Update app.py

app.py

@@ -2,20 +2,14 @@ import gradio as gr
 import pandas as pd
 import joblib
 from transformers import pipeline
-import torch
 
-# Load models once at startup
+# Load all ML models
 product_models = joblib.load('models/inventory_forecaster.pkl')
+llm = pipeline("text2text-generation", model="google/flan-t5-base")
 
-# Use a smaller, faster model and specify device
-device = 0 if torch.cuda.is_available() else -1
-llm = pipeline("text2text-generation", 
-               model="google/flan-t5-small",  # smaller model
-               device=device)  # use GPU if available
-
-# Cache the function to avoid recomputing for same inputs
-@gr.cache_examples
+# Function to predict and generate restocking advice
 def inventory_advisor(product_id, current_inventory, last_day_sales):
+    # Select correct model
     if product_id not in product_models:
         return f"❌ Error: Product ID {product_id} not found in models."
 
@@ -24,13 +18,11 @@ def inventory_advisor(product_id, current_inventory, last_day_sales):
 
     prompt = (f"Current inventory is {current_inventory} units. "
               f"Predicted sales for next week is {int(future_sales)} units. "
-              f"Should restocking be done? Give a brief restocking advice in 2 sentences.")
+              f"Should restocking be done? Suggest a human-readable restocking advice.")
+
+    response = llm(prompt, max_length=100)[0]['generated_text']
 
-    response = llm(prompt, 
-                  max_length=50,  # reduced max length
-                  num_beams=2)    # faster beam search
-    
-    return f"🔮 Predicted Sales Next Week: {int(future_sales)} units\n\n🛒 Advice:\n{response[0]['generated_text']}"
+    return f"🔮 Predicted Sales Next Week: {int(future_sales)} units\n\n🛒 Advice:\n{response}"
 
 iface = gr.Interface(
     fn=inventory_advisor,
@@ -41,9 +33,7 @@ iface = gr.Interface(
     ],
     outputs="text",
     title="📦 Real-Time Inventory Management (Multi-Product)",
-    description="Enter product ID, current stock, and yesterday's sales. Get AI-based restocking advice!",
-    examples=[[1, 100, 50], [2, 200, 75]],  # Add example inputs
-    cache_examples=True
+    description="Enter product ID, current stock, and yesterday's sales. Get AI-based restocking advice!"
 )
 
 if __name__ == "__main__":