Update app.py

app.py

@@ -3,6 +3,7 @@ from huggingface_hub import InferenceClient
 import os
 import traceback
 
+# Styling
 fancy_css = """
 #main-container {
     background-color: #f0f0f0;
@@ -39,15 +40,16 @@ fancy_css = """
 }
 """
 
+# System prompt specialized for chord bot
 CHORD_SYSTEM_PROMPT = """You are a music theory expert specialized in chord identification.
 Given a list of notes (like "C E G" or "D F# A C"), identify the chord name.
 Always respond with the chord name and a short explanation of the intervals.
 """
 
-# Read token from Space secrets
+# Hugging Face API token (in Space secrets)
 HF_TOKEN = os.environ.get("HF_TOKEN")
 
-# Create inference client
+# Inference client
 client = InferenceClient(token=HF_TOKEN)
 
 def respond(message, history, system_message, max_tokens, temperature, top_p):
@@ -55,23 +57,25 @@ def respond(message, history, system_message, max_tokens, temperature, top_p):
         return "⚠️ No HF_TOKEN found. Please add it in your Space secrets."
 
     try:
+        # Build a simple prompt
         prompt = f"{system_message}\nUser: {message}\nAnswer:"
 
-        # Non-streaming inference call
-        output = client.text_generation(
-            prompt,
-            model="gpt2",  # ✅ small model, works with Inference API
-            max_new_tokens=max_tokens,
-            temperature=temperature,
-            top_p=top_p,
-            stream=False,   # 🔑 disable streaming
+        # Raw POST to Hugging Face Inference API
+        output = client.post_json(
+            "https://api-inference.huggingface.co/models/gpt2",  # ✅ change model here if needed
+            {
+                "inputs": prompt,
+                "parameters": {
+                    "max_new_tokens": max_tokens,
+                    "temperature": temperature,
+                    "top_p": top_p,
+                },
+            },
         )
 
-        # Extract text
-        if hasattr(output, "generated_text"):
-            response = output.generated_text
-        elif isinstance(output, dict) and "generated_text" in output:
-            response = output["generated_text"]
+        # Parse response
+        if isinstance(output, list) and len(output) > 0 and "generated_text" in output[0]:
+            response = output[0]["generated_text"]
         else:
             response = str(output)
 
@@ -82,6 +86,7 @@ def respond(message, history, system_message, max_tokens, temperature, top_p):
         return f"⚠️ Error: {str(e)}\n\nTraceback:\n{tb}"
 
 
+# Gradio ChatInterface
 chatbot = gr.ChatInterface(
     fn=respond,
     additional_inputs=[
@@ -93,8 +98,9 @@ chatbot = gr.ChatInterface(
     type="messages",
 )
 
+# Layout
 with gr.Blocks(css=fancy_css) as demo:
-    gr.Markdown("<h1 id='title'>🎶 Chord Bot (API-based, GPT-2) 🎶</h1>")
+    gr.Markdown("<h1 id='title'>🎶 Chord Bot (API-based) 🎶</h1>")
     chatbot.render()
 
 if __name__ == "__main__":