Reasoning on BIG-Bench Hard Object Counting (OOD)
48.4AccuracyLOVER
Evaluation results
| Method | Links | |
|---|---|---|
| LOVERBackbone=mistral-7b2026.05 | 48.4 | |
| CoT-DecodingBackbone=mistral-7b2026.05 | 48 | |
| Majority VotingBackbone=mistral-7b2026.05 | 46 |
| Method | Links | |
|---|---|---|
| LOVERBackbone=mistral-7b2026.05 | 48.4 | |
| CoT-DecodingBackbone=mistral-7b2026.05 | 48 | |
| Majority VotingBackbone=mistral-7b2026.05 | 46 |