Research map / Large language models / Language, safety and interpretability
PLaID++: A Preference Aligned Language Model for Targeted Inorganic Materials Design
Andy Xu, Rohan Desai, Larry Wang, Ethan T. Ritz, Gabriel Hope
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 |
|---|---|---|---|
| use Wyckoff text instead of 3D coordinates (ablation) | PLaID++ SFT · MP-20 unconditional | S.U.N. (%): 2.81 -> 3.58 | Improved |
| use Wyckoff text instead of 3D coordinates (ablation) | PLaID++ DPO · MP-20 unconditional | S.U.N. (%): 1.69 -> 6.25 | Improved |
| run more DPO iterations on 3D coordinate representation (ablation) | PLaID++ 3D Coord · MP-20 unconditional | S.U.N. (%): 2.93 -> 2.23 -> 1.69 | Hurt |
| run more DPO iterations on 3D coordinate representation (ablation) | PLaID++ 3D Coord · MP-20 unconditional | stability (%): 10.35 -> 13.47 -> 15.94 | Improved |
| use DPO instead of iterative SFT (ablation) | PLaID++ Wyckoff · MP-20 unconditional | stability (%): 9.85 -> 15.59 | Improved |
| add DPO to Wyckoff SFT model (ablation) | PLaID++ Wyckoff · MP-20 unconditional | S.U.N. (%): 3.58 -> 6.25 | Improved |
| add dynamic temperature adjustment and joint training (ablation) | PLaID++ Stability + Novelty · MP-20 unconditional | stability (%): 16.72 -> 22.27 | Improved |
| add dynamic temperature adjustment and joint training (ablation) | PLaID++ Stability + Novelty · MP-20 unconditional | S.U.N. (%): 6.22 -> 7.74 | Improved |
| train DPO jointly on bulk modulus, stability, novelty and space group instead of bulk modulus only (ablation) | PLaID++ · Bulk modulus >325 GPa | count high BM & S.U.N.: 28 -> 40 | Improved |
| add DPO on space-group pairs plus unconditional data (ablation) | PLaID++ Wyckoff · Space group conditioned generation | S.S.U.N.: 22% -> 47% | Improved |
| apply DPO on low-data space groups (Amm2, P6_3/mmc) | PLaID++ · Space group conditioned generation | S.S.U.N. | No clear effect |
| apply one DPO iteration for band gap preference | PLaID++ · Band gap >3 eV | S.U.N. count: 0 -> 3 | Improved |
Later papers that compared with this method
1 later accepted paper reports PLAID as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- A Dense Subset Index for Collective Query Coverage (ICLR 2026 · better on 2 of 2 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- FlowLLM — FlowLLM: Flow Matching for Material Generation with Large Language Models as Base Distributions (NeurIPS 2024) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Fine-Tuned Language Models Generate Stable Inorganic Materials as Text (ICLR 2024 · similarity 0.90)
- Generative Hierarchical Materials Search (NeurIPS 2024 · similarity 0.90)
- FlowLLM: Flow Matching for Material Generation with Large Language Models as Base Distributions (NeurIPS 2024 · similarity 0.89)
- WyckoffDiff -- A Generative Diffusion Model for Crystal Symmetry (ICML 2025 · similarity 0.88)
- LLM Meets Diffusion: A Hybrid Framework for Crystal Material Generation (NeurIPS 2025 · similarity 0.88)
- All that structure matches does not glitter (NeurIPS 2025 · similarity 0.88)
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.
