Updated with Content msg, product recommendation and history information2

app.py

@@ -39,7 +39,7 @@ else:
 tokenizer = BlenderbotTokenizer.from_pretrained(model_dir)
 model = BlenderbotForConditionalGeneration.from_pretrained(model_dir)
 
-# Static Context based on Hutter Products GmbH home page
+# Static Context
 context_msg = "I am Hutter, your shopping guide for Hutter Products GmbH. I’m here to help you explore our innovative and sustainable product catalog, featuring eco-friendly items like recycled textiles and ocean plastic goods. Let me assist you in finding the perfect sustainable solution!"
 
 # spaCy Setup
@@ -65,15 +65,16 @@ def extract_keywords(text: str) -> List[str]:
 def detect_intent(text: str) -> str:
     doc = nlp(text.lower())
     text_lower = text.lower()
-    if any(token.text in ["buy", "shop", "find", "recommend", "product", "products", "item", "textile", "jacket", "shirt", "shorts"] for token in doc):
+    # General product-related intent based on shopping context
+    if any(token.text in ["buy", "shop", "find", "recommend", "product", "products", "item", "store", "catalog"] for token in doc) or "what" in text_lower.split()[:2]:
         return "recommend_product"
-    elif any(token.text in ["what", "who", "company", "do"] for token in doc):
+    elif any(token.text in ["company", "who", "do"] for token in doc):
         return "company_info"
     elif "name" in text_lower:
         return "ask_name"
     elif re.search(r"\d+\s*[\+\-\*/]\s*\d+", text_lower):
         return "math_query"
-    return "recommend_product"  # Default to product exploration
+    return "recommend_product"  # Default to product focus for scalability
 
 def search_products_by_keywords(keywords: List[str]) -> List[Dict[str, Any]]:
     if not keywords:
@@ -94,22 +95,19 @@ def search_products_by_keywords(keywords: List[str]) -> List[Dict[str, Any]]:
 def get_product_context(products: List[Dict]) -> str:
     if not products:
         return ""
-    product_str = "Here are some products: "
+    product_str = "Available products: "
     product_str += ", ".join([f"'{p['name']}' - {p['description']}" for p in products[:2]])
     return product_str
 
 def format_response(response: str, products: List[Dict], intent: str, input_text: str) -> str:
-    # Highlight products if available
-    if products:
-        product = products[0]  # Prioritize the first matched product
-        product_highlight = f" How about our '{product['name']}'? It’s {product['description'].lower()}."
-        response += product_highlight
-
-    # Intent-specific tweaks
+    # Handle product recommendation intent
     if intent == "recommend_product":
-        return response  # Product info already appended if available
+        if not products:
+            return "I’d love to recommend something from our sustainable catalog! Could you tell me more about what you’re looking for?"
+        product = products[0]
+        return f"Check out our '{product['name']}'—it’s {product['description'].lower()}. Want to explore more options?"
     elif intent == "company_info":
-        return f"{response} Hutter Products GmbH specializes in sustainable product design and production."
+        return "Hutter Products GmbH specializes in sustainable product design and production, offering eco-friendly items like recycled textiles and ocean plastic goods."
     elif intent == "ask_name":
         return "I’m Hutter, your shopping guide for Hutter Products GmbH. How can I assist you today?"
     elif intent == "math_query":
@@ -117,15 +115,19 @@ def format_response(response: str, products: List[Dict], intent: str, input_text
         if match:
             num1, op, num2 = int(match.group(1)), match.group(2), int(match.group(3))
             if op == "+":
-                return f"{response} Also, {num1} plus {num2} is {num1 + num2}."
+                return f"{num1} plus {num2} is {num1 + num2}. Need help with shopping too?"
             elif op == "-":
-                return f"{response} Also, {num1} minus {num2} is {num1 - num2}."
+                return f"{num1} minus {num2} is {num1 - num2}. Anything else I can assist with?"
             elif op == "*":
-                return f"{response} Also, {num1} times {num2} is {num1 * num2}."
+                return f"{num1} times {num2} is {num1 * num2}. Want to explore our products?"
             elif op == "/":
-                return f"{response} Also, {num1} divided by {num2} is {num1 / num2}." if num2 != 0 else f"{response} Can’t divide by zero!"
-        return f"{response} I can handle simple math too—try something like '2 + 2'."
-    return response  # Default case includes products if matched
+                return f"{num1} divided by {num2} is {num1 / num2}." if num2 != 0 else "Can’t divide by zero! How about some sustainable products instead?"
+        return "I can do simple math—try '2 + 2'. What else can I help you with?"
+    # Fallback with product nudge if available
+    if products:
+        product = products[0]
+        return f"{response} By the way, how about our '{product['name']}'? It’s {product['description'].lower()}."
+    return response if response else "How can I assist you with our sustainable products today?"
 
 # Endpoints
 @app.get("/")
@@ -156,7 +158,7 @@ async def process_prompt(request: PromptRequest):
         logger.info("Input tokenized successfully.")
 
         logger.info("Generating model response...")
-        outputs = model.generate(**inputs, max_length=50, num_beams=1, no_repeat_ngram_size=2)
+        outputs = model.generate(**inputs, max_length=50, num_beams=4, length_penalty=1.0, no_repeat_ngram_size=2)
         logger.info("Model generation complete.")
 
         logger.info("Decoding model output...")