Alessandro Stolfo

prof_pic.jpeg

OAT Y 24

Andreasstrasse 5

8092 Zürich, Switzerland

Hi! I am a doctoral student in the Institute for Machine Learning at ETH Zürich, where I am advised by Prof. Mrinmaya Sachan, and co-advised by Prof. Yonatan Belinkov (Technion).

I am interested in studying and improving machine learning models for natural language processing. I am excited about exploring their capabilities on complex tasks such as solving arithmetic problems and reasoning over factual and commonsense knowledge.

In summer 2023, I interned in the Machine Learning Research Group at Oracle Labs in Burlington, MA, where I worked with Ari Kobren.

Before starting my doctoral studies, I obtained a Master’s degree in Data Science at ETH Zürich, and I worked at Rethink-Resource on the development of Circado. I completed my undergraduate studies in Computer Engineering at Politecnico di Milano.

I am grateful to be a recipient of the CYD Doctoral Fellowship.

news

Nov 22, 2023 I am attending the ML Alignment & Theory Scholars (MATS) Program, mentored by Neel Nanda.
Jul 17, 2023 I started my internship in the ML Research Group at Oracle Labs in Burlington, MA.
Apr 22, 2022 I answered a couple of questions for this EPFL News article. Check it out!

selected publications

  1. EMNLP 2023
    A Mechanistic Interpretation of Arithmetic Reasoning in Language Models using Causal Mediation Analysis
    A. Stolfo, Y. Belinkov, and M. Sachan
    EMNLP 2023
    arXiv
  2. EMNLP 2023
    Towards a Mechanistic Interpretation of Multi-Step Reasoning Capabilities of Language Models
    Y. Hou, J. Li, Y. Fei, A. Stolfo, W. Zhou, G. Zeng, A. Bosselut, and M. Sachan
    EMNLP 2023
    arXiv
  3. ACL 2023
    A Causal Framework to Quantify the Robustness of Mathematical Reasoning with Language Models
    A. Stolfo*, Z. Jin*, K. Shridhar, B. Schölkopf, and M. Sachan
    ACL 2023
    ACL arXiv
  4. ACL 2023
    Distilling Multi-Step Reasoning Capabilities of Large Language Models into Smaller Models via Semantic Decompositions
    K. Shridhar*, A. Stolfo*, and M. Sachan
    ACL 2023 (Findings)
    ACL arXiv