Santiago Acevedo — accepted AI-conference papers

2026

Differential syntactic and semantic encoding in LLMs

Santiago Acevedo, Alessandro Laio, Marco Baroni

ICML 2026poster

We study how syntactic and semantic information is encoded in inner layer representations of Large Language Models (LLMs), focusing on the very large DeepSeek-V3. We find that, by averaging hidden-representation vectors of sentences sharing syntactic structure or meaning, we obtain vectors that capt…

Cited by 6 Source ↗Scholar ↗

2025

Intrinsic Dimension Correlation: uncovering nonlinear connections in multimodal representations

Lorenzo Basile, Santiago Acevedo, Luca Bortolussi, Fabio Anselmi, Alex Rodriguez

ICLR 2025poster

To gain insight into the mechanisms behind machine learning methods, it is crucial to establish connections among the features describing data points. However, these correlations often exhibit a high-dimensional and strongly nonlinear nature, which makes them challenging to detect using standard met…

Cited by 3 Source ↗PDF ↗code ↗Scholar ↗