Sign in

takelab.bsky.social

@takelab.bsky.social
10 followers 7 following 7 posts
PostsRepliesMedia
takelab.bsky.social @takelab.bsky.social · 30/06/2026
Evaluating Pluralism in LLMs through Latent Perspectives 📜 arxiv.org/abs/2606.13254 🗓️ Sat, Jul 11, Workshop on Pluralistic Alignemnt Laura Majer @mtutek.bsky.social
arxiv.org
Evaluating Pluralism in LLMs through Latent Perspectives
The growing need to represent diverse perspectives has increased interest in pluralistic LLM generation. Although difficult to operationalize, identifying perspectives expressed in text would provide ...
020
takelab.bsky.social @takelab.bsky.social · 30/06/2026
BioAgent Bench: An AI Agent Evaluation Suite for Bioinformatics 📜 arxiv.org/abs/2601.21800 🗓️Tue, Jul 7, Poster Session 1 🗓️Sat, Jul 11, Workshop on Agents in the Wild Dionizije Fa (honorary member) @camuljak.bsky.social
130
takelab.bsky.social @takelab.bsky.social · 30/06/2026
Old Habits Die Hard: How Conversational History Geometrically Traps LLMs 📜 arxiv.org/abs/2603.03308 🗓️ Wed, Jul 8, Poster Session 3 🗓️ Fri, Jul 10, Workshop on Mechanistic Interpretability @mtutek.bsky.social
arxiv.org
Old Habits Die Hard: How Conversational History Geometrically Traps LLMs
How does the conversational past of large language models (LLMs) influence their future performance? Recent work suggests that LLMs are affected by their conversational history in unexpected ways. For...
110
takelab.bsky.social @takelab.bsky.social · 30/06/2026
At ICML '26, TakeLab members are presenting three papers. Make sure to stop by if interested in language model interpretability, coding agent evaluations, or pluralistic alignment. Details below 👇🧵 @icmlconf.bsky.social #ICML2026 🇰🇷
141
takelab.bsky.social @takelab.bsky.social · 26/03/2026
[WASSA] A Multi-Aspect Evaluation Framework for Synthetic Data: Case Study on Irony and Sarcasm 📊⚖️🐈 📜 aclanthology.org/2026.wassa-1... 📍 Poster Hall 🗓️ Sunday, March 29th, 16:30-17:30
aclanthology.org
A Multi-Aspect Evaluation Framework for Synthetic Data: Case Study on Irony and Sarcasm
Laura Majer, Ana Barić, Florijan Sandalj, Ivan Unković, Bojan Puvača, Jan Šnajder. The Proceedings for the 15th Workshop on Computational Approaches to Subjectivity, Sentiment Social Media Analysis (W...
010
takelab.bsky.social @takelab.bsky.social · 26/03/2026
[Findings] Sequence Repetition Enhances Token Embeddings and Improves Sequence Labeling with Decoder-Only Language Models 🔁🔁🔁 📜 aclanthology.org/2026.finding... 📍 Poster Hall 🗓️ Friday, March 27th, 9:00-10:30
aclanthology.org
Sequence Repetition Enhances Token Embeddings and Improves Sequence Labeling with Decoder-only Language Models
Matija Luka Kukić, Marko Čuljak, David Dukić, Martin Tutek, Jan Šnajder. Findings of the Association for Computational Linguistics: EACL 2026. 2026.
110
takelab.bsky.social @takelab.bsky.social · 26/03/2026
At EACL '26, TakeLab members are presenting two papers. Feel free to stop by if you'd like to chat about simple tricks to improve LLM token embeddings, principled evaluation of synthetic data, coffee, or life in general. Details below 👇🧵 @eaclmeeting.bsky.social #EACL2026 #NLProc 🇲🇦
172
Reposted by @takelab.bsky.social
Martin Tutek @mtutek.bsky.social · 04/02/2026
ManagerBench was accepted to ICLR! @iclr-conf.bsky.social #ICLR2026 LLMs are still either unsafe, or completely harm avoidant - even when the harm affects furniture 🛋️ Check out our benchmark, online or in Rio 🇧🇷
041
Reposted by @takelab.bsky.social
camuljak.bsky.social @camuljak.bsky.social · 02/02/2026
We already know prompt repetition is a handy hack to improve a decoder-only LM’s performance as it allows the model to “see” bidirectionally, an ability otherwise suppressed by the causal mask. But what happens if we increase the number of repetitions? 🤔🧵 @eaclmeeting.bsky.social #EACL2026
154
Reposted by @takelab.bsky.social
Martin Tutek @mtutek.bsky.social · 07/11/2025
Very honored to be one out of seven outstanding papers at this years' EMNLP :) Huge thanks to my amazing collaborators @fatemehc.bsky.social @anamarasovic.bsky.social @boknilev.bsky.social , this would not have been possible without them!
2236
Reposted by @takelab.bsky.social
Akhil Arora @akhilarora.bsky.social · 21/07/2025
Back from #ICML2025 🛬, and off 🚄 to Norrköping 🇸🇪 for #ic2s2 CLAN (cs.au.dk/~clan/) members are presenting 2 papers: 1 spotlight and 1 oral. See 🧵for posters and summaries 👋 Reach out to chat about observational studies, causality, LLM agents, human-centered AI etc.
cs.au.dk
CLAN for AI Research on Language and Networks
CLAN for AI Research on Language and Networks, or **CLAN** for short, is a research group and a part of the larger [DIS Group](https://cs.au.dk/research/data-intensive-systems) in the [Department of C...
142
Reposted by @takelab.bsky.social
David Dukić @ddaviddukic.bsky.social · 15/07/2025
📣📣 New preprint alert!! Despite events in the world becoming bleaker, the news is… more positive? We conduct a diachronic study of word embeddings trained on 10M Croatian news articles spanning 25 years and find some surprising results! arxiv.org/abs/2506.13569
122