _is_accelerate_prepared issue
• cpu memory leak in recent version

See more issues on GitHub