TIGER-AI-Lab/VL-Rethinker

The official code of "VL-Rethinker: Incentivizing Self-Reflection of Vision-Language Models with Reinforcement Learning" [NeurIPS25]

37
/ 100
Emerging

This project helps improve the reasoning abilities of AI models that understand both images and text. It takes an existing vision-language AI model and makes it 'think' more deeply by explicitly encouraging self-reflection. The output is a more accurate AI model capable of solving complex problems in science, math, and other multidisciplinary fields. This is designed for AI researchers and developers working on advanced multimodal AI systems.

184 stars. No commits in the last 6 months.

Use this if you are developing or fine-tuning vision-language models and want to enhance their slow-thinking and self-reflection capabilities for better performance on complex multimodal reasoning tasks.

Not ideal if you are a general user looking for a ready-to-use application, or if you only work with text-based or image-based AI models without multimodal interaction.

AI model development multimodal AI reinforcement learning AI reasoning model evaluation
Stale 6m No Package No Dependents
Maintenance 2 / 25
Adoption 10 / 25
Maturity 16 / 25
Community 9 / 25

How are scores calculated?

Stars

184

Forks

8

Language

Python

License

Apache-2.0

Last pushed

Jun 05, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/transformers/TIGER-AI-Lab/VL-Rethinker"

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