Research map / Image generation / Autoregressive and latent generation
InertialAR: Autoregressive 3D Molecule Generation with Inertial Frames
Haorui Li, weitao Du, Yuqiang Li, Hongyu Guo, Shengchao 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 GeoPE (ManhattanPE and EuclideanPE) to attention (ablation) | InertialAR · QM9 unconditional | generation quality | Improved |
| use Diffusion Loss instead of direct L2 regression for coordinates (ablation) | InertialAR · QM9 unconditional | generation quality | Improved |
| add deterministic canonical atom reordering (ablation) | InertialAR · QM9 unconditional | validity and uniqueness | Improved |
| use inertial-frame canonical tokenization with hierarchical AR instead of Geo2Seq | Transformer AR · B3LYP-1M | Molecule stability (%): 0.0 -> 24.7 | Improved |
| replace Geo2Seq with InertialAR on QM9 | Transformer AR · QM9 | Valid&Uni (%): 92.5 vs 81.7 favors InertialAR; but below UniGEM 93.2 | Hurt |
| replace InertialAR with GeoLDM on GEOM-Drugs | InertialAR · GEOM-Drugs | Valid (%): 98.0 vs GeoLDM 99.3 | Hurt |
| apply classifier-free guidance with class labels | InertialAR · QM9 class-conditional | Hit rate (%): average 83.3 | Improved |
| raise CFG guidance scale by 0.3 (ablation) | InertialAR · QM9 editing | validity and functional group compliance | Improved |
Closest papers by meaning
- NEAT: Neighborhood-Guided, Efficient, Autoregressive Set Transformer for 3D Molecular Generation (NeurIPS 2026 · similarity 0.90)
- 3D Molecule Generation from Rigid Motifs via $\mathrm{SE}(3)$ Flows (NeurIPS 2026 · similarity 0.88)
- Beyond Atoms: Enhancing Molecular Pretrained Representations with 3D Space Modeling (ICML 2025 · similarity 0.88)
- All-atom Diffusion Transformers: Unified generative modelling of molecules and materials (ICML 2025 · similarity 0.88)
- Symphony: Symmetry-Equivariant Point-Centered Spherical Harmonics for 3D Molecule Generation (ICLR 2024 · similarity 0.88)
- Towards Unified and Lossless Latent Space for 3D Molecular Latent Diffusion Modeling (NeurIPS 2025 · similarity 0.87)
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.
