Beyond Learning on Molecules by Weakly Supervising on Molecules
Not on the map. Why?
Abstract (arXiv)
Molecular representations are inherently task-dependent, yet most pre-trained molecular encoders are not. Task conditioning promises representations that reorganize based on task descriptions, but existing approaches rely on expensive labeled data. We show that weak supervision on programmatically derived molecular motifs is sufficient. Our Adaptive Chemical Embedding Model (ACE-Mol) learns from hundreds of motifs paired with natural language descriptors that are cheap to compute, trivial to scale. Conventional encoders slowly search the embedding space for task-relevant structure, whereas ACE-Mol immediately aligns its representations with the task. ACE-Mol achieves state-of-the-art performance across molecular property prediction benchmarks with interpretable, chemically meaningful representations.
Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 32 of its 51 references that are in the map. First posted on arXiv 4 Feb 2026.
Builds on
32 of 51 references.
References · 32
- 2017Attention Is All You Needcited by 59,932
- 2017Neural Message Passing for Quantum Chemistrycited by 3,460
- 2016Understanding intermediate layers using linear classifier probescited by 1,279
- 2020Graph Contrastive Learning with Augmentationscited by 1,203
- 2017MoleculeNet: A Benchmark for Molecular Machine Learningcited by 1,037
- 2018Adafactor: Adaptive Learning Rates with Sublinear Memory Costcited by 896
- 2022Galactica: A Large Language Model for Sciencecited by 736
- 2024Smarter, Better, Faster, Longer: A Modern Bidirectional Encoder for Fast, Memory Efficient, and Long Context Finetuning and Inferencecited by 527
- 2019Graph Transformer Networkscited by 520
- 2022GraphMAE: Self-Supervised Masked Graph Autoencoderscited by 443
- 2020ChemBERTa: Large-Scale Self-Supervised Pretraining for Molecular Property Predictioncited by 332
- 2019A Multiscale Visualization of Attention in the Transformer Modelcited by 292
20 more
- 2021Large-Scale Chemical Language Representations Capture Molecular Structure and Propertiescited by 203
- 2022Translation between Molecules and Natural Languagecited by 173
- 2022Multi-modal Molecule Structure-text Model for Text-based Retrieval and Editingcited by 165
- 2024Rephrasing the Web: A Recipe for Compute and Data-Efficient Language Modelingcited by 156
- 2022A Molecular Multimodal Foundation Model Associating Molecule Graphs with Natural Languagecited by 129
- 2022ChemBERTa-2: Towards Chemical Foundation Modelscited by 123
- 2024A Review of Large Language Models and Autonomous Agents in Chemistrycited by 96
- 2020Molecular representation learning with language models and domain-relevant auxiliary taskscited by 91
- 2019SMILES Transformer: Pre-trained Molecular Fingerprint for Low Data Drug Discoverycited by 89
- 2023A Text-guided Protein Design Frameworkcited by 61
- 2025Deep Learning is Not So Mysterious or Differentcited by 55
- 2017"Found in Translation": Predicting Outcomes of Complex Organic Chemistry Reactions using Neural Sequence-to-Sequence Modelscited by 54
- 2025Training a Scientific Reasoning Model for Chemistrycited by 54
- 2022Regression Transformer: Concurrent sequence regression and generation for molecular language modelingcited by 49
- 2023Enhancing Activity Prediction Models in Drug Discovery with the Ability to Understand Human Languagecited by 44
- 2025Benchmarking Pretrained Molecular Embedding Models For Molecular Representation Learningcited by 15
- 2020Data-Driven Discovery of Molecular Photoswitches with Multioutput Gaussian Processescited by 14
- 2024Rephrasing natural text data with different languages and quality levels for Large Language Model pre-trainingcited by 8
- 2025ChemPile: A 250GB Diverse and Curated Dataset for Chemical Foundation Modelscited by 7
- 2025General-Purpose Models for the Chemical Sciences: LLMs and Beyondcited by 6
Built on by
Not cited yet.