fix passing vllm model config's max_len to _generate_token_mode#3

wangshangsam merged 1 commit intoCentML:mlperf-inf-mm-q3vl-v6.0CentML/dynamo:mlperf-inf-mm-q3vl-v6.0from

soodoshll:qidongs/fix-max-tokenssoodoshll/dynamo:qidongs/fix-max-tokensCopy head branch name to clipboard