Best AI Tools for Machine Learning & Data Science

LMCache

AI GitHub Repos

High-performance key-value cache layer designed to enhance the efficiency of large language models (LLMs). It optimizes data retrieval speeds, ensuring faster…LMCache is a high-performance key-value cache layer designed to enhance the efficiency of large language models (LLMs). It optimizes data retrieval speeds, ensuring faster…

Runanywhere

AI GitHub Repos

Runanywhere SDKs is a production-ready toolkit designed to run AI models locally. It facilitates the development and deployment of AI applications by providing essential tools and…

PixelRAG

AI GitHub Repos

Open-source project that introduces scalable pixel-native search, aiming to revolutionize web parsing. It offers a new approach to handling and searching pixel data…PixelRAG is an open-source project that introduces scalable pixel-native search, aiming to revolutionize web parsing. It offers a new approach to handling and searching pixel data…

OpenRLHF

AI GitHub Repos

Scalable and high-performance agentic reinforcement learning framework based on Ray. It supports various algorithms like PPO, DAPO, and REINFORCE++, enabling…OpenRLHF is a scalable and high-performance agentic reinforcement learning framework based on Ray. It supports various algorithms like PPO, DAPO, and REINFORCE++, enabling…

Model Context Protocol

AI GitHub Repos

The Model Context Protocol repository provides comprehensive specifications and documentation for implementing the Model Context Protocol. It serves as a guide for developers…

kimi-k3-in-c

AI GitHub Repos

Kimi K3 in C is a 2.78-trillion-parameter model capable of running inference on a single CPU with just 8.24 GB of RAM. It is implemented in portable C99, requiring no BLAS,…

VAR

AI GitHub Repos

Visual Autoregressive Modeling is a state-of-the-art codebase for scalable image generation. It offers a user-friendly implementation of next-scale prediction, making it ideal for…

bitsandbytes

AI GitHub Repos

GitHub project focused on making large language models more accessible through k-bit quantization for PyTorch. It aims to optimize model performance and reduce…Bitsandbytes is a GitHub project focused on making large language models more accessible through k-bit quantization for PyTorch. It aims to optimize model performance and reduce…

Soup

AI GitHub Repos

GitHub repository that enables the fine-tuning of large language models (LLMs) using a single YAML file. It utilizes layer streaming to train an 8 billion parameter…Soup is a GitHub repository that enables the fine-tuning of large language models (LLMs) using a single YAML file. It utilizes layer streaming to train an 8 billion parameter…

GenieX

AI GitHub Repos

Enables the execution of frontier LLMs and VLMs locally on Qualcomm devices, utilizing NPU, GPU, and CPU. It simplifies the process with minimal code, offering efficient AI…GenieX enables the execution of frontier LLMs and VLMs locally on Qualcomm devices, utilizing NPU, GPU, and CPU. It simplifies the process with minimal code, offering efficient AI…

GPTCache

AI GitHub Repos

Semantic cache designed for large language models (LLMs), fully integrated with LangChain and llama_index. It optimizes the performance of LLMs by caching semantic…GPTCache is a semantic cache designed for large language models (LLMs), fully integrated with LangChain and llama_index. It optimizes the performance of LLMs by caching semantic…

Utopia

AI GitHub Repos

World's first open-source enterprise world model. It allows developers to contribute and enhance its capabilities by creating an account on GitHub. This project aims…Utopia is the world's first open-source enterprise world model. It allows developers to contribute and enhance its capabilities by creating an account on GitHub. This project aims…

Lora

AI GitHub Repos

GitHub repository focused on using low-rank adaptation to efficiently fine-tune diffusion models. It provides a method for enhancing model performance without extensive…Lora is a GitHub repository focused on using low-rank adaptation to efficiently fine-tune diffusion models. It provides a method for enhancing model performance without extensive…

OpenEvolve

AI GitHub Repos

Open-source implementation of AlphaEvolve, designed to facilitate the development and evolution of AI models. Hosted on GitHub, it invites contributions from…OpenEvolve is an open-source implementation of AlphaEvolve, designed to facilitate the development and evolution of AI models. Hosted on GitHub, it invites contributions from…

