Ke Zhao

> Thank you very much for your answer > > The error is caused by the fact that in the evaluate_coll method, the index tensor (ti[m1], yi[m1], xi[m1]) and the...

OK，so the sharegpt4v only support for bench_throughput Do we have any suggestions for multi-modal test with bench_serving?

@ywang96 Hi, thanks for the quick reply! So use the vision-arena-bench dataset could solve this. The [README](https://github.com/vllm-project/vllm/blob/main/benchmarks/README.md) here is a little bit misleading. I wonder if we could use some...