vllm
67da5720 - [PERF] Speed up Qwen2.5-VL model by speed up rotary position embedding (#17973)

Commit

213 days ago

[PERF] Speed up Qwen2.5-VL model by speed up rotary position embedding (#17973) Signed-off-by: Vadim Gimpelson <vadim.gimpelson@centml.ai>

References

Author

vadiklyutiy

Parents