Coding & Development

Transformer-MM-Explainability is an official PyTorch implementation for Generic Attention-model Explainability for Interpreting Bi-Modal and Encoder-Decoder Transformers. This open-source project offers a novel method to visualize and understand the decision-making processes of any Transformer-based network. It includes practical examples for popular models such as DETR, VQA, CLIP, and LXMERT, making it a valuable resource for researchers and developers working with multi-modal and encoder-decoder architectures. The tool provides notebooks for easy experimentation and reproduction of results, with clear instructions for setting up environments and running examples on GPUs, including Colab support.

SoTA-Point-Cloud is a GitHub repository offering an extensive survey of deep learning techniques applied to 3D point clouds. Published in IEEE TPAMI 2020, this resource covers major tasks such as 3D shape classification, 3D object detection, and 3D point cloud segmentation. It provides comparative results on numerous publicly available datasets, including ModelNet, KITTI, and Semantic3D. The repository also offers insightful observations and outlines future research directions, making it an invaluable resource for researchers and practitioners in the field of 3D computer vision. The maintainers regularly update the page with new results and suggestions.

UAV_Obstacle_Avoiding_DRL is a comprehensive open-source project focused on developing deep reinforcement learning algorithms for autonomous obstacle avoidance in Unmanned Aerial Vehicles (UAVs). It addresses both static and dynamic environments, offering multiple approaches for each. For static environments, the project explores Multi-Agent Reinforcement Learning (MADDPG, DDPG, TD3) combined with artificial potential field algorithms. In dynamic settings, it utilizes disturbed flow field algorithms alongside single-agent reinforcement learning (PPO+GAE, TD3, DDPG, SAC). The project also includes implementations of traditional path planning methods like A* search, RRT, Ant Colony Algorithm, and D* algorithm for comparison, highlighting the superior performance of reinforcement learning approaches. It provides both MATLAB and Python implementations for various algorithms, making it a valuable resource for researchers and developers in UAV navigation.

This project implements a Stock Trading Bot utilizing Deep Reinforcement Learning, specifically Deep Q-learning. It's designed for learning and experimentation, keeping the implementation simple and close to the algorithm discussed in research papers. The bot allows users to create intelligent agents that learn from market data, making decisions to buy, sell, or hold based on observed states. It incorporates several improvements to the Q-learning algorithm, including Vanilla DQN, DQN with fixed target distribution, Double DQN, Prioritized Experience Replay, and Dueling Network Architectures. Users can train the agent on historical data and evaluate its performance, with visualizations available for model evaluations. It's a valuable resource for those interested in applying reinforcement learning to financial trading.

ChessGPT.ai provides an engaging platform for chess enthusiasts to challenge an AI opponent. By integrating ChatGPT with a chessboard, the tool offers a unique conversational gameplay experience, allowing users to interact with the AI. The website claims the AI can beat almost any human, including the user, and even states that ChessGPT evolved into Stockfish with an ELO of 249. Users can play games, view the board, and submit their scores to a leaderboard. The tool also features an audio track for an immersive experience, suggesting the use of headphones for optimal enjoyment.

Postlog offers web traffic analysis tools designed to improve website rankings and click-through rates (CTR). Users can utilize a free Post Log diagnostics analyzer to instantly uncover factors hindering their traffic, rankings, or CTR. The tool provides quick triage, pointing users to the fastest free fixes like internal linking or CTR optimization. It helps operators identify bottlenecks and offers access to operational tools that drive measurable growth. Postlog focuses on providing organized intelligence and ROI-focused recommendations, ensuring that every featured tool contributes to business outcomes. It categorizes SAAS analytics tools to help users find the right solutions for specific use cases, emphasizing action-ready comparisons and direct links to save research time.

zh-NER-TF is an open-source project offering a straightforward character-based BiLSTM-CRF model specifically designed for Chinese Named Entity Recognition (NER). This TensorFlow-based tool aims to identify three key entity types: PERSON, LOCATION, and ORGANIZATION within Chinese text. The model utilizes a look-up layer for character embeddings, a BiLSTM layer to extract features from both past and future input, and a CRF layer to ensure grammatically correct tag sequences, addressing limitations of simpler Softmax layers. It includes preprocessed data files and a vocabulary for easy setup, and users can train, test, or demo the model with their own datasets after transforming them into the specified format. The repository provides instructions for running the model and evaluating its performance.

