Copy hybim 492 eval monitoring#4

Open

shuningc wants to merge 13 commits intomainfrom

copy-HYBIM-492-eval-monitoring

Commits on Feb 10, 2026

initial plan for eval monitoring feature

zhirafovod
authored and
shuningc
committed
TOR=: git rebase --continue

zhirafovod
authored and
shuningc
committed
update status

zhirafovod
authored and
shuningc
committed
batched deepeval evaluator

zhirafovod
authored and
shuningc
committed
restore original deepeval logic, make batched optional

zhirafovod
authored and
shuningc
committed
add test examples for eval

zhirafovod
authored and
shuningc
committed
fix(deepeval): Fix batched evaluator for local LLM providers

zhirafovod
authored and
shuningc
committed
feat(evals): Add LLMJudgeEvaluator with batched and non-batched modes

zhirafovod
authored and
shuningc
committed
Rename LLMJudgeEvaluator to NativeEvaluator, update env vars

zhirafovod
authored and
shuningc
committed
docs: add performance benchmark results for evaluator modes

zhirafovod
authored and
shuningc
committed
chore: bump versions and update changelogs

zhirafovod
authored and
shuningc
committed
docs: document decision to keep monitoring separate from Emitters

zhirafovod
authored and
shuningc
committed
fix(evals-deepeval): change default implementation back to deepeval for backward compatibility

zhirafovod
authored and
shuningc
committed