Navigating Simply, Aligning Deeply: Winning Solutions for Mouse vs. AI 2025

Phu-Hoa Pham, Chi-Nguyen Tran, Dao Sy Duy Minh, Nguyen Lam Phu Quy, Huynh Trung Kiet · 2026

Read Navigating Simply, Aligning Deeply: Winning Solutions for Mouse vs. AI 2025 by Phu-Hoa Pham, Chi-Nguyen Tran, Dao Sy Duy Minh, Nguyen Lam Phu Quy, Huynh Trung Kiet (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-00982.txt.

Description

Phu-Hoa Pham and colleagues report the work from Team HCMUS_TheFangs in the NeurIPS 2025 Mouse vs. AI contest. For visual robustness, a two-layer CNN with Gated Linear Units and normalization scores 95.4% in the test. For neural alignment, a ResNet-like net with 16 layers and GLU-based gating takes top-1 neural prediction rank. A study of checkpoints finds scores do not rise with training time, with the best mark near 200K steps. The work argues that simple builds win on robustness, deeper builds on neural fit.

Catalog record

CREATORPhu-Hoa Pham, Chi-Nguyen Tran, Dao Sy Duy Minh, Nguyen Lam Phu Quy, Huynh Trung Kiet
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE49712
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602.00982

Browse the full library · About the Tower of Babel