DeepLightRAG: High-performance Document Indexing and Retrieval System (use with any LLM)
Project description
DeepLightRAG
DeepLightRAG is a high-performance document indexing and retrieval system designed to work with any Large Language Model (LLM). It features a dual-layer graph architecture (Visual-Spatial and Entity-Relationship) to provide context-aware and visually-grounded retrieval.
Features
- Dual-Layer Graph: Combines visual layout awareness with semantic entity relationships.
- Visual-Grounded Retrieval: Retrieves not just text, but visual regions and their spatial context.
- Robust OCR: Integrated with DeepSeek-OCR and EasyOCR fallback for reliable text extraction.
- Advanced NER: Uses GLiNER for zero-shot entity recognition.
- Flexible LLM Support: Compatible with OpenAI, Google Gemini, Anthropic, and local LLMs via MLX/Ollama.
Installation
pip install deeplightrag
Usage
Index a document:
deeplightrag index document.pdf
Query the index:
deeplightrag query "What is the main topic?"
License
MIT License
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file deeplightrag-1.0.5.tar.gz.
File metadata
- Download URL: deeplightrag-1.0.5.tar.gz
- Upload date:
- Size: 142.6 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.13.7
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
fcb193ea8bee8fdfbb114c3c0357cd3bfaa24d79ee58c4da52792204463f0226
|
|
| MD5 |
0e14d3b2420f69e847da49699007e731
|
|
| BLAKE2b-256 |
6f75bec7c5c42a7d7066151b9755d4ddd402ba43bb0cf35be98872138fcdf68e
|
File details
Details for the file deeplightrag-1.0.5-py3-none-any.whl.
File metadata
- Download URL: deeplightrag-1.0.5-py3-none-any.whl
- Upload date:
- Size: 156.1 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.13.7
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
f41b747b9f4d7514580901343ad738138de37985f3f6fce9de221a4769d10d79
|
|
| MD5 |
fa49a8f341a794092745de29d5f0836b
|
|
| BLAKE2b-256 |
93eab0880b86bd7e3182f4b1c74830751bb1cd142cf679be73d20f7cd2f9ac94
|
