Coding & Development

ICML2023 Papers is a Hugging Face Space designed to provide access to research papers presented at the ICML2023 conference. While the tool's primary function is to serve as a repository for academic research, it is currently paused. Users interested in utilizing this resource are directed to the community tab to request its restart from the author(s). This tool would typically be valuable for academic research, literature reviews, and information retrieval within the field of AI, catering to researchers and students.

nips14-ssl offers the foundational code necessary to replicate key findings from the NIPS 2014 paper, "Semi-Supervised Learning with Deep Generative Models." This open-source repository is designed for researchers and developers interested in semi-supervised learning (SSL) using deep generative models. It includes scripts for training various models like M1 (Variational Auto-Encoder) and M1+M2 for semi-supervised experiments, as well as M2 with fully observed labels. The code also supports qualitative results such as flying through latent space and analogies, and quantitative evaluation of test-set errors. Users can configure the environment with Python, Numpy, and Theano to run experiments on datasets like MNIST and SVHN.

Passerine is a small, extensible, open-source programming language developed by Isaac Clayton and the Passerine Community. Rooted in Scheme and ML-flavored languages, it emphasizes minimalistic and concise expression. The language is powered by a VM written in Rust and incorporates lambda-calculus with pattern-matching, structural types, fiber-based concurrency, and syntactic extension. While currently strongly and dynamically typed, the developers are working towards a compile-time type-checker based on Hindley-Milner type inference. Passerine is still in early development stages with frequent breaking changes, making it not yet ready for production use. The project aims to eventually adopt MiniVM Assembly (VASM) or LLVM IR as compiler targets and explores advanced concepts like vaporization memory management and kind-based macros.

Picolibc is a C library specifically engineered for embedded 32- and 64-bit systems, offering standard C library APIs. It was developed by blending code from Newlib and AVR Libc, with a focus on optimizing for small embedded systems that have limited RAM. Key differentiators from Newlib include its design and testing for embedded systems, a re-engineered stdio implementation, efficient thread-local-storage support, and a comprehensive integrated test suite. Picolibc also provides bare-metal startup code and linker scripts for various architectures, uses a Meson-based build system for faster compilation, and clearly documents its narrow POSIX OS requirements. It also features built-in semihosting support for many architectures and maintains clear BSD licensing.

Pwnagotchi is an A2C-based "AI" that utilizes bettercap to learn from its surrounding WiFi environment, aiming to maximize the capture of crackable WPA key material. This material is collected as PCAP files, supporting various handshake types compatible with hashcat, including PMKIDs and full/half WPA handshakes. Unlike typical reinforcement learning simulations, Pwnagotchi learns over time, with performance improving across epochs as it explores and tunes parameters for optimal WiFi pwning in different environments. It uses an LSTM with MLP feature extractor as its policy network for the A2C agent. Multiple Pwnagotchi units can also cooperate by communicating and dividing channels for enhanced pwnage.

pytorch-dense-correspondence is the reference implementation for the paper "Dense Object Nets: Learning Dense Visual Object Descriptors By and For Robotic Manipulation." This project focuses on learning dense descriptor networks for previously unseen, potentially deformable objects, and classes of objects. It demonstrates the use of Dense Object Nets for robotic manipulation tasks, including grasping specific points on objects across deformed configurations and transferring grasps across object classes. The repository provides a PyTorch implementation, with a tutorial available for getting started, and supports multi-object descriptor learning. It also includes novel contributions for acquiring descriptors that generalize across object classes or are distinct for each object instance.

PythonMonkey is an innovative tool that integrates the Mozilla SpiderMonkey JavaScript engine directly into the Python Virtual Machine. This unique embedding allows Python to act as the JavaScript host environment, facilitating a seamless interchange between the two languages. Developers can leverage JavaScript Array and Object methods on Python Lists and Dictionaries, and vice-versa, with full adaptation to the respective contexts. Key features include fast and memory-efficient data interchange, where strings share immutable backing stores and TypedArrays share mutable backing stores to minimize memory-copy overhead. It supports CommonJS module systems for both JavaScript and Python, and provides a Python host environment with an event loop, EventEmitter, setTimeout, and XMLHttpRequest. PythonMonkey has reached MVP status and is actively maintained, welcoming external contributions.

