Excited to kick off a 3-month research visit at Rycolab (ETH Zurich)! 🇨🇭

My research focuses on RL, alignment, multilingual LMs, reasoning, and RAG. If you're exploring any of these areas, feel free to reach out or say hi!

#NLP #RL #AIAlignment #Multilinguality

Our lab's papers at #EMNLP2025, spanning themes of #Interpretability, #Transparency, #Explainability, #Multilinguality, and #InformationRetrieval.

I also wrote a #SPARQL quer to see the linguistic composition of the periodical press until 1930 at all locations with titles published in languages of the Eastern Mediterranean: #Arabic, #Ottoman, #Armenian, #Coptic, #Greek, #Farsi, #Ladino, #Azerbaijani […]

How do you know a text about China published in an Arabic journal in Cairo in 1907 is a translation from a Western language in Latin script? The Arabic transcriptions of Chinese names do not make sense phonetically. Take for example "جريدة سهان باو التي تصدر في مدينة شنغاي وهي التي أسست سنة […]

Adding a bunch of tags for discoverability: #machinetranslation #flores #seed #languages #multilinguality #ai #nlp #mt

Second day of "Disciplinary Transformations" by @chnunn.bsky.social starts with the "Dark Sides of DH." First up was Alíz Horváth showing us the multilingual transformations of DH over the last few years. We are more multilingual than we think! #multilingualism #multilinguality

📝 Haastatteluaineistoa kertyi 25,5 tuntia tai 135 878 sanaa kirjoitettuna.

👩🏼‍💻 Käytin temaattista analyysia aineiston koodaamiseen, analysointiin, järjestämiseen ja ryhmittelyyn.

#tiedetaivas
#yliopistojuttuja
#kaksikielisyys
#monikielisyys
#bilingual
#kielet
#kieli
#multilinguality
#vanhemmuus

CroissantLLM: A Truly Bilingual French-English Language Model

Manuel Faysse, Patrick Fernandes, Nuno M Guerreiro et al.

Action editor: Tao Qin

https://openreview.net/forum?id=uA19Xo1o31

#bilingual #multilinguality #frenchbench

CoCo-CoLa: Evaluating Language Adherence in Multilingual LLMs Multilingual Large Language Models (LLMs) develop cross-lingual abilities despite being trained on limited parallel data. However, they often struggle to generate responses in the intended language, f...

Read our latest work CoCo-CoLa on measuring and improving language adherence in multilingual LLMs with efficient fine-tuning. 📖 arxiv.org/abs/2502.12476
Thanks to my co-authors @alirezaziabari.bsky.social and @mdehghani.bsky.social
for their collaboration!
#LLMs #NLP #Multilinguality