List your company

Preference Validation on PPE

57.8AccuracyMRRG
30.13637.31844.551.682Jul 2, 2026

Evaluation results

MethodLinks
MRRGBackbone=GPT-OSS-120B2026.0757.8
MRRGBackbone=GPT-OSS-20B2026.0756.3
Chasing the TailBackbone=GPT-OSS-120B2026.0754.4
MRRGBackbone=Qwen2.5-32B-Instruct2026.0753.9
Chasing the TailBackbone=GPT-OSS-20B2026.0753.2
SVRG w/ SRBackbone=GPT-OSS-120B2026.0752.8
SVRG w/ SRBackbone=GPT-OSS-20B2026.0752.2
Chasing the TailBackbone=Qwen2.5-32B-Instruct2026.0749.5
MRRGBackbone=Qwen2.5-7B-Instruct2026.0748.5
SVRG w/ SRBackbone=Qwen2.5-32B-Instruct2026.0748.2
MRRGBackbone=Qwen2.5-3B-Instruct2026.0746.2
SVRG w/o SRBackbone=Qwen2.5-32B-Instruct2026.0744
SVRG w/o SRBackbone=GPT-OSS-120B2026.0743.3
SVRG w/ SRBackbone=Qwen2.5-7B-Instruct2026.0742.3
SVRG w/o SRBackbone=GPT-OSS-20B2026.0740.1
Chasing the TailBackbone=Qwen2.5-7B-Instruct2026.0737.5
SVRG w/ SRBackbone=Qwen2.5-3B-Instruct2026.0736.3
SVRG w/o SRBackbone=Qwen2.5-7B-Instruct2026.0734
Chasing the TailBackbone=Qwen2.5-3B-Instruct2026.0731.3
SVRG w/o SRBackbone=Qwen2.5-3B-Instruct2026.0731.2