Maksim Bobrin

Publications

Google Scholar

2026

Zero-Shot Off-Policy Learning

A. Asadulaev, M. Bobrin^*, S. Lahlou, D. Dylov, F. Karray, M. Takáč

ICML 2026

Diagram of zero-shot off-policy adaptation

Dual Advantage Fields

A. Zemtsov, M. Bobrin, A. Nikulin, D. V. Dylov, et al.

arXiv:2606.04188

Plan, Don't Pose: Long Composite Motion Generation with Text-Aligned BFM

N. Shvetsov, M. Bobrin, N. Buzun, A. Bozhedarov, D. V. Dylov

arXiv:2605.29906

Convex Compositional Reasoning Models

M. Roketlishvili, S. Semenov, M. Bobrin, et al.

arXiv:2605.23395

2025

Zero-Shot Adaptation of Behavioral Foundation Models to Unseen Dynamics

M. Bobrin, I. Zisman, A. Nikulin, V. Kurenkov, D. Dylov

ICLR 2026

Paper OpenReview

Sphere visualizations of sampled policy-encoding vectors

Random Direct Preference Optimization for Radiography Report Generation

V. Samokhin, B. Shirokikh, M. Goncharov, D. Umerenkov, M. Bobrin, et al.

MICCAI 2025 Workshop on Foundation Models for General Medical AI

HOTA: Hamiltonian Framework for Optimal Transport Advection

N. Buzun, D. Shlenskii, M. Bobrin, D. V. Dylov

ICLR 2026

HOTA transport trajectories on smooth and non-smooth datasets

2024

Ascend HiFloat8 Format for Deep Learning

Y. Luo, Z. Zhang, R. Wu, H. Liu, …, M. Bobrin, et al.

arXiv:2409.16626

HiFloat8 significand precision over exponent plot

Spotlight paper

ENOT: Expectile Regularization for Fast and Accurate Training of Neural Optimal Transport

N. Buzun, M. Bobrin^*, D. V. Dylov

NeurIPS 2024 · Spotlight

Paper OpenReview Project Blog OTT-JAX

Align Your Intents: Offline Imitation Learning via Optimal Transport

M. Bobrin, N. Buzun, D. Krylov, D. V. Dylov

arXiv:2402.13037

Shared first author