Chinese LLaMA-Alpaca-2

AI GitHub Repos

GitHub project focused on developing large language models with extended context capabilities. It offers models like LLaMA-2 and Alpaca-2, designed to…Chinese LLaMA-Alpaca-2 is a GitHub project focused on developing large language models with extended context capabilities. It offers models like LLaMA-2 and Alpaca-2, designed to…

FlashInfer

AI GitHub Repos

Kernel library designed for serving large language models (LLMs). It facilitates efficient deployment and management of LLMs, enabling developers to contribute and…FlashInfer is a kernel library designed for serving large language models (LLMs). It facilitates efficient deployment and management of LLMs, enabling developers to contribute and…

Qwen-VL

AI GitHub Repos

Large vision language model developed by Alibaba Cloud. It offers advanced capabilities for chat and vision tasks, providing a robust platform for AI applications in…Qwen-VL is a large vision language model developed by Alibaba Cloud. It offers advanced capabilities for chat and vision tasks, providing a robust platform for AI applications in…

Tree of Thoughts

AI GitHub Repos

Research project focused on enhancing problem-solving capabilities using large language models. It explores structured reasoning approaches to improve AI…Tree of Thoughts is a research project focused on enhancing problem-solving capabilities using large language models. It explores structured reasoning approaches to improve AI…

Shimmy

AI GitHub Repos

Pure-Rust WebGPU inference engine that is OpenAI-API compatible and GGUF native. It runs on any GPU without the need for Python or llama.cpp, offering a single binary…Shimmy is a pure-Rust WebGPU inference engine that is OpenAI-API compatible and GGUF native. It runs on any GPU without the need for Python or llama.cpp, offering a single binary…

Baichuan-7B

AI GitHub Repos

Large-scale 7 billion parameter pretraining language model developed by BaiChuan-Inc. It is designed to handle complex language tasks and provide robust AI…Baichuan-7B is a large-scale 7 billion parameter pretraining language model developed by BaiChuan-Inc. It is designed to handle complex language tasks and provide robust AI…

OpenChat

AI GitHub Repos

Open-source project focused on advancing language models using imperfect data. It aims to improve the accessibility and performance of AI language models for…OpenChat is an open-source project focused on advancing language models using imperfect data. It aims to improve the accessibility and performance of AI language models for…

Tiny-Universe

AI GitHub Repos

Comprehensive guide for constructing large models from scratch. It offers detailed instructions and resources for developers to build and contribute to the…Tiny-Universe is a comprehensive guide for constructing large models from scratch. It offers detailed instructions and resources for developers to build and contribute to the…

EmbeddingGemma 2

AI Models

Multimodal embedding model developed by Google DeepMind. It maps text, images, video, and audio into a unified vector space, supporting over 100 languages…EmbeddingGemma 2 is a multimodal embedding model developed by Google DeepMind. It maps text, images, video, and audio into a unified vector space, supporting over 100 languages…

Gemma 4 12B Unified Model

AI Models

Gemma 4 12B Unified Model by Google DeepMind is a multimodal AI model designed for efficient local execution. It supports text, image, and audio inputs, offering advanced…

DiffusionGemma

LLMs

Generative model by Google DeepMind, built on the Gemma 4 architecture. It processes text, image, and video inputs to generate text outputs, optimizing for…DiffusionGemma is a generative model by Google DeepMind, built on the Gemma 4 architecture. It processes text, image, and video inputs to generate text outputs, optimizing for…

YuE2-3B

AI Models

Open music generation model that transforms lyrics and style prompts into complete songs with vocals and accompaniment. It allows users to edit melodies and chords…YuE2-3B is an open music generation model that transforms lyrics and style prompts into complete songs with vocals and accompaniment. It allows users to edit melodies and chords…

Krea 2 Raw

AI Models

Text-to-image diffusion model designed for generating images from natural-language descriptions. It serves as a base for fine-tuning and post-training, supporting…Krea 2 Raw is a text-to-image diffusion model designed for generating images from natural-language descriptions. It serves as a base for fine-tuning and post-training, supporting…

