[Frontend] [Core] perf: Automatically detect vLLM-tensorized model, update tensorizer to version 2.9.0 (#4208)

vllm/engine/arg_utils.py

@@ -167,8 +167,8 @@ class EngineArgs:
             '* "dummy" will initialize the weights with random values, '
             'which is mainly for profiling.\n'
             '* "tensorizer" will load the weights using tensorizer from '
-            'CoreWeave which assumes tensorizer_uri is set to the location of '
-            'the serialized weights.')
+            'CoreWeave. See the Tensorize vLLM Model script in the Examples'
+            'section for more information.\n')
         parser.add_argument(
             '--dtype',
             type=str,