Malte Højmark-Bertelsen

Trying NOT to create Terminator while trying to harvest the benefits of AI for the society.

[Malte Højmark-Bertelsen @malteH_B]

@fusaroli Absolutely well-deserved! Aarhus Uni and all those “inspiring students”are so freaking lucky to have you. Huge congratulations Professor Fusaroli! 🍾

[MrBeast @MrBeast]

I’m gonna give 10 random people that repost this and follow me $25,000 for fun (the $250,000 my X video made)

I’ll pick the winners in 72 hours

[Malte Højmark-Bertelsen @malteH_B]

@LeonDerczynski Sounds awesome! Let me know if you want some help of collaboration! 🙌

[Malte Højmark-Bertelsen @malteH_B]

@LeonDerczynski To me this kind of research, where a conglomerate of language models are trained, seems like an amazing opportunity to increase the number of language models in scarce domains. May I ask if you ever considered using DAGW in your explorations?

[Malte Højmark-Bertelsen @malteH_B]

@LeonDerczynski Really awesome work you made with this Leon!

Will definitely use the knowledge for any future model training. 🤗

[Malte Højmark-Bertelsen @malteH_B]

@LeonDerczynski @anmarasovic @pdasigi I definitely believe you did! I really loved your notion of understanding why annotations are made, and increasing the variation of individuals annotating. To me it seems that this could potentially mitigate some of the discriminatory bias in supervised NLP models?

[Ana Marasović @anmarasovic]

#nlphighlights 131: @pdasigi and I had a pleasure to talk with Nanna Inie and Leon Derczynski @LeonDerczynski about opportunities and barriers between HCI and NLP. A lot of lessons learned, thank you!

https://t.co/FbxTwt4gM8

[Malte Højmark-Bertelsen @malteH_B]

@KCEnevoldsen Awesome!

[Malte Højmark-Bertelsen @malteH_B]

@hjian42 @Adinseg @fusaroli @mcxfrank @viveksck Yes! This was my exact intuition, and I think that utilising contextualised embedding a from e.g. transformer-based language models, could reveal more intricate features in CDS.

[Malte Højmark-Bertelsen @malteH_B]

@KCEnevoldsen Maybe because of the fact that the government has prevented companies from going bankrupt? The rate of bankruptcy actually fell by more than 30% in 2020, probably due to the aiding packages (hjælpepakkerne)?

[Riccardo Fusaroli @fusaroli]

Cogsci students at aarhus impress again. @LudvigOlsen ‘s work is amazing, both his packages and his yet unpublished deep learning work

[Malte Højmark-Bertelsen @malteH_B]

@fishnets88 @fusaroli Amazing - cheers!

[Malte Højmark-Bertelsen @malteH_B]

@fishnets88 @fusaroli Ahh cool! Definitely going to check this out!

[Malte Højmark-Bertelsen @malteH_B]

@fishnets88 @fusaroli If you wish to load LaBSE, you can simply use:

from sentence_transformers import SentenceTransformer
model = SentenceTransformer('sentence-transformers/LaBSE')

I hope this helps (but you probably already knew this😁)!

[Malte Højmark-Bertelsen @malteH_B]

@fishnets88 @fusaroli Sadly not, but i do recommend you using the sentence-transformers library: https://t.co/8xU5YxAaIX.

[Malte Højmark-Bertelsen @malteH_B]

@fishnets88 @fusaroli Feel free to reach out if you have any questions or if I can help with anything! 🙌

[Anthony Hopkins @AnthonyHopkins]

With gratitude, I celebrate 45 years of sobriety.

[Leon Derczynski ⚒️☁️🏔️🌲 @LeonDerczynski]

Our paper on data for abusive language detection is here. Highlights:

- Best practices for making abusive language data
- Advice for machine learning practitioners
- Emerging research problems
- Comprehensive corralling of the state-of-the-art

#nlproc https://t.co/NHSgcQNltO

[Malte Højmark-Bertelsen @malteH_B]

Oh, the irony continues: https://t.co/CCPx1tY5aR is yet another website that does not support the unique Danish characters 'Æ', 'Ø' and 'Å'. Anyhow - Merry Christmas!
May your holidays be filled with 'Ableskiver', 'Ag', lots of 'Flode' and 'Gas'... 😉🎄