Update handler.py

handler.py

@@ -5,7 +5,7 @@ import io
 class EndpointHandler:
     def __init__(self, model_dir=None):
         # Set model size, assuming installation has been done with appropriate model files and setup
-        model_size = "large-v2" if model_dir is None else model_dir
+        model_size = "medium" if model_dir is None else model_dir
         # Change to 'cuda' to use the GPU, and set compute_type for faster computation
         self.model = WhisperModel(model_size, device="cuda", compute_type="float16")