The paper claims that a fully spectral wavelet-domain model can reach near-Transformer accuracy on GLUE tasks while using 72% fewer parameters and no attention or convolution layers.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Wavelet Logic Machines: Learning and Reasoning in the Spectral Domain Without Neural Networks
The paper claims that a fully spectral wavelet-domain model can reach near-Transformer accuracy on GLUE tasks while using 72% fewer parameters and no attention or convolution layers.