Coding & Development

The Audio to Stems to MIDI Converter is a web-based AI tool hosted on Hugging Face Spaces that allows users to upload audio files and perform two key functions. First, it can separate the uploaded audio into distinct stems, such as vocals, drums, bass, or other instrumental tracks. This is particularly useful for remixing, sampling, or isolating specific parts of a song. Second, it offers the optional capability to convert these separated audio stems into MIDI files. This feature can be invaluable for musicians, producers, and audio engineers who wish to analyze melodies, harmonies, or rhythms, or to use the MIDI data for further composition and production in digital audio workstations. The tool leverages deep learning models for its audio processing capabilities, providing a streamlined workflow for audio manipulation and musical analysis.

burpgpt is a Burp Suite extension designed to integrate OpenAI's GPT capabilities directly into your security analysis workflow. This tool significantly enhances vulnerability detection by performing passive scans to identify unique and bespoke vulnerabilities that might otherwise be missed. Beyond static analysis, burpgpt also facilitates traffic-based analysis, allowing security professionals to gain deeper insights into application behavior and potential weaknesses. By leveraging the power of GPT, burpgpt aims to improve the accuracy and efficiency of security assessments, making it a valuable addition for developers and security researchers focused on identifying and mitigating complex security threats.

Chatgpt Detector is a web-based application hosted on Hugging Face Spaces, designed to determine the likelihood of a text being generated by ChatGPT. Users can input a question and an answer, and the tool will analyze the provided text to assess if it exhibits characteristics commonly found in AI-generated content. This tool is particularly useful for educators, content managers, and anyone needing to verify the originality of written material. It provides a straightforward interface for quick analysis, making it accessible for identifying potential AI plagiarism.

Hyperbolic is an open-access AI cloud platform designed for developers, researchers, and enterprises. It provides affordable on-demand GPUs for training, scaling, and serving AI models, offering H100 or H200 instances in under a minute. The platform features serverless inference for running state-of-the-art AI models at a fraction of legacy cloud costs, with full OpenAI API compatibility. Users can also secure reserved clusters for guaranteed capacity and dedicated endpoints for high-throughput inference. Hyperbolic aims to be an end-to-end infrastructure solution, supporting various models like Llama, Qwen, DeepSeek, SDXL, and Flux, and offering AI consulting services for fast-scaling teams.

emotion3D, now acquired by indie, offers advanced in-cabin perception software designed for real-time driver and passenger monitoring within vehicles. Their CABIN EYE software stack utilizes cameras to analyze human characteristics and movement, enabling innovative safety features, enhanced user experiences, and intelligent automation capabilities. Key functionalities include Driver Monitoring Systems (DMS) for a broad range of safety and UX use cases, Driver & Occupant Monitoring Systems that combine both functions with a single camera, and Shuttle Passenger Monitoring for safety and comfort in shared mobility. The technology addresses the needs of current and future occupants, setting new standards in driving safety and in-cabin user experience for the automotive industry.

audio-super-res is an open-source project that implements an audio super-resolution model using neural networks. Based on research from NeurIPS 2019 and ICLR 2017, this Python-based tool allows users to enhance the quality of audio files by generating high-resolution versions from low-resolution inputs. It supports training models on datasets like VCTK for single or multi-speaker scenarios and provides scripts for data preparation and model evaluation. The tool is suitable for researchers and developers interested in audio upscaling, time series tasks, and exploring the effects of different low-pass filters on super-resolution performance.

Greater Than provides AI-driven risk intelligence solutions for road safety and climate impact. The platform analyzes driving data to generate a Crash Probability Score and a Climate Impact Score, helping organizations understand and mitigate risks. Key features include the Crash Probability Dashboard for risk analytics, the Crash Probability Hub for risk management, and an AI Coach for behavior-based feedback. It also offers a Climate Impact Dashboard and an ESG Compliance Package for CSRD compliance. The AI technology, trained for 20 years on over 7 billion real-world patterns, measures driver influence on safety and sustainability, enabling businesses to reduce crashes, lower emissions, and optimize telematics data for insurance pricing and risk control. The tool is ideal for fleet management companies, fleet operators, brokers, and telematics partners.

NEPI by Numurus is an open-source edge AI platform designed to accelerate the development and deployment of smart systems. It enables engineers to deploy edge AI on NVIDIA Jetson and x86 hardware without building infrastructure from scratch. NEPI provides essential components such as plug-and-play hardware drivers, comprehensive AI model management, and low-code automation capabilities, all built on ROS 2. This platform handles hardware integration, AI deployment, and automation, allowing teams to build smart systems in days rather than months. It supports various applications in robotics, autonomous systems, industrial inspection, and research, running fully offline with no cloud dependency for critical real-time operations.

