---
license: other
language:
- en
size_categories:
- 10K<n<100K
task_categories:
- visual-question-answering
pretty_name: Future-L1-50K
tags:
- video
- multimodal
- visual-reasoning
- video-event-prediction
- future-prediction
- latent-visual-reasoning
- supervised-fine-tuning
---

# Future-L1-50K

Future-L1-50K is the **SFT training set** used by **Future-L1: Imagine
Before You Predict**, a framework for video event prediction with interleaved
language and latent visual reasoning.

- Project repository: https://github.com/OpenGVLab/Future-L1
- Paper: https://arxiv.org/abs/2606.05769
- Dataset file: `Future-L1-50K.json`
- Video download references:
  https://github.com/snap-research/Panda-70M/tree/main/dataset_dataloading
  and https://huggingface.co/datasets/Liu-Junhua/TwiFF-2.7M


## Citation

If you use this dataset, please cite the paper:

```bibtex
@article{jiang2026imagine,
  title={Imagine Before You Predict: Interleaved Latent Visual Reasoning for Video Event Prediction},
  author={Jiang, Tianxiang and Wu, Linquan and Xia, Sheng and Li, Songze and Yan, Ziang and Yang, Haoyu and Qiao, Yu and Wang, Yi},
  journal={arXiv preprint arXiv:2606.05769},
  year={2026}
}
```