Add paper and code links, task category image-text-to-text

README.md

@@ -1,13 +1,14 @@
 ---
-license: cc-by-nc-sa-4.0
-pretty_name: DriveHierarchy
-task_categories:
-- visual-question-answering
-- question-answering
 language:
 - en
+license: cc-by-nc-sa-4.0
 size_categories:
 - 10K<n<100K
+task_categories:
+- image-text-to-text
+- visual-question-answering
+- question-answering
+pretty_name: DriveHierarchy
 tags:
 - autonomous-driving
 - vision-language-models
@@ -16,6 +17,8 @@ tags:
 
 # 🚗 DriveHierarchy
 
+[Paper](https://huggingface.co/papers/2609.31814) | [Code](https://github.com/PerfectXu88/DriveHierarchy)
+
 DriveHierarchy is a hierarchical research benchmark for evaluating vision-language models in autonomous-driving settings. It separates capability assessment into four ranks, progressing from visual grounding to interactive execution:
 
 | Rank | Capability | Evaluation setting |
@@ -218,4 +221,4 @@ Users should combine benchmark evaluation with independent robustness testing, c
 | DriveHierarchy-authored source code | [Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0) |
 | Third-party and derived material | Original source terms; not relicensed by DriveHierarchy |
 
-See [`LICENSE`](./LICENSE) for the authoritative component-specific notices, links, attribution requirements, and restrictions, including those applicable to LingoQA, DRAMA, WOMD-Reasoning, and Waymo-derived material.
+See [`LICENSE`](./LICENSE) for the authoritative component-specific notices, links, attribution requirements, and restrictions, including those applicable to LingoQA, DRAMA, WOMD-Reasoning, and Waymo-derived material.