MiniCPM3-4B: An edge-side LLM that surpasses GPT-3.5-Turbo.

Meta Lingua: a lean, efficient, and easy-to-hack codebase to research LLMs.

Official code for "SWARM Parallelism: Training Large Models Can Be Surprisingly Communication-Efficient"

Decentralized deep learning in PyTorch. Built to train models on thousands of volunteers across the world.

Best practices & guides on how to write distributed pytorch training code

📜 LaTeX Templates for Notes, Reports, CV/Resumes, and Beamers

✍️ A way to integrate LaTeX, VS Code, and Inkscape in macOS

Animation engine for explanatory math videos

prime (previously called ZeroBand) is a framework for efficient, globally distributed training of AI models over the internet.

NanoGPT (124M) quality in 2.67B tokens

What would you do with 1000 H100s...

Model components of the Llama Stack APIs

The simplest, fastest repository for training/finetuning medium-sized GPTs.

OLMoE: Open Mixture-of-Experts Language Models

Scaling is a distributed training library and installable dependency designed to scale up neural networks, with a dedicated module for training large language models.

深度学习经典、新论文逐段精读

Dermatology ddx dataset, Jax implementations of Monte Carlo conformal prediction, plausibility regions and statistical annotation aggregation from our recent work on uncertain ground truth (TMLR'23…

Distributed Training Over-The-Internet

An extremely fast Python package and project manager, written in Rust.

Efficient Triton Kernels for LLM Training

Deep learning for dummies. All the practical details and useful utilities that go into working with real models.

Xwin-LM: Powerful, Stable, and Reproducible LLM Alignment

OpenDiLoCo: An Open-Source Framework for Globally Distributed Low-Communication Training

DoubleML - Double Machine Learning in Python

Official codebase used to develop Vision Transformer, SigLIP, MLP-Mixer, LiT and more.

This is a library to use with Robinhood Financial App. It currently supports trading crypto-currencies, options, and stocks. In addition, it can be used to get real time ticker information, assess …