🪐 The Sebulba architecture to scale reinforcement learning on Cloud TPUs in JAX

Official codebase for "SWE-RL: Advancing LLM Reasoning via Reinforcement Learning on Open Software Evolution"

A project to improve skills of large language models

Letting Claude Code develop his own MCP tools :)

Humanity's Last Exam

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

Lighteval is your all-in-one toolkit for evaluating LLMs across multiple backends

Search-R1: An Efficient, Scalable RL Training Framework for Reasoning & Search Engine Calling interleaved LLM based on veRL

Autonomous coding agent right in your IDE, capable of creating/editing files, executing commands, using the browser, and more with your permission every step of the way.

Multi-Agent Verification: Scaling Test-Time Compute with Multiple Verifiers

Solve puzzles. Learn CUDA.

Official Repo for Fine-Tuning Large Vision-Language Models as Decision-Making Agents via Reinforcement Learning

Computer gaming agents that run on your PC and laptops.

Official PyTorch implementation for "Large Language Diffusion Models"

Pokemon terminal themes.

Tailor-made Pokémon themes for your Hyper terminal

Tabular In-Context Learning

EasyR1: An Efficient, Scalable, Multi-Modality RL Training Framework based on veRL

DeepGEMM: clean and efficient FP8 GEMM kernels with fine-grained scaling

Playing Pokemon Red with Reinforcement Learning

Code for Paper: Training Software Engineering Agents and Verifiers with SWE-Gym

FlashMLA: Efficient MLA decoding kernels

DeepEP: an efficient expert-parallel communication library

Everything you want to know about Google Cloud TPU

A fast MoE impl for PyTorch

[ICLR2025 Spotlight🔥] Official Implementation of TokenFormer: Rethinking Transformer Scaling with Tokenized Model Parameters