v0.8 release, community growth 📈, and more!
Thanks for being part of the rasa NLU mailing list. Make sure you tell your friends about rasa!
Updates in version 0.8
Dn’ot yu hate it whn users type badlyy? Our new character n-gram models can make your intent recognition more robust to typos. Add it to your pipeline and train with some noisy examples.
Not much training data for your entities? CRFs (conditional random fields) are here! You can activate the ner_crf component in your pipeline instead of the spaCy or MITIE ner modules.
Duckling! Lots of users wanted the ability to turn “next Thursday at 8am” into a proper DateTime object. Since duckling already does that and more out of the box, we’ve added an ner_duckling component 🐦. Kudos to the wit.ai team for open-sourcing this awesome tool! 🎊
Community
We are so stoked on the amount of activity on gitter and github, especially seeing some more experienced members helping out by answering questions from new users. You are amazing! We're still sending out limited edition 80s-style rasa NLU t-shirts, so let us know what you've been up to :)
We’re seeing hundreds of developers clone rasa NLU for the first time every week - that’s a lot of newbies to support! Thanks to everyone for being so welcoming.
Neil Stoker built a super cool project called LockeBot with rasa NLU (love the reference, Neil!) 👑
Detailed how-to guide on setting up rasa NLU on AWS by Ravindra Prasad.
Sneak preview of Paul Aschmann's rasa GUI project.
Ask
We are putting time into writing some more tutorials and expanding the documentation. What topics would you most like to see? (click to vote)
Collecting / cleaning training data
Deep dive on entity extraction
Running in production
If you have a better idea - please email us! hi@rasa.ai
Keep building!
- your rasa team.
|
|
|
|