All repositories

gpt-neox

EleutherAI's library for training large language models from scratch across many GPUs: ZeRO and 3D parallelism, launching via Slurm and MPI. Its own README says to use it only for models with billions of parameters.

Who it is for

For researchers pretraining large models on clusters.

How to start

  1. Clone the repo and install dependencies: pip install -r requirements/requirements.txt.
  2. Prepare a model config from the configs folder.
  3. Start training: python3 deepy.py train.py /path/to/configs/my_model.yml.

Steps are taken from the README. Check the current version in the repository before running them.

Stars over the last 30 days

+2Oct 22 — Nov 19
7,4737,475

Author's description

An implementation of model parallel autoregressive transformers on GPUs, based on the Megatron and DeepSpeed libraries

ollama

ollama

The easiest way to run an open model on your own machine: one command downloads and starts it, plus a local REST API and libraries for Python and JavaScript.

188Kstars+3.3K in 29 dGo

huggingface

transformers

The core library for working with models: one codebase for text, vision, audio and multimodal tasks, for both inference and training. Most new open models ship through it.

169Kstars+1.3K in 29 dPython

open-webui

open-webui

A self-hosted ChatGPT-style interface that connects to Ollama and any OpenAI-compatible API. Installs with a single Docker command and runs on your own server.

159Kstars+2.5K in 28 dSvelte

ggml-org

llama.cpp

Growing fastest

A C/C++ engine that runs language models on ordinary hardware: laptops, phones, GPU-less servers. Much of local AI, Ollama included, is built on it.

136Kstars+3.7K in 29 dC++

pytorch

pytorch

The foundation under most modern neural networks: GPU-accelerated tensors and automatic differentiation, wrapped in ordinary Python. Transformers, Llama and nearly every open model are written on it.

106Kstars+674 in 30 dPython

deepseek-ai

DeepSeek-V3

Repository of the large open DeepSeek-V3 mixture-of-experts model: description, benchmark results and run instructions. It showed an open model can stand next to closed ones.

105Kstars+303 in 29 dPython