As an ML Acolyte, Even I can tell that it’s a well known phenomenon that ensemble methods are more resilient to overfitting issues. What’s fascinating is humanity’s penchant to forget the past and rediscover stuff.
Yeah this is an interesting way to compose the ensemble though. Rather than averaging over sub-networks they’re synthesizing the panel of responses:
A judge model reads every panel response and produces structured analysis: consensus points, contradictions, partial coverage, unique insights, blind spots
Whereas beam search injects variance by trialling candidate sentences and mixture of experts has competing sub-models here we’re reconciling different ideas.
Notice though that they’re always using Opus 4.8 as the judge so I think the claim (of surpassing frontier models) is over-inflated. I’d instead characterise it as outsourcing legwork to cheaper models.
I’m still optimistic about ensembles of smaller models though. You could imagine a specialist synthesis model or advantages from combining latent activations instead of text responses
As an ML Acolyte, Even I can tell that it’s a well known phenomenon that ensemble methods are more resilient to overfitting issues. What’s fascinating is humanity’s penchant to forget the past and rediscover stuff.
Yeah this is an interesting way to compose the ensemble though. Rather than averaging over sub-networks they’re synthesizing the panel of responses:
Whereas beam search injects variance by trialling candidate sentences and mixture of experts has competing sub-models here we’re reconciling different ideas.
Notice though that they’re always using Opus 4.8 as the judge so I think the claim (of surpassing frontier models) is over-inflated. I’d instead characterise it as outsourcing legwork to cheaper models.
I’m still optimistic about ensembles of smaller models though. You could imagine a specialist synthesis model or advantages from combining latent activations instead of text responses