Skip to content

GitLab

Unverified Commit 43f80884 authored by Qiaolin Yu's avatar Qiaolin Yu Committed by GitHub
Browse files

Fix accept rate in speculative decoding metrics (#11572)

parent 60b05032
Hide whitespace changes
Inline Side-by-side
Showing with 1 addition and 1 deletion
python/sglang/srt/managers/scheduler_metrics_mixin.py
View file @ 43f80884
...@@ -259,7 +259,7 @@ class SchedulerMetricsMixin: ...@@ -259,7 +259,7 @@ class SchedulerMetricsMixin:
) )
# Calculate acceptance rate: accepted tokens / total draft tokens # Calculate acceptance rate: accepted tokens / total draft tokens
total_draft_tokens = self.spec_num_total_forward_ct * ( total_draft_tokens = self.spec_num_total_forward_ct * (
self.server_args.speculative_num_steps or 1 (self.server_args.speculative_num_steps or 0) + 1
) )
spec_accept_rate = ( spec_accept_rate = (
self.spec_num_total_accepted_tokens / total_draft_tokens self.spec_num_total_accepted_tokens / total_draft_tokens
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment