intel / neural-compressor

SOTA low-bit LLM quantization (INT8/FP8/MXFP8/INT4/MXFP4/NVFP4) & sparsity; leading model compression techniques on PyTorch, TensorFlow, and ONNX Runtime

openvinotoolkit / nncf

Neural Network Compression Framework for enhanced OpenVINO™ inference