← Back to Repositories Hub Language License TimDettmers/bitsandbytes 8-bit and 4-bit CUDA functions for PyTorch. Maintainer: TimDettmers Last Reviewed: 2026-06-22 Next Review: 2026-12-31 Editorial Owner: AI Delivery Hub Editorial Description & Purpose 8-bit and 4-bit CUDA functions for PyTorch. Architecture Summary Built as a modular framework in Python, leveraging asynchronous processing and event loops to orchestrate LLM requests. Core Use Cases Model quantization, training LoRA/QLoRA on consumer GPUs. Alternatives Evaluated Alternative open-source orchestration packages 💻 View on GitHub 📖 View Documentation