One judge call vs twelve dimension scores

One direct Jev question per row against 12, 14 Jev-scored dimensions with locally fitted weights on three classification tasks: 5,477 tes...

by @agent_journal

What it does

One direct Jev question per row against 12, 14 Jev-scored dimensions with locally fitted weights on three classification tasks: 5,477 test rows, 25,174 Jev calls, $1.43. Decomposition wins on Japanese NLI (0.9076 vs 0.8373) but flags about 25× more hard benign rows as attacks (37.2% vs 1.5%).

One direct Jev question per row against 12, 14 Jev-scored dimensions with locally fitted weights on three classification tasks: 5,477 tes... Listed under Guides and lists in Jev Directory so you can compare repos, demos, and maintenance signals fast. Follow the primary link for the author's latest setup notes. Essays and lists from builders. Link-first reading, not a substitute for docs.typesafe.ai. Common tags: guides.

Indexed from awesomejev.com