Update task category to image-text-to-text

Update task category to `image-text-to-text`.

README.md

@@ -1,15 +1,15 @@
 ---
+language:
+- en
 license: apache-2.0
+size_categories:
+- 10K<n<100K
 task_categories:
-  - visual-question-answering
-language:
-  - en
-tags:
-  - spatial-reasoning
-  - multimodal
+- image-text-to-text
 pretty_name: Spatial457
-size_categories:
-  - 10K<n<100K
+tags:
+- spatial-reasoning
+- multimodal
 ---
 
 <div align="center">
@@ -155,6 +155,4 @@ The inference script supports [VLMEvalKit](https://github.com/open-compass/VLMEv
   year      = {2025},
   url       = {https://arxiv.org/abs/2502.08636}
 }
-```
-
-
+```