Nemotron 3 Diarization

AI Models

Speaker diarization model that identifies 'who spoke when' in audio recordings. It supports up to eight speakers and offers both streaming and offline…Nemotron 3 Diarization is a speaker diarization model that identifies 'who spoke when' in audio recordings. It supports up to eight speakers and offers both streaming and offline…

Qwen-AgentWorld-35B-A3B

LLMs

Language world model designed for simulating agentic environments. It covers seven interaction domains, offering scalable and controllable simulations…Qwen-AgentWorld-35B-A3B is a language world model designed for simulating agentic environments. It covers seven interaction domains, offering scalable and controllable simulations…

NVIDIA NemotronLabs VoiceChat

AI Models

Real-time, full-duplex speech model for conversational AI. It integrates speech understanding and generation in one architecture, reducing…NVIDIA NemotronLabs VoiceChat is a real-time, full-duplex speech model for conversational AI. It integrates speech understanding and generation in one architecture, reducing…

PaddleOCR-VL-1.6

AI Models

Advanced document parsing model that enhances region refinement and post-training processes. It achieves state-of-the-art accuracy on OmniDocBench,…PaddleOCR-VL-1.6 is an advanced document parsing model that enhances region refinement and post-training processes. It achieves state-of-the-art accuracy on OmniDocBench,…

Ideogram 4

AI Models

Cutting-edge text-to-image model offering unparalleled control over image composition, style, and layout. It supports native 2k resolution and multilingual text…Ideogram 4 is a cutting-edge text-to-image model offering unparalleled control over image composition, style, and layout. It supports native 2k resolution and multilingual text…

TabFM 1.0.0 (PyTorch)

AI Models

Zero-shot tabular foundation model by Google Research, designed for classification and regression on structured data. It requires no fine-tuning and supports…TabFM 1.0.0 is a zero-shot tabular foundation model by Google Research, designed for classification and regression on structured data. It requires no fine-tuning and supports…

NVIDIA Nemotron 3.5 Lightning 30B A3B BF16

LLMs

Large language model designed for customization and domain adaptation. It supports multiple languages and is optimized for research…NVIDIA Nemotron 3.5 Lightning 30B A3B BF16 is a large language model designed for customization and domain adaptation. It supports multiple languages and is optimized for research…

Cosmos3-Edge

AI Models

Cosmos3-Edge by NVIDIA is a multimodal AI model designed for Physical AI applications. It generates dynamic outputs from text, images, and videos, supporting robotics, autonomous…

Qwen3-ASR-1.7B-hf

AI Models

Advanced open-source ASR model supporting language identification and speech recognition for 52 languages and dialects. It offers high-quality, robust…Qwen3-ASR-1.7B-hf is an advanced open-source ASR model supporting language identification and speech recognition for 52 languages and dialects. It offers high-quality, robust…

Whistle — Multilingual Speech-to-Text Model

AI Models

Compact speech-to-text model designed for various devices, offering transcription, word timestamps, and speech embedding in multiple languages. It operates…Whistle is a compact speech-to-text model designed for various devices, offering transcription, word timestamps, and speech embedding in multiple languages. It operates…

Moondream Parakeet Redux

AI Models

Compact version of the parakeet-tdt-0.6b-v3 model, optimized for speed and efficiency. It offers improved performance on multilingual tasks while…Moondream Parakeet Redux is a compact version of the parakeet-tdt-0.6b-v3 model, optimized for speed and efficiency. It offers improved performance on multilingual tasks while…

Kumo Tabular

AI Models

NVIDIA's pretrained tabular foundation model designed for classification and regression tasks. It offers advanced capabilities in predicting class probabilities…Kumo Tabular is NVIDIA's pretrained tabular foundation model designed for classification and regression tasks. It offers advanced capabilities in predicting class probabilities…

