List your company
ResearchFormerly WizWand

Role-Playing on RoleBench (test)

88.82LLM-as-a-Judge ScoreLlama-3.1-8B-Instruct
-2.512821.198644.9168.6214Jun 12, 2026
Updated 4mo ago

Evaluation results

MethodLinks
Llama-3.1-8B-InstructRatio=0%, Base Model=Llama-3.1-8B-Instruct2026.0688.82
Persona-PrunerRatio=25%, Recovery finetuning=w/, Base Model=Llama-3.1-8B-Instruct2026.0684.33
Persona-PrunerRatio=25%, Recovery finetuning=w/, Base Model=Llama-3.2-3B-Instruct2026.0684.26
Llama-3.2-3B-InstructRatio=0%, Base Model=Llama-3.2-3B-Instruct2026.0683.35
Persona-PrunerRatio=25%, Recovery finetuning=w/o, Base Model=Llama-3.1-8B-Instruct2026.0682.62
Persona-PrunerRatio=25%, Recovery finetuning=w/o, Base Model=Llama-3.2-3B-Instruct2026.0681.2
Persona-PrunerRatio=50%, Recovery finetuning=w/, Base Model=Llama-3.1-8B-Instruct2026.0680.01
Persona-PrunerRatio=50%, Recovery finetuning=w/, Base Model=Llama-3.2-3B-Instruct2026.0670.37
Persona-PrunerRatio=50%, Recovery finetuning=w/o, Base Model=Llama-3.1-8B-Instruct2026.0667.86
Persona-PrunerRatio=50%, Recovery finetuning=w/o, Base Model=Llama-3.2-3B-Instruct2026.0652.82
LLM-PrunerRatio=25%, Recovery finetuning=w/o, Base Model=Llama-3.2-3B-Instruct2026.0648.43
LLM-PrunerRatio=25%, Recovery finetuning=w/o, Base Model=Llama-3.1-8B-Instruct2026.0647.98
Adapt-PrunerRatio=25%, Recovery finetuning=w/o, Base Model=Llama-3.2-3B-Instruct2026.0639.65
Depth PruningRatio=25%, Recovery finetuning=w/o, Base Model=Llama-3.1-8B-Instruct2026.0633.56
LLM-PrunerRatio=25%, Recovery finetuning=w/, Base Model=Llama-3.1-8B-Instruct2026.0633.41
Adapt-PrunerRatio=25%, Recovery finetuning=w/o, Base Model=Llama-3.1-8B-Instruct2026.0633.25
LLM-PrunerRatio=50%, Recovery finetuning=w/, Base Model=Llama-3.1-8B-Instruct2026.0630.26
LLM-PrunerRatio=25%, Recovery finetuning=w/, Base Model=Llama-3.2-3B-Instruct2026.0630
Depth PruningRatio=25%, Recovery finetuning=w/, Base Model=Llama-3.2-3B-Instruct2026.0627.75
Adapt-PrunerRatio=25%, Recovery finetuning=w/, Base Model=Llama-3.1-8B-Instruct2026.0627.3
Depth PruningRatio=50%, Recovery finetuning=w/, Base Model=Llama-3.1-8B-Instruct2026.0625.72
Adapt-PrunerRatio=25%, Recovery finetuning=w/, Base Model=Llama-3.2-3B-Instruct2026.0625.66
LLM-PrunerRatio=50%, Recovery finetuning=w/o, Base Model=Llama-3.2-3B-Instruct2026.0625.61
Depth PruningRatio=25%, Recovery finetuning=w/, Base Model=Llama-3.1-8B-Instruct2026.0624.63
LLM-PrunerRatio=50%, Recovery finetuning=w/, Base Model=Llama-3.2-3B-Instruct2026.0623.95
LLM-PrunerRatio=50%, Recovery finetuning=w/o, Base Model=Llama-3.1-8B-Instruct2026.0620.53
Adapt-PrunerRatio=50%, Recovery finetuning=w/, Base Model=Llama-3.2-3B-Instruct2026.0617.58
Depth PruningRatio=50%, Recovery finetuning=w/, Base Model=Llama-3.2-3B-Instruct2026.0616.31
Adapt-PrunerRatio=50%, Recovery finetuning=w/, Base Model=Llama-3.1-8B-Instruct2026.0614.29
SliceGPTRatio=25%, Recovery finetuning=w/o, Base Model=Llama-3.1-8B-Instruct2026.0613.99
Depth PruningRatio=25%, Recovery finetuning=w/o, Base Model=Llama-3.2-3B-Instruct2026.0611.4
SliceGPTRatio=25%, Recovery finetuning=w/, Base Model=Llama-3.2-3B-Instruct2026.0610.95
Adapt-PrunerRatio=50%, Recovery finetuning=w/o, Base Model=Llama-3.2-3B-Instruct2026.069.5
SliceGPTRatio=25%, Recovery finetuning=w/, Base Model=Llama-3.1-8B-Instruct2026.066.76
SliceGPTRatio=25%, Recovery finetuning=w/o, Base Model=Llama-3.2-3B-Instruct2026.065.91
SliceGPTRatio=50%, Recovery finetuning=w/, Base Model=Llama-3.2-3B-Instruct2026.063.76
SliceGPTRatio=50%, Recovery finetuning=w/o, Base Model=Llama-3.1-8B-Instruct2026.061.78
Depth PruningRatio=50%, Recovery finetuning=w/o, Base Model=Llama-3.2-3B-Instruct2026.061.41
SliceGPTRatio=50%, Recovery finetuning=w/o, Base Model=Llama-3.2-3B-Instruct2026.061.35
Depth PruningRatio=50%, Recovery finetuning=w/o, Base Model=Llama-3.1-8B-Instruct2026.061.21
SliceGPTRatio=50%, Recovery finetuning=w/, Base Model=Llama-3.1-8B-Instruct2026.061
Adapt-PrunerRatio=50%, Recovery finetuning=w/o, Base Model=Llama-3.1-8B-Instruct2026.061