Update app.py

app.py

@@ -4,6 +4,7 @@ from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStream
 import os
 from threading import Thread
 import requests
+import time
 
 # Define model path for caching (Avoids reloading every app restart)
 MODEL_PATH = "/mnt/data/Phi-4-Hindi"
@@ -17,7 +18,7 @@ def load_model():
         try:
             if not os.path.exists(MODEL_PATH):
                 model = AutoModelForCausalLM.from_pretrained(
-                    MODEL_NAME, token=TOKEN, trust_remote_code=True, torch_dtype=torch.bfloat16
+                    MODEL_NAME, token=TOKEN, trust_remote_code=True, torch_dtype=torch.bfloat16, timeout=30
                 )
                 tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME, token=TOKEN)
                 model.save_pretrained(MODEL_PATH)
@@ -28,6 +29,9 @@ def load_model():
         except requests.exceptions.ConnectionError:
             st.error("⚠️ Connection error! Unable to download the model. Please check your internet connection and try again.")
             return None, None
+        except requests.exceptions.ReadTimeout:
+            st.error("⚠️ Read Timeout! The request took too long. Please try again later.")
+            return None, None
     
     return model, tokenizer