You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@madlib.apache.org by "Advitya Gemawat (Jira)" <ji...@apache.org> on 2020/08/18 19:22:00 UTC
[jira] [Commented] (MADLIB-1447) DL: Hyperband phase 3 - logic for
diagonal runs
[ https://issues.apache.org/jira/browse/MADLIB-1447?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17180024#comment-17180024 ]
Advitya Gemawat commented on MADLIB-1447:
-----------------------------------------
[~fmcquillan] This was changed to story 2 (and vice versa) on the tracker story and it's already done.
> DL: Hyperband phase 3 - logic for diagonal runs
> ------------------------------------------------
>
> Key: MADLIB-1447
> URL: https://issues.apache.org/jira/browse/MADLIB-1447
> Project: Apache MADlib
> Issue Type: New Feature
> Reporter: Frank McQuillan
> Assignee: Advitya Gemawat
> Priority: Major
> Fix For: v1.18.0
>
>
> Python code to do some version of this is in https://github.com/apache/madlib-site/blob/asf-site/community-artifacts/Deep-learning/automl/hyperband-diag-cifar10-v1.ipynb
> **Story***
> Implement the diagonal logic without doing the actual training, i.e., don't call fit. Print out to console so we can see the logic is correct.
> **Acceptance**
> 1) For `R=81, eta=3` print out the s, i, n_i and r_i values to prove that the diagonal logic is working properly
> 2) Set skip_last =1 and check logic again
> 3) Try multiple other values and check that logic is correct.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)