List your company

LLM Alignment on Evol-Instruct

51.4Win RateHighAvg.
14.89624.37333.8543.327May 29, 2025

Evaluation results

MethodLinks
HighAvg.Backbone=Mistral-7B, % Train=33%, Alignment Method=DPO2025.0551.46.56
FullBackbone=Mistral-7B, % Train=100%, Alignment Method=DPO2025.0549.56.5
HighAvg.Backbone=LLaMA-3-8B, % Train=33%, Alignment Method=SimPO2025.0547.96.54
RandomBackbone=Mistral-7B, % Train=33%, Alignment Method=DPO2025.0547.56.54
LowAvg.Backbone=Mistral-7B, % Train=33%, Alignment Method=DPO2025.0546.86.43
FullBackbone=LLaMA-3-8B, % Train=100%, Alignment Method=DPO2025.0546.16.64
FullBackbone=LLaMA-3-8B, % Train=100%, Alignment Method=SimPO2025.0544.76.36
RandomBackbone=LLaMA-3-8B, % Train=33%, Alignment Method=DPO2025.0542.56.13
HighAvg.Backbone=LLaMA-3-8B, % Train=33%, Alignment Method=DPO2025.0542.26.28
LowAvg.Backbone=LLaMA-3-8B, % Train=33%, Alignment Method=DPO2025.0539.46.13
HighAvg.Backbone=Mistral-7B, % Train=33%, Alignment Method=SimPO2025.0538.55.71
HighVar.Backbone=LLaMA-3-8B, % Train=33%, Alignment Method=DPO2025.0535.66.11
HighVar.Backbone=Mistral-7B, % Train=33%, Alignment Method=DPO2025.0533.96.07
RandomBackbone=LLaMA-3-8B, % Train=33%, Alignment Method=SimPO2025.0530.25.91
LowAvg.Backbone=LLaMA-3-8B, % Train=33%, Alignment Method=SimPO2025.0530.25.93
FullBackbone=Mistral-7B, % Train=100%, Alignment Method=SimPO2025.0527.55.65
HighVar.Backbone=LLaMA-3-8B, % Train=33%, Alignment Method=SimPO2025.0526.16.02
ZeroshotBackbone=LLaMA-3-8B, % Train=0%, Alignment Method=DPO2025.0523.45.8
ZeroshotBackbone=LLaMA-3-8B, % Train=0%, Alignment Method=SimPO2025.0523.45.8
RandomBackbone=Mistral-7B, % Train=33%, Alignment Method=SimPO2025.0517.75.08
HighVar.Backbone=Mistral-7B, % Train=33%, Alignment Method=SimPO2025.0516.75.46
ZeroshotBackbone=Mistral-7B, % Train=0%, Alignment Method=DPO2025.0516.34.87
ZeroshotBackbone=Mistral-7B, % Train=0%, Alignment Method=SimPO2025.0516.34.87
LowAvg.Backbone=Mistral-7B, % Train=33%, Alignment Method=SimPO2025.0516.35.1