Math Word Problems on MWPs
83R ScoreLlama-3-70B
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Llama-3-70BProtocol=Direct Ask, Backbone=Llama-3-70B2025.11 | 83 | 32.5 | |
| ChatProtectProtocol=Detection → Revision (with rationale), Backbone=GPT-3.5-turbo2025.11 | 80.5 | 37.9 | |
| HalluCleanProtocol=Detection → Revision (with rationale), Backbone=GPT-3.5-turbo2025.11 | 75.5 | 45 | |
| HalluCleanProtocol=Detection → Revision (with rationale), Backbone=DeepSeek-V32025.11 | 75.5 | 41 | |
| Plan-and-SolveProtocol=Detection → Revision (with rationale), Backbone=GPT-3.5-turbo2025.11 | 53 | 44.3 | |
| GPT-4o-miniProtocol=Direct Ask, Backbone=GPT-4o-mini2025.11 | 47.5 | 14 | |
| Step-by-StepProtocol=Detection → Revision (with rationale), Backbone=GPT-3.5-turbo2025.11 | 44 | 37.5 | |
| GPT-3.5-turboProtocol=Direct Ask, Backbone=GPT-3.5-turbo2025.11 | 42 | 13 | |
| DeepSeek-R1Protocol=Direct Ask, Backbone=DeepSeek-R12025.11 | 42 | 28 | |
| DeepSeek-V3Protocol=Direct Ask, Backbone=DeepSeek-V32025.11 | 40 | 17 |