Update app.py

app.py

@@ -1,25 +1,67 @@
 from flask import Flask, request, jsonify
+import requests
 import logging
+import os
 
 logging.basicConfig(level=logging.INFO)
+logger = logging.getLogger(__name__)
+
 app = Flask(__name__)
 
+# Your model endpoint
+MODEL_URL = "https://api-inference.huggingface.co/models/kacperbb/phi-3.5-merged-lora"
+
 @app.route('/generate', methods=['POST'])
 def generate():
     try:
         data = request.json
         prompt = data.get('inputs', '')
-        return jsonify([{"generated_text": f"Response to: {prompt}"}])
+        max_tokens = data.get('parameters', {}).get('max_new_tokens', 100)
+        
+        # Get HF token from environment variable (you'll need to set this in Space settings)
+        hf_token = os.environ.get('HF_TOKEN')
+        
+        if hf_token:
+            # Call your actual model via HF Inference API
+            headers = {"Authorization": f"Bearer {hf_token}"}
+            payload = {
+                "inputs": prompt,
+                "parameters": {"max_new_tokens": max_tokens}
+            }
+            
+            try:
+                response = requests.post(MODEL_URL, headers=headers, json=payload, timeout=30)
+                if response.status_code == 200:
+                    return jsonify(response.json())
+                else:
+                    logger.warning(f"HF API returned status {response.status_code}")
+            except requests.exceptions.RequestException as e:
+                logger.error(f"Request failed: {e}")
+        
+        # Fallback response if API call fails
+        return jsonify([{"generated_text": f"AI Response to '{prompt}': This would be generated by your fine-tuned Phi 3.5 model."}])
+        
     except Exception as e:
+        logger.error(f"Error: {e}")
         return jsonify({"error": str(e)}), 500
 
 @app.route('/health', methods=['GET'])
 def health():
-    return jsonify({"status": "healthy"})
+    has_token = bool(os.environ.get('HF_TOKEN'))
+    return jsonify({
+        "status": "healthy", 
+        "model": "kacperbb/phi-3.5-merged-lora",
+        "has_token": has_token
+    })
 
 @app.route('/')
 def home():
-    return jsonify({"message": "API Running", "model": "phi-3.5"})
+    return jsonify({
+        "message": "Phi 3.5 Fine-tuned API Running!", 
+        "model": "kacperbb/phi-3.5-merged-lora",
+        "endpoint": "/generate"
+    })
 
 if __name__ == '__main__':
+    logger.info("Starting Phi 3.5 Fine-tuned API...")
     app.run(host='0.0.0.0', port=7860)