Simpler approach to loading entrypoints in hubconf works properly

Author: rwightman

hubconf.py

@@ -1,10 +1,4 @@
 dependencies = ['torch']
-
 from timm.models import registry
 
-current_module = __import__(__name__)
-current_module.__dict__.update(registry._model_entrypoints)
-#for fn_name in registry.list_models():
-#    fn = registry.model_entrypoint(fn_name)
-#    setattr(current_module, fn_name, fn)
-
+globals().update(registry._model_entrypoints)