🌤️ Weather-R1: Multimodal Reasoning in Meteorology

This repository contains the checkpoints for Weather-R1, as presented in the paper Weather-R1: Logically Consistent Reinforcement Fine-Tuning for Multimodal Reasoning in Meteorology.

Paper (ArXiv) | Code (GitHub)

🌤️ Introduction

While Vision Language Models (VLMs) show advancing reasoning capabilities, their application in meteorology is constrained by a domain gap and a reasoning faithfulness gap. Mainstream Reinforcement Fine-Tuning (RFT) can induce Self-Contradictory Reasoning (Self-Contra), where the reasoning process contradicts the final answer, which is unacceptable in this high-stakes domain.

To address these challenges, we construct WeatherQA, a multimodal multiple-choice benchmark for meteorology comprising 15,400 entries that cover four themes and seven imaging modality tasks. We propose Logically Consistent Reinforcement Fine-Tuning (LoCo-RFT), which introduces a logical consistency reward to resolve Self-Contra. Based on this paradigm and WeatherQA, we present Weather-R1, the first reasoning VLM with logical faithfulness in meteorology, to the best of our knowledge. Weather-R1 (7B) achieves 52.9% accuracy on WeatherQA, a 9.8 percentage point gain over the baseline model Qwen2.5-VL-7B; it surpasses Supervised Fine-Tuning and RFT baselines, exceeds the original Qwen2.5-VL-32B, and improves out-of-domain ScienceQA performance by 4.98 percentage points.

Response Comparison.

🗂️ Folder Structure

This repository provides model checkpoints organized by training strategy and task:

Weather-R1/
├─ LoCo-RFT/                 # Weather-R1 checkpoints
│  ├─ WeatherQA-500hPa/
│  ├─ WeatherQA-850hPa/
│  ├─ WeatherQA-Land/
│  ├─ WeatherQA-Max-Temp/
│  ├─ WeatherQA-Min-Temp/
│  ├─ WeatherQA-Phenom/
│  └─ WeatherQA-Rain/
├─ RFT/                      # Standard RFT checkpoints
│  ├─ WeatherQA-500hPa/
│  ├─ WeatherQA-850hPa/
│  ├─ WeatherQA-Land/
│  ├─ WeatherQA-Max-Temp/
│  ├─ WeatherQA-Min-Temp/
│  ├─ WeatherQA-Phenom/
│  └─ WeatherQA-Rain/
└─ asserts/                  # Figures used in README

Each task folder contains HuggingFace-style model files such as config.json, tokenizer.json, and sharded weights like model-00001-of-00004.safetensors.

🚀 Training and Evaluation

Please refer to our official repository: Weather-R1

🙏 Acknowledgements

Training code is built on EasyR1.

📝 Citation

If you use Weather-R1 resources, please cite the following paper:

@misc{wu2026weatherr1logicallyconsistentreinforcement,
      title={Weather-R1: Logically Consistent Reinforcement Fine-Tuning for Multimodal Reasoning in Meteorology}, 
      author={Kaiyu Wu and Pucheng Han and Hualong Zhang and Naigeng Wu and Keze Wang},
      year={2026},
      eprint={2601.14044},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2601.14044}, 
}