Concept icon
Concept

When Repeated Validation Checks Distort Model Selection

If a team keeps using the dev set to judge many ideas during development, the model can gradually become tuned to that specific split. A strong dev-set score at the end, paired with a noticeably worse test-set score, is a warning sign. In that situation, the safest response is to create a new dev set.

0

1

Concept icon
Updated 2026-08-12

Contributors are:

Who are from:

Tags

Machine Learning

Deep Learning

Machine Learning Strategy

Supervised Learning

Dive into Deep Learning @ D2L

Data Science

Machine Learning Yearning @ DeepLearning.AI

Related