Fine-tuning DeepSeek-OCR-2 for Molecular Structure Recognition

Haocheng Tang, Xingyu Dang, Junmei Wang · 2026

Read Fine-tuning DeepSeek-OCR-2 for Molecular Structure Recognition by Haocheng Tang, Xingyu Dang, Junmei Wang (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-2604-03476.txt.

Description

Tang, Dang, and Wang adapt in a 2026 paper DeepSeek-OCR-2 for reading molecular diagrams as SMILES strings. A two-stage fine-tuning strategy starts with parameter-efficient LoRA, then moves to selective full-parameter tuning with split learning rates, trained on synthetic PubChem renderings and real patent images. The resulting MolSeek-OCR matches the best image-to-sequence model on exact matches but trails image-to-graph leaders.

Catalog record

CREATORHaocheng Tang, Xingyu Dang, Junmei Wang
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE20665
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.03476

Browse the full library · About the Tower of Babel