Share your thoughts, 1 month free Claude Pro on usSee more

In-Context Learning on Garg (d=10) (held-out tasks)

0.996R2

Transformer (continuous attn)

Updated 22d ago

Evaluation Results

Method	Links
Transformer (continuous attn) 2026.07		0.996
Spikformer (spike attn) 2026.07		0.977
DendriCNN (continuous ablation) 2026.07		0.824
DendriCL (L = 2) 2026.07		0.811
DendriCL (single-layer) 2026.07		0.807
Pure LIF 2026.07		0.801
Active Dendrites 2026.07		0.668
GRU 2026.07		0.666
LSNN 2026.07		0.011