You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@beam.apache.org by "Younghee Kwon (JIRA)" <ji...@apache.org> on 2017/01/09 21:25:58 UTC
[jira] [Resolved] (BEAM-1233) Implement TFRecordIO (Reading/writing
Tensorflow Standard format)
[ https://issues.apache.org/jira/browse/BEAM-1233?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Younghee Kwon resolved BEAM-1233.
---------------------------------
Resolution: Fixed
Fix Version/s: Not applicable
The PR that adds TFRecordIO is pushed to python-sdk branch.
> Implement TFRecordIO (Reading/writing Tensorflow Standard format)
> -----------------------------------------------------------------
>
> Key: BEAM-1233
> URL: https://issues.apache.org/jira/browse/BEAM-1233
> Project: Beam
> Issue Type: New Feature
> Components: sdk-py
> Reporter: Younghee Kwon
> Assignee: Ahmet Altay
> Fix For: Not applicable
>
>
> Tensorflow is an open source Machine Learning project, which is getting lots of attention these days. Apache Beam can be used as a good preprocessing tool for this Machine Learning tool, however Tensorflow supports limited number of input file formats -- only csv and its own record format (so called TFRecord).
> On the other hand, Apache Beam doesn't support reading/writing in TFRecord format. This would be useful once it supports TFRecordIO natively.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)