xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-03-19 22:07:42 +08:00
Code Issues Packages Projects Releases Wiki Activity

updated

Browse Source 
Signed-off-by: Robert Shaw <robshaw@redhat.com>
This commit is contained in:
Robert Shaw 2025-07-20 17:17:15 +00:00
parent f477b50493
commit 5ea4fa206d
1 changed files with 5 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
vllm/v1/metrics/loggers.py
View File

@ -168,7 +168,8 @@ class PrometheusStatLogger(StatLoggerBase):
model_name = vllm_config.model_config.served_model_name
max_model_len = vllm_config.model_config.max_model_len
if len(self.engine_indexes) > 1:
if (len(self.engine_indexes) > 1
and vllm_config.speculative_config is not None):
raise NotImplementedError("Prometheus metrics with Spec Decoding "
"with >1 EngineCore per AsyncLLM is not "
"supported yet.")
@ -511,9 +512,9 @@ class PrometheusStatLogger(StatLoggerBase):
self.counter_prefix_cache_hits[engine_idx].inc(
scheduler_stats.prefix_cache_stats.hits)
# if scheduler_stats.spec_decoding_stats is not None:
# self.spec_decoding_prom.observe(
# scheduler_stats.spec_decoding_stats)
if scheduler_stats.spec_decoding_stats is not None:
self.spec_decoding_prom.observe(
scheduler_stats.spec_decoding_stats)
if iteration_stats is None:
return