AI Trending

Project directory

Projects

Open-source AI projects tracked by the leaderboard, ordered by GitHub stars.

Unknown 2,360 stars

weihaox/awesome-neural-rendering

Resources of Neural Rendering

### Overview: This repository is a curated list of resources for neural rendering, including surveys, course notes, inverse rendering, neural rerendering, differentiable rendering, implicit neural representation, volumetric performance capture, semantic photo synthesis, texture and surface embedding, novel-view synthesis, light and reflectance, dubbing and talking-head animation, motion transfer, and retargeting. ### Key Features: - Surveys and course notes on neural rendering. - Inverse rendering techniques. - Neural rerendering methods. - Differentiable rendering approaches. - Implicit neural representation and rendering models. - Volumetric performance capture and free viewpoint video. - Semantic photo synthesis and manipulation. - Texture and surface embedding or mapping. - Novel-view synthesis for objects and scenes. - Light, reflectance, illuminance, and shade considerations. - Dubbing and talking-head animation techniques. - Motion transfer, retargeting, and reenactment methods. ### Tech Stack: None specified. ### Why it Matters: This repository serves as a comprehensive resource for researchers and practitioners interested in neural rendering, providing a wealth of information on various techniques and applications in the field.
Unknown 2,360 stars

zebbern/no-cost-ai

80+ free AI services for chat, image, video, voice & APIs (may sometimes include access to lead gen ai models for free)

Overview: This repository offers 80+ free AI services for chat, image, video, voice, and APIs, including access to lead gen AI models for free. It is a living index for developers, researchers, and curious builders. Key features: - Chat interfaces - Media generation - Voice/music - APIs - IDEs Tech stack: Node.js, TypeScript, and various AI models. Why it matters: This repository provides a comprehensive set of free AI services, making it an invaluable resource for developers, researchers, and AI enthusiasts looking to experiment with various AI models without any sign-up requirements.
JavaScript 2,358 stars

ErickWendel/semana-javascript-expert07

JS Expert Week 7.0 - 🙅🤏🏻 Controlling Streaming Platforms using Eye and Hand Detection 👁🖐

Overview: This project is the JavaScript Expert Week 7.0 repository, focusing on controlling streaming platforms using eye and hand detection. It includes a live demo, pre-reqs, running instructions, and a checklist of features. Key features: - Implementing gesture control for streaming platforms - Hand and eye detection for interactive UI elements - Scroll and click functionality based on hand gestures - Video player with machine learning processing via Web worker Tech stack: JavaScript Why it matters: This project is designed to help developers learn and practice JavaScript skills by applying them to real-world scenarios, such as controlling streaming platforms and enhancing user interactions. Repository: ErickWendel/semana-javascript-expert07 GitHub stars: 2,358 GitHub description: JS Expert Week 7.0 - 🙅🤏🏻 Controlling Streaming Platforms using Eye and Hand Detection 👁🖐
Jupyter Notebook 2,358 stars

google-research-datasets/Objectron

Objectron is a dataset of short, object-centric video clips. In addition, the videos also contain AR session metadata including camera poses, sparse point-clouds and planes. In each video, the camera moves around and above the object and captures it from different views. Each object is annotated with a 3D bounding box. The 3D bounding box describes the object’s position, orientation, and dimensions. The dataset contains about 15K annotated video clips and 4M annotated images in the following categories: bikes, books, bottles, cameras, cereal boxes, chairs, cups, laptops, and shoes

