Skip to content

Support overriding metric configs - #223

Merged
JosephMarinier merged 1 commit into
mainfrom
pr/joseph/support-overriding-metric-configs
Sep 24, 2026
Merged

JosephMarinier merged 1 commit into
mainfrom
pr/joseph/support-overriding-metric-configs

Conversation

@JosephMarinier

Copy link
Copy Markdown
Collaborator

Support overriding metric configs via CLI arguments or environment variables.

Examples:

EVA_METRIC_CONFIGS='{"faithfulness": {"judge_model": "gpt-5.6-terra", "judge_params": {"reasoning_effort": "none"}}}'
EVA_METRIC_CONFIGS__FAITHFULNESS__JUDGE_MODEL=gpt-5.6-terra
EVA_METRIC_CONFIGS__FAITHFULNESS__JUDGE_PARAMS__REASONING_EFFORT=none

@JosephMarinier JosephMarinier self-assigned this Sep 2, 2026
@JosephMarinier
JosephMarinier force-pushed the pr/joseph/support-overriding-metric-configs branch from a61b552 to 89f6cde Compare September 2, 2026 23:32
Comment thread src/eva/run_benchmark.py Outdated
via CLI arguments or environment variables.

@gabegma gabegma left a comment

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM! Should we document somewhere?

@JosephMarinier
JosephMarinier force-pushed the pr/joseph/support-overriding-metric-configs branch from 60b40f3 to 1903b0a Compare September 24, 2026 12:55
Comment thread src/eva/models/config.py
@JosephMarinier
JosephMarinier force-pushed the pr/joseph/support-overriding-metric-configs branch from 1903b0a to d8d5748 Compare September 24, 2026 16:11
@JosephMarinier
JosephMarinier added this pull request to the merge queue Sep 24, 2026
Merged via the queue into main with commit 9d11761 Sep 24, 2026
2 checks passed
@JosephMarinier
JosephMarinier deleted the pr/joseph/support-overriding-metric-configs branch September 24, 2026 16:16
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants