InternLM/CapRL

[ICLR 2026] An official implementation of "CapRL: Stimulating Dense Image Caption Capabilities via Reinforcement Learning"

43
/ 100
Emerging

This project helps anyone needing highly detailed, accurate descriptions for images, especially those containing charts, infographics, or documents. You provide an image, and it outputs a well-structured, comprehensive text caption covering all visual information, with fewer inaccuracies than other models. Marketers, researchers, data analysts, or educators can use this to quickly generate descriptions for visual content.

193 stars.

Use this if you need to automatically generate very detailed and accurate textual descriptions for images, particularly those with complex visual information like charts, diagrams, or dense documents.

Not ideal if you only need very brief, high-level descriptions for natural images, or if your primary concern is extremely low computational cost over descriptive richness.

image-description data-visualization-analysis document-intelligence visual-content-analysis digital-asset-management
No Package No Dependents
Maintenance 10 / 25
Adoption 10 / 25
Maturity 15 / 25
Community 8 / 25

How are scores calculated?

Stars

193

Forks

6

Language

Python

License

Apache-2.0

Last pushed

Feb 08, 2026

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/transformers/InternLM/CapRL"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.