[
https://issues.apache.org/jira/browse/OPENNLP-1639?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Martin Wiesner updated OPENNLP-1639:
------------------------------------
Summary: Add basic ud-train.conf file and reference it in ud-train.sh
(was: Add basic ud-training.conf file and reference it in ud-train.sh)
> Add basic ud-train.conf file and reference it in ud-train.sh
> ------------------------------------------------------------
>
> Key: OPENNLP-1639
> URL: https://issues.apache.org/jira/browse/OPENNLP-1639
> Project: OpenNLP
> Issue Type: Sub-task
> Components: Models
> Affects Versions: 2.4.0
> Reporter: Martin Wiesner
> Priority: Major
> Fix For: 2.5.1
>
>
> We want to provide and reference a file with default training parameters.
> With the help of that file, we want to customize certain parameters, such as
> {+}Threads{+}, {+}Iterations{+}, etc., to speed up training runs on certain
> hardware.
> Currently, it seems, training of model files is single-threaded, resulting in
> slow Lemmatizer model training...
> Moreover, only 100 iterations will be conducted, by default. In that case,
> some models could provide better performance if, say, 300 iterations could be
> configured.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