### Overview: Objectron is a comprehensive dataset of short, object-centric video clips, providing detailed annotations including camera poses, sparse point-clouds, and surface planes. The dataset comprises 15,000 annotated videos and over 4 million annotated images across various categories such as bikes, books, bottles, cameras, cereal boxes, chairs, cups, laptops, and shoes. It is designed for research and development in computer vision and 3D object detection, offering a wide range of features and evaluation metrics. ### Key Features: - **15,000 Annotated Videos and 4 Million Annotated Images**: Comprehensive coverage with high-resolution images, object poses, camera poses, point-clouds, and surface planes. - **Multi-View Observations**: Videos showcase the same object from multiple angles, providing a rich dataset for research. - **3D Bounding Box Annotations**: Accurate 3D bounding boxes describe the object’s position, orientation, and dimensions. - **Cross-Platform Compatibility**: Ready to use in TensorFlow/PyTorch format, making it easy to integrate into various ML frameworks. - **Geo-Diverse Samples**: Data collected from 10 countries across five continents, ensuring global diversity. ### Tech Stack: - **Data Storage**: Google Cloud Storage, Objectron bucket. - **Data Formats**: tf.records, tf.SequenceExample, sharded and shuffled datasets. ### Why It Matters: - **Comprehensive Research Tool**: Ideal for researchers and developers looking to enhance their 3D object detection models. - **High-Resolution and Detailed Annotations
C 2,357 stars

ScanNet/ScanNet

### Overview: ScanNet is a comprehensive RGB-D video dataset containing over 2.5 million views across 1500 scans, featuring 3D camera poses, surface reconstructions, and instance-level semantic segmentations. ### Key Features: - **RGB-D Sequence Organization**: Data is organized by RGB-D sequence, with each sequence stored in a directory named `scene<spaceId>_<scanId>` or `scene%04d_%02d`. - **High-Quality Meshes**: Contains high-quality reconstructed meshes (`scene%04d_%02d.ply`), cleaned and decimated meshes for semantic annotations (`scene%04d_%02d.ply`), and aggregated instance-level semantic annotations (`scene%04d_%02d.aggregation.json`). - **Annotation Metadata**: Includes camera poses, surface reconstructions, and instance-level semantic segmentations. ### Tech Stack: C ### Why it Matters: ScanNet provides a rich dataset for 3D reconstruction, semantic segmentation, and instance recognition, making it invaluable for research and development in computer vision, robotics, and other fields. Repository: ScanNet/ScanNet GitHub Stars: 2355 GitHub Description: No description provided.
Dart 2,357 stars

cogwheel0/conduit

Native iOS and Android client for Open WebUI, direct OpenAI-compatible, Ollama, and Hermes agents.

Overview: Conduit is a native mobile client for Open WebUI, enabling direct communication with OpenAI-compatible, Ollama, and Hermes agents, as well as self-hosted agents. Key features: - Direct connections to OpenAI-compatible, Ollama, and Hermes agents - Self-hosted agent functionality - Open WebUI integration - Cross-platform support (Android and iOS) Tech stack: Dart Why it matters: Conduit simplifies access to advanced AI models, making them accessible directly from the device, enhancing user experience and flexibility. Repository: cogwheel0/conduit GitHub stars: 2354 GitHub description: Native iOS and Android client for Open WebUI, direct OpenAI-compatible, Ollama, and Hermes agents.
Python 2,354 stars

AGI-Edgerunners/LLM-Agents-Papers

A repo lists papers related to LLM based agent

### Overview: This repository is a curated collection of papers related to Large Language Model-based Agents (LLM-Agents). It serves as a valuable resource for researchers, developers, and enthusiasts interested in the field of artificial intelligence and its applications. ### Key Features: - **Survey**: A comprehensive survey of evaluation tools for AI assistants and agents. - **Technique For Enhancement**: Includes techniques for improving LLM-based agents, such as planning, memory mechanisms, feedback and reflection, Retrieval-Augmented Generation (RAG), and search algorithms. - **Interaction**: Covers various interaction types, including role-playing, conversation, game playing, human-agent interaction, tool usage, and simulation. - **Application**: Explores applications across different fields such as math, chemistry, biology, physics, geography, art, medicine, finance, software engineering, and research. - **Automation**: Discusses workflow automation and automatic evaluation techniques. - **Training**: Focuses on fine-tuning, reinforcement learning (RL), and differentiable policy optimization (DPO). - **Scaling**: Covers single-agent frameworks and multi-agent systems. - **Stability**: Addresses issues of safety, bias, and hallucination. - **Infrastructure**: Includes benchmark evaluation, environment and platform setup, and dataset considerations. - **Others**: Contains additional resources and recommendations for further reading. ### Tech Stack: Python ### Why It Matters: This repository provides a wealth of information and insights into the development and application of LLM-based agents, making it an essential resource for anyone interested in this field. The comprehensive coverage of various techniques, applications
Python 2,354 stars

google-deepmind/optax

Optax is a gradient processing and optimization library for JAX.

