BPE-knockout just got outperformed by an algorithm that modifies BPE tokenisers in a feedback loop to make them absorb more and more constraints. It doesn't even need more data to do that. It uses the tokeniser itself as a dataset. 🧵

LAGoM will present several papers at #EACL2026 in Rabat next week! Our work at this year’s conference spans tokenisation, multilingual evaluation, and model design.

New EACL paper (with @mdlhx.bsky.social)! We tested if comparing perplexity of parallel data across languages is fair. Turns out: it depends. We show the choice of test set (even with consistent meaning) can flip conclusions about which language is easier to model. Paper: arxiv.org/abs/2601.10580

When is a language hard to model? Previous research has suggested that morphological complexity both does and does not play a role, but it does so by relating the performance of language models to corpus statistics of words or subword tokens in isolation.

Our group has two papers at #acl2025: * (Main) GRaMPa: Subword Regularisation by Skewing Uniform Segmentation Distributions with an Efficient Path-counting Markov Model by Thomas Bauwens, David Kaczér and @mdlhx.bsky.social, presented by Thomas. URL: aclanthology.org/2025.acl-lon...

The submission deadline for #CLIN35 has been extended by one week! New deadline: June 20th. 🔊 Spread the word! More info: clin35.ccl.kuleuven.be/call-for-abs...

Reminder, a few more days to apply!

📅 Don't forget! The deadline for submitting your abstract to the #CLIN conference in Leuven is coming: 13th of June! Submitting is easy: name, title of your work, 500-word abstract, done! #nlp #nlproc #compling #llm #ai #dutch clin35.ccl.kuleuven.be

We are hiring in #nlproc!!

I’m looking for a postdoc, to start ideally ASAP! The work would be in the EU-funded TrustLLM project, focusing on modularisation and language adaptation of LLMs, tokenization, and evaluation benchmarks for multilingual LLMs. The position would be full-time for 2 years with no teaching obligation.

@wpoelman.bsky.social and @mdlhx.bsky.social 's 🔥 hot takes on multilingual LLM evaluation, to appear @nodalida.bsky.social is up on arXiv: arxiv.org/abs/2412.08392

🚨 New Account Alert! This is the official account of the *MilaNLP group*. We had to recreate it because it was not indexed. If you were following us before, please follow us again. If not, now’s the perfect time to start!

There's too many starter packs. 👇 Here's a list, mostly for NLP, ML, and related areas.

Our work on quality estimation of non-English Wikipedia articles is on arXiv! 🎉 arxiv.org/abs/2411.055... By Kushal Tatariya, Artur Kulmizev, Wessel Poelman, @estherploeger.bsky.social, @bollmann.me, Johannes Bjerva, Jiaming Luo, Heather Lent and @mdlhx.bsky.social ✨

We're slowly growing, tell your friends!! #nlp

Thanks for the shoutout to our pixology work 🤩🤩

The NLP labs starter pack is here! go.bsky.app/LKGekew Let us know if you want to be added!

Our team at #EMNLP2024 this week!

Our EMNLP papers are out on the arXiv! 2 main conference papers, about pixology: arxiv.org/abs/2410.12011, about typological diversity: arxiv.org/abs/2402.04222 and one workshop paper about zero-shot pos tagging: arxiv.org/abs/2410.10576. Looking forward to connecting in Miami!

We have a new logo 🤩

New #NLProc paper on ArXiv by E. Ploeger, W. Poelman, M. de Lhoneux and J. Bjerva! More and more papers in NLP claim to evaluate on ‘typologically diverse’ languages. But what does this even mean? We systematically investigate such claims. arxiv.org/abs/2402.04222

New paper: 'Sociolinguistically Informed Interpretability: A Case Study on Hinglish Emotion Classification' t.co/MpOmz4uUsd Kushal will present it at SIGTYP @ EACL 2024! #NLProc

Hello bluesky! We have a website and a bsky account, therefore we exist! (We are also in the bad place but will try to always post on here first)