GATECH-EIC/ShiftAddLLM

Official Pytorch repository for Extreme Compression of Large Language Models via Additive Quantization https://arxiv.org/pdf/2401.06118.pdf and PV-Tuning: Beyond Straight-Through Estimation for Extreme LLM Compression https://arxiv.org/abs/2405.14852

An easy-to-use LLMs quantization package with user-friendly apis, based on GPTQ algorithm.

Gumbel-Softmax post-training quantization for LLMs (1–3 bit scalar, INT/GGUF-compatible).

Accessible large language models via k-bit quantization for PyTorch.

LLM model quantization (compression) toolkit with HW acceleration support for Nvidia, AMD, Intel GPU and Intel/AMD/Apple CPU via HF, vLLM, and SGLang.

Official implementation of LittleBit (NeurIPS 2025) and its follow-up LittleBit-2 (ICML 2026)

Statistically-Lossless Quantization of Large Language Models

Design hardware-friendly model architectures and migrate existing LLMs with minimal performance loss

PrismQuant: Optimal Null-Space Rotations for Grouped Quantizers