TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Ask HN: Testing Machine Learning Pipelines

1 pointsby ml_basicsover 2 years ago
I&#x27;m looking to learn about best practices for testing in the context of training large machine learning models.<p>For example, before launching a model to prod, I&#x27;ve commonly seen that teams have a suite of datasets&#x2F;evals, and a model has to be at least as good as the existing model in production before replacing it.<p>But what about the training pipelines? You can have unit tests for all the individual components, but the analogy of &quot;integration tests&quot; would involve just training a full model, which might be very expensive for large neural networks.<p>One option would be to have a special tiny dataset and tiny model that can be trained fast on CPU. But this would be an imperfect test of the real system. I&#x27;ve also never seen this in practice, so maybe it wouldn&#x27;t actually be useful.<p>Curious if any practitioners could share their wisdom.

no comments

no comments