iiinova — where your digital self keeps evolving

Reposted by Leonie Weissweiler

Lukas Edman @lukasnlp.bsky.social · 03/10/2026

Great news! Our workshop what-llms-can-not-do.org is going to COLING 2027 in Macau! Hope to see you there!

what-llms-can-not-do.org

What LLMs Can(not) Do

A living survey of benchmarks that compare large language models with humans.

0185

Leonie Weissweiler @weissweiler.bsky.social · 01/10/2026

📢I'm hiring a PhD student to work on interpretability for Protein Language Models! 🗓️There are now two weeks left to apply, find out more on the LION Lab website: lionlabnlp.github.io/jobs/ai4pf/

lionlabnlp.github.io

PhD Student Position in Interpretability for Protein Language Models — LION Lab

Fully funded PhD student position at LION Lab, Leipzig University, on interpretability for protein language models.

186

Reposted by Leonie Weissweiler

ELLIS @ellis.eu · 25/09/2026

🏹 Job alert: PhD student position in Interpretability for Protein Language Models at Leipzig University 📍 Leipzig 🇩🇪 ⏰ 15 October 🔗 bit.ly/3V63Gpm

lionlabnlp.github.io

PhD Student Position in Interpretability for Protein Language Models — LION Lab

Fully funded PhD student position at LION Lab, Leipzig University, on interpretability for protein language models.

054

Reposted by Leonie Weissweiler

Lukas Edman @lukasnlp.bsky.social · 17/09/2026

Ever feel like it's too hard to keep track of what LLMs cannot do as well as humans? We're making your life easier over at: what-llms-can-not-do.github.io We're compiling a list of papers testing the abilities of LLMs against humans. Check it out! And you can help contribute too!

what-llms-can-not-do.github.io

What LLMs Can(not) Do

A living survey of benchmarks that compare large language models with humans.

24516

Reposted by Leonie Weissweiler

LION Lab @lionlab.bsky.social · 17/09/2026

🦁LION Lab is hiring! 🧑‍🔬One fully funded PhD student (TVL-13 100%) for 3 years 💉Topic: Interpretability for Protein Language Models 👥Advised by @weissweiler.bsky.social together with Clara Schoeder 🌍Leipzig, Germany 🔗Apply by Oct 15: lionlabnlp.github.io/jobs/ai4pf/ Please share! #NLProc #NLP

0109

Leonie Weissweiler @weissweiler.bsky.social · 15/09/2026

#COLING2027 has joined forces with #EACL2027 and #NAACL2027 for the call for tutorials! 📢Joint Call for Tutorial Proposals (EACL/NAACL/COLING) 2027 🗓️Submission deadline: Friday, 2 October 2026 📝Submit via softconf: softconf.com/p/eacl-tutor... 🔗Full call here: 2027.coling-iccl.org/calls/tutori...

2027.coling-iccl.org

Joint Call for Tutorial Proposals 2027

COLING 2027 Tutorials

073

Reposted by Leonie Weissweiler

LION Lab @lionlab.bsky.social · 31/08/2026

We hope to see the UniDivers again soon in other settings 😊 Fun fact: the old town of Bucharest has been called the Lipscani District since the 17th century, because most of the wares sold on the high street at the time originated from Leipzig!

041

Reposted by Leonie Weissweiler

LION Lab @lionlab.bsky.social · 31/08/2026

It was great seeing everyone again at the last Unidive General Meeting in Bucharest! Jacob Lee Suchardt presented a poster about his EMNLP 2025 paper on the Transcription of Unseen Languages with Speech-to-IPA Models, and we celebrated the past four years of UniDive and all the friends we made.

151

Reposted by Leonie Weissweiler

LION Lab @lionlab.bsky.social · 26/08/2026

Yesterday we had @anja.re from Technion visiting us to give a talk on Interpretability for Generative Information Retrieval. The talk was a cool example to show that interpretability can also be actionable. Thank you very much for joining us, Anja!

