Coding & Development

stock-rnn is an open-source project designed to predict stock market prices using Recurrent Neural Network (RNN) models. It specifically employs multilayer Long Short-Term Memory (LSTM) cells, a type of RNN architecture well-suited for sequential data like stock prices. The tool also offers optional multi-stock embeddings, allowing for more complex analysis across different stocks. Its primary purpose is to serve as a practical demonstration for building and training an RNN model within the Tensorflow framework, providing a hands-on platform for users to experiment with stock price prediction methodologies.

IntelliSafe Analytics has developed patented, human-centered AI wearable technology designed to collect and analyze physiological and behavioral data from workers. This advanced system aims to predict and prevent workplace accidents and injuries by understanding human factors, which account for 80% or more of incidents. Unlike traditional safety measures, IntelliSafe doesn't require knowledge of specific hazards; instead, it interprets the human body's responses to any encountered hazard. The technology captures data at high rates, up to 64 readings per second, and uses this information to power its artificial intelligence models. It also notifies safety personnel of incident locations for quick response, protecting workers from common issues like slips, trips, overexertion, and heat stress.

The-NLP-Pandect is a comprehensive reference for Natural Language Processing (NLP) topics, designed to serve as an encyclopedia for users seeking online resources. It aggregates links to various open-source projects, insightful blog posts, and curated collections of NLP-related materials. This tool aims to simplify the discovery of relevant information and tools within the NLP domain, making it easier for researchers, students, and developers to find what they need. By centralizing these resources, The-NLP-Pandect helps users navigate the vast landscape of NLP, fostering learning and development in the field.

Jiva.ai is a no-code AI platform designed for non-technical users to create, test, and deploy AI models. It allows businesses to turn documents into data, classify content, and predict outcomes using AI trained on their specific examples. The platform emphasizes data sovereignty, enabling models to be trained and run directly within a user's environment, ensuring data never leaves their infrastructure. Jiva.ai offers features like document extraction, content classification, and predictive AI, with the option to build visually or use an AI assistant (SIA) for guidance. It's suitable for various industries, including healthcare, insurance, finance, and government, providing solutions for tasks like invoice processing, tailored search, and transaction categorization.

tanuki.py is a tool designed to assist developers in building and optimizing applications powered by Large Language Models (LLMs). Its core focus is on enhancing the efficiency and cost-effectiveness of LLM-based applications. The tool incorporates prompt engineering techniques to refine LLM interactions and improve output quality. Additionally, it supports test-driven alignment, a methodology that ensures the LLM's behavior aligns with desired outcomes through systematic testing. This approach helps developers to iteratively improve the performance and reliability of their LLM applications, making them more robust and production-ready.

TensorLayerX is a versatile multi-backend AI framework designed for deep learning and reinforcement learning. It supports popular frameworks like TensorFlow, PyTorch, MindSpore, PaddlePaddle, OneFlow, and Jittor, enabling users to run their models on diverse hardware such as Nvidia-GPU, Huawei-Ascend, and Cambricon. Key features include compatibility across various AI chips and platforms, a Model Zoo offering classic and state-of-the-art models for CV, NLP, and RL, and deployment support via ONNX protocol for model export, import, and deployment. Developed by researchers from leading universities, TensorLayerX simplifies the process of defining models and switching between backends with minimal code changes.

TextBox 2.0 is an open-source text generation library built with Python and PyTorch, designed to provide a unified and standardized pipeline for applying pre-trained language models (PLMs) to various text generation tasks. It supports 13 common text generation tasks, including translation, story generation, and style transfer, across 83 widely-used datasets. The library integrates 47 pre-trained language models/modules, covering categories such as general, translation, Chinese, dialogue, controllable, distilled, prompting, and lightweight models. TextBox 2.0 also offers four pre-training objectives and four efficient training strategies, including distributed data parallel and hyper-parameter optimization, making it a comprehensive tool for researchers and developers in natural language processing.

Jump AI Technologies provides an AI platform designed to seamlessly integrate artificial intelligence capabilities into current business operations. The platform focuses on assisting businesses in their transition from conventional methodologies to more efficient, AI-enhanced processes. By leveraging AI, Jump AI Technologies aims to unlock new opportunities for growth and deliver tangible results across various business functions. While specific features are not detailed on the available website content, the overarching goal is to empower organizations to adopt AI for improved performance and innovation.

