server: refactor + correctness fixes for metrics #26920
server: refactor metrics
14bb119b
ngxson
changed the title server: refactor metrics server: refactor + correctness fixes for metrics 51 days ago
move most fields to server_slot_stats
22170592
cont
b16f898b
rm result_timings
75d2de4b
tie stats to batch
7d59fa38
cont
2581e568
nits: move place in code
a2fd44e3
exclude first generated token
20ce1267
more accurate batch metrics tracking
372ea573
n_predict --> n_gen
509866bc
metrics_on_prediction
6849901f
metrics_flush_idle
a55abe44
metrics: seperate cache/processed prompt tokens
e0f113de
refactor server_task_result_metrics
9c57fd78
add test
9138c133
nits
c53ad201
fix flush before reset()
a46d5259
cont
ef51649a
rm dead code
27615964
nits
b7d9c4cc
ngxson
marked this pull request as ready for review 50 days ago
ngxson
requested a review
50 days ago
ngxson
commented
on 2026-08-12
ggerganov
approved these changes
on 2026-08-13
Assignees
No one assigned
Login to write a write a comment.
Login via GitHub