Mathematical Reasoning on MetaMathQA
89.57ScoreDA
Evaluation results
| Method | Links | |
|---|---|---|
| DAsource_model=Qwen3-32B, training_dataset=gsm8k2026.02 | 89.57 | |
| PADsource_model=Qwen3-32B, training_dataset=gsm8k2026.02 | 89.15 | |
| PADsource_model=Qwen3-8B, training_dataset=gsm8k2026.02 | 88.48 | |
| PADsource_model=Gemma3-27B-it, training_dataset=gsm8k2026.02 | 88.48 | |
| DAsource_model=Qwen3-8B, training_dataset=gsm8k2026.02 | 88.1 | |
| RSFTsource_model=Qwen3-8B, training_dataset=gsm8k2026.02 | 87.72 | |
| RSFTsource_model=Gemma3-27B-it, training_dataset=gsm8k2026.02 | 87.72 | |
| DAsource_model=Gemma3-27B-it, training_dataset=gsm8k2026.02 | 86.2 | |
| RSFTsource_model=Qwen3-32B, training_dataset=gsm8k2026.02 | 85.6 | |
| RSFTsource model=qwen3-8b, train dataset=gsm8k, target model=Gemma-7B2026.02 | 75.28 | |
| DAsource model=qwen3-8b, train dataset=gsm8k, target model=Gemma-7B2026.02 | 74.75 | |
| DAsource_model=qwen3-32b, train_dataset=gsm8k, base_model=Qwen3-1.7B2026.02 | 73.89 | |
| DAsource model=qwen3-32b, train dataset=gsm8k, target model=Gemma-7B2026.02 | 73.78 | |
| RSFTsource model=gemma3-27b-it, train dataset=gsm8k, target model=Gemma-7B2026.02 | 73.46 | |
| DAsource model=gemma3-27b-it, train dataset=gsm8k, target model=Gemma-7B2026.02 | 71.34 | |
| PADsource_model=qwen3-32b, train_dataset=gsm8k, base_model=Qwen3-1.7B2026.02 | 70.28 | |
| PADsource_model=gemma3-27b-it, train_dataset=gsm8k, base_model=Qwen3-1.7B2026.02 | 69.98 | |
| DAsource_model=gemma3-27b-it, train_dataset=gsm8k, base_model=Qwen3-1.7B2026.02 | 69.94 | |
| RSFTsource_model=qwen3-32b, train_dataset=gsm8k, base_model=Qwen3-1.7B2026.02 | 69.85 | |
| PADsource_model=qwen3-8b, train_dataset=gsm8k, base_model=Qwen3-1.7B2026.02 | 69.78 | |
| RSFTsource_model=qwen3-8b, train_dataset=gsm8k, base_model=Qwen3-1.7B2026.02 | 69.67 | |
| DAsource_model=qwen3-8b, train_dataset=gsm8k, base_model=Qwen3-1.7B2026.02 | 69.29 | |
| RSFTsource_model=gemma3-27b-it, train_dataset=gsm8k, base_model=Qwen3-1.7B2026.02 | 69.07 | |
| RSFTsource model=qwen3-32b, train dataset=gsm8k, target model=Gemma-7B2026.02 | 67.25 | |
| PADsource model=qwen3-32b, train dataset=gsm8k, target model=Gemma-7B2026.02 | 63.6 | |
| PADsource_model=Qwen3-32B, training_dataset=medmcqa2026.02 | 61.53 | |
| PADsource_model=Gemma3-27B-it, training_dataset=medmcqa2026.02 | 59.86 | |
| RSFTsource_model=Gemma3-27B-it, training_dataset=medmcqa2026.02 | 59.77 | |
| PADsource_model=Qwen3-8B, training_dataset=medmcqa2026.02 | 59.71 | |
| RSFTsource_model=Qwen3-8B, training_dataset=medmcqa2026.02 | 59.6 | |
| DAsource_model=Qwen3-32B, training_dataset=medmcqa2026.02 | 58.33 | |
| RSFTsource_model=Qwen3-32B, training_dataset=medmcqa2026.02 | 57.95 | |
| DAsource_model=Qwen3-8B, training_dataset=medmcqa2026.02 | 57.49 | |
| DAsource_model=Gemma3-27B-it, training_dataset=medmcqa2026.02 | 57.18 | |
| PADsource model=gemma3-27b-it, train dataset=gsm8k, target model=Gemma-7B2026.02 | 53.24 | |
| PADsource model=qwen3-8b, train dataset=gsm8k, target model=Gemma-7B2026.02 | 52.34 | |
| PADsource model=gemma3-27b-it, train dataset=medmcqa, target model=Gemma-7B2026.02 | 50.77 | |
| RSFTsource model=qwen3-32b, train dataset=medmcqa, target model=Gemma-7B2026.02 | 49.76 | |
| PADsource model=qwen3-32b, train dataset=medmcqa, target model=Gemma-7B2026.02 | 49.61 | |
| PADsource model=qwen3-8b, train dataset=medmcqa, target model=Gemma-7B2026.02 | 49.37 | |
| DAsource model=qwen3-32b, train dataset=medmcqa, target model=Gemma-7B2026.02 | 48.74 | |
| RSFTsource model=gemma3-27b-it, train dataset=medmcqa, target model=Gemma-7B2026.02 | 45.66 | |
| RSFTsource model=qwen3-8b, train dataset=medmcqa, target model=Gemma-7B2026.02 | 45.04 | |
| RSFTsource_model=qwen3-32b, train_dataset=medmcqa, base_model=Qwen3-1.7B2026.02 | 44.32 | |
| RSFTsource_model=qwen3-8b, train_dataset=medmcqa, base_model=Qwen3-1.7B2026.02 | 43.22 | |
| DAsource_model=qwen3-32b, train_dataset=medmcqa, base_model=Qwen3-1.7B2026.02 | 43.03 | |
| DAsource model=qwen3-8b, train dataset=medmcqa, target model=Gemma-7B2026.02 | 42.84 | |
| DAsource_model=qwen3-8b, train_dataset=medmcqa, base_model=Qwen3-1.7B2026.02 | 42.65 | |
| DAsource_model=gemma3-27b-it, train_dataset=medmcqa, base_model=Qwen3-1.7B2026.02 | 42.58 | |
| PADsource_model=gemma3-27b-it, train_dataset=medmcqa, base_model=Qwen3-1.7B2026.02 | 42.46 | |
| PADsource_model=qwen3-32b, train_dataset=medmcqa, base_model=Qwen3-1.7B2026.02 | 42.43 | |
| PADsource_model=qwen3-8b, train_dataset=medmcqa, base_model=Qwen3-1.7B2026.02 | 42.41 | |
| DAsource model=gemma3-27b-it, train dataset=medmcqa, target model=Gemma-7B2026.02 | 42.34 | |
| RSFTsource_model=gemma3-27b-it, train_dataset=medmcqa, base_model=Qwen3-1.7B2026.02 | 41.84 |