mxfp8 inference does not work with vllm bench throughput #3549
Description
stack trace: https://gist.github.com/vkuzo/b15ec478ee0a04d274ddb46acfa6d209
repro: #3473
{{ message }}
stack trace: https://gist.github.com/vkuzo/b15ec478ee0a04d274ddb46acfa6d209
repro: #3473