Neural Architecture Search for CTR Prediction
Data Skeptic
00:00
Is There a Data Set You Can Compare Against?
All our models were meeting what we considered to be state of the art performance. There was noh fundamentally wrong with our architecture spaces, which was a good thing to see. We found this large compression from cardinality search, so that was nice to see. It's hard to say if there's really ever going to be a ceiling on how well c t r prediction can perform. But i don't know if there's a limit, but i hope not.
Transcript
Play full episode
