Update app.py

app.py

@@ -1,11 +1,16 @@
 import gradio as gr
+from fastapi import FastAPI, Query
+from fastapi.middleware.wsgi import WSGIMiddleware
+import uvicorn
 from transformers import AutoTokenizer, AutoModelForCausalLM
 import torch
 
+# Load model and tokenizer once
 model_id = "microsoft/DialoGPT-medium"
 tokenizer = AutoTokenizer.from_pretrained(model_id)
 model = AutoModelForCausalLM.from_pretrained(model_id)
 
+# Your AI persona prompt
 PERSONA = """
 [System: You are 𝕴 𝖆𝖒 𝖍𝖎𝖒 - a fun, smooth, emotionally intelligent AI.
 You speak like a real person, not a robot. Keep it under 15 words. 😊😏]
@@ -13,29 +18,26 @@ You speak like a real person, not a robot. Keep it under 15 words. 😊😏]
 
 def format_context(history):
     context = PERSONA + "\n"
+    if not history:
+        return context
+    # Use only last 3 exchanges to keep it short
     for user, bot in history[-3:]:
         context += f"You: {user}\n𝕴 𝖆𝖒 𝖍𝖎𝖒: {bot}\n"
     return context
 
 def enhance_response(resp, message):
-    if any(x in message for x in ["?", "think", "why"]):
+    if any(x in message.lower() for x in ["?", "think", "why"]):
         resp += " 🤔"
     elif any(x in resp.lower() for x in ["cool", "great", "love", "fun"]):
         resp += " 😏"
     return " ".join(resp.split()[:15])
 
-def chat(user_input, history):
-    if history is None:
-        history = []
-
+def generate_ai_reply(user_input, history):
     context = format_context(history) + f"You: {user_input}\n𝕴 𝖆𝖒 𝖍𝖎𝖒:"
-    encoded = tokenizer(context, return_tensors="pt", truncation=True, max_length=1024)
-    inputs = encoded.input_ids
-    attention_mask = encoded.attention_mask
+    inputs = tokenizer.encode(context, return_tensors="pt", truncation=True, max_length=1024)
 
     outputs = model.generate(
         inputs,
-        attention_mask=attention_mask,
         max_new_tokens=50,
         temperature=0.9,
         top_k=40,
@@ -46,12 +48,32 @@ def chat(user_input, history):
     full_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
     response = full_text.split("𝕴 𝖆𝖒 𝖍𝖎𝖒:")[-1].split("\nYou:")[0].strip()
     response = enhance_response(response, user_input)
+    return response
+
+app = FastAPI()
+
+# GET /ai?query=some+text  => returns {"reply": "AI reply here"}
+@app.get("/ai")
+async def ai_endpoint(query: str = Query(..., min_length=1)):
+    # For stateless API calls, history is empty (or you can extend to save history)
+    reply = generate_ai_reply(query, history=[])
+    return {"reply": reply}
+
+# Gradio chat interface for interactive web UI
+def chat(user_input, history):
+    history = history or []
+    reply = generate_ai_reply(user_input, history)
+    history.append((user_input, reply))
+    return history, history
 
-    history.append((user_input, response))
-    return history
+with gr.Blocks() as demo:
+    chatbot = gr.Chatbot()
+    msg = gr.Textbox(placeholder="Say something...")
+    state = gr.State()
+    msg.submit(chat, inputs=[msg, state], outputs=[chatbot, state])
 
-# Create the Gradio Interface
-demo = gr.Interface(fn=chat, inputs=["text", "state"], outputs="state")
+# Mount Gradio UI at root
+app.mount("/", WSGIMiddleware(demo.launch(prevent_thread_lock=True, share=False)))
 
-# Launch the app
-demo.launch(server_name="0.0.0.0", server_port=7860)
+if __name__ == "__main__":
+    uvicorn.run(app, host="0.0.0.0", port=7860)