Showing 2024Show all
3 papers · 1 filter
cs.CL2024
PACE: Procedural Abstractions for Communicating Efficiently
Jonathan D. Thomas, Andrea Silvi, Devdatt Dubhashi +1
A central but unresolved aspect of problem-solving in AI is the capability to introduce and use abstractions, something humans excel at. Work in cognitive science has demonstrated…
cs.CL2024
Learning Efficient Recursive Numeral Systems via Reinforcement Learning
Andrea Silvi, Jonathan Thomas, Emil Carlsson +2
It has previously been shown that by using reinforcement learning (RL), agents can derive simple approximate and exact-restricted numeral systems that are similar to human ones (Ca…
cs.LG2024
Reasoning in Transformers -- Mitigating Spurious Correlations and Reasoning Shortcuts
Daniel Enström, Viktor Kjellberg, Moa Johansson
Transformer language models are neural networks used for a wide variety of tasks concerning natural language, including some that also require logical reasoning. However, a transfo…