Update app.py

app.py

@@ -9,7 +9,7 @@ from transformers import pipeline, AutoTokenizer, AutoModelForCausalLM
 import uvicorn
 import asyncio
 
-# ✅ Use float16 to reduce memory usage
+# ✅ Reduce memory usage by setting float16 precision
 torch.set_default_dtype(torch.float16)
 
 # ✅ Hugging Face API Token
@@ -17,9 +17,12 @@ HF_TOKEN = os.getenv("HF_TOKEN")
 MODEL_NAME = "hpyapali/tinyllama-workout"
 event_store = {}  # Store AI responses for polling fallback
 
+# ✅ Webhook URL (Your Vapor Webhook Server)
+WEBHOOK_URL = "https://694a-50-35-76-93.ngrok-free.app/fineTuneModel"
+
 app = FastAPI()
 
-# ✅ Lazy Load AI Model (to prevent Space timeout)
+# ✅ Lazy Load AI Model (prevents timeout on Hugging Face)
 pipe = None
 
 def get_pipeline():
@@ -85,11 +88,9 @@ async def process_workout_request(request: Request):
         event_store[event_id] = response_text  
 
         # ✅ Send AI response to Vapor Webhook
-        webhook_url = "https://694a-50-35-76-93.ngrok-free.app/fineTuneModel"
-
         async with httpx.AsyncClient() as client:
             try:
-                webhook_response = await client.post(webhook_url, json={"event_id": event_id, "data": [response_text]})
+                webhook_response = await client.post(WEBHOOK_URL, json={"event_id": event_id, "data": [response_text]})
                 webhook_response.raise_for_status()
                 print(f"✅ Webhook sent successfully: {webhook_response.json()}")
             except Exception as e: