List your company
ResearchFormerly WizWand

Instruction Following on Evol-Inst

92.2Win RateTeacher
16.581636.213355.84575.4767May 29, 2025Jul 21, 2025Sep 12, 2025Nov 5, 2025Dec 28, 2025Feb 19, 2026Apr 14, 2026
Updated 5mo ago

Evaluation results

MethodLinks
TeacherModel=Qwen2.5-14B-Instruct2025.1092.2—
TeacherModel=Qwen2.5-7B-Instruct, Role=Teacher2025.1089.6—
DistiLLM-2α=-5, Teacher Model=Qwen2.5-14B-Instruct, Student Model=Qwen2.5-1.5B-Instruct2025.1071.1—
DistiLLM-2alpha=-5, Base Model=Qwen2.5-1.5B-Instruct2025.1071—
DistiLLM-2α=-1, Teacher Model=Qwen2.5-14B-Instruct, Student Model=Qwen2.5-1.5B-Instruct2025.1070.4—
MT (Teacher)Distillation Pair (Teacher Model -> Student Model)=Qwen2.5-7B-Inst -> Qwen2.5-1.5B2026.0469.67—
DistiLLM-2alpha=-1, Base Model=Qwen2.5-1.5B-Instruct2025.1069—
MT (Teacher)Distillation Pair (Teacher Model -> Student Model)=Qwen2-7B-Inst -> Qwen2-1.5B2026.0459.75—
VCRDDistillation Pair (Teacher Model -> Student Model)=Qwen2.5-7B-Inst -> Qwen2.5-1.5B2026.0457.8—
DistillLM-2Distillation Pair (Teacher Model -> Student Model)=Qwen2.5-7B-Inst -> Qwen2.5-1.5B2026.0456.77—
DistilLLMDistillation Pair (Teacher Model -> Student Model)=Qwen2.5-7B-Inst -> Qwen2.5-1.5B2026.0455.38—
KDDistillation Pair (Teacher Model -> Student Model)=Qwen2.5-7B-Inst -> Qwen2.5-1.5B2026.0454.59—
StudentModel=Qwen2.5-1.5B-Instruct2025.1047.2—
SeqKDDistillation Pair (Teacher Model -> Student Model)=Qwen2.5-7B-Inst -> Qwen2.5-1.5B2026.0446.33—
StudentModel=Qwen2.5-1.5B-Instruct, Role=Student2025.1046.2—
VCRDDistillation Pair (Teacher Model -> Student Model)=Qwen2-7B-Inst -> Qwen2-1.5B2026.0444.49—
DistillLM-2Distillation Pair (Teacher Model -> Student Model)=Qwen2-7B-Inst -> Qwen2-1.5B2026.0437.27—
DistilLLMDistillation Pair (Teacher Model -> Student Model)=Qwen2-7B-Inst -> Qwen2-1.5B2026.0435.66—
KDDistillation Pair (Teacher Model -> Student Model)=Qwen2-7B-Inst -> Qwen2-1.5B2026.0432.45—
MS (Student)Distillation Pair (Teacher Model -> Student Model)=Qwen2.5-7B-Inst -> Qwen2.5-1.5B2026.0430.73—
SeqKDDistillation Pair (Teacher Model -> Student Model)=Qwen2-7B-Inst -> Qwen2-1.5B2026.0428.32—
DPOOptimization Method=DPO, % Train=100%, Train set=Full, Backbone=LLaMA-3-8B2025.0526.85.93
SimPOOptimization Method=SimPO, % Train=33%, Train set=Random, Backbone=LLaMA-3-8B2025.0526.65.97
SimPOOptimization Method=SimPO, % Train=33%, Train set=LowAvg., Backbone=LLaMA-3-8B, Data Map Partition=Alignment Data Map2025.0526.15.92
SimPOOptimization Method=SimPO, % Train=100%, Train set=Full, Backbone=LLaMA-3-8B2025.0526.15.88
DPOOptimization Method=DPO, % Train=33%, Train set=LowAvg., Backbone=LLaMA-3-8B, Data Map Partition=Alignment Data Map2025.0525.95.91
SimPOOptimization Method=SimPO, % Train=33%, Train set=HighVar., Backbone=LLaMA-3-8B, Data Map Partition=Alignment Data Map2025.0525.95.94
DPOOptimization Method=DPO, % Train=33%, Train set=HighAvg., Backbone=LLaMA-3-8B, Data Map Partition=Alignment Data Map2025.0525.75.97
SimPOOptimization Method=SimPO, % Train=33%, Train set=HighAvg., Backbone=LLaMA-3-8B, Data Map Partition=Alignment Data Map2025.0525.55.94
DPOOptimization Method=DPO, % Train=33%, Train set=HighVar., Backbone=LLaMA-3-8B, Data Map Partition=Alignment Data Map2025.0525.25.92
DPOOptimization Method=DPO, % Train=33%, Train set=Random, Backbone=LLaMA-3-8B2025.05255.9
DPOOptimization Method=DPO, % Train=0%, Train set=Zeroshot, Backbone=LLaMA-3-8B2025.0523.45.8
SimPOOptimization Method=SimPO, % Train=0%, Train set=Zeroshot, Backbone=LLaMA-3-8B2025.0523.45.8
MS (Student)Distillation Pair (Teacher Model -> Student Model)=Qwen2-7B-Inst -> Qwen2-1.5B2026.0419.49—