File size: 705 Bytes
51aa9df 1736018 51aa9df 1736018 51aa9df 8f18d06 51aa9df 1736018 477cc28 51aa9df 477cc28 1736018 477cc28 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 |
---
base_model:
- Qwen/Qwen2.5-VL-7B-Instruct
datasets:
- TIGER-Lab/PixelReasoner-SFT-Data
language:
- en
library_name: transformers
license: apache-2.0
metrics:
- accuracy
pipeline_tag: image-text-to-text
---
The model is trained with curiosity-driven RL described in [paper](https://arxiv.org/abs/2505.15966).
We have released vllm based inference code at https://github.com/TIGER-AI-Lab/Pixel-Reasoner/.
Project page: https://tiger-ai-lab.github.io/Pixel-Reasoner/
Github repository: https://github.com/TIGER-AI-Lab/Pixel-Reasoner/
We will release a simple hf.generate() based inference code.
Please also play with the cool [interactive demo](https://huggingface.co/spaces/TIGER-Lab/Pixel-Reasoner) |