Does 4-bit quantization break a clinical AI — and can a bigger model fix it?

Models and Hub profiles the arms call

Arms

Judged scores

Scenario × arm  (click a cell)

Recent  (click a row)