FrogNano-4B-2609

AI Models

Compact coding agent developed by Microsoft. It specializes in repository-level software engineering tasks, utilizing reinforcement learning for improved…FrogNano-4B-2609 is a compact coding agent developed by Microsoft. It specializes in repository-level software engineering tasks, utilizing reinforcement learning for improved…

GLiNER2.5-Decide

AI Models

340M parameter English classification model designed for operational decisions. It supports single and multi-label tasks, offering high accuracy across…GLiNER2.5-Decide is a 340M parameter English classification model designed for operational decisions. It supports single and multi-label tasks, offering high accuracy across…

Ming-Image-0.1-Design

AI Models

6B text-to-image model designed for creating UI, infographics, posters, and other text-rich visual designs. It supports RGBA output for transparent…Ming-Image-0.1-Design is a 6B text-to-image model designed for creating UI, infographics, posters, and other text-rich visual designs. It supports RGBA output for transparent…

AuK Model by Tencent

AI Models

AI model hosted on Hugging Face, developed by Tencent. It aims to advance AI capabilities through open-source collaboration and innovation. Users can access it via API…AuK is an AI model hosted on Hugging Face, developed by Tencent. It aims to advance AI capabilities through open-source collaboration and innovation. Users can access it via API…

FLUX 3 Action Base

AI Models

Open-weight 7B world action model designed for robotic applications. It processes camera frames, robot states, and text instructions to generate action…FLUX 3 Action Base is an open-weight 7B world action model designed for robotic applications. It processes camera frames, robot states, and text instructions to generate action…

NeuDecide

AI Models

Compact voice-action model that converts speech directly into function calls. It operates offline, requires no GPU, and adapts to various tools through JSON schemas.NeuDecide is a compact voice-action model that converts speech directly into function calls. It operates offline, requires no GPU, and adapts to various tools through JSON schemas.

Youtu-Parsing-Omni

AI Models

Versatile omni-modal parsing model that converts diverse inputs like documents, images, charts, and videos into structured JSON. It excels in perception…Youtu-Parsing-Omni is a versatile omni-modal parsing model that converts diverse inputs like documents, images, charts, and videos into structured JSON. It excels in perception…

LightOnOCR-3-4B

AI Models

High-performance OCR model offering advanced layout and visual understanding capabilities. It provides accurate transcription and bounding box coordinates for…LightOnOCR-3-4B is a high-performance OCR model offering advanced layout and visual understanding capabilities. It provides accurate transcription and bounding box coordinates for…

pplx-embed-v2-context-9b-preview

AI Models

Contextual embedding model designed for document chunks in RAG systems. It encodes document chunks together, reflecting their surrounding…pplx-embed-v2-context-9b-preview is a contextual embedding model designed for document chunks in RAG systems. It encodes document chunks together, reflecting their surrounding…

Index-Translate-2B

AI Models

Compact multilingual text model supporting translation across 150 languages. It excels in multimodal image translation and follows detailed translation…Index-Translate-2B is a compact multilingual text model supporting translation across 150 languages. It excels in multimodal image translation and follows detailed translation…

Geo-Sleuth

AI GitHub Repos

Agent skill that determines the location where a photo was taken using OpenStreetMap geometry, elevation skylines, satellite imagery, and street view. It…Geo-Sleuth is an agent skill that determines the location where a photo was taken using OpenStreetMap geometry, elevation skylines, satellite imagery, and street view. It…

Jeff

AI GitHub Repos

Open-source model designed for making rapid decisions across various domains. It utilizes a 0.8 billion parameter 'System 1' model to provide calibrated probabilities…Jeff is an open-source model designed for making rapid decisions across various domains. It utilizes a 0.8 billion parameter 'System 1' model to provide calibrated probabilities…

Laya CoreML

AI GitHub Repos

GitHub repository focused on local Laya typed decisions using Apple Core ML and Neural Engine. It offers validated ports and benchmarks for quick decision-making…Laya CoreML is a GitHub repository focused on local Laya typed decisions using Apple Core ML and Neural Engine. It offers validated ports and benchmarks for quick decision-making…

