RO reformulation on Hard (Out-of-Distribution)
94.8AccuracyAutoREM
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| AutoREMBase LLM=DeepSeek-V4-Flash2026.05 | 94.8 | 6,944 | |
| Max ThinkingBase LLM=DeepSeek-V4-Flash2026.05 | 83.3 | 14,902 | |
| Expert PromptBase LLM=DeepSeek-V4-Flash2026.05 | 83.3 | 7,549 | |
| ACEBase LLM=DeepSeek-V4-Flash2026.05 | 81.3 | 5,238 | |
| ReasoningBankBase LLM=DeepSeek-V4-Flash2026.05 | 80.2 | 8,089 | |
| Base LLMBase LLM=DeepSeek-V4-Flash2026.05 | 70.8 | 9,026 |