Hi—I’m Michael Hanna! I’m interested in interpreting and evaluating NLP models from a linguist’s and cognitive scientist’s point of view. I’ve just finished my master’s degrees through the EM LCT dual-degree program at the University of Trento and Charles University. Now, I’ve begun a PhD at the ILLC at the University of Amsterdam, supervised by Sandro Pezzelle and Yonatan Belinkov, through ELLIS!
- December 2022: I will be spending January 2023 participating in Redwood Research’s REMIX program! I’m very excited to meet other interpretability researchers, and apply mechanistic interpretability techniques to transformer language models. Feel free to reach out if you’re in the area!
- September 2022: I graduated from Charles University in Prague (Computer Science)! My thesis is available here; feel free to reach out for a much shorter preprint discussing these results!
- August 2022: New paper at COLING 2022 with Federico Pedeni, Alessandro Suglia, Alberto Testoni, and Prof. Raffaella Bernardi!
- July 2022: I graduated from the University of Trento (Cognitive Science)! Stay tuned for news about my thesis, supervised by Roberto Zamparelli and David Mareček, on using causal interventions to understand subject-verb agreement in large language models.
- November 2021: New paper on BERTScore’s successes and failures with Ondřej Bojar at WMT 2021
- November 2021: New paper on knowledge of hypernymy in BERT with David Mareček at BlackboxNLP 2021
- June 2020: I graduated from the University of Chicago (CS and Linguistics)! I wrote an honors thesis on interpreting variational autoencoders for sentences, advised by Allyson Ettinger.
My main research interest is creating models with human-like abilities. I’m interested in approaching this goal by using linguistically-motivated interpretability and analysis methods to ask “How close are models’ abilities to those of humans?”. I’m also interested in grounded NLP, especially when that grounding goes beyond vision, to include additional modalities and embodiment. I’m excited about using ideas from linguistics and cognitive science to improve deep learning models and training regimes.
- Living Abroad: I’ve been lucky to participate in a number of programs, most of them free, that have allowed me to live in other countries and learn new ways of life. So far, I’ve lived for a year each in Spain (SYA), Czechia, and Italy (LCT), as well as a couple of summers in Korea (NSLI-Y, FLAG). If you’re in high school or university and want to know more about opportunities like this, shoot me an email—I’m happy to share more info about programs that will let you live abroad for free!
- Urbanism: I’m fascinated by the way that our built environment affects our lives. I’m a big fan of walkable neighborhoods, public transportation, and the power of well-designed spaces to bring people together.
- Vocal Performance: During my undergrad, I was a baritone in Run for Cover, an a cappella group at UChicago. Check us out on Spotify or iTunes!