Aggregate Cognitive Performance on Average
76.6Human-Model Similarity (1-nWD)TASKPR
Evaluation Results
| Method | Links | |
|---|---|---|
| TASKPRModel Backbone=Llama 3 8B2026.05 | 76.6 | |
| TASKPRModel Backbone=Claude Opus 4.62026.05 | 60.3 | |
| COMPACTORModel Backbone=Claude Opus 4.62026.05 | 6.9 | |
| COMPACTORModel Backbone=Llama 3 8B2026.05 | 5.4 | |
| MEMPRModel Backbone=Llama 3 8B2026.05 | 1.6 | |
| HUMPRModel Backbone=Llama 3 8B2026.05 | 1 | |
| HUMPRModel Backbone=Claude Opus 4.62026.05 | 0.4 | |
| MEMPRModel Backbone=Claude Opus 4.62026.05 | 0 |