3 ms·
>Outputs are evaluated with a scoring rubric by Claude Sonnet 3.7 I feel it might be beneficial to evaluate by an ensemble of a bunch of models, picking the SO
by anshumankmr 2y ago
>Outputs are evaluated with a scoring rubric by Claude Sonnet 3.7
I feel it might be beneficial to evaluate by an ensemble of a bunch of models, picking the SOTA models cause of the subjectivity of the task at hand.