What types of data can I use to train an engine?
KantanMT engines can be trained using both monolingual and bilingual TMX files.
What file formats are supported?
You can find a full list of supported file types here
Is there a file size limit to what can be uploaded?
Files of up to 50MB can be uploaded onto the KantanMT platform. If you have files that are larger than this you can compress it into a zip file and upload it.
Can I add terminology or glossaries to my engine?
Yes, adding glossaries/terminology can significantly improve your translation output. We would recommend that you upload these with the files you are translating in the client files tab on your dashboard.
Do I need to cleanse my data before uploading?
All data members upload onto the KantanMT platform goes through twelve stages of cleansing, however, we encourage members to carefully assess the quality of the data prior to uploading to ensure the highest quality engines.
Is my training data used by other members?
No. Neither KantanMT or any other members have access to training data uploaded by an individual account. The multi-tenancy architecture of the KantanMT.com platform heightens this security.
Can I save my training data to use again for building new engines?
Not exactly, but you can copy engines, so a suggestion would be to build one or two base engines which can be copied and built upon as needed.
For more information about KantanMT please see www.kantanmt.com