Train your Agent model via our easy and efficient framework

Scalable toolkit for efficient model reinforcement

Democratizing Reinforcement Learning for LLMs

Fully open data curation for reasoning models

Redis for LLMs

Understanding R1-Zero-Like Training: A Critical Perspective

Scalable RL solution for advanced reasoning of language models

Due to the huge vocaburary size (151,936) of Qwen models, the Embedding and LM Head weights are excessively heavy. Therefore, this project provides a Tokenizer vocabulary shearing solution for Qwen…

Towards Economical Inference: Enabling DeepSeek's Multi-Head Latent Attention in Any Transformer-based LLMs

An unofficial cuda assembler, for all generations of SASS, hopefully ：）

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

FlashMLA: Efficient MLA decoding kernels

Official Repo for Open-Reasoner-Zero

Easily fine-tune, evaluate and deploy Qwen3, DeepSeek-R1, Llama 4 or any open source LLM / VLM!

Reproduce R1 Zero on Logic Puzzle

Inject code into running Python processes

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

Fully open reproduction of DeepSeek-R1

Minimal reproduction of DeepSeek R1-Zero

Simple RL training for reasoning

[NAACL 2025] MiLoRA: Harnessing Minor Singular Components for Parameter-Efficient LLM Finetuning

Same as llm.c but in Rust, as I want to get deeper and deeper into Rust programming

A Python package that makes it easy for developers to create AI apps powered by various AI providers.

A self-hosted, ad-free, privacy-respecting metasearch engine