KernelBench: Can LLMs Write GPU Kernels? - Benchmark with Torch -> CUDA problems

A throughput-oriented high-performance serving framework for LLMs

verl: Volcano Engine Reinforcement Learning for LLMs

Control Logic Synthesis: Drawing the Rest of the OWL

Domain-specific language designed to streamline the development of high-performance GPU/CPU/Accelerators kernels

A simple, performant and scalable Jax LLM!

A PyTorch native library for large model training

Tritonbench is a collection of PyTorch custom operators with example inputs to measure their performance.

An Easy-to-use, Scalable and High-performance RLHF Framework (70B+ PPO Full Tuning & Iterative DPO & LoRA & RingAttention & RFT)

Supplemental materials for The ASPLOS 2025 / EuroSys 2025 Contest on Intra-Operator Parallelism for Distributed Deep Learning

HLS implementation of the tracklet pattern reco modules of the Hybrid tracking chain

Github mirror of trition-lang/triton repo.

Efficient Triton Kernels for LLM Training

A playbook for effectively prompting post-trained LLMs

A bibliography and survey of the papers surrounding o1

GPU Performance Advisor

Differentiable Combinatorial Scheduling at Scale (ICML'24). Mingju Liu, Yingjie Li, Jiaqi Yin, Zhiru Zhang, Cunxi Yu.

depyf is a tool to help you understand and adapt to PyTorch compiler torch.compile.

Low-bit LLM inference on CPU with lookup table

Quantized Attention that achieves speedups of 2.1-3.1x and 2.7-5.1x compared to FlashAttention2 and xformers, respectively, without lossing end-to-end metrics across various models.

Sum example for Xilinx VCK5000