THUMT is an open-source toolkit for neural machine translation developed by the Natural Language Processing Group at Tsinghua University. It provides implementations in PyTorch and TensorFlow, supporting models like Transformer, sequence-to-sequence, and standard attention-based models. Key features include multi-GPU training and decoding, multi-worker distributed training, mixed precision training, model ensemble, gradient aggregation, and TensorBoard for visualization. The toolkit is designed for researchers and developers working on machine translation, offering a robust platform for experimenting with and building advanced MT systems.

TensorFlow-LiveLessons is a comprehensive repository designed to accompany the "Deep Learning with TensorFlow" video series. It serves as a valuable resource for individuals looking to understand and implement deep learning concepts using the TensorFlow framework. The repository contains practical code examples and educational materials, making complex topics accessible. It specifically covers applications in natural language processing, providing users with hands-on experience in this critical AI domain. A second edition of the video series is also available, ensuring the content remains current and relevant with the latest advancements in deep learning.

tensor_parallel is a Python library designed to automatically split PyTorch models across multiple GPUs, facilitating both training and inference for large language models (LLMs). This tool allows users to run models that would otherwise exceed the memory capacity of a single GPU, offering potentially linear speedups. It simplifies the process with a single line of code integration and supports memory-efficient dispatch by converting state_dicts. Key features include options for custom parallelism strategies, distributed training with `torch.distributed`, and sharding parameters using the ZeRO-3 algorithm to avoid duplicate parameters. It is particularly useful for quick prototyping on a single machine with multiple GPUs, offering an easier setup compared to more complex distributed training frameworks.

Apache TVM is an open machine learning compilation framework designed for Python-first development, allowing for quick customization of machine learning compiler pipelines. It focuses on universal deployment, enabling models to be integrated into minimum deployable modules. The project has evolved significantly, now featuring TensorIR as a tensor-level representation and Relax as a graph-level representation, with a strong emphasis on Python-first transformations. This design makes ML compilers more accessible by allowing most transformations to be customizable in Python, optimizing computational graphs, tensor programs, and libraries. TVM also serves as a foundational infrastructure for building Python-first vertical compilers, particularly for domains like Large Language Models (LLMs).

Twitter-Insight-LLM is an open-source project designed for comprehensive Twitter data management and analysis. It facilitates fetching liked tweets using Selenium, saving this data into structured JSON and Excel files for easy access. Beyond basic data ingestion, the tool supports initial data analysis, allowing users to gain insights from their collected Twitter data. A standout feature is its experimental embedding-based image search, which enables natural language queries for unlabeled images without requiring GPU support. This functionality supports multiple languages, enhancing its utility for diverse users. The project also integrates with OpenAI API for image captioning, providing a robust solution for understanding and organizing visual content from Twitter.

OPNTEC GmbH specializes in developing open AI technologies and IoT solutions, focusing on sustainable open-source software and open-hardware products. Based in Berlin, they collaborate with a global community to create technologies that benefit customers and society. Their offerings include the Pocket Science Lab for collecting environmental data, controlling robots, and industrial automation, featuring instruments like oscilloscopes and multimeters. For AI, OPNTEC provides open-source voice AI solutions and data visualization models, enabling companies to automate processes and enhance customer experience while keeping data in-house. They offer customized AI solutions for various business problems, including search, big data, and interactive voice assistants, with a strong emphasis on data privacy and independence from external cloud providers. OPNTEC also provides expertise in IoT hardware development, machine learning, and industry-specific AI solutions like image and voice recognition.

Medius.si is a boutique custom software development company with cross-industry expertise, focusing on accelerating out-of-the-box thinking for medium and enterprise companies. They specialize in data-driven technologies, including enterprise applications, big data, machine learning & AI, IIoT & IoT, blockchain development, SaaS/PaaS development, and marketplace solutions. Medius.si also engages in R&D projects and development for startups. Their solutions are used across various sectors such as manufacturing, transportation, telecom, electricity and gas, government, and insurance, with a strong track record of successful enterprise projects and a high rate of returning customers.

Soterix Systems introduces NexaiQ, an advanced AI Cloud Computing Platform designed to revolutionize security, operations, and decision-making. NexaiQ integrates seamlessly with existing cameras, IoT devices, and access control systems, leveraging true AI to reduce costs, enhance security, and provide real-time insights without requiring new hardware. The platform offers 100% browser-based management, minimizing false alarms, detecting anomalies, and automating responses. It extends the life of current infrastructure, ensuring a low total cost of ownership and maximizing ROI. NexaiQ provides AI-enhanced monitoring for video surveillance, manages devices and permissions with advanced automation, and is enterprise-ready with flexible cloud deployment. Its capabilities include advanced security threat detection, business intelligence for operational optimization, security intelligence for incident detection and compliance, License Plate Recognition (LPR), loitering and behavior analysis, and smart attribute classification.

whishper is an open-source, local-first tool designed for transcribing and translating audio to text. It operates entirely on your local machine, ensuring privacy and efficiency by not sending data to external servers. The tool provides a user-friendly web interface where users can upload audio files, generate transcriptions, and then translate them. A key feature is the ability to edit subtitles directly within the application, offering granular control over the output. Powered by whisper models, whishper delivers accurate results and is ideal for individuals or organizations that require secure, offline audio processing capabilities for various content creation needs.

