Role-playing Evaluation on CoSER (Mean Rank)
1.59Human Mean RankPersona-Pruner
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Persona-PrunerBase Model=Qwen2.5-32B-Instruct, Sparsity Ratio=50%2026.06 | 1.59 | 1.05 | 1.1 | |
| LLM-PrunerBase Model=Qwen2.5-32B-Instruct, Sparsity Ratio=50%2026.06 | 2.59 | 2.95 | 2.44 | |
| Adapt-PrunerBase Model=Qwen2.5-32B-Instruct, Sparsity Ratio=50%2026.06 | 2.73 | 3.22 | 3.05 | |
| Depth PruningBase Model=Qwen2.5-32B-Instruct, Sparsity Ratio=50%2026.06 | 3.69 | 3.58 | 3.79 | |
| SliceGPTBase Model=Qwen2.5-32B-Instruct, Sparsity Ratio=50%2026.06 | 4.41 | 4.35 | 4.25 |