052

Reposted by Leonie Weissweiler

LION Lab @lionlab.bsky.social · 11/08/2026

🗣️Today we had the pleasure of hosting @bbunzeck.bsky.social for a talk on how LLMs are not usage-based (yet). We learned a lot about the challenges of training smaller language models on dialogue data and evaluating from a production-based perspective. Thanks for coming, Bastian!

051

Reposted by Leonie Weissweiler

LION Lab @lionlab.bsky.social · 21/07/2026

We had a great time yesterday listening to @davidrmortensen.bsky.social explain how The Reconstruction Will Not Be Supervised — because the Comparative Method of reconstructing the pronunciation of earlier languages can be automated using agents! Thanks for visiting, David!

051

Reposted by Leonie Weissweiler

LION Lab @lionlab.bsky.social · 15/07/2026

🗣️We were delighted yesterday to welcome Nina Böbel to our lab for a talk on Language as a Linguistic Network. Thanks for coming and patiently teaching us about construction networks, Nina!

061

Leonie Weissweiler @weissweiler.bsky.social · 29/06/2026

📋To help us plan, we'd appreciate expressions of interest if you're planning to eventually submit a proposal, by July 24. 🔗You can find more information on the call and how to submit your expression of interest here: leonieweissweiler.github.io/coling2027.pdf

leonieweissweiler.github.io

010

Leonie Weissweiler @weissweiler.bsky.social · 29/06/2026

📢Call for Expressions of Interest: COLING 2027 Tutorials Tutorial chairs @microth.bsky.social and I are inviting expressions of interest for tutorials to be held at #COLING2027 in Macau in all areas of CL, NLP and related disciplines, in the categories 🚀cutting-edge, 🌱introductory, and 🤝community.

163

Leonie Weissweiler @weissweiler.bsky.social · 27/06/2026

Congratulations again! 🎓🌳🥳

120

Leonie Weissweiler @weissweiler.bsky.social · 23/06/2026

Thanks, Adele! I might steal that as a motto 😊

010

Reposted by Leonie Weissweiler

Adele Goldberg @adelegoldberg.bsky.social · 23/06/2026

Congratulations Leonie! Hear her roar!

181

Leonie Weissweiler @weissweiler.bsky.social · 23/06/2026

🥳I'm excited to introduce everyone to my group at Leipzig University, the LION (Linguistically-Oriented NLP) Lab! 🦁 Looking forward to exciting work on computational linguistics, interpretability, and multilinguality! Find out more on our website lionlabnlp.github.io

lionlabnlp.github.io

Home — LION Lab

LION Lab — Linguistically-Oriented NLP research group at the University of Leipzig.

1263

Reposted by Leonie Weissweiler

ELLIS @ellis.eu · 21/01/2026

Meet @weissweiler.bsky.social, Postdoc at Uppsala Uni 🇸🇪 and ELLIS Member, working on #NLProc, computational linguistics & LM interpretability. She’s accepted an Assistant Prof position at Leipzig Uni 🇩🇪—the first female CompSci professor there—and is eager to support more women 👏 #WomenInELLIS

0171

Reposted by Leonie Weissweiler

Leonie Weissweiler @weissweiler.bsky.social · 11/12/2025

🧑‍🔬I’m recruiting PhD students in Natural Language Processing @unileipzig.bsky.social Computer Science, together with @scadsai.bsky.social! Topics include, but aren’t limited to: 🔎Linguistic Interpretability 🌍Multilingual Evaluation 📖Computational Typology Please share! #NLProc #NLP

14225

Reposted by Leonie Weissweiler

ELLIS @ellis.eu · 18/12/2025

🏹 Job alert: Two fully funded PhD positions in Natural Language Processing at University of Leipzig 📍 Leipzig 🇩🇪 📅 Apply by Jan 15th 🔗 ellis.eu/research/jobs/2025-12-16-t…

ellis.eu

Two fully funded PhD positions in Natural Language Processing at University of Leipzig | European Laboratory for Learning and Intelligent Systems

0108

Reposted by Leonie Weissweiler

ScaDS.AI Dresden/Leipzig @scadsai.bsky.social · 17/12/2025

@weissweiler.bsky.social is setting up her own group at @unileipzig.bsky.social and @scadsai.bsky.social as Assistant Professor for #NaturalLanguageProcessing and is advertising two open PhD positions. 🥳 👉 Full announcement: leonieweissweiler.github.io/phd_leipzig....

