List your company
ResearchFormerly WizWand

Question Answering on SciQA

76.2Normalized AccuracyGatedFWA-NSA
72.97673.81374.6575.487Dec 8, 2025
Updated 8mo ago

Evaluation results

MethodLinks
GatedFWA-NSAClass=Attention, Implementation=Triton, Linear=true2025.1276.2
HGRN2Class=RNN-Like, Implementation=Triton, Linear=true2025.1275.6
GatedFWAClass=Attention, Implementation=Triton, Linear=true2025.1274.86
Transformer (LLaMA) + SWA + NSAClass=Attention, Implementation=Triton, Linear=true2025.1274.75
DeltaNetClass=RNN-Like, Implementation=Triton, Linear=true2025.1274.3
Transformer (LLaMA)Class=Attention, Implementation=CUDA, Linear=false2025.1273.9
Transformer (LLaMA) + SWAClass=Attention, Implementation=CUDA, Linear=true2025.1273.79
MambaClass=RNN-Like, Implementation=CUDA, Linear=true2025.1273.7
GLAClass=RNN-Like, Implementation=Triton, Linear=true2025.1273.2
RetNetClass=RNN-Like, Implementation=CUDA, Linear=true2025.1273.1