Update app.py

app.py

@@ -66,35 +66,44 @@ class CurriculumChatbot:
     
     def _setup_llm(self):
         try:
-            # Use a smaller, faster model for Hugging Face Spaces
-            model_name = "microsoft/DialoGPT-medium"  # Smaller model for faster inference
+            # Use Llama 3.1 8B for better quality answers
+            model_name = "meta-llama/Meta-Llama-3.1-8B-Instruct"
             pipe = pipeline(
                 "text-generation",
                 model=model_name,
                 max_new_tokens=200,
-                temperature=0.7,
+                temperature=0.3,
                 do_sample=True,
-                pad_token_id=50256
+                top_p=0.9,
+                repetition_penalty=1.1,
+                device_map="auto" if torch.cuda.is_available() else None
             )
             self.llm = HuggingFacePipeline(pipeline=pipe)
             
-            # Create QA prompt template for DialoGPT
-            qa_template = """You are a helpful programming tutor. Answer the following question based on the curriculum content provided.
+            # Create QA prompt template for Llama 3.1
+            qa_template = """<|begin_of_text|><|start_header_id|>system<|end_header_id|>
 
-Curriculum Content:
-{filled_context}
+You are a helpful AI programming tutor. Answer questions about programming concepts clearly and educationally. If the question is about curriculum content, use the provided context. If not, provide a general programming answer.
+
+<|eot_id|><|start_header_id|>user<|end_header_id|>
 
 Question: {question}
 
-Provide a clear, educational answer explaining the concept:"""
+{filled_context}
+
+<|eot_id|><|start_header_id|>assistant<|end_header_id|>"""
             
             self.qa_chain = LLMChain(llm=self.llm, prompt=PromptTemplate(
                 input_variables=["question", "filled_context"],
                 template=qa_template
             ))
             
-            # Create slide selection prompt template for DialoGPT
-            slide_selection_template = """You are an AI that analyzes curriculum slides to find the best one for teaching a concept.
+            # Create slide selection prompt template for Llama 3.1
+            slide_selection_template = """<|begin_of_text|><|start_header_id|>system<|end_header_id|>
+
+You are an AI that analyzes curriculum slides to find the best one for teaching a concept. Return ONLY the filename and page number.
+
+<|eot_id|><|start_header_id|>user<|end_header_id|>
 
 Question: {question}
 
@@ -107,33 +116,37 @@ Which slide is the BEST for teaching this concept to a student? Consider:
 - Which slide explains the concept most clearly?
 - Which slide would be most helpful for learning?
 
-Return ONLY the filename and page number like this: "filename.pdf - Page X"
+Return only: "filename.pdf - Page X"
 
-Answer:"""
+<|eot_id|><|start_header_id|>assistant<|end_header_id|>"""
             
             self.slide_selection_chain = LLMChain(llm=self.llm, prompt=PromptTemplate(
                 input_variables=["question", "slide_contents"],
                 template=slide_selection_template
             ))
             
-            # Create focused answer prompt template
-            focused_qa_template = """You are a helpful programming tutor. Answer the question based on the specific slide content provided.
+            # Create focused answer prompt template for Llama 3.1
+            focused_qa_template = """<|begin_of_text|><|start_header_id|>system<|end_header_id|>
+
+You are a helpful AI programming tutor. Answer questions about programming concepts clearly and educationally based on the provided slide content.
+
+<|eot_id|><|start_header_id|>user<|end_header_id|>
 
 Slide Content:
 {slide_content}
 
 Question: {question}
 
-Provide a clear, educational answer based on this slide:"""
+<|eot_id|><|start_header_id|>assistant<|end_header_id|>"""
             
             self.focused_qa_chain = LLMChain(llm=self.llm, prompt=PromptTemplate(
                 input_variables=["question", "slide_content"],
                 template=focused_qa_template
             ))
             
-            print("✅ Llama 3.1-8B loaded successfully!")
+            print("✅ Llama 3.1 8B loaded successfully!")
         except Exception as e:
-            print(f"Warning: Could not load Llama 3.1-8B: {e}")
+            print(f"Warning: Could not load Llama 3.1 8B: {e}")
             print("Falling back to basic search mode...")
             self.llm = None
             self.qa_chain = None