Selective Prediction on GSM8K
90.4PRREvo-Anth-1
Evaluation results
| Method | Links | |
|---|---|---|
| Evo-Anth-1Generator=Claude2026.04 | 90.4 | |
| Evo-Anth-8Generator=Claude2026.04 | 90.4 | |
| Evo-Anth-22Generator=Claude2026.04 | 90.3 | |
| SP×ExpectedRankGenerator=GPT, Optimization=Evolved specifically for selective prediction2026.04 | 90.2 | |
| SP×MeanEntropyGenerator=GPT, Optimization=Evolved specifically for selective prediction2026.04 | 90.2 | |
| Evo-Anth-30Generator=Claude2026.04 | 90.1 | |
| Evo-Anth-15Generator=Claude2026.04 | 90.1 | |
| Evo-Gpt-9Generator=GPT2026.04 | 89.9 | |
| Evo-Gpt-7Generator=GPT2026.04 | 89.9 | |
| SequenceProbability2026.04 | 89.9 | |
| Evo-Gpt-13Generator=GPT2026.04 | 89.8 | |
| Evo-Gpt-5Generator=GPT2026.04 | 89.1 | |
| Evo-Gpt-1Generator=GPT2026.04 | 89 | |
| RAUQ2026.04 | 88.5 | |
| Perplexity2026.04 | 88.4 | |
| MeanTokenEntropy2026.04 | 88.4 | |
| AttentionScore2026.04 | 87.5 | |
| Evo-Gpt-11Generator=GPT2026.04 | 86.7 | |
| Evo-Gpt-3Generator=GPT2026.04 | 86 | |
| Evo-Gpt-15Generator=GPT2026.04 | 85.8 |