← all releases
Allen Institute for AI (Ai2)·Oct 4, 2026·v3.0.0·open source
Olmo-core
Olmo-core is a specialized Python library designed to provide essential building blocks for large-scale distributed training of language models using PyTorch. It serves as the foundational framework for the OLMo ecosystem, enabling researchers to train, evaluate, and interact with OLMo models efficiently.
LLMPyTorchDistributed TrainingOpen Source
overview
Olmo-core is a specialized Python library designed to provide essential building blocks for large-scale distributed training of language models using PyTorch. It serves as the foundational framework for the OLMo ecosystem, enabling researchers to train, evaluate, and interact with OLMo models efficiently.
key features
- 01Distributed training support for large-scale PyTorch models
- 02Integration with high-performance backends like Flash-Attention and TransformerEngine
- 03Support for advanced training techniques including float8 and dropless MoE
- 04Official training scripts for OLMo-2 and OLMo-3 model families
- 05Native autoregressive generation and chat-loop capabilities
related productsbrowse all →
Protenixopen sourceProtenix is an open-source, trainable PyTorch implementation of AlphaFold 3, designed for high-accuracy biomolecular structure prediction. It aims to advance accessible and extensible research tools for the computational biology community. ([github.com](https://github.com/bytedance/Protenix?utm_source=openai))Z-Imageopen sourceZ-Image is an efficient image generation foundation model developed by Tongyi-MAI, designed to produce high-quality, diverse, and stylistically versatile images. It serves as a robust base for creators, researchers, and developers seeking advanced image generation capabilities.