Add model card

README.md

@@ -0,0 +1,28 @@
+---
+pipeline_tag: image-text-to-text
+---
+
+# SwimBird: Eliciting Switchable Reasoning Mode in Hybrid Autoregressive MLLMs
+
+[Paper](https://huggingface.co/papers/2602.06040) | [Project Page](https://accio-lab.github.io/SwimBird) | [GitHub](https://github.com/Accio-Lab/SwimBird)
+
+SwimBird is a reasoning-switchable Multimodal Large Language Model (MLLM) that dynamically switches among three reasoning modes conditioned on the input:
+1. **Text-only reasoning**: Standard textual Chain-of-Thought.
+2. **Vision-only reasoning**: Utilizes continuous hidden states as "visual thoughts" for vision-intensive tasks.
+3. **Interleaved vision-text reasoning**: A combination of both textual and visual thinking modalities.
+
+By enabling flexible, query-adaptive mode selection, SwimBird preserves strong textual logic while substantially improving performance on vision-dense tasks.
+
+## Method
+SwimBird adopts a hybrid autoregressive formulation that unifies next-token prediction for textual thoughts with next-embedding prediction for visual thoughts. To enable this capability, the authors designed a systematic reasoning-mode curation strategy to construct **SwimBird-SFT-92K**, a diverse supervised fine-tuning dataset covering all three reasoning patterns. Experiments across diverse benchmarks covering textual reasoning and challenging visual understanding demonstrate that SwimBird achieves state-of-the-art results and robust gains over prior fixed-pattern multimodal reasoning methods.
+
+## Citation
+
+```bibtex
+@article{tong2025swimbird,
+  title={SwimBird: Eliciting Switchable Reasoning Mode in Hybrid Autoregressive MLLMs},
+  author={Tong, Jintao and Yan, Shilin and Xue, Hongwei and Tang, Xiaojun and Shi, Kunyu and Zhang, Guannan and Li, Ruixuan and Zou, Yixiong},
+  journal={arXiv preprint arXiv:2602.06040},
+  year={2025}
+}
+```