Update pipeline tag to image-to-video and add paper link (#1)

- Update pipeline tag to image-to-video and add paper link (77184e5ebbf78771c4a4e8b31766d5655a3ea34f)


Co-authored-by: Niels Rogge <nielsr@users.noreply.huggingface.co>

README.md

@@ -1,8 +1,8 @@
 ---
-license: apache-2.0
-library_name: pytorch
-pipeline_tag: text-to-video
 base_model: Wan-AI/Wan2.2-I2V-A14B
+library_name: pytorch
+license: apache-2.0
+pipeline_tag: image-to-video
 tags:
 - video-generation
 - image-to-video
@@ -16,7 +16,7 @@ tags:
 Official checkpoints for SlotMem, a character-addressable internal memory framework for multi-character narrative long video generation. SlotMem represents recurring characters with compact role-wise memory slots, updates them during autoregressive generation, and injects them only into localized character-relevant tokens.
 
 - Repository: [YilaiLiu-HKU/SlotMem](https://github.com/YilaiLiu-HKU/SlotMem)
-- Paper: coming soon
+- Paper: [SlotMem: Character-Addressable Internal Memory for Narrative Long Video Generation](https://huggingface.co/papers/2607.15772)
 
 <p align="center">
   <img src="https://raw.githubusercontent.com/YilaiLiu-HKU/SlotMem/main/assets/jigsaw2.png" alt="SlotMem method overview" width="95%">
@@ -56,4 +56,11 @@ For training, data curation, benchmarking, and sample data format, please refer
 
 ## Citation
 
-The BibTeX entry will be updated when the paper is public.
+```bibtex
+@article{liu2026slotmem,
+  title={SlotMem: Character-Addressable Internal Memory for Narrative Long Video Generation},
+  author={Liu, Yilai and others},
+  journal={arXiv preprint arXiv:2607.15772},
+  year={2026}
+}
+```