low-precision

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

Low Precision Arithmetic Simulation in PyTorch

Stable and Efficient Reinforcement Learning for Trillion-Parameter LLMs

A script to convert floating-point CNN models into generalized low-precision ShiftCNN representation

Low Precision(quantized) Yolov5

CUDA/HIP header-only library for low-precision (16 bit, 8 bit) and vectorized GPU kernel development

JAX Scalify: end-to-end scaled arithmetics

Code for DNN feature map compression paper

Low Precision Floating Point Types in Fortran

Open, bit-reproducible low-precision number format (b-posit / AI-Posit) for AI inference — 30–50% less memory, with results identical to the bit on any GPU, CPU, or RISC-V. Reference + conformance suite + CORE-ET RTL

Library of specialized algebra and performance primitives

超度量数系技术栈（SGN / HC）：基于超度量空间的自定义数值类型系统，C11/C++ 实现。含完整 ABI 主线（v0.1）与神经网络量化推理扩展（v0.2 SBE Conv2d/VNNI），已完结。Apache-2.0。Hypermetric codec stack: custom ultrametric number system in C, with neural-network quantization inference extensions (SBE, VNNI, Conv2d). Completed, archived.

NVFP4-DiT: Theory and Practice of 4-Bit Low-Precision Training and Inference for Audio-Guided Video Diffusion Transformers

Bounded Posits at IEEE Tensor-Core Throughput on Commodity NVIDIA Blackwell, with Bit-Exact Reproducibility

整数量化的严格数学验证：随机舍入、误差反馈、位拆分、精度分配、噪声整形、梯度恒等式、网络属性，NumPy/PyTorch 双库互证，开箱即跑。Independent, reproducible, dual-library (NumPy + PyTorch) verification of integer-quantization mathematics. Clone and run.

Implemented post-training quantisation (PTQ) on transformer-based reasoning models using 8-bit and 4-bit weight quantisation (INT8, INT4) with frameworks like PyTorch and Hugging Face Transformers. Leveraged libraries such as bitsandbytes to reduce model size and accelerate inference, while evaluating performance degradation on reasoning tasks. Com

SGN（Structured Gradient Network）：以整数/量化路径为特色的独立神经网络框架。tape 自动微分、标准层与损失、STE/GEF/SR 量化反向、MSint 整数编码、Level 精度调度，CPU SIMD（AVX2/AVX-VNNI）加速，无需 PyTorch。

Block-scaled FP8 / FP4 / INT4 tensor primitive with Triton scaled-matmul at FP32 parity on H100. NumPy / PyTorch / MLX / JAX backends.

LinearCosine: Adding beats multiplying for lower-precision efficient cosine similarity

QuantLab-8bit is a reproducible benchmark of 8-bit quantization on compact vision backbones. It includes FP32 baselines, PTQ (dynamic & static), QAT, ONNX exports, parity checks, ORT CPU latency, and visual diagnostics.