Breaking Bad Molecules: Are MLLMs Ready for Structure-Level Molecular Detoxification?
cs.AI, cs.CL
Submitted: 2025-06-12
Updated: 2026-11-04
Comments: Accepted by KDD 2026 (AI for Sciences Track, Oral)
Journal ref: Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2 (KDD '26), Jeju Island, Republic of Korea, pp. 11399-11410, 2026
Code: https://github.com/HydroSophy/ToxiMol
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Yi: Open Foundation Models by 01.AI
- LIDDIA: Language-based Intelligent Drug Discovery Agent
- Intern-S1: A Scientific Multimodal Foundation Model
- Qwen3-VL Technical Report
- Qwen2.5-VL Technical Report
- InternLM2 Technical Report
- IMG2SMI: Translating Molecular Structure Images to Simplified Molecular-input Line-entry System
- Tx-LLM: A Large Language Model for Therapeutics
- MiMo-V2-Flash Technical Report
- Seed1.5-VL Technical Report
- MV-CLAM: Multi-View Molecular Interpretation with Cross-Modal Projection via Language Model
- ChemEval: A Comprehensive Multi-Level Chemical Evaluation for Large Language Models
- Structural Reasoning Improves Molecular Understanding of LLM
- A Dataset for Distilling Knowledge Priors from Literature for Therapeutic Design
- Kimi-VL Technical Report
- $\texttt{MiniMol}$: A Parameter-Efficient Foundation Model for Molecular Learning
- Mol-LLM: Multimodal Generalist Molecular LLM with Improved Graph Utilization
- LLaVA-OneVision: Easy Visual Task Transfer
- How to Detect and Defeat Molecular Mirage: A Metric-Driven Benchmark for Hallucination in LLM-based Molecular Comprehension
- Speak-to-Structure: Evaluating LLMs in Open-domain Natural Language-Driven Molecule Generation
Related papers
- Multi-Objective Structured Pruning of LLMs for Latency and Model Size Optimization
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Embeddings for Preferences, Not Semantics
- Can We Trust Item Response Theory for AI Evaluation?
- Diagnosing Task Insensitivity in Language Agents