Mathematical Reasoning (Missing Information Detection) on GSM8K v1 (test)
96.19Overall AccuracyRT-ICA + GPT-5
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| RT-ICA + GPT-5Base Model=GPT-5, Framework=RT-ICA2025.12 | 96.19 | 94.23 | 98.11 | |
| RT-ICA + DeepSeek-V3Base Model=DeepSeek-V3, Framework=RT-ICA2025.12 | 94.29 | 92.31 | 96.23 | |
| GPT-5Base Model=GPT-52025.12 | 93.33 | 90.38 | 96.23 | |
| DeepSeek-V3Base Model=DeepSeek-V32025.12 | 80.95 | 84.62 | 77.36 | |
| RT-ICA + GPT-3.5-turboBase Model=GPT-3.5-turbo, Framework=RT-ICA2025.12 | 72.38 | 82.69 | 62.26 | |
| GPT-3.5-turboBase Model=GPT-3.5-turbo2025.12 | 44.76 | 30.77 | 58.49 |