### Overview: Optax is a gradient processing and optimization library designed for JAX, providing simple, well-tested, efficient implementations of core components and enabling the easy combination of low-level ingredients into custom optimizers. ### Key features: - Implementations of popular optimizers - Loss functions for various tasks - Easy integration and recombination of components - Focus on readability and standard equation structuring ### Tech stack: Python ### Why it matters: Optax simplifies research by providing building blocks that can be easily combined into custom solutions, improving research productivity and accelerating adoption of new ideas.
Python 2,348 stars

dot-agent/nextpy

🤖Self-Modifying Framework from the Future 🔮 World's First AMS

### Overview: Nextpy is a self-modifying framework designed for dynamic, AI-driven software development. It offers precise control over AI systems, optimized prompt engineering, and efficient session management with LLMs. ### Key Features: - **Set clear boundaries:** Precise definition of AI system capabilities. - **More effective prompt engineering:** Offers greater control over LLMs compared to traditional methods. - **Pre-compiling prompts:** Eliminates unnecessary redundant LLM processing. - **Session state with LLMs:** Maintains state with LLMs, reducing redundant generations and speeding up code generation. - **Optimized tokens:** Transforms output tokens into prompt token batches for faster and cheaper processing. - **Speculative sampling (WIP):** Enhances token generation speed using a smaller model as an assistant. - **Better AI generations:** Precise output control, optimized for code generation, and syntax error detection. - **Modular and extensible:** Runs on various platforms and can be integrated with Python or plain English. - **Developer-first:** Transfers fundamental Python knowledge and enhances development skills. ### Tech Stack: Python ### Why it Matters: Nextpy provides a framework for dynamic software development, offering precise control over AI systems and optimized prompt engineering, making it a valuable tool for developers looking to enhance their AI-driven projects.
Python 2,347 stars

D-X-Y/Awesome-AutoDL

Automated Deep Learning: Neural Architecture Search Is Not the End (a curated list of AutoDL resources and an in-depth analysis)

### Overview: Automated Deep Learning: Neural Architecture Search Is Not the End is a curated list of resources related to automated deep learning, focusing on Neural Architecture Search (NAS) and Hyperparameter Optimization (HPO). The repository aims to provide a comprehensive overview of the latest research and resources in the field, inspired by other popular awesome lists such as awesome-deep-vision, awesome-adversarial-machine-learning, and awesome-architecture-search. ### Key Features: - **PyGlove**: A library for Neural Architecture Search. - **NASLib**: A library for Neural Architecture Search. - **Keras Tuner**: A library for Neural Architecture Search. - **NNI**: A library for Neural Architecture Search. - **AutoGluon**: A library for Neural Architecture Search. - **Auto-PyTorch**: A library for Neural Architecture Search. - **AutoDL-Projects**: A library for Neural Architecture Search. - **Determined**: A library for Neural Architecture Search. - **TPOT**: A library for Neural Architecture Search. ### Tech Stack: Python ### Why it Matters: This repository serves as a valuable resource for researchers, developers, and practitioners interested in Neural Architecture Search and Hyperparameter Optimization. It provides a curated list of resources, including libraries, papers, and benchmarks, making it easier to stay updated with the latest developments in the field.
TypeScript 2,347 stars

