Coding & Development

Nemotron is an open-source family of large language models created by NVIDIA. It serves as a robust foundation for developers and researchers, enabling them to build and deploy custom generative AI applications. The models are designed to facilitate innovation across various AI tasks, particularly in natural language processing and understanding. Nemotron aims to provide the necessary tools and resources for advancing generative AI capabilities.

Untether AI specializes in delivering AI inference acceleration solutions designed for deployment across various environments, from edge devices to cloud infrastructure. The company's core offering revolves around an energy-centric architecture that efficiently supports a wide range of neural network models. A key focus is on significantly reducing data movement and overall power consumption, which are critical factors in AI workload efficiency. Untether AI empowers enterprises to execute their AI workloads effectively in diverse locations, including those outside traditional data centers.

godot-dodo is a project dedicated to enhancing large language models specifically for generating GDScript. It offers a comprehensive pipeline designed to facilitate the training of open-source language models. This training leverages human-created, language-specific code sourced directly from GitHub, ensuring high-quality and relevant data. The primary application of this technology is to enable the efficient generation of GDScript code, which is crucial for various game development purposes.

VisRAG is a parsing-free Retrieval-Augmented Generation (RAG) tool that leverages Vision Language Models (VLMs). Its primary function is to support evidence-guided multi-image reasoning within visual RAG tasks. The tool is designed to assist researchers and engineers in developing and exploring advanced VLM applications. It is particularly useful for tasks that necessitate reasoning across multiple images, providing a robust framework for complex visual data analysis and generation.

GPT_Vuln-analyzer is a proof-of-concept application designed to automate the generation of vulnerability reports. It integrates various AI models, including ChatGPT API, Bard API, and Llama2, to process and analyze security data. The tool also incorporates Python-Nmap for network scanning, DNS Recon for domain information gathering, PCAP for packet analysis, and JWT recon modules. Specifically, it utilizes the GPT-3 model to analyze Nmap scan data and DNS information, and it has the capability to perform subdomain enumeration, streamlining the initial stages of security assessments.

AI & Computer Vision Lab (iVision) is a research institution at the Institute of Space Technology dedicated to exploring and advancing the fields of computer vision, machine learning, and signal processing. The lab serves as a hub for researchers to collaborate on fundamental advancements in these areas. iVision also actively engages with both public and private sector industries, applying its expertise to socio-economic projects and fostering practical applications of its research.

Llama2-Code-Interpreter is a tool designed to enhance Large Language Models (LLMs) by giving them the capability to interact with code. It allows LLMs to generate new code, execute it to test its functionality, and debug any issues that arise. A key feature is the ability for LLMs to receive feedback on their code, save successful code snippets, and reuse them across different programming challenges. This tool supports multiple programming languages and frameworks, offering versatility for a wide range of coding requirements and making LLMs more effective in software development tasks.

LiveCodeBench serves as a comprehensive repository designed for the holistic and contamination-free evaluation of large language models (LLMs) specifically in the domain of code. It offers a dedicated platform for benchmarking the coding capabilities of various LLMs, ensuring rigorous and unbiased assessment. The tool continuously gathers new programming problems from contests over time, keeping its evaluation dataset fresh and relevant. This resource is particularly valuable for researchers and developers involved in the advancement and refinement of AI models, providing critical insights into their performance.

Llama Coder provides a self-hosted alternative to GitHub Copilot, specifically designed for VS Code users. It leverages Ollama and Code Llama to deliver code autocomplete functionality directly on the user's local hardware. The tool prioritizes privacy and speed, ensuring that no telemetry or user data ever leaves the device. This makes it an ideal solution for developers seeking a secure and efficient coding experience without reliance on external services.

LLM-Adapters provides a framework for parameter-efficient fine-tuning (PEFT) of Large Language Models (LLMs). It integrates various adapter techniques directly into LLMs, allowing users to apply adapter-based PEFT methods for a wide range of tasks. This tool is built as an extension of HuggingFace's popular PEFT library, leveraging its existing capabilities while introducing additional functionalities for adapter management and application.

awesome-ChatGPT-resource-zh is a comprehensive, curated list of resources specifically tailored for OpenAI's ChatGPT, with a strong emphasis on Chinese-language tools and information. This repository serves as a valuable collection of links and references related to ChatGPT and GPT-3, designed to assist users in navigating and leveraging these powerful AI models within Chinese-speaking environments. It aims to streamline the process of finding relevant information and applications for a Chinese audience.

LLMBox is a comprehensive library designed for the implementation and utilization of Large Language Models (LLMs). It provides a unified training pipeline, streamlining the process of developing and deploying LLMs. The platform also includes tools for comprehensive model evaluation, ensuring the performance and reliability of the trained models. LLMBox aims to be a one-stop solution, offering both flexibility and efficiency for users working with LLMs.

