#81 · Primary category: Computer Vision

sparrow

agentic-ai computer-vision documentai huggingface-transformers llm machinelearning vllm

Structured data extraction, instruction calling and agentic workflows with ML, LLM and Vision LLM

Project last updated:08/26/26

GitHub Stars

5.2K

Forks

519

Contributors

4

License

GPL-3.0

Why we included this project

Sparrow is a self-hosted document processing platform: you submit a document over REST and get back schema-validated JSON, with nothing sent to a third-party cloud. It pairs OCR preprocessing with vision and text LLMs, and the pipeline is pluggable, so you can choose a vision-based parser, an instruction-calling model, or orchestrated agents depending on the task. Backends span MLX, vLLM, Ollama, and Hugging Face, and the instruction and agent APIs go beyond extraction to handle validation, decision-making, and multi-step workflows. A web UI and dashboard cover monitoring. If your team routinely turns invoices, receipts, bank statements, forms, or tables into structured records, this is a practical way to do it on your own hardware.

Articles for this project

No articles for this project yet.

To suggest a topic or contribute an article, contact us.

Related projects in this category