📄 Paper · 🌐 Project Page · 🤖 Models · 🤗 Dataset · 🏆 Leaderboard · 💻 Code · 🎬 Video

IntReal-Bagel-SFT-RL

The Bagel checkpoint fine-tuned on IntReal-100K for multimodal interleaved generation.

Inference

Use our Bagel inference script. See the inference guide for environment setup and usage.

Evaluation

Evaluate the model with the IntReal Evaluation Suite.

Citation

@misc{yu2026intrealrealworlddatasetbenchmark,
      title={IntReal: A Real-World Dataset and Benchmark for Multimodal Interleaved Generation}, 
      author={Shaohan Yu and Jie Qin and Wende Tan and Haiwen Diao and Limeng Qiao and Yueming Lyu and Chenyang Si},
      year={2026},
      eprint={2610.38425},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2610.38425}, 
}