Upload README.md with huggingface_hub

README.md

@@ -10,15 +10,16 @@ task_categories:
   - image-text-to-text
 size_categories:
   - 1K<n<10K
+arxiv: 2606.01247
 ---
 
 # TVR-SFT-VA
 
-Visual-Action SFT training data for **Target Viewpoint Reproduction (TVR)**, from the paper **"Where to Look: Can Foundation Models Reach a Target Viewpoint Through Active Exploration?"**.
+Visual-Action SFT training data for **Target Viewpoint Reproduction (TVR)**, from the paper **"Where to Look: Can Foundation Models Reach a Target Viewpoint Through Active Exploration?"** [[arXiv]](https://arxiv.org/abs/2606.01247).
 
 ## Dataset Description
 
-Expert trajectories collected by a rule-based planner. Each trajectory is a multi-turn conversation where the agent observes its current view and the target view, then takes an action to reduce the viewpoint discrepancy.
+Expert trajectories collected by a rule-based Dijkstra planner in [AI2-THOR](https://ai2thor.allenai.org/) indoor environments. Each trajectory is a multi-turn conversation where the agent observes its current view and the target view, then takes an action to reduce the viewpoint discrepancy.
 
 - **Samples**: 1,600 trajectories (multi-turn, average ~14 turns each)
 - **Images**: 22,335 frames (640×360, with CURRENT/TARGET labels burned in)
@@ -62,3 +63,17 @@ Each sample in `sft_train.json`:
 ## Usage
 
 See the [TVRBench repository](https://github.com/aim-uofa/TVRBench) for training configs and full instructions.
+
+## Citation
+
+```bibtex
+@misc{li2026lookfoundationmodelsreach,
+      title={Where to Look: Can Foundation Models Reach a Target Viewpoint Through Active Exploration?}, 
+      author={Liyang Li and Muzhi Zhu and Zhiyue Zhao and Hengyu Zhao and Ke Liu and Linhao Zhong and Hao Chen and Chunhua Shen},
+      year={2026},
+      eprint={2606.01247},
+      archivePrefix={arXiv},
+      primaryClass={cs.CV},
+      url={https://arxiv.org/abs/2606.01247}, 
+}
+```