Optuna Hyperband Algorithm Not Following Expected Model Training Scheme #5383
Unanswered
d-sutariya
asked this question in
Q&A
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Expected behavior
I have observed an issue while using the Hyperband algorithm in Optuna. According to the Hyperband algorithm, when min_resources = 5, max_resources = 20, and reduction_factor = 2, the search should start with an initial space of 4 models for bracket 1, with each model receiving 5 epochs in the first round. Subsequently, the number of models is reduced by a factor of 2 in each round and search space should also reduced by factor of 2 for next brackets i.e bracket 2 will have initial search space of 2 models, and the number of epochs for the remaining models is doubled in each subsequent round. so total models should be 11 is expected .
link of the article:- https://arxiv.org/pdf/1603.06560.pdf
Environment
Error messages, stack traces, or logs
Steps to reproduce
#5382
Beta Was this translation helpful? Give feedback.
All reactions