A compact implementation of SGLang, designed to demystify the complexities of modern LLM serving systems.

Code and dataset for paper: DeepPlanner: Scaling Planning Capability for Deep Research Agents via Advantage Shaping

slime is an LLM post-training framework for RL Scaling.

(best/better) practices of megatron on veRL and tuning guide

A repo for open research on building large reasoning models

Official implementation for DenseMixer: Improving MoE Post-Training with Precise Router Gradient

The evaluation framework for training-free sparse attention in LLMs

A version of verl to support diverse tool use [TMLR 2026]

Awesome List for Agentic RL

PyTorch bindings for CUTLASS grouped GEMM.

Implementation and datasets for "Training Language Models to Generate Quality Code with Program Analysis Feedback"

Official repository for FLAME-MoE: A Transparent End-to-End Research Platform for Mixture-of-Experts Language Models

Code for our tutorial on Discrete Variational Autoencoders

[TMLR] A curated list of language modeling researches for code (and other software engineering activities), plus related datasets.

[ACL2025 Oral] Cuckoo: A Series of IE Free Riders Using LLM's Resources to Scale up Themselves.

A bibliography and survey of the papers surrounding o1

GRadient-INformed MoE

The road to hack SysML and become an system expert

A framework for the evaluation of autoregressive code generation language models.

Tools for merging pretrained large language models.

Large Language Model Text Generation Inference

GPU programming related news and material links

CUDA Templates and Python DSLs for High-Performance Linear Algebra

Welcome to the Llama Cookbook! This is your go to guide for Building with Llama: Getting started with Inference, Fine-Tuning, RAG. We also show you how to solve end to end problems using Llama mode…

Best practice for training LLaMA models in Megatron-LM

MII makes low-latency and high-throughput inference possible, powered by DeepSpeed.

DeepSpeed is a deep learning optimization library that makes distributed training and inference easy, efficient, and effective.

The ChatGPT Retrieval Plugin lets you easily find personal or work documents by asking questions in natural language.

A curated list of papers and applications on tool learning.

A repo lists papers related to LLM based agent