Reasoning on 7 reasoning datasets
65.74Reasoning AccuracyPASER
Evaluation results
| Method | Links | |
|---|---|---|
| PASERPruning=SliceGPT ratio=25%, Base Model=LLaMA2-7B2025.02 | 65.74 | |
| PASERPruning=LLM-Pruner ratio=25%, Base Model=LLaMA2-7B2025.02 | 63.79 | |
| w/o TrainingPruning=w/o pruning, Base Model=LLaMA2-7B2025.02 | 62.91 | |
| Full DataPruning=LLM-Pruner ratio=25%, Base Model=LLaMA2-7B2025.02 | 62.68 | |
| NuggetsPruning=LLM-Pruner ratio=25%, Base Model=LLaMA2-7B2025.02 | 60.99 | |
| NuggetsPruning=SliceGPT ratio=25%, Base Model=LLaMA2-7B2025.02 | 60.96 | |
| RandomPruning=LLM-Pruner ratio=25%, Base Model=LLaMA2-7B2025.02 | 60.93 | |
| Instruction MiningPruning=LLM-Pruner ratio=25%, Base Model=LLaMA2-7B2025.02 | 60.76 | |
| IFDPruning=LLM-Pruner ratio=25%, Base Model=LLaMA2-7B2025.02 | 60.65 | |
| IFDPruning=SliceGPT ratio=25%, Base Model=LLaMA2-7B2025.02 | 58.94 | |
| Full DataPruning=SliceGPT ratio=25%, Base Model=LLaMA2-7B2025.02 | 58.31 | |
| w/o TrainingPruning=LLM-Pruner ratio=25%, Base Model=LLaMA2-7B2025.02 | 57.78 | |
| Instruction MiningPruning=SliceGPT ratio=25%, Base Model=LLaMA2-7B2025.02 | 57.53 | |
| RandomPruning=SliceGPT ratio=25%, Base Model=LLaMA2-7B2025.02 | 56.68 | |
| w/o TrainingPruning=SliceGPT ratio=25%, Base Model=LLaMA2-7B2025.02 | 54.27 |