Language model tokenization at GB/s

Training library for Megatron-based models with bidirectional Hugging Face conversion capability

Arctic Training and Inference Platform

MOSS-Audio-Tokenizer is a Causal Transformer-based audio tokenizer built on the CAT architecture. Trained on 3M hours of diverse audio, it supports streaming and variable bitrates, delivering SOTA …

Unofficial fairseq-free PyTorch implementation of UTMOS (v1, 2022), matching the original system.

Aurora optimizer release

Documentation sites for Python packages: start simple, go deep

The world's most sophisticated street level image geolocation software

Late Interaction Models Training & Retrieval

Tooling for exact and MinHash deduplication of large-scale text datasets

New generation of CLIP with strong fine grained discrimination capability, ICML2026 and ICML2025

Synthetic data annotation for retrieval evaluations by ZeroEntropy

Developer-friendly OSS embedded retrieval library for multimodal AI. Search More; Manage Less.

Voice Activity Detector (VAD) : low-latency, high-performance and lightweight

Implementation of papers in 100 lines of code.

General plug-and-play inference library for Recursive Language Models (RLMs), supporting various sandboxes.

High-quality single file implementation of Deep Reinforcement Learning algorithms with research-friendly features (PPO, DQN, C51, DDPG, TD3, SAC, PPG)

PyTorch-native post-training at scale

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

Fully Open Framework for Democratized Multimodal Training

Nano vLLM

A Conversational Speech Generation Model

Minimal and annotated implementations of key ideas from modern deep learning research.

A Comprehensive Toolkit for High-Quality PDF Content Extraction

Heuristic based boilerplate removal tool