Upload README.md with huggingface_hub

README.md

@@ -1,14 +1,43 @@
 ---
 library_name: swirl-wine-recommendations
 tags:
-- model_hub_mixin
-- pytorch_model_hub_mixin
 - recommendation
 - two-tower
 - wine
+- pytorch
 ---
 
-This model has been pushed to the Hub using the [PytorchModelHubMixin](https://huggingface.co/docs/huggingface_hub/package_reference/mixins#huggingface_hub.PyTorchModelHubMixin) integration:
-- Code: [More Information Needed]
-- Paper: [More Information Needed]
-- Docs: [More Information Needed]
+# Two-Tower Wine Recommendation Model
+
+Neural network for personalized wine recommendations.
+
+## Usage
+
+This model is designed to be used with HuggingFace Inference Endpoints.
+The custom `handler.py` processes requests in this format:
+
+```json
+{
+    "inputs": {
+        "user_wines": [
+            {"embedding": [768 floats], "rating": 4.5},
+            ...
+        ],
+        "candidate_wine": {
+            "embedding": [768 floats],
+            "color": "red",
+            "type": "still"
+        }
+    }
+}
+```
+
+Returns: `{"score": 75.5}`
+
+## Architecture
+
+- **User Tower**: Encodes user preferences from reviewed wines
+- **Wine Tower**: Encodes wine characteristics
+- **Output**: Match score 0-100 via dot product
+
+See [Swirl Wine App](https://swirlyourwines.com) for more info.