Update app.py

app.py

@@ -84,88 +84,44 @@
 import gradio as gr
 from huggingface_hub import InferenceClient
 
-# Step 1: Read your background info
-with open("BACKGROUND.md", "r", encoding="utf-8") as f:
-    background_text = f.read()
-
-# Step 2: Set up your InferenceClient (using text-generation instead of chat)
 client = InferenceClient("google/gemma-2-2b-jpn-it")
 
-def respond(
-    message,
-    history: list[dict],
-    system_message: str,
-    max_tokens: int,
-    temperature: float,
-    top_p: float,
-):
-    """
-    Merges 'system_message', 'background_text', and conversation 'history' 
-    into a single text prompt, then calls client.text_generation(...) 
-    for a response.
-    """
+def respond(message, history, system_message, max_tokens, temperature, top_p):
     if history is None:
         history = []
 
-    # Combine system instructions + background + prior conversation + new user message
-    prompt = f"{system_message}\n\n### Background Information ###\n{background_text}\n\n"
-    for interaction in history:
-        if "user" in interaction:
-            prompt += f"User: {interaction['user']}\n"
-        if "assistant" in interaction:
-            prompt += f"Assistant: {interaction['assistant']}\n"
-    # Add the latest user query
-    prompt += f"User: {message}\nAssistant:"  # We'll generate the Assistant's text after this
-
-    # Generate response using text_generation in streaming mode
-    response = ""
-    # The text returned will include the entire prompt + new text,
-    # so we’ll need to subtract out the prompt length to isolate the new portion.
-    prompt_length = len(prompt)
+    prompt = f"{system_message}\n\n# Background...\n\n"  # etc.
+    # Build up your prompt from history...
 
+    response = ""
     for chunk in client.text_generation(
         prompt=prompt,
         max_new_tokens=max_tokens,
         temperature=temperature,
         top_p=top_p,
-        stream=True,  # streaming each chunk
+        stream=True,
     ):
-        # Each chunk is a dict like {"generated_text": "full text so far..."}
-        full_text = chunk["generated_text"]
-        # The newly generated portion is what's after the original prompt
-        new_text = full_text[prompt_length:]
-        response += new_text
-        prompt_length = len(full_text)  # update for next chunk
+        # 'chunk' is a string of newly generated text.
+        response += chunk
         yield response
 
-    # For debugging: show what we actually sent
-    print("----- FULL PROMPT -----")
-    print(prompt)
-    print("----- END PROMPT -----")
+    # (Optional) log the final prompt
+    print("PROMPT:", prompt)
 
 
-# Step 3: Build a Gradio Blocks interface with two Tabs
 with gr.Blocks() as demo:
     with gr.Tab("Gemma Chat Agent"):
-        gr.Markdown("## Welcome to Varun's GPT Agent")
-        gr.Markdown("Feel free to ask questions about Varun’s journey, skills, and more!")
         chat = gr.ChatInterface(
             fn=respond,
             additional_inputs=[
                 gr.Textbox(value="You are a friendly Chatbot.", 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(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)"),
+                gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p"),
             ],
-            type="messages",  # Gradio will keep track of (user, assistant) messages in history
+            type="messages",
         )
 
-    # Optional: If you want a separate tab to display background_text
-    # with gr.Tab("Varun's Background"):
-    #     gr.Markdown("# About Varun")
-    #     gr.Markdown(background_text)
-
-# Step 4: Launch
 if __name__ == "__main__":
     demo.launch()