Downstream Task Evaluation on Downstream Tasks Aggregate
60.49AccuracyMeSH
Evaluation Results
| Method | Links | |
|---|---|---|
| MeSHModel Backbone=Pythia-6.9B, Scheme=Recursive (-31.25%), Layers=6+10R2+6, Variant=+mesh, Evaluation Protocol (Few-shot setting)=5-shot2025.10 | 60.49 | |
| Recursive baselineModel Backbone=Pythia-6.9B, Scheme=Recursive (-31.25%), Layers=6+10R2+6, Variant=base, Evaluation Protocol (Few-shot setting)=5-shot2025.10 | 59.43 | |
| MeSHModel Backbone=Pythia-6.9B, Scheme=Recursive (-31.25%), Layers=6+10R2+6, Variant=+mesh, Evaluation Protocol (Few-shot setting)=0-shot2025.10 | 58.83 | |
| MeSHModel Backbone=Pythia-2.8B, Scheme=Recursive (-31.25%), Layers=6+10R2+6, Variant=+mesh, Evaluation Protocol (Few-shot setting)=5-shot2025.10 | 56.85 | |
| Recursive baselineModel Backbone=Pythia-6.9B, Scheme=Recursive (-31.25%), Layers=6+10R2+6, Variant=base, Evaluation Protocol (Few-shot setting)=0-shot2025.10 | 56.67 | |
| Recursive baselineModel Backbone=Pythia-2.8B, Scheme=Recursive (-31.25%), Layers=6+10R2+6, Variant=base, Evaluation Protocol (Few-shot setting)=5-shot2025.10 | 54.92 | |
| MeSHModel Backbone=Pythia-2.8B, Scheme=Recursive (-31.25%), Layers=6+10R2+6, Variant=+mesh, Evaluation Protocol (Few-shot setting)=0-shot2025.10 | 54.71 | |
| SPIRALFORMER-LModel Scale=1.4B, Topology=MeSH, Evaluation Protocol=5-shot2026.02 | 54.37 | |
| Recursive baselineModel Backbone=Pythia-2.8B, Scheme=Recursive (-31.25%), Layers=6+10R2+6, Variant=base, Evaluation Protocol (Few-shot setting)=0-shot2025.10 | 52.49 | |
| SPIRALFORMER-LModel Scale=1.4B, Topology=MeSH, Evaluation Protocol=0-shot2026.02 | 51.75 | |
| BASELINE (PYTHIA)Model Scale=160M, Evaluation Protocol=0-shot2026.02 | 39.88 |