[Bugfix][Core] Preserve speculative metrics when coalescing outputs (#55708)

Signed-off-by: git-jxj <65210887+git-jxj@users.noreply.github.com>
Co-authored-by: OpenAI Codex <codex@openai.com>
This commit is contained in:
jxj
2026-10-07 08:35:14 +08:00
committed by GitHub
co-authored by OpenAI Codex
parent d77cf81156
commit 6c2925fe5f
2 changed files with 53 additions and 0 deletions
+4
View File
@@ -212,6 +212,10 @@ class RequestOutput:
completion.sampling_mask.token_ids.extend(
next_completion.sampling_mask.token_ids
)
if next_completion.spec_decode_metrics is not None:
completion.spec_decode_metrics = (
next_completion.spec_decode_metrics
)
completion.cumulative_logprob = (
next_completion.cumulative_logprob
)