Reaction Feasibility Prediction on FREA Overall 1.0 (test)
99.9AccuracyTheoretical Upper Bound
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Theoretical Upper Bound2026.05 | 99.9 | 0.998 | |
| ARMORCategory=LLM-based Methods, Backbone=T3Q-Qwen-14B2026.05 | 91.62 | 0.8323 | |
| HarderMoECategory=Dynamic Methods2026.05 | 89.5 | 0.7919 | |
| BERTCategory=Single-Tool Methods2026.05 | 87.9 | 0.7669 | |
| ToolEyesCategory=LLM-based Methods2026.05 | 87.25 | 0.7536 | |
| Weighted VotingCategory=Statistical Methods2026.05 | 86.35 | 0.7304 | |
| Claude-Sonnet-4.6Category=LLM-based Methods2026.05 | 85.55 | 0.7288 | |
| Majority VotingCategory=Statistical Methods2026.05 | 82.88 | 0.6577 | |
| KNORA-ECategory=Dynamic Methods2026.05 | 82.88 | 0.6577 | |
| MolecularCategory=Single-Tool Methods2026.05 | 82.3 | 0.6621 | |
| ChemformerCategory=Single-Tool Methods2026.05 | 81.88 | 0.6628 | |
| KNORA-UCategory=Dynamic Methods2026.05 | 81.85 | 0.6373 | |
| DeepSeek-v4-flashCategory=LLM-based Methods2026.05 | 80.9 | 0.6198 | |
| DRFPCategory=Single-Tool Methods2026.05 | 80.62 | 0.6398 | |
| GPT-5.4-miniCategory=LLM-based Methods2026.05 | 75.98 | 0.5352 | |
| DES-KNNCategory=Dynamic Methods2026.05 | 75.8 | 0.5161 | |
| StaticSelCategory=Statistical Methods2026.05 | 74.12 | 0.4882 | |
| DES-ClusteringCategory=Dynamic Methods2026.05 | 72.2 | 0.4591 | |
| MolecularT3QCategory=Single-Tool Methods, Backbone=T3Q2026.05 | 71.93 | 0.4442 | |
| ChemformerT3QCategory=Single-Tool Methods, Backbone=T3Q2026.05 | 69.98 | 0.4085 | |
| Random SelectionCategory=Statistical Methods2026.05 | 67.48 | 0.3497 | |
| ICLT3QCategory=Single-Tool Methods, Backbone=T3Q2026.05 | 63.38 | 0.277 | |
| MolecularLlamaCategory=Single-Tool Methods, Backbone=Llama2026.05 | 60.93 | 0.2494 | |
| ChemformerLlamaCategory=Single-Tool Methods, Backbone=Llama2026.05 | 60.55 | 0.2645 | |
| PromptingT3QCategory=Single-Tool Methods, Backbone=T3Q2026.05 | 58.07 | 0.1618 | |
| ICLLlamaCategory=Single-Tool Methods, Backbone=Llama2026.05 | 55.12 | 0.1426 | |
| PromptingLlamaCategory=Single-Tool Methods, Backbone=Llama2026.05 | 54.02 | 0.1017 | |
| Dora_xgbCategory=Single-Tool Methods2026.05 | 50.8 | 0.0196 |