Red is a powerful, next-generation programming language inspired by Rebol, designed for a broad range of applications from system programming to high-level scripting and cross-platform reactive GUI development. It features a native-code compiler, enabling the creation of small, dependency-free executables (less than 1MB). Red supports modern concurrency and multi-core CPUs, and utilizes a DSL-oriented approach with built-in dialects like Red/System for C-level programming, Parse for PEG parsing, VID for GUI layout, and Draw for 2D vector graphics. The toolchain is self-contained, offering an encapper, compiler, interpreter, and linker, all within a single, zero-install file. It supports cross-compilation to various platforms including Windows, Linux, macOS, and Android, making it highly versatile for developers.

InVesalius is a free and open-source software designed for 3D medical imaging reconstruction, capable of generating detailed models from 2D DICOM files obtained through CT or MRI equipment. It boasts multi-platform compatibility, running on GNU Linux, Windows, and MacOS, and is internationalized with support for numerous languages including English, Portuguese, French, and German. The software provides a comprehensive suite of tools for medical professionals, encompassing DICOM support, image manipulation (zoom, pan, brightness/contrast), various segmentation methods (2D slices, watershed, pre-defined thresholds), and precise measurement tools. Users can create and export 3D surfaces in formats like STL, PLY, OBJ, and VRML, alongside high-quality volume rendering projections and picture exportation options.

rust-smallvec is a Rust library designed to optimize memory usage and performance by implementing a "small vector" optimization. This technique allows a small, predefined number of items to be stored directly on the stack, avoiding heap allocations for common, smaller data sets. If the collection grows beyond this small capacity, it automatically transitions to using heap storage. This approach is particularly beneficial in Rust applications where memory efficiency and predictable performance are critical. The library provides a `SmallVec` type that behaves like a standard `Vec` but with this stack-based optimization, allowing developers to use normal slice indexing and methods for accessing its contents.

Jsonformer is an AI tool specifically designed to automate various tasks, with a strong focus on content generation and educational applications. It aims to simplify complex processes by leveraging artificial intelligence. The tool is accessible to a broad audience, as it is available for free and deployed on the HuggingFace platform, making it easy to access and utilize for different projects.

chatgpt-failures is a GitHub repository dedicated to collecting and documenting instances where ChatGPT and other large language models exhibit failures. This archive acts as a valuable resource for researchers, developers, and AI enthusiasts interested in understanding the limitations, biases, and vulnerabilities inherent in these advanced AI systems. Users can leverage this collection for comparative analysis with alternative models, to identify common failure patterns, and to generate synthetic data for robust testing and training of new AI models. It provides a practical dataset for improving the reliability and safety of language models.

SpacetimeGaussians is an open-source implementation of Spacetime Gaussian Feature Splatting, a technique for real-time dynamic view synthesis. Developed by OPPO US Research Center, this tool is designed for researchers and developers working on advanced 3D scene reconstruction and rendering. It supports various datasets including Neural 3D, Technicolor, and Google Immersive, offering both 'lite' and 'full' models for different computational needs. The project includes comprehensive installation guides for Linux and Windows (via WSL2), detailed instructions for data preprocessing, training, and testing. A real-time viewer, based on SIBR and Gaussian Splatting, is also provided, along with pre-trained models to facilitate immediate use and experimentation. The codebase is modular, allowing users to customize Gaussian representations, rendering pipelines, and even integrate new datasets.

stable-diffusion-webui-dataset-tag-editor is an extension specifically designed for the Stable Diffusion Web UI, particularly for use with the AUTOMATIC1111 web UI. Its primary function is to facilitate the editing of captions within training datasets. Users can modify and manage the text captions associated with images that are intended for training Stable Diffusion models. This tool is provided as a free, open-source extension, making it accessible for those working with Stable Diffusion.

LDM Faces Generator is an AI-powered tool designed for generating diverse facial images. Leveraging Latent Diffusion Models, it can produce both realistic and stylized faces. This tool is freely accessible on Hugging Face, making it a valuable resource for individuals interested in AI image creation. Its capabilities cater to a broad audience, including AI enthusiasts, digital image creators, and researchers exploring diffusion models and synthetic media.

