Accelerating RL for LLM Reasoning with Optimal Advantage Regression

Modular, scalable library to train ML models

Accelerating MoE with IO and Tile-aware Optimizations

MSLK (Meta Superintelligence Labs Kernels) is a collection of PyTorch GPU operator libraries that are designed and optimized for GenAI training and inference, such as FP8 row-wise quantization and …

[Preprint] RLVE: Scaling Up Reinforcement Learning for Language Models with Adaptive Verifiable Environments

Async RL Training at Scale

Kernel sources for https://huggingface.co/kernels-community

A scalable asynchronous reinforcement learning implementation with in-flight weight updates.

verl: Volcano Engine Reinforcement Learning for LLMs

A framework to study AI models in Reasoning, Alignment, and use of Memory (RAM).

Train transformer language models with reinforcement learning.

A benchmark to evaluate search-augmented LLMs

Scalable data pre processing and curation toolkit for LLMs

Reference PyTorch implementation and models for DINOv3

Tool for generating high quality Synthetic datasets

gpt-oss-120b and gpt-oss-20b are two open-weight language models by OpenAI

Simple and efficient DeepSeek V3 SFT using pipeline parallel and expert parallel, with both FP8 and BF16 trainings

Jan is an open source alternative to ChatGPT that runs 100% offline on your computer.

Implementation for our COLM paper "Off-Policy Corrected Reward Modeling for RLHF"

An action for automatically labelling pull requests