Graph VQ-Transformer (GVT): Fast and Accurate Molecular Generation via High-Fidelity Discrete Latents

Haozhuo Zheng, Cheng Wang, Yang Liu · 2025

Read Graph VQ-Transformer (GVT): Fast and Accurate Molecular Generation via High-Fidelity Discrete Latents by Haozhuo Zheng, Cheng Wang, Yang Liu (2025) 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-2512-02667.txt.

Description

Haozhuo Zheng, Cheng Wang, and Yang Liu generate new molecules with a two stage transformer system. Designing molecules with wanted traits is slow when each candidate must be built and tested. Diffusion models generate well but cost heavy compute, while step by step models can drift into errors.

Their Graph VQ-Transformer first compresses molecules into high fidelity discrete codes. A transformer then writes new codes, which decode into full molecular graphs. The design aims for diffusion level accuracy at lower cost, helping chemists search larger chemical spaces for drugs and materials. Codes that keep the shape of the whole graph keep the traits that make a drug work or fail.

Catalog record

CREATORHaozhuo Zheng, Cheng Wang, Yang Liu
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE34 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-02667

Browse the full library · About the Tower of Babel