Update app.py

app.py

@@ -36,6 +36,8 @@ from langgraph.graph import START, END, StateGraph
 from langchain_groq import ChatGroq
 from langchain_google_genai import ChatGoogleGenerativeAI
 from langchain_community.llms import HuggingFaceHub
+from langchain_huggingface import ChatHuggingFace, HuggingFaceEndpoint
+
 
 # RAG
 from langchain_text_splitters import RecursiveCharacterTextSplitter
@@ -1055,34 +1057,35 @@ REMEMBER: One tool per turn. No reasoning without tools. Exact answer format.
             #    max_new_tokens=2048, # Increased token limit for potentially longer reasoning/tool use
             #    temperature=0.01, # Keep temperature low for factual tasks     
             # Use tool_choice="any" to FORCE tool usage
-        print("Initializing Groq LLM...")   
-        try:
-            self.llm_with_tools = ChatGroq(
-                temperature=0,
-                groq_api_key=GROQ_API_KEY,
-                model_name="llama-3.1-8b-instant",
-                max_tokens=4096,
-                timeout=60
-                ).bind_tools(self.tools, tool_choice="auto")  
-            print("✅ LLM initialized without FORCED tool usage.")
-
-        except Exception as e:
-            print(f"❌ Error initializing HuggingFace: {e}")
-            raise
-#        print("Initializing LLM Endpoint...")
-#        llm = HuggingFaceEndpoint(
-#            repo_id="Qwen/Qwen2.5-Coder-32B-Instruct", # Changed model
-#            huggingfacehub_api_token=HUGGINGFACEHUB_API_TOKEN,
-#            max_new_tokens=2048, # Increased token limit for potentially longer reasoning/tool use
-#            temperature=0.01, # Keep temperature low for factual tasks
-#        )
-#        chat_llm = ChatHuggingFace(llm=llm)
-#        print("✅ LLM Endpoint initialized.")
+#        print("Initializing Groq LLM...")   
+#        try:
+#            self.llm_with_tools = ChatGroq(
+#                temperature=0,
+#                groq_api_key=GROQ_API_KEY,
+#                model_name="llama-3.1-8b-instant",
+#                max_tokens=4096,
+#                timeout=60
+#                ).bind_tools(self.tools, tool_choice="auto")  
+#            print("✅ LLM initialized without FORCED tool usage.")
 #
-#        # 6. Bind tools to the LLM
-#        # Ensure the LLM knows how to format calls for the tools
-#        self.llm_with_tools = chat_llm.bind_tools(self.tools)
-#        print("✅ Tools bound to LLM.")
+#        except Exception as e:
+#            print(f"❌ Error initializing HuggingFace: {e}")
+#            raise
+#        print("Initializing LLM Endpoint...")
+        print("Initializing HuggingFace LLM...")
+        
+        llm = HuggingFaceEndpoint(
+            repo_id="meta-llama/Llama-3.1-70B-Instruct",  # Free on HF Inference API
+            huggingfacehub_api_token=HUGGINGFACEHUB_API_TOKEN,
+            max_new_tokens=4096,
+            temperature=0.01,
+        )
+        chat_llm = ChatHuggingFace(llm=llm)
+        print("✅ HuggingFace LLM Endpoint initialized.")
+        
+        # Bind tools to the LLM
+        self.llm_with_tools = chat_llm.bind_tools(self.tools, tool_choice="auto")
+        print("✅ Tools bound to LLM.")
 #        print("Initializing Google Gemini LLM...")
 #        try:
 #            self.llm_with_tools = ChatGoogleGenerativeAI(