Sep 29, 2026	We released a new preprint: EMPIRIC: Experiment-Driven Learning of Residual World Models for Robot Planning. This was a great project led by Yichao Liang!
Sep 25, 2026	We released a new preprint: Coding Agents for Generalized Task and Motion Planning Problems, with Tom Silver’s group at Princeton. Check out the project website for videos of the programs the agents write!
Sep 24, 2026	QVal was accepted to the NeurIPS 2026 Evaluations and Datasets Track as a poster! Congratulations to Sergio and all the authors!
Sep 02, 2026	Three papers accepted to EMNLP 2026! DecSelfMask as a main conference paper, ViPlan and SAGE as findings papers. Congratulations to all the authors!
Jul 02, 2026	We released a new preprint: QVal: Cheaply Evaluating Dense Supervision Signals for Long-Horizon LLM Agents, led by Sergio Hernández.
Jun 13, 2026	We released a new preprint: DecSelfMask: Leveraging Unlabeled Text via Self-Relevance-Guided Masking for Decoder-Only Classification, led by Pietro Ferrazzi.
Nov 25, 2025	A preliminary version of our latest work, Guiding Reinforcement Learning with Selective Vision-Language Model Supervision, has been published in the ECAI 2025 CAIPI workshop. We are currently extending this work into a full-length paper for a conference submission.
May 20, 2025	We released a new preprint: ViPlan: A Benchmark for Visual Planning with Symbolic Predicates and Vision-Language Models.
Mar 05, 2025	My Master’s Thesis was awarded as one of the three best at the Aalto University School of Science in 2024.
Sep 25, 2024	Our paper “Generating Code World Models with Large Language Models Guided by Monte Carlo Tree Search” has been accepted at NeurIPS 2024.
Jul 08, 2024	Our paper “In-Context Symbolic Regression: Leveraging Large Language Models for Function Discovery” has been accepted at the ACL 2024 Student Research Workshop.