cogentapps/chat-with-gpt

An open-source ChatGPT app with a voice

Overview: Chat with GPT is an open-source, unofficial ChatGPT app that connects ChatGPT with ElevenLabs to provide a realistic human voice. It includes features like fast response times, searchability, customizable System Prompts, temperature adjustment, realistic voice, speech recognition, sharing of chat sessions, copy-and-paste, editing, and markdown support. Key features: - 🔎 Search through past chat conversations - 📄 View and customize System Prompt - 🌡 Adjust temperature for creativity - 🎤 Realistic human voice with ElevenLabs - 📄 Share chat sessions online - ✏️ Edit messages - 🖼 Full markdown support - 🫰 Pay for API usage Tech stack: TypeScript + React Why it matters: This app offers a more personalized and interactive ChatGPT experience, with features like real voice and customizable settings, making it useful for developers and enthusiasts looking for a more advanced ChatGPT interface. Repository: cogentapps/chat-with-gpt Primary language: TypeScript GitHub stars: 2347 GitHub description: An open-source ChatGPT app with a voice README excerpt:
Python 2,345 stars

JasonKessler/scattertext

Beautiful visualizations of how language differs among document types.

### Overview: Scattertext is a tool designed to visualize how language differs among document types, specifically focusing on the 2012 American political conventions. It uses an interactive HTML scatter plot to display terms used by different parties, making it easy to compare and analyze the linguistic differences. ### Key Features: - **Interactive Scatter Plot**: Visualizes terms used in different parties as points in an interactive scatter plot. - **Scalable and Customizable**: Supports mouse-wheel or trackpad zooming and drag panning for better visualization. - **PMI Threshold**: Allows for the threshold coefficient to be adjusted for better visualization. - **Metadata Support**: Uses metadata from the corpus to label terms. ### Tech Stack: Python ### Why It Matters: Scattertext provides a user-friendly and interactive way to visualize how language differs among different document types, making it an essential tool for linguists, researchers, and anyone interested in analyzing text corpora. ### Repository: JasonKessler/scattertext ### Primary Language: Python ### GitHub Stars: 2344 ### GitHub Description: Beautiful visualizations of how language differs among document types.
Python 2,345 stars

RasaHQ/rasa_core

Rasa Core is now part of the Rasa repo: An open source machine learning framework to automate text-and voice-based conversations

Overview: Rasa Core is a Python-based framework for building conversational AI applications. Key features: - Natural language processing - Dialog management - Model training - Integration with various backends Tech stack: Python Why it matters: Rasa Core is a crucial component of the Rasa project, offering a robust framework for creating intelligent chatbots and voice assistants.
Python 2,340 stars

achillesrasquinha/bulbea

:boar: :bear: Deep Learning based Python Library for Stock Market Prediction and Modelling

Overview: Bulbea is a Python library designed for deep learning-based stock market prediction and modeling. Key features: - Deep learning models for stock market prediction and analysis - Support for various neural network architectures - Integration with popular financial data sources like Quandl - Sentiment analysis capabilities using Twitter API Tech stack: Python Why it matters: Bulbea simplifies the process of building and deploying deep learning models for financial analysis, making it accessible to developers without extensive machine learning expertise.
Elm 2,340 stars

eikek/docspell

Assist in organizing your piles of documents, resulting from scanners, e-mails and other sources with miminal effort.

Overview: Docspell is a document management system designed for organizing documents from various sources, including scanners, emails, and other files. It simplifies the process of converting physical papers into digital files and provides a user-friendly interface for managing and searching through these documents. Key features: - OCR (Optical Character Recognition) on scanned documents - Fulltext search functionality - Machine learning for tagging and metadata suggestions - Integration with email clients - REST API for accessing documents via HTTP Tech stack: Elm, Scala Steward, Docker, Stanford Core NLP library Why it matters: Docspell streamlines the organization of documents, making it easier to find and manage files from various sources. Its machine learning capabilities help in tagging documents and suggesting metadata, making it a valuable tool for personal and small business use.
Python 2,337 stars

moorcheh-ai/memanto

Memory that AI Agents Love!

