List your company
ResearchFormerly WizWand

Mathematical Reasoning on GSM8K ID (test)

54.59AccuracyWanda-Struct
8.497220.463632.4344.3964May 9, 2026
Updated 5mo ago

Evaluation results

MethodLinks
Wanda-StructSparsity=30%, Backbone=Qwen-2.5-7B, Fine-tuning steps=150, PEFT Method=LoRA2026.0554.59
AGFSparsity=30%, Backbone=Qwen-2.5-7B, Fine-tuning steps=150, PEFT Method=LoRA2026.0546.1
RKUSparsity=30%, Backbone=Qwen-2.5-7B, Fine-tuning steps=150, PEFT Method=LoRA2026.0545.53
Taylor-FOSparsity=30%, Backbone=Qwen-2.5-7B, Fine-tuning steps=150, PEFT Method=LoRA2026.0540.45
Wanda-StructSparsity=40%, Backbone=Qwen-2.5-7B, Fine-tuning steps=150, PEFT Method=LoRA2026.0539.95
AGFSparsity=40%, Backbone=Qwen-2.5-7B, Fine-tuning steps=150, PEFT Method=LoRA2026.0533.06
RKUSparsity=40%, Backbone=Qwen-2.5-7B, Fine-tuning steps=150, PEFT Method=LoRA2026.0530.48
Wanda-StructSparsity=50%, Backbone=Qwen-2.5-7B, Fine-tuning steps=150, PEFT Method=LoRA2026.0529.49
Taylor-FOSparsity=40%, Backbone=Qwen-2.5-7B, Fine-tuning steps=150, PEFT Method=LoRA2026.0523.01
AGFSparsity=50%, Backbone=Qwen-2.5-7B, Fine-tuning steps=150, PEFT Method=LoRA2026.0518.57
RKUSparsity=50%, Backbone=Qwen-2.5-7B, Fine-tuning steps=150, PEFT Method=LoRA2026.0515.69
Taylor-FOSparsity=50%, Backbone=Qwen-2.5-7B, Fine-tuning steps=150, PEFT Method=LoRA2026.0510.27