GitHub is a leading platform for software development, offering robust tools for version control, collaboration, and project management. It enables developers to host public and private repositories, automate CI/CD pipelines with GitHub Actions, and secure their code with features like Dependabot and Advanced Security. The platform supports various team sizes, from individual developers to large enterprises, providing features like code review, issue tracking, and instant dev environments with Codespaces. GitHub also offers AI-powered tools like Copilot for code creation and GitHub Models for integrating AI into workflows, making it a versatile solution for modern software development.

SmoothQuant is an open-source project from MIT Han Lab, designed to provide accurate and efficient post-training quantization for large language models (LLMs). This tool enables INT8 model inference and W8A8 quantization, significantly reducing the memory footprint and computational cost of LLMs like Llama, Falcon, Mistral, and Mixtral, all while maintaining minimal loss in accuracy. It is particularly valuable for developers and researchers working with large models who need to optimize performance for deployment on resource-constrained hardware or to achieve faster inference speeds. The project is available on GitHub, making it accessible for community contributions and widespread adoption in AI development.

Web Stable Diffusion is an innovative project that enables stable diffusion models to run directly within web browsers, eliminating the need for server-side computation. This client-side execution offers significant benefits such as reduced operational costs for service providers, enhanced personalization, and improved privacy protection. The project leverages WebAssembly and WebGPU to achieve native GPU execution in the browser, making advanced AI capabilities accessible without specialized client applications. It provides a Python-first, hackable, and composable workflow for developing and optimizing these models, ensuring universal deployment across various environments, including the web. The system is built on open-source technologies like PyTorch, Hugging Face diffusers, and Apache TVM Unity, allowing for efficient model import, optimization, and deployment.

Prompt Token Counter for OpenAI Models is an online tool designed to help users accurately count tokens for their prompts when working with OpenAI language models. This is crucial for staying within the models' token limits, which prevents requests from being rejected and helps manage costs effectively, as many models charge based on token usage. The tool allows users to input their prompt and instantly see the token count across various OpenAI models, including GPT-4o, GPT-4, and others. It also provides valuable information on what tokens are, why counting them is important, and best practices for optimizing prompts to fit within limits, ensuring efficient and cost-effective interactions with AI models.

X-Adapter is an open-source tool designed to bridge the compatibility gap between diffusion model plugins and upgraded diffusion models. It enables plugins, such as ControlNet and LoRA, that were originally trained on older versions (e.g., Stable Diffusion 1.5) to function seamlessly with newer, upgraded models like SDXL, eliminating the need for extensive retraining. This significantly enhances the flexibility and utility of existing plugins, allowing developers and researchers to leverage their work across different model generations. The project, presented at CVPR 2024, provides inference code, setup instructions, and guidance for integrating various plugins, making it a valuable resource for those working with diffusion models.

Tidyquant is a strategic AI consultancy focused on assisting enterprises in the design and deployment of advanced artificial intelligence systems. Their expertise lies in creating AI solutions that are not only scalable but also adhere to principles of governance, explainability, and security. The company aims to help businesses integrate AI effectively, ensuring that these systems are robust, transparent, and trustworthy. Tidyquant's services are tailored to meet the complex demands of large organizations looking to leverage AI for strategic advantage, emphasizing practical implementation and long-term viability.

Canopy Security offers advanced monitoring solutions specifically designed for pickup truck beds. The core product, Pickup Cam, provides HD live streaming, AI-powered intrusion notifications, and cloud video evidence. The system is engineered for easy DIY setup and features continuous power via a rechargeable battery that charges from the vehicle's OBD-II port. Canopy's AI is calibrated to monitor the truck bed specifically, intelligently ignoring activity outside of it, and automatically saves video clips of potential intrusions to the cloud. While the Pickup Cam is no longer available for sale due to Ford's acquisition of Canopy, the technology aims to provide peace of mind for tradespeople and truck owners by securing their valuable tools and equipment.