Add video-text-to-text pipeline tag to metadata

Hi! I'm Niels from the Hugging Face community science team.

This pull request adds the `pipeline_tag: video-text-to-text` to your model card's metadata. This ensures that the model is correctly categorized on the Hugging Face Hub, making it more discoverable for users looking for video-to-text models.

No other changes have been made to the markdown content of your README. Feel free to merge this PR if it looks good!

README.md

@@ -2,20 +2,21 @@
 base_model: Qwen/Qwen2.5-VL-7B-Instruct
 library_name: peft
 license: apache-2.0
+pipeline_tag: video-text-to-text
 tags:
-  - peft
-  - lora
-  - vision-language
-  - video
-  - optical-flow
-  - uav
-  - drone
-  - motion
-  - embodied-ai
-  - spatial-intelligence
-  - vlm
-  - vla
-  - world-model
+- peft
+- lora
+- vision-language
+- video
+- optical-flow
+- uav
+- drone
+- motion
+- embodied-ai
+- spatial-intelligence
+- vlm
+- vla
+- world-model
 ---
 
 # SIS-Motion
@@ -202,7 +203,7 @@ Place SIS-Bench under the registered data root or override the paths directly:
 
 ```text
 data/SIS-Bench/SIS-Bench.jsonl
-data/SIS-Bench/frames/
+data/SIS-Bench/frames/\
 ```
 
 ### 4. Run inference
@@ -260,4 +261,4 @@ If you use SIS-Motion in your research, please cite:
 
 For questions or collaboration requests:
 
-[choucisan@gmail.com](mailto:choucisan@gmail.com)
+[choucisan@gmail.com](mailto:choucisan@gmail.com)