bitsandbytes

`bitsandbytes` enables accessible large language models via k-bit quantization for PyTorch. We provide three main features for dramatically reducing memory consumption for inference and training: * 8-bit optimizers uses block-wise quantization to maintain 32-bit performance at a small fraction of the memory cost. * LLM.int8() or 8-bit quantization enables large language model inference with only half the required memory and without any performance degradation. This method is based on vector-wise quantization to quantize most features to 8-bits and separately treating outliers with 16-bit matrix multiplication. * QLoRA or 4-bit quantization enables large language model training with several memory-saving techniques that don't compromise performance. This method quantizes a model to 4-bits and inserts a small set of trainable low-rank adaptation (LoRA) weights to allow training. The library includes quantization primitives for 8-bit & 4-bit operations, through `bitsandbytes.nn.Linear8bitLt` and `bitsandbytes.nn.Linear4bit` and 8-bit optimizers through `bitsandbytes.optim` module. ## System Requirements bitsandbytes has the following minimum requirements for all platforms: * Python 3.9+ * [PyTorch](https://pytorch.org/get-started/locally/) 2.2+ * _Note: While we aim to provide wide backwards compatibility, we recommend using the latest version of PyTorch for the best experience._ #### Accelerator support:

Platform	Accelerator	Hardware Requirements	Support Status
🐧 Linux, glibc >= 2.24
x86-64	◻️ CPU	AVX2	〰️ Partial Support
	🟩 NVIDIA GPU `cuda`	SM50+ minimum SM75+ recommended	✅ Full Support
	🟥 AMD GPU `cuda`	CDNA: gfx90a, gfx942 RDNA: gfx1100, gfx1200	🚧 In Development
	🟦 Intel GPU `xpu`	Data Center GPU Max Series Arc A-Series (Alchemist) Arc B-Series (Battlemage)	🚧 In Development
	🟪 Intel Gaudi `hpu`	Gaudi1, Gaudi2, Gaudi3	🚧 In Development
aarch64	◻️ CPU		〰️ Partial Support
	🟩 NVIDIA GPU `cuda`	SM75, SM80, SM90, SM100	✅ Full Support
🪟 Windows 11 / Windows Server 2019+
x86-64	◻️ CPU	AVX2	〰️ Partial Support
	🟩 NVIDIA GPU `cuda`	SM50+ minimum SM75+ recommended	✅ Full Support
	🟦 Intel GPU `xpu`	Arc A-Series (Alchemist) Arc B-Series (Battlemage)	🚧 In Development
🍎 macOS 13.1+
arm64	◻️ CPU	Apple M1+	🚧 In Development
	⬜ Metal `mps`	Apple M1+	🚧 In Development

## :book: Documentation * [Official Documentation](https://huggingface.co/docs/bitsandbytes/main) * 🤗 [Transformers](https://huggingface.co/docs/transformers/quantization/bitsandbytes) * 🤗 [Diffusers](https://huggingface.co/docs/diffusers/quantization/bitsandbytes) * 🤗 [PEFT](https://huggingface.co/docs/peft/developer_guides/quantization#quantize-a-model) ## :heart: Sponsors The continued maintenance and development of `bitsandbytes` is made possible thanks to the generous support of our sponsors. Their contributions help ensure that we can keep improving the project and delivering valuable updates to the community.

## License `bitsandbytes` is MIT licensed. We thank Fabio Cannizzo for his work on [FastBinarySearch](https://github.com/fabiocannizzo/FastBinarySearch) which we use for CPU quantization. ## How to cite us If you found this library useful, please consider citing our work: ### QLoRA ```bibtex @article{dettmers2023qlora, title={Qlora: Efficient finetuning of quantized llms}, author={Dettmers, Tim and Pagnoni, Artidoro and Holtzman, Ari and Zettlemoyer, Luke}, journal={arXiv preprint arXiv:2305.14314}, year={2023} } ``` ### LLM.int8() ```bibtex @article{dettmers2022llmint8, title={LLM.int8(): 8-bit Matrix Multiplication for Transformers at Scale}, author={Dettmers, Tim and Lewis, Mike and Belkada, Younes and Zettlemoyer, Luke}, journal={arXiv preprint arXiv:2208.07339}, year={2022} } ``` ### 8-bit Optimizers ```bibtex @article{dettmers2022optimizers, title={8-bit Optimizers via Block-wise Quantization}, author={Dettmers, Tim and Lewis, Mike and Shleifer, Sam and Zettlemoyer, Luke}, journal={9th International Conference on Learning Representations, ICLR}, year={2022} } ```