Update app.py

app.py

@@ -2,35 +2,33 @@ import gradio as gr
 import torch
 import yfinance as yf
 import numpy as np
+from chronos import ChronosPipeline
+
+# Initialize model once for the app
+print("Loading Chronos model...")
+pipeline = ChronosPipeline.from_pretrained(
+    "amazon/chronos-t5-base",
+    device_map="auto",
+    torch_dtype=torch.bfloat16,
+)
 
 def predict_stock_numeric(ticker_symbol, timeframe):
     try:
-        # Map timeframe to yfinance period/interval
         tf_map = {
-            "5 mins": ("1d", "5m"),
-            "15 mins": ("1d", "15m"),
-            "1 hr": ("1mo", "1h"),
-            "1 day": ("1y", "1d"),
-            "1 week": ("2y", "1wk"),
-            "1 month": ("5y", "1mo"),
+            "5 mins": ("1d", "5m"), "15 mins": ("1d", "15m"),
+            "1 hr": ("1mo", "1h"), "1 day": ("1y", "1d"),
+            "1 week": ("2y", "1wk"), "1 month": ("5y", "1mo"),
             "1 year": ("max", "1y")
         }
-
         period, interval = tf_map.get(timeframe, ("2y", "1d"))
 
-        # 1. Fetch Data
         df = yf.download(ticker_symbol, period=period, interval=interval)
-        if df.empty:
-            return "Error: Ticker not found or no data for this timeframe."
+        if df.empty: return "Error: Ticker not found."
 
         prices_series = df['Close'].values.flatten()
-
-        # 2. Inference
         context_tensor = torch.tensor(prices_series)
-        pred_len = 1
-        forecast_result = pipeline.predict(context_tensor, pred_len)
+        forecast_result = pipeline.predict(context_tensor, 1)
 
-        # 3. Process Result
         median_pred = np.median(forecast_result[0].numpy(), axis=0)[0]
         current_price = prices_series[-1]
         change = median_pred - current_price
@@ -46,26 +44,20 @@ def predict_stock_numeric(ticker_symbol, timeframe):
     except Exception as e:
         return f"Error: {str(e)}"
 
-# Create GUI
 with gr.Blocks() as demo:
     gr.Markdown("# Strix Technologies Predictor")
     with gr.Row():
-        # Dropdown with allow_custom_value lets users pick or type
         ticker_input = gr.Dropdown(
             choices=["AAPL", "TSLA", "GOOGL", "MSFT", "AMZN", "NVDA", "BTC-USD", "ETH-USD"],
-            label="Select or Type Ticker",
-            value="AAPL",
-            allow_custom_value=True
+            label="Select or Type Ticker", value="AAPL", allow_custom_value=True
         )
         timeframe_input = gr.Dropdown(
             choices=["5 mins", "15 mins", "1 hr", "1 day", "1 week", "1 month", "1 year"],
-            label="Timeframe",
-            value="1 day"
+            label="Timeframe", value="1 day"
         )
     btn = gr.Button("Predict Next Close")
     output_text = gr.Textbox(label="Prediction Result", lines=6)
-
     btn.click(fn=predict_stock_numeric, inputs=[ticker_input, timeframe_input], outputs=output_text)
 
-# share=True creates a temporary public link (72h)
-demo.launch(debug=True, share=True)
+if __name__ == '__main__':
+    demo.launch()