bitnet

Large-scale Self-supervised Pre-training Across Tasks, Languages, and Modalities

Official implementation for BitVLA: 1-bit Vision-Language-Action Models for Robotics Manipulation (CoRL 2026, Spotlight)

A ternary, zero-heap tiny language model that runs inside a $2 microcontroller — bit-exact Python <-> C99 <-> Cortex-M3 (QEMU) parity. Apache-2.0.

A custom LLM inference engine for low-cost GPUs: NVIDIA cards from a 1 GB Quadro K420 to an RTX 3080, the AMD Radeon 780M and the CPU. Runs the Bonsai 2B to 27B models and Nemotron with an OpenAI-compatible server.

Running Microsoft's BitNet via Electron, React & Astro

CPU-optimized LLM inference engine (C)

BitNet-Stack - Run a small BitNet model on local machine with one Docker command, and chat in browser.

Implementing the BitNet model in Rust

Running Microsoft's BitNet inference framework via FastAPI, Uvicorn and Docker.

pure pytorch implementation of Microsoft's BitNet b1.58 2B4T

Foundational ternary-model inference & training library — CUDA + CPU, BitNet/TQ ternary kernels

All the important tools for BITNET/HNET programs like relay chat, xyzzy, YWAKEUP, IUCVTRAP, CLRSCRN

Official implementation of BitMamba-2. A scalable 1.58-bit State Space Model (Mamba-2 + BitNet) trained from scratch on 150B tokens. Includes JAX training code and high-performance C++ inference engine.

Run BitNet b1.58 ternary LLMs with WebGPU — in browsers and native apps

Software side of an LLM running inside a DRAM chip via charge-sharing PIM (BitNet b1.58 on DRAM-Bender silicon).

Ultra-lightweight C++ inference engine for BitMamba-2 (1.58-bit SSM). Runs 1B models on consumer CPUs at 50+ tok/s using <700MB RAM. No heavy dependencies.

BitNet: Learning-Based-Bit-Depth-Expansion

Portable C inference engine for BitMamba-2 models

Single-launch batch-1 decode engine for PrismML Bonsai 27B (ternary and 1-bit) on NVIDIA GPUs. 1.76x the vendor llama.cpp fork on H100, same outputs.

Rust bindings for bitnet.cpp based on llama-cpp-4