Coding & Development

Awesome-Adaptation-of-Agentic-AI is a curated repository featuring a comprehensive list of academic papers focused on the adaptation strategies of agentic AI systems. This resource is designed for researchers and practitioners interested in the evolving field of agentic AI, offering insights into various adaptation methods. The repository categorizes papers based on agent adaptation (tool execution signaled, agent output signaled) and tool adaptation (agent-agnostic, agent-supervised), detailing development timelines, methods, venues, tasks, tools, agent backbones, and tuning techniques. It serves as a valuable reference for understanding the latest advancements and research trends in making AI agents more adaptive and intelligent.

Ragnexus specializes in building customized personal assistants powered by Retriever-Augmented Generation (RAG) technology. These bespoke AI systems are designed to deliver highly personalized and contextually relevant responses by utilizing private customer information. The platform aims to improve efficiency and productivity by providing accurate information quickly, enhance customer experience through tailored solutions, and reduce costs by automating repetitive tasks. Ragnexus integrates seamlessly with over 40 existing platforms, including Asana, Confluence, Dropbox, GitHub, Google Drive, Jira, Notion, Salesforce, Slack, AWS S3, and Zendesk, eliminating the need for internal AI infrastructure development.

chess-alpha-zero is an open-source project dedicated to chess reinforcement learning, implementing methods inspired by DeepMind's AlphaGo Zero. It allows users to train AI models to play chess through self-play, supervised learning, and distributed training. The project provides a modular architecture with 'self' for data generation, 'opt' for model training, and 'eval' for model evaluation. It supports Python 3.6.3, TensorFlow-GPU, and Keras, making it suitable for developers and researchers interested in AI game development and machine learning applications in strategic games. The tool also offers a Universal Chess Interface (UCI) for integration with chess GUIs, allowing users to observe and interact with the trained AI.

Business Machine Learning (BML) and Business Data Science (BDS) Applications is a comprehensive, open-source resource available on GitHub, offering a curated list of practical applications across diverse business functions. This repository provides insights and examples for Accounting, Customer, Employee, Legal, Management, and Operations, making it a valuable reference for professionals and researchers. It details specific projects such as predictive modeling with GitHub logs, satellite data analysis for financial forecasting, and data imputation techniques. The resource also highlights opportunities for collaboration with Sov.ai, a company focused on integrating advanced machine learning with financial data analysis, and includes a wide range of research and project opportunities.

C-Plus-Plus is an open-source repository on GitHub providing a comprehensive collection of algorithms implemented in C++. Designed for educational purposes, it covers a wide range of topics including mathematics, machine learning, computer science, and physics. The repository features well-documented source code with detailed explanations, making it a valuable resource for both educators and students. Each algorithm implementation is atomic, utilizing STL classes without external library dependencies, which allows for in-depth study of the fundamentals. The code adheres to the C++17 standard, ensuring portability across various operating systems and embedded systems like ESP32 and ARM Cortex. It also includes self-checks for implementation correctness and modular designs for easy integration into other applications. Online documentation is generated directly from the source code, offering snippets, execution details, diagrams, and links to C++ STL library functions.

mdlm is an open-source masked discrete diffusion language model (MDLM) that features a novel substitution-based parameterization. This approach simplifies the absorbing state diffusion loss to a mixture of classical masked language modeling losses, leading to state-of-the-art perplexity numbers on LM1B and OpenWebText among diffusion models. It also achieves competitive zero-shot perplexity with state-of-the-art autoregressive models on various datasets. The repository provides the MDLM framework, simplified loss calculation, baseline implementations, and efficient samplers that make MDLM significantly faster than existing diffusion models, including semi-autoregressive generation capabilities.

DeepCTR-Torch is a comprehensive, open-source Python package designed for building and experimenting with deep learning-based Click-Through Rate (CTR) models, leveraging the PyTorch framework. It offers a modular and extensible architecture, allowing users to easily implement and customize a wide range of CTR models, including popular architectures like DeepFM, xDeepFM, and Wide & Deep. The package includes numerous core component layers, enabling data scientists and researchers to construct their own custom models efficiently. With its user-friendly API, DeepCTR-Torch simplifies the process of training and predicting with complex models using standard `model.fit()` and `model.predict()` functions, making it an invaluable tool for recommendation systems and advertising applications.

