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

Ongoing research training transformer models at scale

My learning notes/codes for ML SYS.

Nano vLLM

verl: Volcano Engine Reinforcement Learning for LLMs

Supercharge Your LLM with the Fastest KV Cache Layer

Collect every awesome work about r1!

Triton Documentation in Chinese Simplified / Triton 中文文档

Solve puzzles. Learn CUDA.

A PyTorch native platform for training generative AI models

MS-Agent: Lightweight Framework for Empowering Agents with Autonomous Exploration in Complex Task Scenarios

PyTorch distributed training acceleration framework

flash attention tutorial written in python, triton, cuda, cutlass

Use PEFT or Full-parameter to CPT/SFT/DPO/GRPO 500+ LLMs (Qwen3, Qwen3-MoE, Llama4, GLM4.5, InternLM3, DeepSeek-R1, ...) and 200+ MLLMs (Qwen3-VL, Qwen3-Omni, InternVL3.5, Ovis2.5, Llava, GLM4v, Ph…

本项目旨在分享大模型相关技术原理以及实战经验（大模型工程化、大模型应用落地）

AISystem 主要是指AI系统，包括AI芯片、AI编译器、AI推理和训练框架等AI全栈底层技术

how to optimize some algorithm in cuda.

fastllm是后端无依赖的高性能大模型推理库。同时支持张量并行推理稠密模型和混合模式推理MOE模型，任意10G以上显卡即可推理满血DeepSeek。双路9004/9005服务器+单显卡部署DeepSeek满血满精度原版模型，单并发20tps；INT4量化模型单并发30tps，多并发可达60+。

llm deploy project based mnn. This project has merged into MNN.

LMDeploy is a toolkit for compressing, deploying, and serving LLMs.

Universal LLM Deployment Engine with ML Compilation

🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.

Langchain-Chatchat（原Langchain-ChatGLM）基于 Langchain 与 ChatGLM, Qwen 与 Llama 等语言模型的 RAG 与 Agent 应用 | Langchain-Chatchat (formerly langchain-ChatGLM), local knowledge based LLM (like ChatGLM, Qwen and…

MLJ: Libsignal: an open library for traffic signal control

An API standard for multi-agent reinforcement learning environments, with popular reference environments and related utilities

A toolkit for developing and comparing reinforcement learning algorithms.

Reinforcement Learning environments for Traffic Signal Control with SUMO. Compatible with Gymnasium, PettingZoo, and popular RL libraries.

Elegant and powerful theme for Hexo.

Awesome Deep Learning papers for industrial Search, Recommendation and Advertisement. They focus on Embedding, Matching, Pre-Ranking, Ranking (CTR/CVR prediction), Post Ranking, Relevance, LLM, Rei…