KernelBench

KernelBench A benchmark designed to evaluate the ability of LLMs to generate efficient GPU kernels for optimizing neural network performance Version [07-21-2025] This HF dataset version has been updated to v0.1 Citation @misc{ouyang2024kernelbench, title={KernelBench: Can LLMs Write GPU Kernels?}, author={Anne Ouyang a

Papers using KernelBench (17)

CUDA Agent: Large-Scale Agentic RL for High-Performance CUDA Kernel Generation2026HTAM: Hierarchical Transition-Attended Memory for Operator Optimization2026CUDA-L1: Improving CUDA Optimization via Contrastive Reinforcement Learning2025 · 1 citesCUDA-L1: Improving CUDA Optimization via Contrastive Reinforcement Learning2025 · 1 citesKernelBench: Can LLMs Write Efficient GPU Kernels?2025 · 1 citesStitchCUDA: An Automated Multi-Agents End-to-End GPU Programing Framework with Rubric-based Agentic Reinforcement Learning2026Making LLMs Optimize Multi-Scenario CUDA Kernels Like Experts2026KernelFoundry: Hardware-aware evolutionary GPU kernel optimization2026Fine-Tuning GPT-5 for GPU Kernel Generation2026DICE: Diffusion Large Language Models Excel at Generating CUDA Kernels2026Dr. Kernel: Reinforcement Learning Done Right for Triton Kernel Generations2026MaxCode: A Max-Reward Reinforcement Learning Framework for Automated Code Optimization2026CudaForge: An Agent Framework with Hardware Feedback for CUDA Kernel Optimization2025STARK: Strategic Team of Agents for Refining Kernels2025TritonRL: Training LLMs to Think and Code Triton Without Cheating2025AutoTriton: Automatic Triton Programming with Reinforcement Learning in LLMs2025AutoTriton: Automatic Triton Programming with Reinforcement Learning in LLMs2025