Event banner
Event details
With advances in post-editing learning, would it be possible to train a smaller model on an organization's post-editing changes rather than re-training the translation engine on a large sample of parallel documents?
- diegobartolome
Microsoft
You can use https://docs.microsoft.com/en-us/azure/cognitive-services/translator/custom-translator/overview to train with the source sentences and the post-edited sentences.- moelghazThough no ETA at this time, we have Custom Translator features to reduce the 10K parallel sentences minimum requirements and to enable incremental retraining.