dl-docker offers an all-in-one Docker image designed for deep learning, simplifying the setup process by pre-packaging popular frameworks such as TensorFlow, Caffe, Theano, Keras, and Torch. It supports both CPU and GPU configurations, with the GPU version including CUDA 8.0 and cuDNN v5. The image also comes with essential libraries like iPython/Jupyter Notebook, Numpy, SciPy, Pandas, Scikit Learn, Matplotlib, and OpenCV. Users can either pull pre-built CPU images from Docker Hub or build both CPU and GPU versions locally. This solution addresses the 'dependency hell' often encountered when installing multiple deep learning frameworks, providing an isolated and fully functional OS environment for development.

deep-learning-uncertainty is an open-source repository dedicated to predictive uncertainty estimation in deep learning models. It offers a comprehensive literature survey, detailed paper reviews, and experimental setups for various baseline methods. The repository also includes a collection of implementations, making it a valuable resource for researchers and engineers. This tool is designed to help users understand, quantify, and improve the reliability of predictions made by deep learning models, addressing critical aspects of model trustworthiness and robustness. It serves as a central hub for exploring established and emerging techniques in uncertainty quantification.

Domain Transfer Network (DTN) is a TensorFlow-based implementation for unsupervised cross-domain image generation. This tool enables users to transfer image characteristics from one domain to another, such as converting SVHN images to MNIST, without requiring paired training data. It is designed for researchers and developers interested in image synthesis and domain adaptation, providing a practical framework for experimenting with generative models. The repository includes Python scripts for dataset download, preprocessing, model pretraining, training, and evaluation, making it a comprehensive resource for those working with generative adversarial networks (GANs) and similar architectures.

MOA (Massive Online Analysis) is a popular open-source framework designed for Big Data stream mining. It provides a comprehensive suite of machine learning algorithms, including classification, regression, clustering, outlier detection, concept drift detection, and recommender systems. Built in Java, MOA is related to the WEKA project but is specifically engineered to handle more demanding, large-scale, and real-time data stream processing challenges. The framework is extensible, allowing users to integrate new mining algorithms, stream generators, or evaluation measures, and serves as a benchmark suite for the stream mining community.

faceID_beta is an open-source project available on GitHub that provides an implementation of iPhone X's FaceID technology. It leverages face embeddings and siamese networks, processing RGBD images for facial recognition. The project is primarily presented as a Jupyter Notebook file, with an automatically generated Python file also available. This makes it particularly suitable for developers and researchers interested in understanding and experimenting with advanced facial recognition techniques. The repository includes details on the implementation and encourages users to explore the notebook version for a clearer understanding of the code's structure and functionality.

DriveLM is an open-source project focused on advancing autonomous driving research through Graph Visual Question Answering (GVQA). It provides comprehensive datasets, DriveLM-Data, built upon nuScenes and CARLA, specifically designed for driving with language. The project also offers DriveLM-Agent, a VLM-based baseline approach for jointly performing GVQA and end-to-end driving. DriveLM serves as a main track in the CVPR 2024 Autonomous Driving Challenge, offering a baseline, test data, submission format, and evaluation pipeline. It addresses the community's challenges by providing a benchmark for driving with language, exploring embodied applications of LLMs/VLMs, and investigating closed-loop planning with language.

ml5-library is an open-source JavaScript library designed to make machine learning accessible to a broad audience, including artists, creative coders, and students. It provides pre-built functions and models for various machine learning tasks, allowing developers to integrate capabilities like image recognition, pose estimation, and sound analysis directly into web applications. The library is built on top of TensorFlow.js and emphasizes ethical computing, with documentation addressing data bias and responsible usage. ml5.js is heavily inspired by Processing and p5.js, fostering a friendly and welcoming environment for contributors and users alike. It offers code examples, tutorials, and sample datasets to aid in learning and implementation.

Graph-Adversarial-Learning is a comprehensive, curated collection of resources dedicated to adversarial attacks and defenses on graph data. This GitHub repository serves as a valuable hub for researchers and developers interested in understanding and mitigating vulnerabilities in graph-based machine learning models. It categorizes papers by year, venue, and includes those with associated code, spanning from 2017 to 2023. The collection covers various aspects such as attack techniques, defense strategies, robustness certification, and stability. It also provides a quick look at recently updated papers, making it an essential reference for staying current in the field of graph adversarial learning.

Notebooks On The Hub is an AI application hosted on Hugging Face, designed to provide users with a platform for accessing and exploring AI notebooks. It enables users to create and customize static web pages by directly editing HTML files within the platform. This functionality is accessible through the Files and versions tab, allowing for immediate viewing of changes on the web page. The tool is part of the Hugging Face Spaces ecosystem, indicating its focus on community and collaborative development within the AI domain. It is particularly useful for individuals looking to experiment with or share AI-related code and demonstrations in an easily accessible web environment.

