From d392ac836e17e87f5fec6594ef5ee5b56db72a91 Mon Sep 17 00:00:00 2001 From: Michele Campi <215741962+MicheleCampi@users.noreply.github.com> Date: Mon, 14 Sep 2026 23:15:28 +0200 Subject: [PATCH] [Doc][Metrics] Fix spec-decode PromQL examples to use the exposed names (#55016) Signed-off-by: Michele Campi --- vllm/v1/spec_decode/metrics.py | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/vllm/v1/spec_decode/metrics.py b/vllm/v1/spec_decode/metrics.py index a3ccfb29e7..826dfbd16a 100644 --- a/vllm/v1/spec_decode/metrics.py +++ b/vllm/v1/spec_decode/metrics.py @@ -187,12 +187,12 @@ class SpecDecodingProm: 1 + ( rate(vllm:spec_decode_num_accepted_tokens_total[$interval]) / - rate(vllm:spec_decode_num_drafts[$interval])) + rate(vllm:spec_decode_num_drafts_total[$interval])) A per-position acceptance rate vector can be computed using - vllm:spec_decode_num_accepted_tokens_per_pos[$interval] / - vllm:spec_decode_num_drafts[$interval] + vllm:spec_decode_num_accepted_tokens_per_pos_total[$interval] / + vllm:spec_decode_num_drafts_total[$interval] """ _counter_cls = prometheus_client.Counter