SiLK (Simple Learned Keypoint) is a self-supervised deep learning keypoint model from Facebook Research, designed for learning keypoints with an emphasis on simplicity and flexibility. It provides state-of-the-art and competitive results across various benchmarks. The framework includes pre-trained models and has been tested on Linux with specific GPU requirements. Users can set up the Python environment, integrate datasets, train SiLK, add custom backbones, run evaluation pipelines, and perform inference. The project also supports converting SiLK to TorchScript and importing features into COLMAP, making it a versatile tool for researchers and developers working with computer vision and deep learning.

Sureel is a platform designed for creators, media rights owners, and AI companies to navigate the AI revolution. It offers solutions to protect, control, and monetize media by allowing owners to define how their content can be used for AI training. Key features include showing AI companies what content they can and cannot train on, setting granular rules for media usage and alteration, and dynamically licensing media based on its impact on AI outputs. Sureel also provides attribution tools, enabling creators to opt-in to share approved media for protection and monetization, or opt-out to prevent specific content from being used as training data. Real-time attribution reporting and analysis of media influence on AI creations are also core functionalities.

Awesome-AutoML-and-Lightweight-Models is a comprehensive GitHub repository that curates high-quality and recent works in the field of Automated Machine Learning (AutoML) and lightweight models. It serves as a valuable resource for researchers and practitioners, categorizing information into key areas such as Neural Architecture Search, Lightweight Structures, Model Compression, Quantization and Acceleration, Hyperparameter Optimization, and Automated Feature Engineering. The repository includes links to papers and associated code repositories (often in PyTorch or TensorFlow), making it easy to explore and implement the discussed techniques. It is continuously updated, welcoming contributions to ensure it remains a current and relevant resource for the AutoML research community.

Wallaroo.AI is a comprehensive AI inference platform designed for deploying, serving, observing, and optimizing AI models in production at scale. It supports any model and hardware, from CPUs to GPUs, across various environments including cloud, multi-cloud, on-premise, and edge locations. The platform offers features like automated resource orchestration, scaling, load balancing, and centralized monitoring for AI inference pipelines. It aims to significantly reduce engineering time, infrastructure costs, and accelerate time-to-value by providing up to 12X faster inferencing and 80% lower costs. Wallaroo.AI integrates seamlessly with existing AI toolchains via Python SDK and API, supporting complex workflows and enabling continuous optimization of live models.

ContribHub is a dedicated platform designed to connect developers and enthusiasts with open source projects seeking contributions. It streamlines the process of finding relevant projects by allowing users to search based on specific technologies and interests. The platform aims to foster a vibrant open source community by making it easier for individuals to discover opportunities to contribute their skills and for projects to gain valuable support. ContribHub serves as a central hub for exploring various open source initiatives, promoting collaboration, and helping users build their portfolios through meaningful contributions.

Awesome-Spiking-Neural-Networks is a comprehensive, open-source repository dedicated to collecting and organizing research papers, associated code, and relevant websites pertaining to spiking neural networks (SNNs). The project is diligently maintained and continuously updated, ensuring that users have access to the most recent advancements in the field. It features papers from prestigious conferences like AAAI, ICLR, NeurIPS, ICML, and CVPR, as well as top-tier journals such as Nature, Science, and Cell. This resource is invaluable for researchers, academics, and students looking to stay current with the cutting edge of SNN development, providing a centralized hub for critical information and resources.

awesome-openai-vision-api-experiments is a comprehensive, open-source repository designed for developers and AI enthusiasts looking to explore and build upon the OpenAI Vision API. It serves as a central hub for innovative experiments, showcasing a diverse range of applications from fundamental image classifications to sophisticated zero-shot learning models. The resource helps users understand the API's capabilities and limitations, offering solutions for challenges like object detection and image segmentation by combining GPT-4V with foundational models such as GroundingDINO or Segment Anything (SAM). It includes practical examples like WebcamGPT, HotDogGPT, and zero-shot object detection, alongside a curated list of must-read papers and blogs to deepen understanding and foster collaboration within the visual AI community.

Andesite provides an AI-powered Security Operations Center (SOC) solution designed to augment human cybersecurity teams. It enables automation and acceleration of investigations, connecting disparate data silos to provide a unified view of threats. The platform helps teams make critical decisions, assess risks, and focus on prevention by correlating, contextualizing, and analyzing structured, unstructured, and semi-structured data from various sources. Andesite is built for enterprise environments, offering high-complexity and high-risk security, and features a "Decision Fabric" for scalable data insights. It boasts robust security and compliance, including FedRAMP High Authorization, SOC 2 Type II, and ISO 27001 certifications, with deployment options for single-tenancy SaaS or self-managed, air-gapped systems.

