Add paper link and task category

This PR updates the dataset card to include:
- A link to the official paper: [Mobile-Bench-v2: A More Realistic and Comprehensive Benchmark for VLM-based Mobile Agents](https://huggingface.co/papers/2505.11891).
- The `image-text-to-text` task category for better discoverability.
- Updated license metadata to `cc-by-nc-sa-4.0` to match the dataset's specific license mentioned in the description.

README.md

@@ -1,9 +1,14 @@
 ---
-license: apache-2.0
+license: cc-by-nc-sa-4.0
+task_categories:
+- image-text-to-text
 ---
+
 # Mobile-Bench v2
-<!-- ![Mobile-Bench v2 Overview](Mobile_bench_v2.png)  
-*Figure 1: Mobile-Bench v2 Overview* -->
+
+[**Paper**](https://huggingface.co/papers/2505.11891)
+
+Mobile-Bench-v2 is a comprehensive benchmark for VLM-based mobile agents designed to address the limitations of existing benchmarks. It introduces more realistic scenarios including multi-path evaluation for GUI tasks, a noisy split (containing apps with pop-ups and ads), and an ambiguous instruction split to evaluate proactive interaction capabilities.
 
 ### Dataset Description:
 - **anbu-instruction.zip** is a subset of the ambiguous category split in Mobile-Bench-v2.