KushoAI offers AI-native infrastructure designed to enhance software reliability and security by integrating autonomous agents directly into CI/CD pipelines. It automates critical software maintenance tasks such as API contract testing, continuous security scanning, and comprehensive end-to-end workflow validation across APIs, databases, and UI layers. A key differentiator is its self-healing testing infrastructure, which automatically adapts tests as APIs evolve, preventing test breakage and ensuring the test suite remains current. KushoAI also provides release intelligence with AI-computed risk scores, helping engineering leaders make confident ship or no-go decisions. It supports enterprise-grade security, governance, and offers both cloud and on-premise deployment options.

write-you-a-vector-db is a comprehensive tutorial designed to guide users through the process of integrating vector capabilities into relational database systems. The tutorial is built upon modified versions of educational database systems, specifically CMU-DB's BusTub for the C++ variant and RisingLight for the upcoming Rust version. Users will learn to implement vector storage, vector expressions, and vector indexes. This resource is ideal for those looking to deepen their understanding of vector database implementation, offering practical, hands-on experience. The project is actively developed and encourages community participation through a dedicated Discord server.

Ant Design X is an open-source project focused on simplifying AI interface development, offering a rich set of atomic components for various interaction stages based on the RICH interaction paradigm. It helps developers build excellent AI interfaces and pioneer intelligent new experiences. The tool includes `@ant-design/x-sdk` for managing AI application data streams, `@ant-design/x-markdown` for a streaming-friendly Markdown renderer, `@ant-design/x-card` for dynamic card rendering based on the A2UI protocol, and `@ant-design/x-skill` for an intelligent skill library to improve development efficiency. It is widely used in AI-driven user interfaces within Ant Group.

xLearn is a robust, high-performance machine learning package developed in C++ for maximum CPU and memory utilization. It includes implementations of linear models (LR), factorization machines (FM), and field-aware factorization machines (FFM), making it ideal for solving large-scale machine learning problems, particularly with high-dimensional sparse data common in recommendation systems. The package is designed for ease of use, requiring no third-party libraries for compilation and offering simple Python and CLI interfaces. xLearn also boasts scalability, supporting out-of-core training to handle terabytes of data by leveraging disk storage, and includes features like cross-validation and early-stop mechanisms.

Yellowbrick is an open-source suite of visual diagnostic tools, known as "Visualizers," designed to enhance the machine learning model selection process. It seamlessly integrates with scikit-learn and matplotlib, allowing users to generate insightful visualizations for their machine learning workflows. The tool supports various visualizers for feature analysis, such as Rank2D for pairwise feature comparisons, and model evaluation, like ROCAUC for classifier sensitivity and specificity. Yellowbrick is compatible with Python 3.4 or later and can be easily installed via pip or conda. It also provides access to several datasets for examples and testing, making it a comprehensive solution for data scientists and developers looking to visually steer their model development.

The Yi series models are a collection of open-source large language models developed from scratch by 01.AI. These models are designed to be bilingual, trained on a 3T multilingual corpus, and excel in language understanding, commonsense reasoning, and reading comprehension. The Yi-34B-Chat model has demonstrated strong performance, ranking highly on leaderboards like AlpacaEval. The series includes both chat-optimized and base models, with options for different parameter sizes (6B, 9B, 34B) and context window lengths (up to 200K). Yi models are built on the Transformer architecture, similar to Llama, but are not derivatives, utilizing independently created training datasets and infrastructure. They are available for deployment via pip, Docker, conda-lock, and llama.cpp, and can be fine-tuned or quantized for specific needs.

ZynqNet is an open-source project stemming from a Master Thesis, focusing on FPGA-accelerated embedded convolutional neural networks. It provides a comprehensive solution for image classification on embedded systems, featuring the ZynqNet CNN, an optimized and customized CNN topology, and the ZynqNet FPGA Accelerator, an FPGA-based architecture for its evaluation. The project also includes the Netscope CNN Analyzer, a custom tool for visualizing, analyzing, and editing CNN topologies. ZynqNet is designed for high efficiency, achieving 84.5% top-5 accuracy with minimal computational complexity, making it ideal for real-time and power-constrained applications. The repository offers the full project report, CNN prototxt, pretrained weights, HLS C++ source code for the accelerator, and firmware for the Zynq XC-7Z045 ARM processors.