Aura.com provides comprehensive digital safety for individuals and families, leveraging AI to protect against identity theft, scams, and online threats. The platform offers a suite of features including credit monitoring across all three major bureaus, identity theft protection with alerts for compromised SSN and online accounts, and suspicious transaction alerts for banking and retirement accounts. For families, Aura includes robust parental controls, safe gaming with cyberbullying alerts, and online wellbeing insights. Additional features encompass a VPN, antivirus, password manager, spam call protection, and a digital vault for sensitive files. All plans come with a $1M insurance policy for eligible losses and 24/7 U.S. based support.

The MCP Toolkit, hosted on Hugging Face, is an advanced AI tool designed for deepfake detection and forensic analysis. Users can upload images to the platform, which then employs multiple sophisticated models and forensic techniques to assess whether the image is AI-generated or a real photograph. The toolkit provides a detailed analysis of its findings and offers a consensus label, helping users to identify potential manipulations. This makes it a valuable resource for researchers, security professionals, and anyone interested in combating disinformation by verifying the authenticity of visual media.

Encrypted Anonymization Using Fully Homomorphic Encryption is a tool hosted on Hugging Face Spaces by zama-fhe, designed for secure text data anonymization. This application enables users to encrypt sensitive text data before sending it to a server for processing, ensuring that personally identifiable information (PII) remains protected during transit and computation. After processing, the data can be decrypted to reveal the anonymized output. This approach leverages fully homomorphic encryption (FHE) to maintain data privacy throughout the entire workflow, making it suitable for scenarios where data confidentiality is paramount, such as privacy-preserving machine learning or secure data analysis. The tool is available as an open-source solution, promoting transparency and community contributions.

Data-Analysis-and-Machine-Learning-Projects is an open-source repository offering a collection of teaching materials, code, and data specifically designed for data analysis and machine learning projects. Each project typically corresponds to a blog post by Randal S. Olson, providing practical examples and insights. Users can access documentation, often in IPython Notebook format, to understand the analysis, data usage terms, and more. The instructional materials are licensed under Creative Commons Attribution, allowing for sharing and adaptation, even commercially, with proper attribution. The software components are provided under the MIT license, granting broad permissions for use, modification, and distribution. This resource is ideal for those looking to learn from or contribute to real-world data science applications.

CircleGuardBench offers the first benchmark specifically designed for testing the safety and accuracy of LLM guards. This Hugging Face Space provides a comprehensive leaderboard where users can explore various models, including those from AtlaAI and Google's Gemma series. The platform allows for filtering models by access type, customizing displayed columns, and uploading new results files to contribute to the benchmark. Built with Gradio, CircleGuardBench facilitates an interactive experience for researchers and developers to evaluate and compare the performance of different LLM guard implementations.

dlbook_notation offers a collection of LaTeX files specifically designed for the Deep Learning textbook's notation and style. This resource includes the core .tex file for the notation page, along with other essential components like `math_commands.tex`, `natbib.bst`, and `notation.bib`. The project aims to provide a consistent and professional presentation for academic work in deep learning. Users can leverage these files to maintain the same aesthetic and notational conventions as the widely recognized Deep Learning book, making it an invaluable asset for researchers, students, and educators in the field. All files are freely available, promoting open access and reusability.

Normal Computing specializes in AI-accelerated co-design for silicon engineering teams, offering solutions for both EDA (Electronic Design Automation) and ASICs (Application-Specific Integrated Circuits). Their Normal EDA platform helps structure chip representations, intelligently plan verification, and generate stimulus for simulations, continuously learning from team feedback to optimize engineering artifacts. For hardware, Normal ASICs explores physics-based designs that relax traditional computing assumptions, enabling stochastic, stateful, and asynchronous computing for orders-of-magnitude more efficient compute. The company aims to converge software and silicon physics, transforming hardware development and enabling ultra-efficient ASICs for various workloads. Founded by former Google Brain and Google X members, Normal Computing focuses on solving the scale, complexity, and energy demands of modern AI models.

DeepSeek-Math is an advanced open-source language model specifically engineered for mathematical reasoning. Initialized with DeepSeek-Coder-v1.5 7B, it undergoes extensive pre-training on a vast corpus of math-related tokens from Common Crawl, alongside natural language and code data. The model, particularly DeepSeekMath 7B, has demonstrated impressive performance, achieving a 51.7% score on the competition-level MATH benchmark without relying on external toolkits. It also offers strong tool-use capabilities for solving and proving mathematical problems by writing programs. DeepSeek-Math is available in base, instruct, and RL models, supporting both academic and commercial research.