Description
Building RAG or fine-tuning data means material spread across PDFs, tables, images, audio, video and web pages, each needing its own parser. OmniParse turns all of that unstructured data into clean Markdown ready for language models.
It runs fully locally without external APIs, fits on a T4 GPU, supports about 20 file types, and offers a Gradio UI and Docker deployment.
Many formats:Documents, tables, images, audio, video and web pages.
Tables and images:Extract tables and images, with image captioning.
Transcription:Transcribe audio and video content.
Local:No external APIs, runs via Docker or Colab.
It runs fully locally without external APIs, fits on a T4 GPU, supports about 20 file types, and offers a Gradio UI and Docker deployment.
Features
Many formats:Documents, tables, images, audio, video and web pages.
Tables and images:Extract tables and images, with image captioning.
Transcription:Transcribe audio and video content.
Local:No external APIs, runs via Docker or Colab.
