Add model card with pipeline_tag for Inference API

README.md

@@ -0,0 +1,44 @@
+---
+pipeline_tag: text2text-generation
+library_name: transformers
+tags:
+- t5
+- text2text-generation
+- sql
+- text-to-sql
+- wikisql
+---
+
+# T2SQL V6 Structured - Text to SQL
+
+Fine-tuned T5 model that converts natural language questions to SQL queries.
+
+## Usage
+
+```python
+from transformers import pipeline
+
+pipe = pipeline("text2text-generation", model="RealMati/t2sql_v6_structured")
+result = pipe("translate to SQL: list all users older than 18 | schema: users(id, name, age, email)")
+print(result[0]["generated_text"])
+```
+
+## Inference API
+
+```python
+import requests
+
+API_URL = "https://router.huggingface.co/hf-inference/models/RealMati/t2sql_v6_structured"
+headers = {"Authorization": "Bearer hf_YOUR_TOKEN"}
+
+response = requests.post(API_URL, headers=headers, json={
+    "inputs": "translate to SQL: How many users are there? | schema: users(id, name, age)"
+})
+print(response.json())
+```
+
+## Training
+
+- **Base model:** T5-base
+- **Dataset:** WikiSQL (56k train / 8k val / 15k test)
+- **Task:** Natural language to structured SQL output