A RAG-based chatbot designed to provide hallucination-free answers to academic questions based on the official UAJY academic handbook using Google Gemini and FAISS.

A library that makes custom Docker containers compatible with Amazon SageMaker, simplifying the process of training machine learning models in isolated environments.

An open-source benchmarking platform for image and video diffusion models that automates prompt sweeps, metric calculation, and reproducible run logging.

A fast and generic Julia implementation of the AlphaZero algorithm that enables training AI agents for combinatorial games using self-play on standard desktop hardware.

LeoAI is an open‑source, AI‑augmented red‑team platform that combines a web dashboard, dual Java/PHP Puppet runtimes for remote host control, and LangChain4j‑driven LLM agents to automate post‑exploitation tasks, generate reports, and manage teams—all deployable via a JAR or Docker.

DeepThinkVLA is a Vision‑Language‑Action model that inserts an explicit chain‑of‑thought reasoning step before generating robot actions. Using a hybrid decoder (autoregressive reasoning → bidirectional parallel action generation), the authors train on a newly curated embodied CoT dataset and then refine with outcome‑driven reinforcement learning. The model achieves 97 % average success on the LIBERO benchmark, runs at only 0.175× the latency of a pure autoregressive baseline, and transfers zero‑shot to the newer LIBERO Plus suite.

A cross-platform CLI tool that measures the tokens-per-second throughput of local LLMs running via Ollama based on the user's available system RAM.

A framework for synthesizing music and audio loops by applying diffusion models to mel spectrograms, supporting latent diffusion and conditional generation.

A Node-RED library that allows users to visually build and deploy chatbots for Telegram, Facebook Messenger, WhatsApp, Viber, and Slack with minimal coding.

An automated account registration tool for Grok (x.ai) that uses temporary emails and browser patches to bypass Cloudflare Turnstile.

An open-source PHP library providing tools for professional Arabic language processing, including sentiment analysis, spell checking, and Hijri date management.

Paper Pilot is a desktop (Electron + React) research assistant that crawls 8+ scholarly databases, downloads PDFs, indexes them with SQLite + FTS5 + vector search, and lets you ask a local or hosted LLM for evidence‑grounded answers. All data stays on your machine; the app provides audit‑ready citation traces and a reproducible workflow for literature review.

A standalone Windows utility that allows users to crop and pin interactive sub-windows from any application, featuring advanced OCR, translation, and screenshot tools.

A command-line interface for Hono developers and AI agents that integrates documentation search, server management, and bundle optimization.

A comprehensive collection of C++ and Python implementations of deep learning foundations, machine learning algorithms, and modern AI frameworks like YOLO and Ollama.

A collection of tools and SDKs for implementing and validating attestation in NVIDIA Trusted Computing and confidential computing environments.

ch.at is a single‑binary Go server that lets you chat with an LLM (default GPT‑4o) via HTTP, SSH, DNS, or API, with no JavaScript, no accounts, and all state kept in RAM. It’s privacy‑focused, easy to self‑host, and includes a public board for posting messages over the same lightweight protocols.

A reconciliation-based coding agent that prevents context rotting by using fresh contexts and file-system memory to iteratively align a codebase with a specified goal.

A Kubernetes usage accounting tool that tracks CPU, Memory, and GPU resources to reveal hidden system overhead and enable precise per-namespace cost allocation.

A set of extensions for Unity's AR Foundation that provides native API access to Google ARCore's augmented reality features.

A remote access and monitoring service for 3D printers that includes AI-powered print failure detection to automatically stop failed prints.

A social chatbot that combines learning-based repetition of group chat corpora with LLM capabilities to create a bot that mimics the specific linguistic style of its community.

A comprehensive email processing library that integrates deep learning and deterministic rules to automate email routing, prioritization, and segmentation.

A tokenized payment layer that enables AI agents to safely purchase gift cards and pay bills using pre-funded balances and scoped API keys.

A decision intelligence framework integrating operations research and AI to provide robust, auditable optimization for high-stakes domains like healthcare and finance.

A Python library for black-box optimization of mixed search spaces, providing a unified interface to 23 gradient-free algorithms for hyperparameter tuning and engineering design.

