Datayoo/HuggingFists

What it solves

HuggingFists is a low-code data flow tool designed to simplify the integration and use of Large Language Models (LLMs) and Hugging Face models. It removes the need for extensive coding to build complex data processing pipelines, acting as a visual alternative to frameworks like LangChain.

How it works

Users build workflows by dragging and dropping operators (functional blocks) and connecting their input and output ports. The system manages the entire lifecycle of these flows, from data source connection to execution.

• Model Access: It supports accessing Hugging Face models via the Inference API (cloud) or by pointing to locally deployed model files on a CPU or GPU.
• Data Management: It includes a centralized account and data source management system to handle API tokens, database connections, and file systems.
• Execution & Debugging: Workflows can be executed immediately or scheduled as recurring jobs. The platform provides breakpoint debugging, allowing users to inspect data structures at specific ports to ensure the flow is correct.
• Extensibility: Users can expand the system by adding new connectors (for data sources) and operators (for specific functionalities).

Who it’s for

It is intended for users who want to deploy AI-driven data processing tasks—such as text summarization, sentiment analysis, or image classification—without writing deep integration code.

Highlights

• Visual Workflow Designer: Drag-and-drop interface for building executable data processing logic.
• Hybrid Model Deployment: Supports both Hugging Face Inference API and local model execution.
• Breakpoint Debugging: Ability to set breakpoints on operator ports to verify data flow.
• Comprehensive Data Integration: Built-in connectors for Hugging Face datasets, MySQL, and various file formats (PDF, Word, Excel, etc.).