llm-testbench

mirror of https://github.com/lukaszraczylo/llm-testbench.git synced 2026-10-12 00:45:19 +00:00

Files

T

lukaszraczylo 6ad9dfbe42 fix(runner): record wall-clock latency on error rows

An error row with zero latency reads as an instant rejection when it
can be three exhausted request timeouts. A 651-second thinking-model
generation looked instant in the 2026-08-30 run and misdirected the
diagnosis toward gateway rejection. The error path now records the
wall clock across all attempts.

2026-08-30 16:19:21 +01:00

config

feat(config): raise max_tokens_default to 40000

2026-08-30 14:19:22 +01:00

eval

fix(eval): remove four negation-detector false-positive classes

2026-08-30 14:18:53 +01:00

llm

fix(llm): retry 429 responses and 2xx bodies carrying an error object

2026-08-30 14:19:35 +01:00

report

Update