Saved in:
Bibliographic Details
Main Authors: Shahane, Aditya Hemant, Sirohi, Anuj Kumar, Arora, Devansh, Kumar, Nitin, P, Prathosh A, Kumar, Sandeep
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.24089
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913064583954432
author Shahane, Aditya Hemant
Sirohi, Anuj Kumar
Arora, Devansh
Kumar, Nitin
P, Prathosh A
Kumar, Sandeep
author_facet Shahane, Aditya Hemant
Sirohi, Anuj Kumar
Arora, Devansh
Kumar, Nitin
P, Prathosh A
Kumar, Sandeep
contents Bridging molecular structures and natural language is essential for controllable design. Autoregressive models struggle with long-range dependencies, while standard diffusion processes apply uniform corruption across positions, which can distort structurally informative tokens. We present BiMol-Diff, a unified diffusion framework for the paired tasks of text-conditioned molecule generation and molecule captioning. Our key component is a token-aware noise schedule that assigns position-dependent corruption based on token recovery difficulty, preserving harder-to-recover substructures during the forward process. On ChEBI-20 and M3-20M, BiMol-Diff improves molecule reconstruction with a 15.4% relative gain in Exact Match and achieves strong captioning results, attaining best BLEU and BERTScore among compared baselines. These results indicate token-aware noising improves fidelity in molecular structure-language modelling.
format Preprint
id arxiv_https___arxiv_org_abs_2604_24089
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle BiMol-Diff: A Unified Diffusion Framework for Molecular Generation and Captioning
Shahane, Aditya Hemant
Sirohi, Anuj Kumar
Arora, Devansh
Kumar, Nitin
P, Prathosh A
Kumar, Sandeep
Computation and Language
Bridging molecular structures and natural language is essential for controllable design. Autoregressive models struggle with long-range dependencies, while standard diffusion processes apply uniform corruption across positions, which can distort structurally informative tokens. We present BiMol-Diff, a unified diffusion framework for the paired tasks of text-conditioned molecule generation and molecule captioning. Our key component is a token-aware noise schedule that assigns position-dependent corruption based on token recovery difficulty, preserving harder-to-recover substructures during the forward process. On ChEBI-20 and M3-20M, BiMol-Diff improves molecule reconstruction with a 15.4% relative gain in Exact Match and achieves strong captioning results, attaining best BLEU and BERTScore among compared baselines. These results indicate token-aware noising improves fidelity in molecular structure-language modelling.
title BiMol-Diff: A Unified Diffusion Framework for Molecular Generation and Captioning
topic Computation and Language
url https://arxiv.org/abs/2604.24089