Mathematical Reasoning on GSM8K (Accuracy Improvement)
5.7Accuracy Improvement (GSM8K)Ultra-Dense Prompting
Evaluation results
| Method | Links | |
|---|---|---|
| Ultra-Dense PromptingModel=Gemini 2.02026.04 | 5.7 | |
| Ultra-Dense PromptingModel=Claude 3.72026.04 | 5.1 | |
| Ultra-Dense PromptingModel=Average across Models2026.04 | 4.9 | |
| Ultra-Dense PromptingModel=GPT-4o2026.04 | 4.8 | |
| Ultra-Dense PromptingModel=GPT-4o-mini2026.04 | 4.2 |