Easy to use stem (e.g. instrumental/vocals) separation from CLI or as a python package, using a variety of amazing pre-trained models (primarily from UVR)

Transform geospatial relations into graphs for Graph Neural Networks and network analysis

Rust native ready-to-use NLP pipelines and transformer-based models (BERT, DistilBERT, GPT2,...)

High-Rate Mixout: Revisiting Mixout for Robust Domain Generalization (Accepted at WACV 2025)

Gradient Conductor - Constructs a unified gradient from a set of loss functions using orthogonal projection and momentum smoothing

Official PyTorch Implementation for Fast Adaptive Multitask Optimization (FAMO)

Official implementation of Scalable Bilevel Loss Balancing for Multi-Task Learning.

BiLSTM-CNN-CRF architecture for sequence tagging

RST Discourse Parsers

Retrieval-based Voice Conversion (RVC) implemented with Hugging Face Transformers.

The picky and eager Git hook runner

Python bindings for path_jail, secure filesystem sandbox preventing path traversal attacks

A fast Rust JSON library based on SIMD.

Official JAX implementation of End-to-End Test-Time Training for Long Context

Thrift definitions of parquet

Peek inside Parquet files right from your terminal

parquet dedupe estimator

FastCDC implementation in Rust

deterministic evaluation of performance via instruction counting tool

Prototype record matching database.

The official code for “Geometric Representation Learning for Document Image Rectification”, ECCV, 2022.

MoE training for Me and You and maybe other people

Subspace Power Method

Code for the paper: "Independent mechanism analysis, a new concept?"

Preconditioned ICA for Real Data

A Rust machine learning framework.

Robust and fast topic models with sentence-transformers.

Powerful topic model visualization in Python

🚀 Accelerate inference and training of 🤗 Transformers, Diffusers, TIMM and Sentence Transformers with easy to use hardware optimization tools