Indie Panel offers a centralized dashboard for indie developers to manage all their projects. It provides seamless integration with various databases, including Neon, Supabase, and PostgreSQL, allowing users to track essential metrics such as total users, paid conversions, and growth trends. The tool delivers real-time data with automatic caching and daily snapshots, ensuring up-to-date insights. Security is prioritized with AES-256-GCM encryption for all connection strings. Indie Panel simplifies project management by consolidating user metrics and growth monitoring into one intuitive interface, helping developers make informed decisions about their applications.

WorkPing automates the creation of client-ready progress updates directly from GitHub activity. Designed for freelance developers, it analyzes merged pull requests and commits to generate professional summaries. Users can review and edit these AI-generated updates in a clean editor before copying and sending them via email, Slack, or other platforms. The tool offers secure, read-only access to GitHub repositories, including private ones, and allows for the addition of manual notes for non-code work like meetings or blockers, ensuring comprehensive reporting. WorkPing aims to streamline client communication, allowing developers to focus more on their work and less on administrative tasks.

keras-mmoe provides a TensorFlow Keras implementation of the "Modeling Task Relationships in Multi-task Learning with Multi-gate Mixture-of-Experts" paper (KDD 2018). This open-source repository offers a Python 3.6 implementation, also compatible with Python 2.7, making it accessible for various development environments. It includes an example demo for running the model with the census-income dataset from UCI, which is the same dataset used in Section 6.3 of the original paper. The code is well-documented and designed for easy extension, encouraging contributions from the community for performance improvements, benchmark accuracy, and training on other public datasets. This tool is ideal for developers and researchers working on deep learning and multi-task learning applications.

artyom.js is a robust and constantly updated open-source JavaScript library that wraps the webkitSpeechRecognition and speechSynthesis APIs. It enables developers to integrate voice control, voice commands, speech recognition, and speech synthesis into their web applications. Key features include quick recognition of voice commands, easy addition of dynamic commands, smart commands with wildcards and regular expressions, and the ability to convert voice to text. The library supports synthesizing large blocks of text and works on both desktop browsers and mobile devices. It offers support for multiple languages and provides options for continuous listening, soundex algorithm for accuracy, and a remote command processor. Developers can create custom voice assistants similar to Siri, Google Now, or Cortana within their websites.

temperature_scaling is an open-source Python module designed to calibrate neural networks by adjusting their confidence scores. Originally created as a demonstration for PyTorch 0.3, it implements temperature scaling, a post-processing technique that divides logits by a learned scalar parameter to minimize negative log-likelihood on a validation set. This helps address the common issue of neural networks outputting overconfident probabilities, ensuring that confidence scores better match true correctness likelihood. While the repository is unmaintained, it offers a clear example of how to integrate temperature scaling into a project for improved model calibration.

Applying_EANNs is a 2D Unity simulation designed to showcase how cars can learn to navigate various courses. The cars are controlled by a feedforward neural network, whose weights are optimized using a modified genetic algorithm. This project provides a practical demonstration of evolutionary artificial neural networks in a simulated environment. Users can tinker with simulation parameters in the Unity Editor or run the built executable with default settings. The neural network architecture includes an input layer, two hidden layers, and an output layer, with its training managed by a customizable genetic algorithm. The user interface displays real-time data for the best performing car, including neural network output, evaluation value, and a generation counter, along with a visual representation of the neural network's weights.

AskEllyn is the world's first AI companion specifically designed for individuals diagnosed with breast cancer and their caregivers. This non-medical, private, and free tool offers knowledge, wisdom, and advice from survivors, aiming to ensure no one goes through breast cancer alone. It functions as an empathetic digital friend, validating experiences and empowering users to self-advocate. AskEllyn speaks every language and is committed to remaining free forever through The Lyndall Project, a registered non-profit organization. Beyond the AI chatbot, the platform also features a blog with articles on breast cancer, a best-selling book by Ellyn Winters-Robinson, and speaking engagement opportunities.

Logparser provides a comprehensive machine learning toolkit designed for automated log parsing, a critical step in structured log analytics. It enables users to automatically extract event templates from unstructured logs and transform raw log messages into a sequence of structured events. This process is also known as message template extraction, log key extraction, or log message clustering. The toolkit includes various log parsers, such as SLCT, AEL, IPLoM, LKE, Spell, Drain, and DivLog, each backed by academic research. It supports Python 3 and offers benchmarks for evaluating parsing accuracy, making it suitable for both research and practical application in log analysis.

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.