### Overview: Memanto is a Memory Agent designed to manage the memories of your AI agents, ensuring they maintain accurate and consistent knowledge across different platforms. ### Key Features: - **Observes & Extracts**: Watches agent interactions and extracts durable knowledge from ephemeral traffic. - **Consolidates**: Merges extracted memories into a single, canonical estate. - **Reconciles**: Resolves contradictions between new and old knowledge, preserving belief history. - **Forgets**: Executes policies for decay, expiry, and deliberate deletion. ### Tech Stack: Python ### Why it Matters: Memanto ensures that your agents' memories are managed efficiently and accurately across different platforms, preventing conflicts and maintaining consistency.
Python 2,335 stars

dvmazur/mixtral-offloading

Run Mixtral-8x7B models in Colab or consumer desktops

Overview: This project implements efficient inference of Mixtral-8x7B models, enabling faster execution in Colab or consumer desktops. Key features: - Efficient inference of Mixtral-8x7B models - Mixed quantization with HQQ - MoE offloading strategy - Support for other quantization methods Tech stack: Python Why it matters: This project provides a practical solution for efficient inference of large language models, improving performance and resource utilization in various computing environments.
Python 2,335 stars

salesforce/decaNLP

The Natural Language Decathlon: A Multitask Challenge for NLP

### Overview: The Natural Language Decathlon is a comprehensive multitask challenge designed to evaluate the performance of natural language processing (NLP) models across ten distinct tasks. This repository, `salesforce/decaNLP`, introduces a novel multitask question answering network (MQAN) that addresses the challenges of single-task and multitask learning in NLP. ### Key features: - Question Answering (SQuAD) - Machine Translation (IWSLT) - Summarization (CNN/DM) - Natural Language Inference (MNLI) - Sentiment Analysis (SST) - Semantic Role Labeling (QA-SRL) - Zero-shot Relation Extraction (QA-ZRE) - Goal-Oriented Dialogue (WOZ) - Semantic Parsing (WikiSQL) - Commonsense Reasoning (MWSC) ### Tech stack: Python ### Why it matters: This repository showcases the effectiveness of the MQAN model in achieving high performance across various NLP tasks, demonstrating the potential of multitask learning in improving model generalization and efficiency. The framework is designed to be adaptable to different single-task or multitask learning approaches, making it a valuable resource for researchers and practitioners in the field of NLP.
Python 2,331 stars

refuel-ai/autolabel

Label, clean and enrich text datasets with LLMs.

Overview: Autolabel is a Python library designed to automatically label, clean, and enrich text datasets using any Large Language Models (LLMs). It leverages state-of-the-art language models like GPT-4 to achieve high accuracy at a fraction of the cost and time compared to manual labeling. Key features: - Automatically labels, cleans, and enriches text datasets - Supports multiple LLMs, including GPT-4 and others - Provides a simple 3-step process for labeling data - Benchmarks models on a provided benchmark dataset for performance evaluation Tech stack: Python Why it matters: By automating the labeling process, Autolabel streamlines the data labeling pipeline, making it more efficient and cost-effective, especially for large datasets. This tool is crucial for researchers and practitioners looking to leverage the power of large language models without the need for extensive manual labor.
Python 2,331 stars

vearch/vearch

Distributed vector search for AI-native applications

### Overview Vearch is a cloud-native distributed vector database designed for efficient similarity search of embedding vectors in AI applications. It offers hybrid search capabilities, fast vector retrieval, and scalability with replication and elastic scaling. ### Key features - **Hybrid search**: Combines vector search and scalar filtering for comprehensive results. - **Performance**: Retrieves millions of objects in milliseconds. - **Scalability & Reliability**: Supports replication and elastic scaling. ### Tech stack Python ### Why it matters Vearch simplifies the integration of vector search capabilities into AI applications, enabling faster and more efficient similarity searches. Its cloud-native architecture ensures scalability and reliability, making it ideal for large-scale applications.