vllm/benchmarks/lib/endpoint_request_func.py · 1e4748c66a156af3ea7b72a9dbe3fb686c8750eb · OpenDAS / vllm_cscc · GitLab

Find file Blame History Permalink

[Bugfix] Fix `vllm bench serve` to count multimodal tokens in "total input tokens" (#38654) · 1e4748c6
Matthias Gehre authored Apr 14, 2026
```
Signed-off-by: Matthias Gehre <matthias.gehre@amd.com>
```
1e4748c6

endpoint_request_func.py 27 KB