bitsandbytes Files

Accessible large language models via k-bit quantization for PyTorch

This is an exact mirror of the bitsandbytes project, hosted at https://github.com/bitsandbytes-foundation/bitsandbytes. SourceForge is not affiliated with bitsandbytes.

The interactive file manager requires Javascript. Please enable it or use sftp or scp.
You may still browse the files here.

Name	Modified	Size	InfoDownloads / Week
Parent folder
0.48.1 source code.tar.gz	2025-10-02	302.7 kB	0
0.48.1 source code.zip	2025-10-02	390.5 kB	0
README.md	2025-10-02	571 Bytes	0
Totals: 3 Items		693.8 kB	0

This release fixes a regression introduced in 0.48.0 related to LLM.int8(). This issue caused poor inference results with pre-quantized checkpoints in HF transformers.

What's Changed

Add trove-classifiers requirement to pyproject.toml by @ccoulombe in https://github.com/bitsandbytes-foundation/bitsandbytes/pull/1774
Fix regression in 8bit parameter device movement by @matthewdouglas in https://github.com/bitsandbytes-foundation/bitsandbytes/pull/1776

Full Changelog: https://github.com/bitsandbytes-foundation/bitsandbytes/compare/0.48.0...0.48.1

Source: README.md, updated 2025-10-02

Other Useful Business Software

Gemini 3 and 200+ AI Models on One Platform

Access Google's best plus Claude, Llama, and Gemma. Fine-tune and deploy from one console.

Build generative AI apps with Vertex AI. Switch between models without switching platforms.

Start Free

Easily Host LLMs and Web Apps on Cloud Run

Run everything from popular models with on-demand NVIDIA L4 GPUs to web apps without infrastructure management.

Run frontend and backend services, batch jobs, host LLMs, and queue processing workloads without the need to manage infrastructure. Cloud Run gives you on-demand GPU access for hosting LLMs and running real-time AI—with 5-second cold starts and automatic scale-to-zero so you only pay for actual usage. New customers get $300 in free credit to start.

Try Cloud Run Free

Featured

Gemini 3 and 200+ AI Models on One Platform

Access Google's best plus Claude, Llama, and Gemma. Fine-tune and deploy from one console.

Build generative AI apps with Vertex AI. Switch between models without switching platforms.

Start Free

Recommended Projects

ChatGLM-6B
ChatGLM-6B: An Open Bilingual Dialogue Language Model
AutoGPTQ
An easy-to-use LLMs quantization package with user-friendly apis
Xtuner
A Next-Generation Training Engine Built for Ultra-Large MoE Models
Mixtral offloading
Run Mixtral-8x7B models in Colab or consumer desktops
xLSTM
Neural Network architecture based on ideas of the original LSTM

bitsandbytes Files

Accessible large language models via k-bit quantization for PyTorch

Get an email when there's a new version of bitsandbytes

What's Changed