AnyJev

AI GitHub Repos

GitHub project that transforms any large language model (LLM) into a Jev-style decision model. It offers typed decisions and real probabilities without requiring…AnyJev is a GitHub project that transforms any large language model (LLM) into a Jev-style decision model. It offers typed decisions and real probabilities without requiring…

claude-opus-5-5-demo

AI GitHub Repos

Public GitHub repository showcasing a demo project. It allows developers to contribute and collaborate on the development of AI models and tools,…claude-opus-5-5-demo is a public GitHub repository showcasing a demo project. It allows developers to contribute and collaborate on the development of AI models and tools,…

Decider

AI GitHub Repos

Family of System One-style models fine-tuned from Qwen3.5. It is designed for making one-pass typed decisions with calibrated probabilities, providing a streamlined…Decider is a family of System One-style models fine-tuned from Qwen3.5. It is designed for making one-pass typed decisions with calibrated probabilities, providing a streamlined…

AI Manager

AI GitHub Repos

Desktop application designed to streamline the use of AI coding tools. It provides developers with an integrated environment to manage and deploy AI models…AI Manager is a desktop application designed to streamline the use of AI coding tools. It provides developers with an integrated environment to manage and deploy AI models…

Bug Bounty Hunting Prompts

AI GitHub Repos

Offers reusable prompts designed to streamline a structured, evidence-first approach to bug bounty hunting. This GitHub repository provides a valuable…Bug Bounty Hunting Prompts offers reusable prompts designed to streamline a structured, evidence-first approach to bug bounty hunting. This GitHub repository provides a valuable…

JevHarness

AI GitHub Repos

GitHub project offering LLM-authored task-specific harnesses. It features optional full-trajectory reward reflection and GEPA evolution, aimed at enhancing AI…JevHarness is a GitHub project offering LLM-authored task-specific harnesses. It features optional full-trajectory reward reflection and GEPA evolution, aimed at enhancing AI…

Official inference framework for 1-bit large language models (LLMs) developed by Microsoft. It enables efficient processing and deployment of LLMs, optimizing…BitNet is an official inference framework for 1-bit large language models (LLMs) developed by Microsoft. It enables efficient processing and deployment of LLMs, optimizing…

Google AI

AI Apps

Develops innovative AI tools and technologies to solve complex challenges and enrich knowledge. It aims to make AI accessible and beneficial for everyone by providing…Google AI develops innovative AI tools and technologies to solve complex challenges and enrich knowledge. It aims to make AI accessible and beneficial for everyone by providing…

SuperCLUE

AI Apps

Comprehensive benchmark for evaluating Chinese general-purpose large models. It assesses models across various dimensions to determine their effectiveness compared…SuperCLUE is a comprehensive benchmark for evaluating Chinese general-purpose large models. It assesses models across various dimensions to determine their effectiveness compared…

LLMEval

AI Apps

Research initiative by Fudan NLP Lab, focusing on developing comprehensive evaluation frameworks for large language models across various disciplines. It offers…LLMEval is a research initiative by Fudan NLP Lab, focusing on developing comprehensive evaluation frameworks for large language models across various disciplines. It offers…

Midjourney Prompt Generator

AI Apps

Helps users create detailed prompts for Midjourney's AI image generation bot. It simplifies the process by allowing users to set parameters and apply…Midjourney Prompt Generator helps users create detailed prompts for Midjourney's AI image generation bot. It simplifies the process by allowing users to set parameters and apply…

Generrated

AI Apps

Comprehensive resource featuring 9,300 images created using DALL·E 2 and DALL·E 3 prompts. It serves as a reference and inspiration tool for users exploring the…Generrated is a comprehensive resource featuring 9,300 images created using DALL·E 2 and DALL·E 3 prompts. It serves as a reference and inspiration tool for users exploring the…

Indra OCR

AI GitHub Repos

This week

Fast CPU-based optical character recognition tool designed for line detection, CTC recognition, reading order, and markdown processing. It is ideal for developers…Indra OCR is a fast CPU-based optical character recognition tool designed for line detection, CTC recognition, reading order, and markdown processing. It is ideal for developers…

