PyTorch native quantization for training and inference

training sparsity cuda inference pytorch transformer llama quantization mx brrr dtypes float8

• Updated Nov 12, 2026
• Python

Pure Go machine learning framework. Train, run, and serve ML models with go build. Zero CGo.

go golang machine-learning deep-learning neural-network transformer fp16 autodiff distributed-training float16 onnx graph-ml ml-framework fp8 float8

• Updated Oct 14, 2026
• Go

Official Code for the paper ELMO : Efficiency via Low-precision and Peak Memory Optimization in Large Output Spaces (in ICML 2025)

chunking multi-label-classification extreme-classification low-precision-training large-output-space float8 triton-kernels gradient-fusion

• Updated Feb 27, 2026
• Python

8-bit and sub-byte floating point types for Rust

rust float4 microfloat float8 subfloat float6

• Updated May 3, 2026
• Rust

A library written in C for converting between float8 (8-bit minifloat numbers) and float32 (single-precision floating-point numbers) formats.

floating-point minifloat float32 float8

• Updated Jun 21, 2025
• C

minifloat (8-bit float) in Golang

golang minifloat float8

• Updated Aug 1, 2024
• Go

Provide a dark-mode skin for Archive of Our Own inspired by Van Gogh’s The Starry Night, enhancing readability with a calming color scheme.

music training app music-library cuda voice inference pytorch llama musicplayer music-manager mx github-config musiclib meida audioplayers brrr float8

• Updated Oct 13, 2026

FP8 (E5M2 and E4M3, any pairing) GEMM, GEMV and fused AdamW for NVIDIA GPUs without native FP8 (sm_50–sm_87, Maxwell to Ampere): exact products, FP32 accumulation, TE HYBRID backward, C API and PyTorch bindings incl. torch._scaled_mm and Float8Tensor

training pascal deep-learning gpu cuda inference pytorch nvidia matrix-multiplication cuda-kernels quantization gemm mixed-precision adamw gemv fp8 float8 transformer-engine e4m3 e5m2

• Updated Oct 21, 2026
• Cuda

FP8 (E4M3) GEMM, GEMV and fused AdamW for NVIDIA GPUs without native FP8 (sm_50–sm_87, Maxwell to Ampere): exact products, FP32 accumulation, C API and PyTorch bindings incl. torch._scaled_mm and Float8Tensor

training pascal deep-learning gpu cuda inference pytorch nvidia matrix-multiplication cuda-kernels quantization gemm nvidia-gpu mixed-precision adamw gemv fp8 float8 transformer-engine e4m3

• Updated Oct 20, 2026
• Cuda