082

Leonie Weissweiler @weissweiler.bsky.social · 11/12/2025

aww thanks for the vote of confidence!

010

Leonie Weissweiler @weissweiler.bsky.social · 11/12/2025

🔗Find more information and apply here: leonieweissweiler.github.io/phd_leipzig....

leonieweissweiler.github.io

000

Leonie Weissweiler @weissweiler.bsky.social · 11/12/2025

14225

Leonie Weissweiler @weissweiler.bsky.social · 10/12/2025

🥳Life Update! I’m thrilled to share that I’ll be starting as assistant professor for Natural Language Processing @unileipzig.bsky.social in April! I’m deeply grateful to everyone who supported me on this journey. I will be recruiting PhD students with @scadsai.bsky.social, stay tuned for details!

8474

Leonie Weissweiler @weissweiler.bsky.social · 19/11/2025

Oops, here are the three caused-motion examples that were meant to go in the first post:

000

Leonie Weissweiler @weissweiler.bsky.social · 19/11/2025

👥Joint work with Abdullatif Köksal and Hinrich Schütze 📰Check out the paper: nejlt.ep.liu.se/article/view... 💻The full dataset and code are available on GitHub: github.com/LeonieWeissw... 🧵7/7

nejlt.ep.liu.se

View of Hybrid Human-LLM Corpus Construction and LLM Evaluation for the Caused-Motion Construction

010

Leonie Weissweiler @weissweiler.bsky.social · 19/11/2025

We find that a range of models indeed struggle with this, but Gemma 27B solves it almost perfectly! In grey are cases where the model struggles to answer both questions, in red the cases where it would have needed to reply on the caused-motion semantics. 🧵6/7

100

Leonie Weissweiler @weissweiler.bsky.social · 19/11/2025

Once we have manually annotated the final dataset in this way, we return to the original question and test if LLMs find the third question below easier than the second, which would indicate difficulties in making use of the semantics of the caused-motion construction. 🧵5/7

100

Leonie Weissweiler @weissweiler.bsky.social · 19/11/2025

Designing the right prompt is tricky and depends on annotation cost. For example, giving examples and asking for a json with the sentences and labels is always a good idea, but using o1 over 4o-mini is only worth it if human annotation costs more than .5$ per sentence! 🧵4/7

100

Leonie Weissweiler @weissweiler.bsky.social · 19/11/2025

But this leaves many FPs, and filtering them by hand would be expensive. To reduce this cost, we use few-shot prompt-based filtering, which greatly reduces the number of FPs that our human annotator will have to sift through, and therefore the annotation cost. 🧵3/7

100

Leonie Weissweiler @weissweiler.bsky.social · 19/11/2025

They are all instances of the so-called caused-motion construction, and collecting enough instances for testing was a challenge, given its rarity! To construct our dataset, we first create a dependency filter based on the syntactic side of the construction. 🧵2/7

100

Leonie Weissweiler @weissweiler.bsky.social · 19/11/2025

📢Out now in NEJLT!📢 In each of these sentences, a verb that doesn't usually encode motion is being used to convey that an object is moving to a destination. Given that these usages are rare, complex, and creative, we ask: Do LLMs understand what's going on in them? 🧵1/7

2153

Reposted by Leonie Weissweiler

Kyle Mahowald @kmahowald.bsky.social · 10/11/2025

Oh cool! Excited this LM + construction paper was SAC-Highlighted! Check it out to see how LM-derived measures of statistical affinity separate out constructions with similar words like "I was so happy I saw you" vs "It was so big it fell over".

0174

Leonie Weissweiler @weissweiler.bsky.social · 10/11/2025

I can confirm that I was indeed the 2nd author!

171

Reposted by Leonie Weissweiler

Paul Lerner @lernerp.bsky.social · 06/11/2025

There's many directions where this could go, multilingual, low-resource language, interpretability, depending on your profile, and the internship may lead to a PhD, provided we get funding!

011