You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@madlib.apache.org by "Frank McQuillan (Jira)" <ji...@apache.org> on 2020/05/20 18:13:00 UTC
[jira] [Created] (MADLIB-1427) DL - cache data to speed training
Frank McQuillan created MADLIB-1427:
---------------------------------------
Summary: DL - cache data to speed training
Key: MADLIB-1427
URL: https://issues.apache.org/jira/browse/MADLIB-1427
Project: Apache MADlib
Issue Type: Improvement
Components: Deep Learning
Reporter: Frank McQuillan
Fix For: v1.18.0
Instead of calling Keras fit() for each buffer (row) in the input table, cache multiple buffers together and make a single call to Keras fit().
Also add a BOOLEAN to fit() in case user wants to turn off caching if the data does not fit in memory on the segment host.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)