Since Transformer LMs were invented, we’ve wanted them to be able to read longer inputs during inference than they saw during training. Our Attention with Linear Biases enables this, in very few lines of code, without requiring extra params or runtime ofir.io/train_short_te… 🧵⬇

meetNLP's 6th edition is here.

29th June. 18:30 CEST. Online.

NLP and beyond is this edition's topic. Tune in for talks on intellectual property & other key legal issues in NLP and what language models could learn from neuroscience. 👏

Sign up here: lnkd.in/dVx8-X5

Can't wait to try this out!🙌

It's Meetup day! Anyone interested in NLP - please join us tonight for the 5th edition of the Hamburg Natural Language Processing Meetup starting at 18:30 CEST.

Info on tonight's amazing speakers below. ⬇️⬇️

Link: meetu.ps/e/JPLw6/KkSxX/i

Experimenting with text generation has to be the most fun to be had trying out #NLP applications. Check out this nifty framework.

Fewer parameters, better? Almost!

Today we describe a #NaturalLanguageProcessing model that achieves near BERT-level performance on text classification tasks, while using orders of magnitude fewer model parameters. Learn all about it below: goo.gle/2FRyHqx

Are you enthusiastic about Natural Language Processing? Are you interested in hearing the latest findings from #NLP experts, discussing with the community about possible applications or just learning more about #NLP ? Then take part in our #Meetup :

meetup.com/de-DE/Hamburg-…