Create README.md

README.md

@@ -0,0 +1,35 @@
+---
+language:
+- en
+- zh
+task_categories:
+- spoken-language-understanding
+- intent-classification
+- slot-filling
+---
+
+# TeleVRSLUBench
+
+## Dataset Description
+
+**TeleVRSLUBench** is a spoken language understanding (SLU) benchmark that incorporates **visual scene information and explicit reasoning processes** for **joint intent detection and slot filling**.
+
+The dataset is proposed in the paper:
+
+> *Introducing Visual Scenes and Reasoning: A More Realistic Benchmark for Spoken Language Understanding*
+
+It is, to the best of our knowledge, the **first SLU benchmark** that integrates scene-level visual context and reasoning information to support more realistic spoken language understanding.
+
+The official code repository is available at:  
+https://github.com/Tele-AI/TeleVRSLUBench
+
+---
+
+## Usage
+
+The dataset can be loaded using the Hugging Face `datasets` library:
+
+```python
+from datasets import load_dataset
+
+dataset = load_dataset("Tele-AI/TeleVRSLU")