Olga Zamaraeva

A graduate student (master's, PhD) or a postdoc at Universidade da Coruña? Join our new Facebook community to get to know each other and exchange information! facebook.com/groups/1332581…

that's right

The Green AI project at University of A Coruña (Spain) has a PhD position open! Your work will be on efficient NLP. Contact me if interested!

Our first two days at #EMNLP2024!

We have presented "Spanish Resource Grammar version 2023" by @OlgaZamaraeva, Lorena S. Allegue and myself, at @LrecColing 2024, Turin. In pic, Olga and I with Montserrat Marimon (the original SRG dev) and Lluís Padró (Freeling dev). Without them this work wouldn't be possible.

I wrote a blog post :) @olga.zamaraeva/writing-a-successful-proposal-4150e29ea210" target="_blank" rel="nofollow noopener">medium.com/@olga.zamaraev…

My paper with Paul Williams "A Confederacy of Models: a Comprehensive Evaluation of LLMs on Creative Writing", to appear in Findings of #EMNLP2023, is now in arXiv! arxiv.org/abs/2310.08433 - we compare 12 LLMs with human writers on a wacky creative writing challenge.

Our paper "4 and 7-bit Labeling for Projective and Non-Projective Dependency Trees" (with Diego Roca and @eusondavid) now in arXiv! We define simple encodings to represent dependency trees in a few bits per word, useful for parsing as sequence labeling. arxiv.org/abs/2310.14319

Super honored to have just hosted the 19th annual 🐬DELPH-IN 🐬 (grammar engineering) summit here at @UDC_gal ! Symbolic #NLProc

Started a series of notes on my MSCA project @olga.zamaraeva/marie-sk%C5%82odowska-curie-fellowship-3ebe58605b63" target="_blank" rel="nofollow noopener">medium.com/@olga.zamaraev…

Are there already trained morphophonological analyzers for Spanish which output EAGLES tags? (In addition to Freeling) #nlproc

My dudes, asking an LLM *any* question about itself (its training data, carbon footprint, abilities, etc.) is just contributing to feeding the hype that attributes self-awareness and intelligence to these models that they simply **do not have**.

And finally, we had to ask #ChatGPT to make up a new Zodiac and the results didn't disappoint. Do you feel more like a penguin, a potato, or a llama?

Thanks to @fralonp, I learned that I was wrong and the data is right. Sorry, La Coruña! So here's a corrected population density map of Spain. One more that includes las Islas Canarias is forthcoming. #rayshader adventures, an #rstats tale

@kilinguistics Probably not possible to answer with one statement, but to me it was mainly about soul-searching, like how one can reach the "bottom", psychologically speaking, but not necessarily die at that point. I used to love that book but this year, I started and then suddenly dropped it.

Are all human languages equally complex? 👇 Meta-analyses of 28 complexity metrics applied to texts written in overall 80 typologically diverse languages: - partially support the trade-off hypothesis - fully support the equi-complexity hypothesis degruyter.com/document/doi/1…

🔥SIGTYP 2023 will be co-located with #EACL2023 (early May 2023; Dubrovnik, Croatia) 💫Keynotes: Natalia Levshina and Ella Rabinovich 💫Shared Task on Cognate and Derivative Detection for Low-Resourced Languages More details to come! Stay tuned!

I think an AI just tried to convince me it failed the Turing Test. Or maybe I failed its Turing Test? Either way, the world is only going to get weirder. A play in four acts (plain text is me, green text is the GPT-3 AI system, using the new language model released yesterday)

Come work with us!

@emilymbender @zehavoc @KarnFort1 I bet it's simply what used to be "locale" in Microsoft world :) I'd bet on a 1:1 mapping of the term usage :)

@zehavoc @KarnFort1 "Only Portuguese is supported for subcultures: pt-PT and pt-BR. All other cultures are supported at the primary culture level." I'm ... not sure what they mean by "culture", but I very much doubt it's based on anything sensible.

Come work with us!