LLM architecture from scratch

AI GitHub Repos

This week

llm-architecture-from-scratch is a GitHub repository by Yeonwoo Sung that focuses on implementing large language model architectures from scratch using PyTorch. It provides a…

Qwen-Image-2.1-Turbo

AI Models

This week

Accelerated version of Qwen-Image-2.1, designed for efficient text-to-image generation and image editing. It features an 8-step denoising process and…Qwen-Image-2.1-Turbo is an accelerated version of Qwen-Image-2.1, designed for efficient text-to-image generation and image editing. It features an 8-step denoising process and…

LIDstral Arabic

AI Models

This week

Language and dialect classifier for Arabic script. It identifies Modern Standard Arabic, various Arabic dialects, and non-Arabic languages using the same…LIDstral Arabic is a language and dialect classifier for Arabic script. It identifies Modern Standard Arabic, various Arabic dialects, and non-Arabic languages using the same…

Step 5 Preview

AI Models

This week

StepFun's flagship AI model designed for agentic work. It excels in software engineering, professional knowledge work, and finance, offering advanced…Step 5 Preview is StepFun's flagship AI model designed for agentic work. It excels in software engineering, professional knowledge work, and finance, offering advanced…

Laya

AI GitHub Repos

This week

Open-source decision model compatible with Jev's System-1, designed for execution in Node.js and TypeScript environments via ONNX Runtime. It facilitates…Laya is an open-source decision model compatible with Jev's System-1, designed for execution in Node.js and TypeScript environments via ONNX Runtime. It facilitates…

MOSS-Image-SFT

AI Models

This week

Supervised fine-tuned text-to-image model from OpenMOSS. It supports natural-language prompts, tag sequences, and JSON layouts for text and object arrangement,…MOSS-Image-SFT is a supervised fine-tuned text-to-image model from OpenMOSS. It supports natural-language prompts, tag sequences, and JSON layouts for text and object arrangement,…

StartLux-Decision

AI GitHub Repos

This week

Offers open decision models ranging from 0.8B to 27B parameters, providing calibrated probabilities for each option. It generates concise thoughts when…StartLux-Decision offers open decision models ranging from 0.8B to 27B parameters, providing calibrated probabilities for each option. It generates concise thoughts when…

Qwen3-8B-DSpine-b16

AI Models

Yesterday

AI model designed for speculative decoding, enabling parallel drafting with adjacent causal injection. It enhances performance by unfolding the causal…Qwen3-8B-DSpine-b16 is an AI model designed for speculative decoding, enabling parallel drafting with adjacent causal injection. It enhances performance by unfolding the causal…

PS5LM

AI GitHub Repos

Yesterday

Enables the execution of GGUF models on a jailbroken PS5, starting with Qwen 3.8. This project leverages llama.cpp to run models natively on the console, offering a unique…PS5LM enables the execution of GGUF models on a jailbroken PS5, starting with Qwen 3.8. This project leverages llama.cpp to run models natively on the console, offering a unique…

Wanhe AI Tools

AI Apps

Today

Offers a local AI toolbox that includes 22 APIs and 31 Edge extensions. It allows users to run AI models on their own computers, ensuring data privacy and offline…Wanhe AI Tools offers a local AI toolbox that includes 22 APIs and 31 Edge extensions. It allows users to run AI models on their own computers, ensuring data privacy and offline…

EdgeRock

AI GitHub Repos

Today

Multimodal model offering strong reasoning capabilities with constant memory usage. It supports O(N) generation and O(1) persistent history, providing efficient…EdgeRock is a multimodal model offering strong reasoning capabilities with constant memory usage. It supports O(N) generation and O(1) persistent history, providing efficient…

MOSS-Image

AI GitHub Repos

Today

5B-parameter base DiT model designed for high-quality text rendering, layout generation, and efficient fine-tuning. It serves as a robust tool for developers and…MOSS-Image is a 5B-parameter base DiT model designed for high-quality text rendering, layout generation, and efficient fine-tuning. It serves as a robust tool for developers and…