TL;DR: Does the next-token logit track the conditional or the joint probability of the whole sequence?I had an invisi...

Training modern large language models (LLMs) has become a veritable smorgasbord of algorithms and datasets designed to elicit particular behaviors, making it critical to develop techniques to understa...

Switch off your brain and answer this:Given three points $\mathbf{X}, \mathbf{Y}, \mathbf{Z}$ sampled from a high-dim...

Recurrent neural networks (RNNs) are powerful models used widely in both machine learning and neuroscience to learn tasks with temporal dependencies and to model neural dynamics. However, despite...

Here’s a secret that I accidentally discovered during my PhD: consuming history-of-science content is an efficient wa...

We design a suite of minimal algorithmic tasks that are a loose abstraction of open-ended real-world tasks. This allows us to cleanly and controllably quantify the creative limits of the present-day l...

Communications Chairs 2025 2021 Conference

I was a kinda awkward teenager. If you are a CS researcher reading this post, then chances are, you were too. How to navigate social situations and make friends is not always intuitive, and has to …

Recently, a large amount of work has focused on improving large language models' (LLMs') performance on reasoning benchmarks such as math and logic. However, past work has largely assumed that tasks a...