Update app.py

app.py

@@ -1,48 +1,53 @@
 import gradio as gr
 import os
 from openai import OpenAI
-import time
 
-# Initialize the OpenAI Client
+# Initialize the OpenAI Client with your API key and endpoint
+api_key = os.environ.get("RUNPOD_API_KEY")  # Ensure your API key is correctly loaded from environment variables
 client = OpenAI(
-    api_key=os.environ.get("RUNPOD_API_KEY"),
+    api_key=api_key,
     base_url="https://api.runpod.ai/v2/vllm-k0g4c60zor9xuu/openai/v1",
 )
 
-def runpod_chat(question, history=None):
-    if history is None:
-        history = []  # Ensure history starts as an empty list if none is provided
-    # Add the role description at the beginning of the session
+def predict(message, history=[]):
+    # Append the system role at the start if history is empty
     if not history:
-        history.append({"role": "system", "content": "You are a history assistant, that provides the best possible answers to any historical questions asked about American History. Be helpful and specific, providing any detailed nuance needed to have a full understanding of the question."})
-    history.append({"role": "user", "content": question})
+        history.append(("system", "You are a history assistant, that provides the best possible answers to any historical questions asked about American History. Be helpful and specific, providing any detailed nuance needed to have a full understanding of the question."))
+    
+    # Prepare messages in the format required by OpenAI
+    history_openai_format = []
+    for human, assistant in history:
+        history_openai_format.append({"role": "user", "content": human})
+        history_openai_format.append({"role": "assistant", "content": assistant})
+    history_openai_format.append({"role": "user", "content": message})
 
+    # Make the API call
     response_stream = client.chat.completions.create(
         model="ambrosfitz/llama-3-history",
-        messages=history,
+        messages=history_openai_format,
         temperature=0,
         max_tokens=150,
         stream=True,
     )
-    
-    # Stream the response and accumulate full response before displaying
-    full_response = "HistoryBot: "
-    for message in response_stream:
-        part = message.choices[0].delta.content if message.choices[0].delta.content is not None else ""
-        full_response += part
 
-    # Append the full response to history once complete
-    history.append({"role": "assistant", "content": full_response})
+    # Accumulate response chunks to form the full message
+    full_message = ""
+    for chunk in response_stream:
+        if chunk.choices[0].delta.content is not None:
+            full_message += chunk.choices[0].delta.content
+            yield full_message
 
-    return full_response, history  # Return full response and updated history to maintain state
+    # Update history with the latest exchange
+    history.append((message, full_message))
 
-# Set up the Gradio interface
+# Set up the Gradio chat interface
 iface = gr.Interface(
-    fn=runpod_chat,
-    inputs=[gr.Textbox(label="Enter your question:"), gr.State()],
-    outputs=[gr.Textbox(label="Responses"), gr.State()],
+    fn=predict,
+    inputs=[gr.Textbox(label="Type your question here..."), gr.State(label="History")],
+    outputs=[gr.Textbox(label="HistoryBot Responses"), gr.State()],
     title="HistoryBot Chat",
-    description="Interact with HistoryBot, a specialized assistant for American History. Ask any historical questions to get detailed and nuanced answers."
+    description="Interact with HistoryBot, a specialized assistant for American History. Ask any historical questions to get detailed and nuanced answers.",
+    allow_flagging="never"
 )
 
 iface.launch()