Tubingen just got ultra-exciting :D
06.08.2025 15:49
👍 3
🔁 0
💬 0
📌 0
Tubingen just got ultra-exciting :D
Submit your latest and greatest papers to the hottest workshop on the block---on cognitive interpretability! 🔥
Excited to announce the first workshop on CogInterp: Interpreting Cognition in Deep Learning Models @ NeurIPS 2025! 📣
How can we interpret the algorithms and representations underlying complex behavior in deep learning models?
🌐 coginterp.github.io/neurips2025/
1/4
I'll be at ICML beginning this Monday---hit me up if you'd like to chat!
Our recent paper may be relevant (arxiv.org/abs/2506.17859)! We take a rational analysis lens to argue that beyond simplicity bias, we must model how well a hypothesis explains the data to yield a *predictive* account of behavior in neural nets! This helps explain learning of more complex functions.
I am definitely not biased :)
Check out one of the most exciting papers of the year! :D
I'll be attending NAACL at New Mexico beginning today---hit me up if you'd like to chat!
Check out our new work on the duality between SAEs and how concepts are organized in model representations!
New preprint alert!
Do Sparse Autoencoders (SAEs) reveal all concepts a model relies on? Or do they impose hidden biases that shape what we can even detect?
We uncover a fundamental duality between SAE architectures and concepts they can recover.
Link: arxiv.org/abs/2503.01822
Oh god, I had no clue this happened. :/
Does bsky not do GIFs?
Very nice paper; quite aligned with the ideas in our recent perspective on the broader spectrum of ICL. In large models, there's probably a complicated, context dependent mixture of strategies that get learned, not a single ability.
Paper co-led with @corefpark, and in collaboration with the ever-awesome @PresItamar and @Hidenori8Tanaka! :)
arXiv link: arxiv.org/abs/2412.01003
Dynamics of attention maps is particularly striking in this competition: e.g., with high diversity, we see a bigram counter forming, but memorization eventually occurs and the pattern becomes uniform! This means models can remove learned components if they are not useful anymore!
Beyond corroborating our phase diagram, LIA confirms a persistent competition underlies ICL: once the induction head forms, with enough diversity bigram-based inference takes over; under low diversity, memorization occurs faster, yielding a bigram-based retrieval solution!
Given optimization and data diversity are independent axes, we then ask if these forces race against each other to yield our observed algorithmic phases. We propose a tool called LIA (linear interpolation of algorithms) for this analysis.
The tests check out! We see before/after a critical # of train steps are met (where induction head emerges), the model relies on unigram/bigram stats. With few chains (less diversity), there is retrieval behavior: we can literally reconstruct transition matrices from MLP neurons!
To test the above claim, we compute the effect of shuffling a sequence on next-token probs: this breaks bigram stats, but preserves unigrams. We check how “retrieval-like” or memorization-based model behavior is by comparing predicted transitions’ KL to a random set of chains.
We claim four algorithms explain the model’s behavior in diff. train/test settings. These algos compute uni-/bi- gram frequency statistics of an input to either *retrieve* a memorized chain or to in-context *infer* the chain used to define the input: latter performs better OOD!
We analyze models trained on a fairly simple task: learning to simulate a *finite mixture* of Markov chains. The sequence modeling nature of this task makes it a better abstraction for studying ICL abilities in LMs, (compared to abstractions of few-shot learning like linear reg.)
New paper–accepted as *spotlight* at #ICLR2025! 🧵👇
We show a competition dynamic between several algorithms splits a toy model’s ICL abilities into four broad phases of train/test settings! This means ICL is akin to a mixture of different algorithms, not a monolithic ability.
bsky.app/profile/ajyl...
Some threads about recent works ;-)
bsky.app/profile/ekde...
Our group, funded by NTT Research, Inc., uniquely bridges industry and academia. We integrate approaches from physics, neuroscience, and psychology while grounding our work in empirical AI research.
Apply from "Physics of AI Group Research Intern": careers.ntt-research.com
Check out our recent work, including 5 ICLR 2025 papers all (co)led by amazing past and current interns: sites.google.com/view/htanaka...
There's never been a more exciting time to explore the science of intelligence! 🧠
What can ideas and approaches from science tell us about how AI works?
What might superhuman AI reveal about human cognition?
Join us for an internship at Harvard to explore together!
1/
Are those... bats?
Now accepted at NAACL! This would be my first time presenting at an ACL conference---I've got almost first-year grad school level of excitement! :P
New paper! “In-Context Learning of Representations”
What happens to an LLM’s internal representations in the large context limit?
We find that LLMs form “in-context representations” to match the structure of the task given in context!
New paper <3
Interested in inference-time scaling? In-context Learning? Mech Interp?
LMs can solve novel in-context tasks, with sufficient examples (longer contexts). Why? Bc they dynamically form *in-context representations*!
1/N
