🚀🚀 「大模型」2小时完全从0训练26M的小参数GPT！🌏 Train a 26M-parameter GPT from scratch in just 2h!

个人构建MoE大模型：从预训练到DPO的完整实践

Community maintained hardware plugin for vLLM on Ascend

Fully open reproduction of DeepSeek-R1

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

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

A bidirectional pipeline parallelism algorithm for computation-communication overlap in DeepSeek V3/R1 training.

Production-tested AI infrastructure tools for efficient AGI development and community-driven innovation

📚LeetCUDA: Modern CUDA Learn Notes with PyTorch for Beginners🐑, 200+ CUDA Kernels, Tensor Cores, HGEMM, FA-2 MMA.🎉

The PYthoN General UnIt Test geNerator is a test-generation tool for Python

LLaMA-Factory使用经验记录

🙌 OpenHands: AI-Driven Development

verl: Volcano Engine Reinforcement Learning for LLMs

Minimal reproduction of DeepSeek R1-Zero

Everything about the SmolLM and SmolVLM family of models

Simple RL training for reasoning

🧑‍🏫 60+ Implementations/tutorials of deep learning papers with side-by-side notes 📝; including transformers (original, xl, switch, feedback, vit, ...), optimizers (adam, adabelief, sophia, ...), ga…

21 Lessons, Get Started Building with Generative AI

DataComp for Language Models

The TinyLlama project is an open endeavor to pretrain a 1.1B Llama model on 3 trillion tokens.

The Open Cookbook for Top-Tier Code Large Language Model

Data processing for and with foundation models! 🍎 🍋 🌽 ➡️ ➡️🍸 🍹 🍷

Implement a ChatGPT-like LLM in PyTorch from scratch, step by step

✨ RepoBench: Benchmarking Repository-Level Code Auto-Completion Systems - ICLR 2024

Minimal, clean code for the Byte Pair Encoding (BPE) algorithm commonly used in LLM tokenization.

Code for the paper "Language Models are Unsupervised Multitask Learners"

llama3 implementation one matrix multiplication at a time