A set of custom skills for Claude Code that provide persistent memory across sessions via structured logging and context preservation, reducing session-restart token costs.

A local-first, file-based knowledge system that enables AI agents to retrieve information via progressive disclosure to manage context window limits.

A training framework for Radiance Meshes that enables volumetric reconstruction of 3D scenes from images, with options for optimizing models for mobile web viewing.

A bridging layer that enables AI agents to control and edit Unreal Engine 5.3+ in real time, featuring AST-based validation to prevent hallucinations and deep integration with Blueprints, VFX, and rigging.

GPT‑NeoX is an open‑source, Megatron‑based library for training billion‑parameter language models on multi‑GPU clusters. It adds DeepSpeed‑style optimizations, supports many launchers (Slurm, MPI, etc.), works on NVIDIA and AMD GPUs, and includes modern features like flash attention, Mixture‑of‑Experts, and preference‑learning fine‑tuning. Ideal for research labs with large compute budgets; for inference‑only use, Hugging Face `transformers` is recommended.

A library of differentiable graphics and geometry layers for TensorFlow that enables self-supervised training of 3D vision models through analysis by synthesis.

A collection of StyleGAN2-based face generators that create high-fidelity, unique virtual human faces for use in film, advertising, and gaming.

A bilingual and crosslingual embedding and reranking framework optimized for RAG, specializing in high-performance retrieval across Chinese and English.

A Python library for tensor learning that provides tools for tensor decomposition and algebra with seamless support for multiple backends like PyTorch, TensorFlow, and JAX.

A video conversation model that combines LLMs with a spatiotemporal visual encoder to enable detailed, natural language discussions about video content.

A distilled, faster version of the Qwen‑Image text‑to‑image model (4‑ or 8‑step diffusion) with fp32/bf16/fp8 checkpoints, LoRA adapters, and ready‑to‑use ComfyUI workflows. Provides 12–25× speed‑up with modest quality loss and integrates with Diffusers, ComfyUI, Nunchaku, and Cache‑dit.

GLaMM is a pixel-grounding large multimodal model that integrates natural language responses with object segmentation masks for precise visual grounding.

EasyJailbreak is an open‑source Python framework that modularises LLM jailbreak attacks. It provides ready‑made recipes (ReNeLLM, GPTFuzz, AutoDAN, etc.) and lets users plug in custom selectors, mutators, constraints, and evaluators to generate, run, and score adversarial prompts against target language models. Install via `pip install easyjailbreak` or from source, then use the high‑level API (e.g., `PAIR` recipe) or build your own attack pipeline. The project includes documentation, a paper, datasets, and downloadable experimental results.

A model merging extension for AUTOMATIC1111's Stable Diffusion WebUI that allows users to merge models in-memory and perform block-by-block weight adjustments without saving to disk.

Torch.rb is a Ruby gem that wraps the PyTorch C++ library (LibTorch), giving Ruby developers a full‑featured deep‑learning API. It supports tensors, autograd, neural‑network modules, optimizers, and GPU (CUDA/MPS) acceleration, while mirroring the PyTorch Python API with Ruby‑style method names. Installation requires a matching LibTorch build, after which you can write and run models (e.g., CNNs, GANs) directly in Ruby.

LLaVA++ enhances LLaVA 1.5 by integrating LLaMA-3 and Phi-3 LLMs to improve visual instruction-following and academic task performance.

A WSL2 appliance for running the Strata SCTL live cycle harness, using a Codex/DeepSeek-backed disposable-worker model to automate software development cycles.

A real-time super-resolution implementation based on SR-GAN that uses pixel shuffle to efficiently upscale low-resolution videos to high-resolution.

OmniVinci is an open-source omni-modal LLM that jointly understands vision, audio, and text using specialized temporal and alignment architectures to improve cross-modal reasoning.

AViD is a framework for fine-tuning Grounding DINO using LoRA and EMA, enabling efficient adaptation of open-vocabulary object detectors to custom datasets.

A comprehensive evaluation framework for testing SAM and SAM 2's ability to segment context-dependent concepts like medical lesions and industrial defects across images and videos.

A lightweight training framework for large-scale language models that focuses on modular configurations and fast iteration across SFT, RLVR, and evaluation.