Gpu Kernels Python Packages

grouped-nf4-gemm

4-bit GPU kernels that compute directly on packed Mixture-of-Experts weights, for training and inference.

graphsignal

GPU Profiler for AI agents

weco

Production-Grade Autoresearch. Ideal for agent harness engineering, prompt engineering, ML model development, GPU kernels, and other optimizable code.

hyperloom-inference-optimizer

An agentic system that auto-optimizes LLM workloads on AMD GPUs.

triton-msl

Apple Silicon (Metal) backend for Triton: write standard @triton.jit kernels on your Mac GPU. The same source runs bit-identical on NVIDIA and AMD, so you develop kernel logic locally and rent a datacenter GPU only for the perf pass.

llvq-tetra

Tetra : Leech lattice vector quantization for 2-bit LLM weights, in Rust, with a fused multi-shell CUDA decoder.

mingru-scans

minGRU (Feng et al.) in PyTorch, plus a measured ladder of state-tracking recoveries — signed, rotation, Givens, and delta (DeltaNet-style Householder products) mixers with time decay, all under one parallel scan — with fused Triton GPU kernels as an optional backend

torchnorm

Fused Triton kernels for normalisation layers (RMSNorm, LayerNorm, GroupNorm, FusedAddRMSNorm)

grouped-mxfp4-gemm

4-bit GPU kernels that compute directly on packed Mixture-of-Experts weights, for training and inference.

gnf4

4-bit GPU kernels that compute directly on packed Mixture-of-Experts weights, for training and inference.

nf4gemm

4-bit GPU kernels that compute directly on packed Mixture-of-Experts weights, for training and inference.

mlx-recurrence

Fused Metal training kernels for linear recurrence on Apple Silicon — SSD (Mamba-2), GLA, RG-LRU, rotational LRU. Flash-linear-attention for MLX: 12-18x less training memory, production-validated.

orda-kernel

ORDA: Triton kernels for logit-level knowledge distillation loss.