Autoresearch for GPU kernels. Give it any PyTorch model, go to sleep, wake up to optimized Triton kernels.
-
Updated
Mar 19, 2026 - Python
#
kernel-optimization
Here are 43 public repositories matching this topic...
A lightweight, general-purpose framework for evaluating GPU kernel and benchmark.
-
Updated
Oct 15, 2026 - Python
Open reproduction of NVIDIA's AVO paper (arXiv:2603.24517): evolutionary search where an autonomous coding agent IS the variation operator — Vary(P)=Agent(P,K,f). Runs on the Claude Code or Codex session you already have.
mcp evolutionary-algorithms reproducibility 2048 codex expectimax paper-implementation llm-agents coding-agents mcp-server claude-code kernel-optimization
-
Updated
Aug 22, 2026 - Python
Agent-queryable ROCm kernel optimization knowledge base for AMD Instinct MI300/gfx942 and MI350/MI355X/gfx950, packaged for Codex CLI and Claude Code with merged-PR provenance, real-silicon validation, and a maintainer-controlled pull-request evidence pipeline.
-
Updated
Jul 27, 2026 - Python
Evidence-driven CUDA, CUTLASS, Triton and GPU workload optimization for ChatGPT · 使用 ChatGPT 驱动 GPU workload 性能优化
performance skills gpu cuda pytorch triton profiling cutlass codex chatgpt vllm llm-inference nsight-compute agent-skills kernel-optimization trtllm
-
Updated
Oct 9, 2026 - Python
Noeris — autonomous kernel fusion discovery + Triton autotuning for LLM kernels and Gemma layer deeper fusion (A100/H100 wins).
benchmarking cuda pytorch triton autotuning gemma gpu-kernels github-actions kernel-fusion llm-training llm-inference kernel-optimization
-
Updated
Aug 19, 2026 - Python
Custom AWS Transform agent that migrates PyTorch/Triton kernels to AWS Trainium NKI (@nki.jit) and compiles, numerically verifies, and profiles every candidate on a real Trainium device before opening a PR.
mcp pytorch triton neuron nki code-migration aws-transform llm-agent kernel-optimization aws-trainium amazon-bedrock-agentcore gpu-to-trainium
-
Updated
Oct 19, 2026 - Python
Compiler MVP that detects Transformer fusion patterns, generates optimized CUDA kernels with WMMA Tensor Cores, and executes them on real GPU hardware — 10.5 TFLOPs on RTX 2070, correctness validated against PyTorch.
performance-engineering deep-learning compiler jinja2 gpu cuda transformer fusion systems code-generation turing gemm auto-tuning tensor-cores llm machine-learning-infrastructure kernel-optimization wmma rtx-2070 sm75
-
Updated
Jul 26, 2026 - Python
Measurement harness for the sliding window attention premium in the vLLM TPU Ragged Paged Attention v3 kernel: per layer decode cost, block size control, throughput, and goodput for Gemma 4 31B on TPU v6e.
google-cloud attention-mechanism gemma pallas gpu-kernels tpu jax inference-optimization tpu-benchmarks tpus tpu-acceleration trillium kernel-optimizers vllm llm-inference paged-attention sliding-window-attention vllm-serve kernel-optimization goodput
-
Updated
Jul 29, 2026 - Python
Reward-hardened evaluation for LLM-generated GPU kernels, built on KernelBench.
-
Updated
Jul 23, 2026 - Python
Automatic Triton kernel generation and optimization for Intel GPU, powered by Claude Code.
-
Updated
Oct 15, 2026 - Python
Fast Triton kernels for multi-scale deformable attention (MSDA) — the core operator behind Deformable DETR, DINO, and Mask2Former — plus a from-scratch course and hands-on labs
education deep-learning gpu cuda pytorch transformer triton object-detection attention-mechanism gpu-kernels gpu-programming deformable-detr deformable-attention triton-lang kernel-optimization
-
Updated
Oct 20, 2026 - Python
Native AscendC Mamba2 selective scan / SSD forward-backward custom operator for Huawei Ascend 910B3 and 950PR, with CANN, torch_npu, A100 benchmarks and msprof profiling.
pytorch ssd mamba cann performance-optimization state-space-models ascend-npu selective-scan ai-infrastructure ascendc kernel-optimization mamba2 huawei-ascend torch-npu custom-operator ascend-910b triton-ascend ascend-950
-
Updated
Oct 7, 2026 - Python
LLM agents that generate, verify, and evolve Triton GPU kernels. Includes a reward-hack-resistant benchmarking harness with strict correctness verification and fresh-input evaluation. Achieves up to 174.7× over PyTorch eager and outperforms FlexAttention (1.48×) and SDPA (1.17×) on selected workloads.
inference openai evolutionary-algorithms automatic-code-generation large-language-models agentic-workflow agentic-ai triton-kernels kernel-optimization gepa harness-engineering agentic-kernel-generation
-
Updated
Aug 12, 2026 - Python
An agentic harness for accelerator kernel generation and optimization, with reusable skills extracted from expert implementations.
-
Updated
Oct 21, 2026 - Python
Hydra-2P, KDA, and CADR: fused PyTorch/Triton kernels for routing across model depth and sequence time.
deep-learning cuda pytorch autograd triton gpu-kernels mixed-precision linear-attention kernel-optimization attention-residuals online-softmax kimi-delta-attention cross-axis-routing delta-routing
-
Updated
Jul 18, 2026 - Python
Portable Ascend C and TileLang operator engineering skill with tested workflows and lessons from DeepGEMM and TileKernels.
-
Updated
Oct 20, 2026 - Python
Paged attention decode kernel for LLM inference, implemented in both Triton and CUDA C++. Split-K (FlashDecoding-style) parallelism, Nsight-driven optimization, benchmarked against FlashInfer.
gpu cuda pytorch triton attention performance-optimization gpu-programming llm-inference flash-attention kernel-optimization
-
Updated
Aug 7, 2026 - Python
Custom CUDA kernels for AWQ 4-bit LLM weight dequantization. A simple learning project.
cuda pytorch gpu-computing quantization gpu-programming machine-learning-systems inference-optimization awq vllm llm-inference kernel-optimization
-
Updated
Jul 11, 2026 - Python
Add this topic to your repo
To associate your repository with the kernel-optimization topic, visit your repo's landing page and select "manage topics."