I will be presenting our ICLR 2026 paper on attributing culture conditioned generations to memorization of pretraining data, on Fri April. 25, Hall 3 + Hall 2B #255! DM me if you want to chat about memorization, culture, or anything else! #ICLR2025 #iclr #ICLR25

1+1=3 2+2=5 3+3=? Many language models (e.g., Llama 3 8B, Mistral v0.1 7B) will answer 7. But why? We dig into the model internals, uncover a function induction mechanism, and find that it’s broadly reused when models encounter surprises during in-context learning. 🧵

Our paper has been accepted to #EMNLP2025 Main Conference🎉

Heading to #COLM2025! While I will not be presenting a paper, I will have all the time to talk about memorization, reasoning and beyond 🤩 p.s. I am also looking for Summer 2026 internship opportunities😁