LMQL is a programming language designed for efficient and constraint-guided programming of Large Language Models (LLMs). Built as a superset of Python, it allows developers to seamlessly integrate traditional programming logic with calls to LLMs. This approach aims to provide a more structured and controlled way to interact with and program LLMs, moving beyond simple templating. It's intended for developers looking for advanced methods to build AI applications with greater precision and efficiency.

Automated Task Solving with Coding & Planning Agents is a system designed to effectively solve tasks by combining coding and planning agents. It utilizes the AutoGen framework, a platform for building multi-agent AI applications. This tool automates coding tasks through AI planning and execution, making it suitable for complex problem-solving scenarios. The system emphasizes the use of collaborative AI agents to achieve its objectives.

ChatGLM-Tuning provides a fine-tuning solution specifically designed for the ChatGLM-6B model, leveraging the LoRA (Low-Rank Adaptation) technique. This approach aims to offer an affordable and accessible method for users to implement a model with capabilities similar to ChatGPT. The tool supports the Alpaca dataset for training and requires a computing environment equipped with a GPU that has at least 16GB of VRAM, along with a Python environment configured with CUDA.

MinkowskiEngine is a specialized neural network library built for handling high-dimensional sparse tensors. It provides auto-differentiable capabilities, making it suitable for advanced deep learning tasks, particularly within the realm of 3D data processing. The library supports a variety of operations tailored for sparse tensors, which allows for more efficient computation and memory usage when dealing with large, sparse datasets common in 3D deep learning applications. Its design focuses on optimizing performance for these specific data structures.

Compose-for-agents is a specialized tool designed for developers and researchers to construct and deploy AI agents efficiently. It leverages Docker Compose to streamline the setup and management of agent environments, providing pre-built examples for integrating open-source Large Language Models (LLMs), various tools, and agent runtimes. The platform is particularly useful for local development and experimentation, requiring Docker Desktop or Docker Engine and a GPU to run models effectively on a local machine.

convnet-drawer is a Python script specifically developed for creating visual representations of Convolutional Neural Networks (CNNs). It enables users to generate detailed illustrations of their CNN models by leveraging Keras-like model definitions as input. The tool provides flexibility in output formats, allowing users to save their generated diagrams as scalable vector graphics (SVG) or Microsoft PowerPoint (PPTX) files. This script is inspired by and builds upon concepts from the draw_convnet project, aiming to simplify the visualization process for CNN architectures.

claude-code-reverse is a specialized tool designed to visualize and understand the internal workings of Claude Code's Large Language Model (LLM). It enables users to reverse engineer and analyze the behavior of this AI model, providing insights into its decision-making processes. The tool features an interactive visualization component, allowing for a dynamic exploration of the reverse engineering analysis results. This capability is particularly useful for understanding how code execution is handled within the Claude environment.

ClawRouter is an agent-native LLM router specifically designed to power AI agents like OpenClaw. Its core function is to intelligently route requests to the most appropriate large language model (LLM) based on various criteria, ensuring optimal performance and cost efficiency. The platform boasts a sophisticated 15-dimension scoring system to evaluate and select models. It supports a wide array of LLMs, with compatibility for over 41 different models, and uniquely allows this without requiring individual API keys for each model. This makes it a powerful tool for developers and organizations looking to streamline their LLM operations and reduce expenses.

FlappyBirdRL is an open-source project designed for experimenting with reinforcement learning by training an AI agent to play the classic Flappy Bird game. It provides the necessary code and instructions for users to set up a local server, run the game, and integrate their AI agent. This tool is ideal for those interested in understanding and applying reinforcement learning concepts in a practical, engaging environment.

Motorhead is a specialized server designed to enhance the memory and information retrieval capabilities of Large Language Models (LLMs). Its primary function is to assist in memory handling, which is crucial for developing robust and efficient chat applications powered by LLMs. By providing a dedicated solution for memory management, Motorhead aims to simplify the development process for engineers working on LLM-based applications, ensuring that conversational AI can maintain context and recall information effectively over time.

ITS42 is a software development company that focuses on leveraging artificial intelligence and blockchain technologies to create advanced platforms for businesses. Their services span the entire product development lifecycle, from initial conceptualization to the delivery of a fully developed product. The company aims to provide innovative solutions that are also user-friendly, catering to businesses looking to integrate cutting-edge AI and blockchain capabilities into their operations.

FastAI.jl is a comprehensive Julia library that facilitates the entire deep learning pipeline. It provides tools to efficiently load and preprocess datasets, and then train deep learning models with ease. The library is built with reusable components, allowing for flexible customization to suit various project needs. It specifically supports common deep learning applications such as computer vision and natural language processing tasks, making it a versatile tool for researchers and developers working with Julia.