Pangea is a fully open multilingual multimodal LLM developed by NeuLab at LTI/CMU, supporting 39 languages. It is designed for research and development in multilingual AI, offering a simple interface for text translation. Users can input text, select source and target languages, and receive a translated version. The tool is available as a Hugging Face Space, making it accessible for experimentation and integration into various projects. Its open-source nature under the Apache 2.0 license encourages diverse language applications and collaborative development within the AI community.

MobiLlama is an open-source small language model (SLM) specifically designed for efficient deployment on edge devices. It addresses the limitations of larger LLMs by focusing on reduced memory footprint, energy efficiency, and faster response times, making it ideal for privacy-sensitive and resource-constrained environments. MobiLlama offers models ranging from 0.5 billion to 1.2 billion parameters, demonstrating superior performance compared to other SLMs in its class. The project provides fully transparent training and evaluation scripts, pre-trained models, and even an Android APK for mobile integration, making it accessible for developers and researchers working on on-device AI applications.

mlops-on-gcp is a GitHub repository by Google Cloud Platform dedicated to MLOps. It offers a comprehensive collection of hands-on labs and code samples, showcasing best practices and effective patterns for implementing and operationalizing production-grade machine learning workflows on Google Cloud Platform. The repository is structured into two main sections: mini-workshops for instructor-led learning and code samples that illustrate various ML Engineering topics. This resource is ideal for developers and data scientists looking to deploy and manage machine learning models efficiently within the Google Cloud ecosystem, providing practical guidance and examples for continuous training, model serving, and more.

The MNIST Adversarial Examples Challenge is a platform designed to explore the adversarial robustness of neural networks using the MNIST dataset. It builds upon recent advancements in adversarial attacks by providing a structured challenge. Researchers are invited to submit attacks against a pre-trained, robust neural network, with the objective of finding adversarial examples that significantly reduce the network's accuracy. The platform provides both the training code and the network architecture, while initially keeping the network weights secret to encourage black-box attack development. A leaderboard tracks the most successful attacks, fostering reproducibility and empirical comparisons in the field of defense mechanisms against adversarial attacks. The challenge has evolved to include white-box attacks after the release of the secret model weights.

nn_robust_attacks is an open-source tool designed to evaluate the robustness of neural networks against adversarial attacks. It provides implementations of three attack algorithms in TensorFlow, enabling researchers and developers to find adversarial examples. The tool supports Python 3 and requires setting up models for MNIST, CIFAR, or Inception. It allows users to create a model class with a predict method to run predictions without softmax, defining image size, number of channels, and labels. The CarliniL2 attack, for instance, can be run with tunable hyperparameters to assess model vulnerabilities. This code is based on the paper "Towards Evaluating the Robustness of Neural Networks" by Nicholas Carlini and David Wagner.

SmartAIConnect offers a comprehensive platform for managing Responsible AI across the project lifecycle, particularly for computer vision initiatives. Its Project Assurance Software Solution (PASS) integrates Governance, Risk & Compliance (GRC) tools, device and system monitoring, and AI model management. Key features include a curated AI model library with risk ratings, AI model cards detailing ethics and bias, compliance questionnaires, and a two-step approval process for AI deployments. The platform supports secure deployment at scale, monitors cameras for unauthorized AI apps, and maintains a full audit trail of AI deployments and data delivery. It caters to various industries, including government, healthcare, transportation, and R&D, ensuring compliance with regulatory requirements.

Podscript is an open-source application hosted on Hugging Face Spaces, developed by Amrrs. It is designed to automate various tasks within the machine learning domain, leveraging the community-driven ML app ecosystem. While the live website currently indicates a runtime error due to hardware capacity issues, its nature as a Hugging Face Space suggests it's intended for exploration, development, and potentially educational projects. As an open-source tool, it offers transparency and the ability for users to inspect and modify its code, making it suitable for those interested in understanding the underlying mechanics of AI applications. The tool is part of a broader movement to make machine learning accessible and collaborative.

pytorch_diffusion offers a PyTorch reimplementation of Denoising Diffusion Probabilistic Models, complete with checkpoints converted from the original TensorFlow implementation. This tool allows users to load diffusion models with pretrained weights for various datasets like CIFAR-10, LSUN-bedroom, LSUN-cat, and LSUN-church. It provides a quickstart guide for running a Streamlit demo, making it accessible for immediate use. Users can also instantiate and configure the U-Net model for denoising independently. The repository includes instructions for producing samples, evaluating results against TensorFlow models, and converting TensorFlow checkpoints to PyTorch, making it a comprehensive resource for researchers and developers working with diffusion models.