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 3.10+ * [PyTorch](https://pytorch.org/get-started/locally/) 2.4+ * _Note: While we aim to provide wide backwards compatibility, we recommend using the latest version of PyTorch for the best experience._ #### Accelerator support: Note: this table reflects the status of the current development branch. For the latest stable release, see the [document in the 0.49.2 tag](https://github.com/bitsandbytes-foundation/bitsandbytes/blob/0.49.2/README.md#accelerator-support). ##### Legend: 🚧 = In Development, 〰️ = Partially Supported, ✅ = Supported, 🐢 = Slow Implementation Supported, ❌ = Not Supported

Platform	Accelerator	Hardware Requirements	LLM.int8()	QLoRA 4-bit	8-bit Optimizers
🐧 Linux, glibc >= 2.24
x86-64	◻️ CPU	Minimum: AVX2 Optimized: AVX512F, AVX512BF16	✅	✅	✅
	🟩 NVIDIA GPU `cuda`	SM60+ minimum SM75+ recommended	✅	✅	✅
	🟥 AMD GPU `cuda`	CDNA: gfx90a, gfx942, gfx950 RDNA: gfx1100, gfx1101, gfx1102, gfx1103, gfx1150, gfx1151, gfx1152, gfx1153, gfx1200, gfx1201	✅	✅	✅
	🟦 Intel GPU `xpu`	Data Center GPU Max Series Arc A-Series (Alchemist) Arc B-Series (Battlemage)	✅	✅	✅
	🟪 Intel Gaudi `hpu`	Gaudi2, Gaudi3	✅	〰️	❌
aarch64	◻️ CPU		✅	✅	✅
	🟩 NVIDIA GPU `cuda`	SM75+	✅	✅	✅
🪟 Windows 11 / Windows Server 2022+
x86-64	◻️ CPU	AVX2	✅	✅	✅
	🟩 NVIDIA GPU `cuda`	SM60+ minimum SM75+ recommended	✅	✅	✅
	🟥 AMD GPU `cuda`	RDNA: gfx1100, gfx1101, gfx1102, gfx1150, gfx1151, gfx1200, gfx1201	✅	✅	✅
	🟦 Intel GPU `xpu`	Arc A-Series (Alchemist) Arc B-Series (Battlemage)	✅	✅	✅
arm64	◻️ CPU		✅	✅	✅
🍎 macOS 14+
arm64	◻️ CPU	Apple M1+	✅	✅	✅
	⬜ Metal `mps`	Apple M1+	🐢	🐢	🚧

## :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. ## 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} } ```