To what extent can LMs perform Bayesian decision making? Check out our new preprint to find out! We combined different layers of analysis and different methods to answer this question - I really enjoyed the cogsci-minded approach! 👇🏻
Is an LM that acts Bayesian also Bayesian inside?
☝️ Only as far as its beliefs allow!
Fine-tuned on an optimal Bayesian model, LMs hold and use better beliefs than usual fine-tuning on golden answers.
Details 👇 or bayeslm.github.io
#interpretability #nlproc (1/🧵)
Polina Tsvilodub retweeted
Small bird, fast wings, Kolibri is here.
78B parameters. 3.46B active. Up to 1M tokens of context. Built in Europe.
Now the weights are yours. Run it on your own hardware, under Apache 2.0.
Polina Tsvilodub retweeted
Have we built machines that think like humans? 🧠🤖
In a new preprint, we propose CogGym, which compares AI and human judgments across 258 commonsense-reasoning experiments from 100 cognitive-science papers.
Paper: arxiv.org/abs/2609.21259
Platform: coggym.org
Polina Tsvilodub retweeted
LMs that "know more" about toxicity are less toxic!
Our #TACL 📄 connects behavior and internals:
💠 LMs amplify toxicity beyond humans
💠 Information about toxicity peaks in lower layers
💠 Bypassing these layers increases toxicity
More details👇 #NLProc #interpretability (1/🧵)
Polina Tsvilodub retweeted
Come work with us!!
Opening for PhD position (3.5 years) on Language Evolution | Computational Modeling | Primate Communication in CRC "Common Ground".
Share, apply, or reach out for details!
tinyurl.com/y2av9hhd
Polina Tsvilodub retweeted
Come work with us!!
PostDoc position in Collaborative Research Center "Common Ground" (3.5y)
Deadline November 10th 2025
Philosophy of Language | Cognitive Science | Pragmatics | Logic | ...
Apply, share, or reach out for details!
tinyurl.com/y33vjwj2
Polina Tsvilodub retweeted
More information about the Common ground project and more Ph.D. positions available at uni-tuebingen.de/de/283575 Apply to join our team!
Polina Tsvilodub retweeted
PostDoc position (3.5y) available in interdisciplinary Collaborative Research Center "Common Ground".
🎯 Topic: Pragmatic reasoning about Common Ground
🎓 linguistics | philosophy of language | cognitive science
📅 Deadline: July 14 tinyurl.com/3vd9wa6p
Please share!
Polina Tsvilodub retweeted
📢 2 PhD positions (Tübingen, Germany) on great ape communication & the evolution of common ground!
🧠 biology, psychology, linguistics, philosophy of language, language evolution, modeling
🗓️ Deadline: Aug 13
🔗 bit.ly/4l8p7hy & bit.ly/46jcfAq
Polina Tsvilodub retweeted
📅 The submission deadline has been extended to June 27th AoE! Looking forward to your ideas :)
Happy to announce the first workshop on Pragmatic Reasoning in Language Models — PragLM @ COLM 2025! 🧠🎉
How do LLMs engage in pragmatic reasoning, and what core pragmatic capacities remain beyond their reach?
🌐 sites.google.com/berkeley.ed…
📅 Submit by June 23rd
📢 We have extended the submission deadline for our PragLM workshop @ COLM‘25 until June 27th end of day!📢
Submit & register - we look forward to seeing you in Montreal!
LLMs are now everyday conversation partners, but key questions remain: which, if any, inferences do they compute for conversation? Are LLMs (non) human-like *pragmatic* language users?
Submit to our workshop and join for talks by @_jennhu @vdemberg @meanwhileina @dan_fried!
Polina Tsvilodub retweeted
Job alert! We have 2 Ph.D. positions available in a #DFG funded project. We are planning to combine acquisition work with computational #modeling to investigate how children and adults understand #irony. Deadline: July 20, 2025. Info linguistlist.org/issues/36/1… Please share.
Polina Tsvilodub retweeted
Imagine a (fully hypothetical) situation: you are terrified of some recent election outcome. When would you say that the election outcome is interesting, rather than calling it horrible? In a new paper with @meanwhileina and Martin Butz we develop a model of #indirect speech.
LLMs are now everyday conversation partners, but key questions remain: which, if any, inferences do they compute for conversation? Are LLMs (non) human-like *pragmatic* language users?
Submit to our workshop and join for talks by @_jennhu @vdemberg @meanwhileina @dan_fried!
Happy to announce the first workshop on Pragmatic Reasoning in Language Models — PragLM @ COLM 2025! 🧠🎉
How do LLMs engage in pragmatic reasoning, and what core pragmatic capacities remain beyond their reach?
🌐 sites.google.com/berkeley.ed…
📅 Submit by June 23rd
Polina Tsvilodub retweeted
We are looking for two PhD students at our institute in Munich.
Both postions are open-topic, so anything between cognitive science and machine learning is possible.
More information: hcai-munich.com/PhDHCAI.pdf
Feel free to share broadly!
Polina Tsvilodub retweeted
arxiv.org/abs/2502.06204
work with the amazing @tsvilodub @gandhikanishk @HaoranZhaoHRZ @jphilipp95 @meanwhileina
Our group led by @meanwhileina is hiring! Check out the great opportunity. Apply & spread the world!
Position opening: #CompLing, #NLP, #CogSci
Junior Group Leader / PostDoc
(4y, German E14 salary scale) @uni_tue
- no teaching obligations
- PhD co-supervision
- dynamic interdisciplinary team in one of Germany’s top research universities
tinyurl.com/jhxb4wu2
1
171
I’m delighted to co-organize this great event! Register & share!
🧠 Join Us for the PhD Symposium on “Understanding Context in Cognition”!
📆 April 7–9, 2025
📍 University of Tübingen, Germany
🌐 Registration is FREE! Register now here: cogsciprag.github.io/context…
Polina Tsvilodub retweeted
I'm recruiting PhD students + postdocs for my lab, coming to @JohnsHopkins in Fall 2025!
Our brand new lab is at the intersection of cognitive science and AI, using computational + behavioral methods to understand how language works in minds and machines.
Details below! (1/4)

















