A native PyTorch Library for large model training

A high-throughput and memory-efficient inference and serving engine for LLMs

Lighteval is your all-in-one toolkit for evaluating LLMs across multiple backends

PyTorch native post-training library

veRL: Volcano Engine Reinforcement Learning for LLM

A modular, primitive-first, python-first PyTorch library for Reinforcement Learning.

Clean, minimal, accessible reproduction of DeepSeek R1-Zero

Janus-Series: Unified Multimodal Understanding and Generation Models

Fully open reproduction of DeepSeek-R1

Tensors and Dynamic neural networks in Python with strong GPU acceleration

Development repository for the Triton language and compiler

Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more

Minimalist ML framework for Rust

Flax is a neural network library for JAX that is designed for flexibility.

Train transformer language models with reinforcement learning.

A framework for few-shot evaluation of language models.

Scripts for fine-tuning Meta Llama with composable FSDP & PEFT methods to cover single/multi-node GPUs. Supports default & custom datasets for applications such as summarization and Q&A. Supporting…

Run PyTorch LLMs locally on servers, desktop and mobile

LeanRL is a fork of CleanRL, where selected PyTorch scripts optimized for performance using compile and cudagraphs.

Distributed Evolutionary Algorithms in Python

🔎 📈 🐍 💰 Backtest trading strategies in Python.

Calendars for various securities exchanges.

You like pytorch? You like micrograd? You love tinygrad! ❤️