Update app.py

app.py

@@ -1,12 +1,8 @@
 import gradio as gr
 from huggingface_hub import InferenceClient
 
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
 client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
 
-
 def respond(
     message,
     history: list[tuple[str, str]],
@@ -27,25 +23,27 @@ def respond(
 
     response = ""
 
-    for message in client.chat_completion(
+    # Get the model's response
+    for response_chunk in client.chat_completion(
         messages,
         max_tokens=max_tokens,
         stream=True,
         temperature=temperature,
         top_p=top_p,
     ):
-        token = message.choices[0].delta.content
-
+        token = response_chunk.choices[0].delta.content
         response += token
-        yield response
 
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
+    # Post-process the response to ensure relevance
+    if "Sorry, I can only answer questions related to the Constitution of India." in response:
+        response = "Sorry, I can only answer questions related to the Constitution of India."
+
+    yield response
+
 demo = gr.ChatInterface(
     respond,
     additional_inputs=[
-        gr.Textbox(value="You are a knowledgeable assistant specializing in the Constitution of India. Answer only questions related to the Constitution.", label="System message"),
+        gr.Textbox(value="You are a knowledgeable assistant specializing in the Constitution of India. Answer only questions related to the Constitution. If you find that a question is not relevant, inform the user accordingly.", label="System message"),
         gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
         gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
         gr.Slider(
@@ -58,6 +56,5 @@ demo = gr.ChatInterface(
     ],
 )
 
-
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()