Research map / Graph learning / Graph self-supervised learning
HSA-Net: Hierarchical and Structure-Aware Framework for Efficient and Scalable Molecular Language Modeling
Zihang Shao, Wentao Lei, Lei Wang, Wen-Cai Ye, Li Liu
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 Graph-Mamba projector to cross-attention projector (no SAF) (ablation) | HSA-Net · Molecule Description (generalist ablation) | BLEU: 37.9 -> 38.8 | Improved |
| add Graph-Mamba projector to cross-attention projector (no SAF) (ablation) | HSA-Net · Molecule Description (generalist ablation) | METEOR: 67.1 -> 68.5 | Improved |
| add SAF module to cross-attention projector (ablation) | HSA-Net · Molecule Description (generalist ablation) | BLEU: 37.9 -> 40.0 | Improved |
| add SAF module to Graph-Mamba projector (ablation) | HSA-Net · Molecule Description (generalist ablation) | BLEU: 37.5 -> 39.4 | Improved |
| add SAF module to Attention + Mamba (ablation) | HSA-Net · Molecule Description (generalist ablation) | BLEU: 38.8 -> 43.5 | Improved |
| add SAF module to Attention + Mamba (ablation) | HSA-Net · Molecule Description (generalist ablation) | METEOR: 68.5 -> 72.1 | Improved |
| add cross-attention projector to Mamba + SAF (ablation) | HSA-Net · Molecule Description (generalist ablation) | BLEU: 39.4 -> 43.5 | Improved |
| add Graph-Mamba projector to Attention + SAF (ablation) | HSA-Net · Molecule Description (generalist ablation) | METEOR: 69.2 -> 72.1 | Improved |
| swap cross-attention for Graph-Mamba projector (no SAF) (ablation) | HSA-Net · Molecule Description (generalist ablation) | BLEU: 37.9 -> 37.5 | Mixed |
| use HSA-Net instead of LLaMo | generalist LLaMA2-7B · Property Prediction (generalist) | MAE: 0.0061 -> 0.0049 | Improved |
| use HSA-Net instead of LLaMo | specialist Galactica-1.3B · IUPAC Prediction (specialist) | METEOR: 70.1 -> 72.8 | Improved |
Closest papers by meaning
- LLaMo: Large Language Model-based Molecular Graph Assistant (NeurIPS 2024 · similarity 0.88)
- Improving Large Molecular Language Model via Relation-aware Multimodal Collaboration (AAAI 2026 · similarity 0.87)
- Hierarchical Graph Tokenization for Molecule-Language Alignment (ICML 2025 · similarity 0.86)
- Bridging Molecular Graphs and Large Language Models (AAAI 2025 · similarity 0.86)
- MOL-Mamba: Enhancing Molecular Representation with Structural & Electronic Insights (AAAI 2025 · similarity 0.85)
- On the Scalability of GNNs for Molecular Graphs (NeurIPS 2024 · similarity 0.85)
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.
