List your company

Model Utility on IFEval

61.6AccuracyVanilla
16.98428.56740.1551.733May 11, 2026

Evaluation results

MethodLinks
VanillaBackbone=Qwen2.5-7B-Instruct2026.0561.6
embedding disruption methodBackbone=Qwen2.5-7B-Instruct2026.0561.6
Perplexity FilterBackbone=Qwen2.5-7B-Instruct2026.0558.1
RESTABackbone=Qwen2.5-7B-Instruct2026.0557
VanillaBackbone=Vicuna-13B2026.0547
embedding disruption methodBackbone=Vicuna-13B2026.0547
Perplexity FilterBackbone=Vicuna-13B2026.0546.4
VanillaBackbone=LLaMA2-7B-Chat2026.0538.3
Perplexity FilterBackbone=LLaMA2-7B-Chat2026.0537.9
embedding disruption methodBackbone=LLaMA2-7B-Chat2026.0537.5
RESTABackbone=LLaMA2-7B-Chat2026.0536.8
Erase-and-CheckBackbone=Qwen2.5-7B-Instruct2026.0534.4
SmoothLLMBackbone=Qwen2.5-7B-Instruct2026.0531.3
RESTABackbone=Vicuna-13B2026.0527.5
Erase-and-CheckBackbone=Vicuna-13B2026.0526.3
SmoothLLMBackbone=Vicuna-13B2026.0524
Erase-and-CheckBackbone=LLaMA2-7B-Chat2026.0521.4
SmoothLLMBackbone=LLaMA2-7B-Chat2026.0518.7