Thermodynamic diffusion inference at production scale is shown using hierarchical bilinear coupling for U-Net skips and a 2,560-parameter digital bottleneck, attaining 0.9906 cosine similarity with theoretical 10^7x energy reduction over GPU.
An efficient probabilistic hardware architecture for diffusion-like models
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.LG 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
Symmetric Equilibrium Propagation provides a local, readout-only training rule for bilinear thermodynamic diffusion models that is unbiased at zero nudge, reduces bias to O(β²) with symmetric nudging, and projects 10³-10⁴× energy savings over GPU baselines.
citing papers explorer
-
Thermodynamic Diffusion Inference with Minimal Digital Conditioning
Thermodynamic diffusion inference at production scale is shown using hierarchical bilinear coupling for U-Net skips and a 2,560-parameter digital bottleneck, attaining 0.9906 cosine similarity with theoretical 10^7x energy reduction over GPU.
-
Symmetric Equilibrium Propagation for Thermodynamic Diffusion Training
Symmetric Equilibrium Propagation provides a local, readout-only training rule for bilinear thermodynamic diffusion models that is unbiased at zero nudge, reduces bias to O(β²) with symmetric nudging, and projects 10³-10⁴× energy savings over GPU baselines.