Research map / Architectures and efficiency / RNNs and normalization
Uncovering Neural Scaling Laws in Molecular Representation Learning
Dingshuo Chen, Yanqiao Zhu, Jieyu Zhang, Yuanqi Du, Zhixun Li, Qiang Liu, Shu Wu, Liang Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add GraphMAE pre-training on PCQM4Mv2 before fine-tuning | GIN · small downstream datasets | ROC-AUC | Improved |
| add GraphMAE pre-training | GIN · PCBA, above 40K training samples | ROC-AUC | Hurt |
| pre-training vs from scratch: scaling exponent | GraphMAE-pretrained GIN · downstream scaling curves | power-law exponent | Hurt |
| use random split instead of scaffold or imbalanced split | GIN · scaling curves | power-law exponent | Improved |
| use graph modality instead of SMILES or fingerprint | GIN vs Transformer · 3 tasks | data efficiency | Improved |
| use SMILES strings instead of graph | 1-layer Transformer · MUV | ROC-AUC at large data | Hurt |
| increase GIN capacity from 128 to 1.5K parameters (ablation) | GIN · HIV and PCBA | ROC-AUC | Mixed |
| use smallest GIN (128 parameters) (ablation) | GIN · MUV | ROC-AUC | Improved |
| increase Transformer layers for SMILES (RoBERTa, 1-layer, 3-layer) (ablation) | Transformer · SMILES modality | ROC-AUC | Mixed |
| replace 1-layer Transformer by MLP for fingerprints (ablation) | fingerprint encoder · scaling curves | power-law adherence | No clear effect |
| data pruning with k-means instead of random | GIN · PCBA uniform, 1% subset | ROC-AUC: 5.2 -> 4.1 | Hurt |
| data pruning with Herding instead of random | GIN · PCBA uniform, 1% subset | ROC-AUC: 5.2 -> 3.7 | Hurt |
Closest papers by meaning
- Learning Over Molecular Conformer Ensembles: Datasets and Benchmarks (ICLR 2024 · similarity 0.88)
- Learning Molecular Representation in a Cell (ICLR 2025 · similarity 0.87)
- MOL-Mamba: Enhancing Molecular Representation with Structural & Electronic Insights (AAAI 2025 · similarity 0.87)
- On the Scalability of GNNs for Molecular Graphs (NeurIPS 2024 · similarity 0.87)
- Exploring Molecular Pretraining Model at Scale (NeurIPS 2024 · similarity 0.87)
- MIPT: Multilevel Informed Prompt Tuning for Robust Molecular Property Prediction (ICML 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
