A collection of ROS-based IMU filters and visualizers that fuse angular velocity, acceleration, and magnetic readings into a stable orientation estimate.

A C++ vector expression template library for OpenCL and CUDA that reduces boilerplate code for GPGPU development.

A collection of microwrappers for reinforcement learning environments that enable easy preprocessing of observations for Gymnasium and PettingZoo.

A probabilistic logic programming toolbox that combines logic programs with probabilistic facts to handle uncertainty and perform complex inference tasks.

An AI-powered code exploration platform that helps developers understand GitHub repositories through an interactive dashboard and a codebase-aware AI chat assistant.

An unofficial PyTorch implementation of the Noise2Noise paper, providing scripts to train and test a U‑Net denoiser on noisy‑only data using Gaussian, Poisson, text‑overlay, or Monte‑Carlo rendering noise.

A low-code data flow tool that enables users to visually build and schedule workflows using LLMs and Hugging Face models without extensive coding.

An extension for Rasa NLU that adds advanced Chinese NLP capabilities, including BERT-based intent classification and BiLSTM+CRF entity extraction.

A web-based AI platform interface that allows users to quickly deploy a multi-model chat UI using Cloudflare Workers AI and AI Gateway.

A local, agentic RAG engine that uses a 9-layer pipeline including GraphRAG and neural reranking to provide high-accuracy document Q&A without cloud dependencies.

A family of large-scale autoregressive vision encoders that provide high-performance backbones for multimodal understanding and image recognition.

Easy Voice Toolkit bundles open‑source speech models (Whisper, GPT‑SoVITS) into a GUI/Colab workflow for cleaning audio, transcribing it, building a voice‑conversion dataset, training a custom voice model, and generating converted speech. It offers a ready‑to‑run Windows portable package and a Colab notebook, with a developer‑friendly install guide. The project is academic‑use only and plans future LLM chatbot integration and Linux support.

combo is a Python toolbox for combining machine learning models and scores. It offers unified APIs for ensemble techniques like stacking, dynamic classifier/ensemble selection, and outlier detector fusion, supporting classification, clustering, and anomaly detection with models from scikit-learn, XGBoost, LightGBM, and pyod.

A terminal utility that generates shell commands from natural language descriptions and inserts them directly into the prompt using various LLM providers.

A 3D-aware generative model that transforms 2D GANs into Multiplane Images to enable 3D view synthesis and mesh extraction.

GaussianCity is a generative 3D city synthesis tool that uses Gaussian Splatting to create unbounded, large-scale urban environments.

A tool that enables running verl RLHF experiments (like PPO and GRPO) on a single NVIDIA GPU by scheduling model roles in phases.

A cross-platform AI chat client that integrates with the Model Context Protocol (MCP) to connect LLMs with external data sources and tools.

VRT is a PyTorch implementation of the Video Restoration Transformer, a transformer‑based model that handles video super‑resolution, deblurring, denoising, frame interpolation and space‑time SR. It uses Temporal Mutual Self‑Attention and parallel warping to capture long‑range temporal dependencies, delivering state‑of‑the‑art PSNR gains on nine benchmarks. The repo includes pretrained weights, test scripts, a Colab demo, and instructions for training on standard video datasets.

A unified framework for visual perception and reasoning that combines image, video, and pixel-level segmentation into a single model, reducing the need for specialized specialist models.

A high-performance ML inference engine that accelerates TensorFlow Lite and ONNX models on Arm Cortex-A CPUs and Arm Mali GPUs.

A lightweight C++17 library providing STL-like generic data structures for the GPU to enable flexible and reliable data management in custom kernels.

An open-vocabulary extension of the Segment Anything Model (SAM) that enables simultaneous interactive segmentation and recognition of thousands of object classes.

