Forward Prediction on Materials Discovery (In-Domain)
51.2Overall ScoreARIA-FULL
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| ARIA-FULL2026.06 | 51.2 | 25.4 | 33.9 | 93 | 92.5 | |
| ARIA-FULL2026.06 | 51.2 | — | — | — | — | |
| ARIA-SEARCH2026.06 | 51.1 | 27.5 | 38.2 | 79.9 | 81.6 | |
| Self-RAG2026.06 | 45.3 | — | — | — | — | |
| ARIA-CORE2026.06 | 41 | 23.5 | 40 | 74.9 | 79.1 | |
| Baseline LLM2026.06 | 34 | 25.2 | 38.5 | 65.9 | 87.7 | |
| NAIVE-KG2026.06 | 33.7 | 28 | 39.4 | 73.2 | 35 | |
| KG-ONLY2026.06 | 23.1 | 6.8 | 14.3 | 60 | 35 |