Goal Prediction on Gather OOD
46.7Classification AccuracyT-CoT
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| T-CoTSize=4, Backbone=GPT-4o2026.06 | 46.7 | 13.3 | |
| SVoTpSize=4, Backbone=Anole2026.06 | 46.7 | 16.7 | |
| SVoToSize=4, Backbone=Anole2026.06 | 43.3 | 10 | |
| T-CoTSize=4, Backbone=Anole2026.06 | 36.7 | 0 | |
| MVoTSize=4, Backbone=Anole2026.06 | 36.7 | 3.3 | |
| T-CoTSize=5, Backbone=GPT-4o2026.06 | 33.3 | 0 | |
| SVoTpSize=5, Backbone=Anole2026.06 | 33.3 | 1.7 | |
| MVoTSize=5, Backbone=Anole2026.06 | 30 | 0 | |
| SVoToSize=5, Backbone=Anole2026.06 | 30 | 0 | |
| SVoTpSize=6, Backbone=Anole2026.06 | 30 | 1.7 | |
| T-CoTSize=6, Backbone=GPT-4o2026.06 | 26.7 | 0 | |
| T-CoTSize=5, Backbone=Anole2026.06 | 25 | 0 | |
| T-CoTSize=6, Backbone=Anole2026.06 | 23.3 | 0 | |
| SVoToSize=6, Backbone=Anole2026.06 | 23.3 | 0 | |
| MVoTSize=6, Backbone=Anole2026.06 | 20 | 0 |