JAX‑CFD is a research‑grade, JAX‑based library for differentiable CFD. It provides finite‑volume and pseudospectral solvers, optional ML‑augmented models, and data utilities, enabling gradient‑based experiments on turbulent flow simulations. The package is installable via pip, comes with Colab demos, and is now unmaintained, with newer alternatives recommended.

CityDreamer is a research‑grade PyTorch framework that generates unlimited, photorealistic 3‑D cityscapes by combining a layout generator, a background‑stuff generator, and a building instance generator. It includes code for training on large OSM/Google‑Earth datasets, pre‑trained checkpoints, a web demo, and a CLI for rendering videos.

DeepLSD is a high-precision line segment detector that combines deep learning with image gradients to extract and refine line segments from real-world images.

A PyTorch implementation of BERT that allows users to build the model from scratch and apply it to various NLP tasks like classification, QA, and NER.

DFormer is a research codebase for RGB‑D semantic segmentation, implementing DFormer, DFormerv2 (geometry‑guided attention) and DFormer++ (efficient high‑accuracy variants). It includes pre‑training scripts, dataset helpers, training/evaluation pipelines, and pretrained weights for NYU‑Depth v2 and SUN‑RGBD. The repo is a genuine AI/ML project.

Subaligner is an open‑source Python/CLI tool that synchronises subtitles with video/audio, can transcribe speech via Whisper, translate subtitles with HuggingFace models, and even train custom alignment models. It supports many subtitle and media formats, offers fast global‑shift (`single`) and high‑accuracy two‑stage (`dual`) alignment, and provides Docker, pip, and optional extras for LLM‑based features.

An open-source Python framework for automatic number plate recognition using YOLOv8 and RNN-based OCR to detect and read license plates across multiple countries.

A text preprocessing library that provides tokenization, detokenization, normalization, and truecasing tools for NLP datasets.

A high-performance SDK from NVIDIA designed to accelerate quantum science simulations, providing specialized libraries for state vector and tensor network computations.

A post-training compression tool that makes transformer networks smaller and faster by slicing off insignificant weight matrix components using orthogonal transformations.

An asymmetric and adaptive refinement of the GRPO reinforcement learning algorithm that uses four independent clipping boundaries to prevent entropy collapse and improve generalization in LLM reasoning tasks.

A powerful agent framework that integrates browser automation, code execution, and MCP tool support across multiple agent backends.

A Vision RAG (V-RAG) demo that embeds PDF pages as images using a VLM to eliminate the need for text chunking during document retrieval.

An acceleration framework for large-scale unsupervised outlier detection that optimizes training and prediction for ensembles of heterogeneous models.

A collection of logits processors for LLMs that allow developers to control output behavior, such as enforcing specific phrases, managing response length, and mitigating hallucinations.

An autonomous task execution system for Claude CLI that monitors usage windows and automatically runs predefined tasks from a Markdown file.

A centralized, plain-text memory layer for AI agents that uses markdown files to share context and persistent knowledge across different tools and projects.

A toolkit for evaluating RAG chat applications by benchmarking generated answers against ground truth data using GPT-based and code-based metrics.

A Go-based RAG system that allows users to create knowledge bases from documents and web pages, featuring multi-path recall and a management GUI.

A self-hosted web application that uses OpenAI's Whisper to locally transcribe audio into text or subtitles in various formats.

A library that integrates Large Language Models into the causal analysis process to suggest confounders, causal relationships, and validation strategies.

A unified framework for Scientific Machine Learning (SciML) providing implementations of PINNs, DeepONets, and KANs for studying complex physical systems.

A large-scale multimodal dataset and benchmark for human activity recognition and reasoning, integrating seven synchronized sensor modalities to enable complex action understanding.

DiffuCoder is a masked diffusion model for code generation that introduces Coupled-GRPO to improve post-training efficiency and learning signals for all tokens.

An end-to-end framework for Matryoshka Diffusion Models that enables efficient training of high-resolution text-to-image synthesis models up to 1024x1024 pixels.