Update README.md

README.md

@@ -1,4 +1,15 @@
-# TABench
+---
+license: apache-2.0
+task_categories:
+- visual-question-answering
+language:
+- en
+- zh
+pretty_name: T
+size_categories:
+- 1K<n<10K
+---
+# TextAnchor-Bench (TABench)
 
 **TABench** is a comprehensive benchmark designed to evaluate the text anchoring capabilities of Vision-Language Models (VLMs). It assesses whether a model can:
 1.  **Region-to-Text (R2T)**: Accurately read the text within a specified region.
@@ -46,7 +57,7 @@ python simple_vllm_infer.py \
     --model_path /path/to/your/model \
     --coords abs \
     --output-dir ./infer_output \
-    --concurrency 64
+    --concurrency 4
 ```
 
 *   `--coords`: Choose `abs` (absolute coordinates) or `rel1000` (relative coordinates).
@@ -78,4 +89,4 @@ If you use TABench in your research, please cite our paper:
   journal={arXiv preprint arXiv:2604.00161},
   year={2026}
 }
-```
+```