Update app.py

app.py

@@ -1,62 +1,71 @@
 import streamlit as st
-from langchain.chains import ConversationChain
-from langchain.memory import ConversationBufferMemory
-from langchain.schema import HumanMessage, AIMessage
-from langchain.chat_models.base import BaseChatModel
 from llama_cpp import Llama
 
-# Initialize the Llama model
-llm = Llama.from_pretrained(
-    repo_id="Divyansh12/check",
-    filename="unsloth.F16.gguf",  
-    verbose=True,
-    n_ctx=32768,
-    n_threads=2,
-    chat_format="chatml"
-)
-
-# Define the LangChain model for Llama
-class LlamaChatModel(BaseChatModel):
-    def _llm_type(self) -> str:
-        return "llama"
-
-    def _generate(self, messages, stop=None):
-        # Prepare prompt from conversation history
-        prompt = "\n".join(
-            f"user: {msg.content}" if isinstance(msg, HumanMessage) else f"assistant: {msg.content}"
-            for msg in messages
-        )
-
-        # Generate response from Llama
-        response = llm.chat(prompt)
-        return [AIMessage(content=response)]
-
-# Initialize memory and chat model
-memory = ConversationBufferMemory(return_messages=True)
-llama_chat_model = LlamaChatModel()
-
-# Create the conversation chain
-conversation = ConversationChain(memory=memory, llm=llama_chat_model)
+# Load the model once per session
+if 'llm' not in st.session_state:
+    st.session_state.llm = Llama.from_pretrained(
+        repo_id="Divyansh12/check",
+        filename="unsloth.F16.gguf",  
+        verbose=True,
+        n_ctx=32768,
+        n_threads=2,
+        chat_format="chatml"
+    )
+
+# Define the function to get responses from the model
+def respond(message, history):
+    messages = []
+
+    for user_message, assistant_message in history:
+        if user_message:
+            messages.append({"role": "user", "content": user_message})
+        if assistant_message:
+            messages.append({"role": "assistant", "content": assistant_message})
+
+    messages.append({"role": "user", "content": message})
+
+    response = ""
+    # Stream the response from the model
+    response_stream = st.session_state.llm.create_chat_completion(
+        messages=messages,
+        stream=True,
+        max_tokens=512,  # Use a default value for simplicity
+        temperature=0.7,  # Use a default value for simplicity
+        top_p=0.95  # Use a default value for simplicity
+    )
+
+    # Collect the response chunks
+    for chunk in response_stream:
+        if len(chunk['choices'][0]["delta"]) != 0 and "content" in chunk['choices'][0]["delta"]:
+            response += chunk['choices'][0]["delta"]["content"]
+
+    return response  # Return the full response
 
 # Streamlit UI
-st.title("Chatbot with LangChain and Llama")
+st.title("Simple Chatbot")
 st.write("### Interact with the chatbot!")
 
 # User input field
 user_message = st.text_area("Your Message:", "")
 
+# Chat history
+if 'history' not in st.session_state:
+    st.session_state.history = []
+
 # Button to send the message
-if st.button("Send") and user_message:
-    # Get the response from the LangChain conversation
-    response = conversation.predict(input=user_message)
+if st.button("Send"):
+    if user_message:  # Check if user has entered a message
+        # Get the response from the model
+        response = respond(user_message, st.session_state.history)
+
+        # Add user message and model response to history
+        st.session_state.history.append((user_message, response))
 
-    # Update chat history in session state
-    st.session_state.history = memory.chat_memory.messages
+        # Clear the input field after sending
+        user_message = ""  # Reset user_message to clear input
 
 # Display the chat history
 st.write("### Chat History")
-for msg in st.session_state.get('history', []):
-    if isinstance(msg, HumanMessage):
-        st.write(f"**User:** {msg.content}")
-    elif isinstance(msg, AIMessage):
-        st.write(f"**Assistant:** {msg.content}")
+for user_msg, assistant_msg in st.session_state.history:
+    st.write(f"**User:** {user_msg}")
+    st.write(f"**Assistant:** {assistant_msg}")