AI Data Processing Framework
3.4k 2026-04-18
towhee-io/towhee
Towhee is a cutting-edge framework designed to simplify and accelerate neural data processing pipelines, particularly for unstructured multimodal data and LLM orchestration.
Core Features
Multi-Modality Support for diverse data types (image, video, text, audio, molecular structures).
LLM Pipeline Orchestration with prompt management, knowledge retrieval, and local model hosting.
Rich Operators and over 140 state-of-the-art models across five AI domains.
Prebuilt ETL Pipelines for common tasks like RAG, text-image search, and video copy detection.
High-performance backend leveraging Triton Inference Server for efficient model serving and deployment.
Quick Start
pip install towhee towhee.modelsDetailed Introduction
Towhee is a powerful framework that streamlines the creation of neural data processing pipelines, focusing on unstructured data and Large Language Model (LLM) orchestration. It excels at extracting insights from diverse data types like text, images, audio, and video, transforming them into specific formats such as embeddings for vector databases. With a user-friendly Pythonic API, Towhee enables developers to rapidly prototype and optimize high-performance AI applications, leveraging generative AI and SOTA deep learning models for efficient production deployment.