Neural Architecture Optimization

Performance Proxy

A low-cost metric or model used to estimate the final performance of an architecture, avoiding a full and lengthy training phase during the search stage.

← Quay lại