takisan/Game-RL-Qwen2.5-VL-7B_fix_img

*This model (GameQA-Qwen2.5-VL-7B) results from training Qwen2.5-VL-7B with GRPO solely on our [GameQA-5K](https://huggingface.co/datasets/Code2Logic/GameQA-5K) (sampled from the full [GameQA-140K](https://huggingface.co/datasets/Gabriel166/GameQA-140K) dataset).*

Evaluation Results on General Vision BenchMarks

<div align=center><img src="https://raw.githubusercontent.com/tongjingqi/Code2Logic/refs/heads/main/assets/evaluationresultsongeneralvision_benchmarks.png"></div>

*(The inference and evaluation configurations were unified across both the original open-source models and our trained models.)*

It's also found that getting trained on 5k samples from our GameQA dataset can lead to better results than on 8k samples from MAVIS and on multimodal-open-r1-8k-verified.

<div align=center><img src="https://raw.githubusercontent.com/tongjingqi/Code2Logic/refs/heads/main/assets/GameQAgeneralizesbetter.png"></div>

Code2Logic: Game-Code-Driven Data Synthesis for Enhancing VLMs General Reasoning

This is the first work, to the best of our knowledge, that leverages *game code to synthesize multimodal reasoning data for training VLMs. Furthermore, when trained with a GRPO strategy solely on GameQA (synthesized via our proposed Code2Logic* approach), multiple cutting-edge open-source models exhibit significantly enhanced out-of-domain generalization.

[📖 Paper] [💻 Code] [🤗 GameQA-140K Dataset] [🤗 GameQA-5K Dataset] [🤗 GameQA-InternVL3-8B ] [🤗 GameQA-InternVL2.5-8B ] [🤗 GameQA-Qwen2.5-VL-7B] [🤗 GameQA-LLaVA-OV-7B ]

Code: https://github.com/tongjingqi/Code2Logic

<div align=center><img src="https://raw.githubusercontent.com/tongjingqi/Code2Logic/refs/heads/main/assets/categorized30games_images.png"></div>

News

• —We've open-sourced the *three* models trained with GRPO on GameQA on Huggingface.