# `bitsandbytes`
[](https://pepy.tech/project/bitsandbytes) [](https://pepy.tech/project/bitsandbytes) [](https://pepy.tech/project/bitsandbytes)
The `bitsandbytes` library is a lightweight Python wrapper around CUDA custom functions, in particular 8-bit optimizers, matrix multiplication (LLM.int8()), and 8 & 4-bit quantization functions.
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.
There are ongoing efforts to support further hardware backends, i.e. Intel CPU + GPU, AMD GPU, Apple Silicon. Windows support is quite far along and is on its way as well.
**Please head to the official documentation page:**
**[https://huggingface.co/docs/bitsandbytes/main](https://huggingface.co/docs/bitsandbytes/main)**
## `bitsandbytes` multi-backend _alpha_ release is out!
🚀 Big news! After months of hard work and incredible community contributions, we're thrilled to announce the **bitsandbytes multi-backend _alpha_ release**! 💥
Now supporting:
- 🔥 **AMD GPUs** (ROCm)
- ⚡ **Intel CPUs** & **GPUs**
We’d love your early feedback! 🙏
👉 [Instructions for your `pip install` here](https://huggingface.co/docs/bitsandbytes/main/en/installation#multi-backend)
We're super excited about these recent developments and grateful for any constructive input or support that you can give to help us make this a reality (e.g. helping us with the upcoming Apple Silicon backend or reporting bugs). BNB is a community project and we're excited for your collaboration 🤗
## 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.
Raw data
{
"_id": null,
"home_page": "https://github.com/bitsandbytes-foundation/bitsandbytes",
"name": "bitsandbytes",
"maintainer": null,
"docs_url": null,
"requires_python": null,
"maintainer_email": null,
"keywords": "gpu optimizers optimization 8-bit quantization compression",
"author": "Tim Dettmers",
"author_email": "dettmers@cs.washington.edu",
"download_url": null,
"platform": null,
"description": "# `bitsandbytes`\n\n[](https://pepy.tech/project/bitsandbytes) [](https://pepy.tech/project/bitsandbytes) [](https://pepy.tech/project/bitsandbytes)\n\nThe `bitsandbytes` library is a lightweight Python wrapper around CUDA custom functions, in particular 8-bit optimizers, matrix multiplication (LLM.int8()), and 8 & 4-bit quantization functions.\n\nThe 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.\n\nThere are ongoing efforts to support further hardware backends, i.e. Intel CPU + GPU, AMD GPU, Apple Silicon. Windows support is quite far along and is on its way as well.\n\n**Please head to the official documentation page:**\n\n**[https://huggingface.co/docs/bitsandbytes/main](https://huggingface.co/docs/bitsandbytes/main)**\n\n## `bitsandbytes` multi-backend _alpha_ release is out!\n\n\ud83d\ude80 Big news! After months of hard work and incredible community contributions, we're thrilled to announce the **bitsandbytes multi-backend _alpha_ release**! \ud83d\udca5\n\nNow supporting:\n- \ud83d\udd25 **AMD GPUs** (ROCm)\n- \u26a1 **Intel CPUs** & **GPUs**\n\nWe\u2019d love your early feedback! \ud83d\ude4f\n\n\ud83d\udc49 [Instructions for your `pip install` here](https://huggingface.co/docs/bitsandbytes/main/en/installation#multi-backend)\n\nWe're super excited about these recent developments and grateful for any constructive input or support that you can give to help us make this a reality (e.g. helping us with the upcoming Apple Silicon backend or reporting bugs). BNB is a community project and we're excited for your collaboration \ud83e\udd17\n\n## License\n\n`bitsandbytes` is MIT licensed.\n\nWe thank Fabio Cannizzo for his work on [FastBinarySearch](https://github.com/fabiocannizzo/FastBinarySearch) which we use for CPU quantization.\n",
"bugtrack_url": null,
"license": "MIT",
"summary": "k-bit optimizers and matrix multiplication routines.",
"version": "0.45.0",
"project_urls": {
"Homepage": "https://github.com/bitsandbytes-foundation/bitsandbytes"
},
"split_keywords": [
"gpu",
"optimizers",
"optimization",
"8-bit",
"quantization",
"compression"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "999af41d252bf8b0bc5969b4dce1274cd04b7ddc541de1060dd27eca680bc1b2",
"md5": "20231a257c6fe773c8b0327a36e5c84a",
"sha256": "0f0323de1ff1fdf8383e79bdad1283516a4c05a6fd2b44a363bf4e059422305b"
},
"downloads": -1,
"filename": "bitsandbytes-0.45.0-py3-none-manylinux_2_24_x86_64.whl",
"has_sig": false,
"md5_digest": "20231a257c6fe773c8b0327a36e5c84a",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": null,
"size": 69084267,
"upload_time": "2024-12-05T16:05:19",
"upload_time_iso_8601": "2024-12-05T16:05:19.724762Z",
"url": "https://files.pythonhosted.org/packages/99/9a/f41d252bf8b0bc5969b4dce1274cd04b7ddc541de1060dd27eca680bc1b2/bitsandbytes-0.45.0-py3-none-manylinux_2_24_x86_64.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "1e90a2bbb9b5f997b9c9aa9c15ee4adf553ee71053bb942f89fd48d920a1aa9d",
"md5": "ba1166c87dd842a6689f46befb3f15f5",
"sha256": "ebbf96e0ecb466716a65ecdeaef3fa1983575447b9ab66b74e5211892507c6ff"
},
"downloads": -1,
"filename": "bitsandbytes-0.45.0-py3-none-win_amd64.whl",
"has_sig": false,
"md5_digest": "ba1166c87dd842a6689f46befb3f15f5",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": null,
"size": 68520043,
"upload_time": "2024-12-05T16:05:25",
"upload_time_iso_8601": "2024-12-05T16:05:25.805201Z",
"url": "https://files.pythonhosted.org/packages/1e/90/a2bbb9b5f997b9c9aa9c15ee4adf553ee71053bb942f89fd48d920a1aa9d/bitsandbytes-0.45.0-py3-none-win_amd64.whl",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-12-05 16:05:19",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "bitsandbytes-foundation",
"github_project": "bitsandbytes",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"lcname": "bitsandbytes"
}
JSON