Weekly github, operating as BongaCams, offers a comprehensive adult entertainment platform with free live sex cams and adult chat. Users can access thousands of webcam models globally, including teens, matures, MILFs, ebony girls, and amateurs, for live XXX shows. The platform supports various categories like Anal, Asian, BBW, Big Ass, Big Tits, and Fetish, allowing users to find specific content. It provides options for free sex chat, live XXX porn shows, and private rooms for more intimate experiences. BongaCams also features HD XXX porn videos uploaded by users and models, available on both desktop and mobile devices, and supports multiple languages.

Llama 2 13b Chat is an AI chatbot specifically developed for conversational AI applications. It focuses on natural language processing and research, making it a valuable tool for those working in these fields. The chatbot is well-suited for the development and experimentation of AI chatbots, providing a platform for building and testing conversational agents. It is available for free, offering an accessible option for developers and researchers.

Trampoline is a static Real-Time Operating System (RTOS) specifically designed for small embedded systems. Its API adheres to the OSEK/VDX OS and AUTOSAR OS 4.2 standards, ensuring compatibility and reliability for automotive and industrial applications. The RTOS supports a wide range of platforms, including Cortex-M (M0+, M3, M4), Cortex-A (A7), Cortex-R (R52), and RISC-V instruction sets, as well as PowerPC, POSIX systems, AVR, and ARM architectures. It provides numerous examples for various microcontrollers and boards, demonstrating functionalities like LED blinking, button polling, and interrupt handling. Trampoline also includes precompiled binaries of goil, an OIL and ARXML compiler, for different operating systems, facilitating configuration and development.

LLaVA++ (LLaMA-3-V) is an AI chatbot specifically designed for research and development purposes. It provides a platform for experimenting with AI models and can be utilized as an educational resource for those interested in AI and chatbot applications. The tool is available for free, making it accessible for a wide range of users looking to explore and develop AI-powered conversational agents.

HeadPoseEstimation-WHENet is an end-to-end head-pose estimation network designed for real-time, fine-grained prediction of Euler angles across the full range of head yaws from a single RGB image. Unlike many existing methods that perform well only for frontal views, WHENet targets head poses from all viewpoints, making it suitable for applications in autonomous driving and retail. The network builds on multi-loss approaches with adapted loss functions and training strategies for wide-range estimation. It also uniquely extracts ground truth labelings of anterior views from a panoptic dataset. WHENet is compact and efficient, making it suitable for mobile devices and applications, and meets or beats state-of-the-art methods for frontal head pose estimation.

The Insight Toolkit (ITK) is an open-source, cross-platform software toolkit designed for N-dimensional scientific image processing, segmentation, and registration. It offers a proven, spatially-oriented architecture for analyzing scientific images, particularly those from medical instrumentation like CT or MRI scanners. Key functionalities include segmentation, which identifies and classifies data in digitally sampled representations, and registration, which aligns or develops correspondences between data, such as combining CT and MRI scans. ITK is distributed as binary Python packages and its C++ core can be built from source using CMake. The project operates under an open governance model, is fiscally sponsored by NumFOCUS, and is distributed under a license enabling both non-commercial and commercial applications.

AnyGen AI, powered by IP Agent AI, is an advanced platform designed to simplify and optimize patent-related tasks. It acts as an intelligent agent, guiding users through the intricacies of intellectual property management. The tool focuses on transforming complex patent workflows into clear, manageable processes, making it easier for users to handle patent data and procedures efficiently. By leveraging AI, AnyGen AI aims to reduce the time and effort typically associated with patent research, drafting, and management, providing a streamlined solution for intellectual property professionals.

Masked Diffusion Transformers (MDT) is an AI tool designed for generating and manipulating images. It provides a platform for users to explore and work with diffusion models, making it particularly suitable for those involved in AI research and development. The tool is available on Hugging Face and is offered free of charge, encouraging experimentation and innovation within the AI community.

MaskGCT TTS Demo is an AI tool designed for text-to-speech conversion, providing a platform for users to explore and utilize GCT models for speech synthesis. Hosted on Hugging Face, this tool is freely accessible, making it suitable for a wide range of applications. It is particularly well-suited for individuals involved in speech synthesis research and those looking to create various forms of audio content.