scitex-nn 0.1.7

Neural network building blocks (BNet, Hilbert, PAC, Wavelet, Filters, …) — standalone module from the SciTeX ecosystem

scitex-nn

PyTorch neural-network building blocks (BNet, Hilbert, PAC, Wavelet, Filters, AxiswiseDropout, …) extracted from the SciTeX ecosystem as a standalone package.

Install

pip install scitex-nn

API

import scitex_nn as nn

m = nn.BNet(...)
m = nn.Filters(...)
m = nn.Hilbert(...)
m = nn.Wavelet(...)
m = nn.PAC(...)

Status

Standalone fork of scitex.nn. The umbrella package's scitex.nn import path is preserved via a sys.modules-alias bridge.

Decoupling notes:

• scitex.{decorators,gen} → scitex_decorators / scitex_gen direct imports.
• scitex.dsp.utils (build_bandpass_filters, init_bandpass_filters, ensure_3d, ensure_even_len, zero_pad, design_filter) → vendored under _vendor_dsp_utils/. Vendor prefers the real scitex.dsp.utils when the umbrella is installed (lockstep behaviour) and falls back to the vendored copy when scitex_nn runs standalone.
• scitex.nn.X self-references rewritten to scitex_nn.X.
• Example if __name__ == "__main__": blocks still reference scitex.{io,plt,session,dsp,ai} — only run when the umbrella is installed; module-level imports do not depend on those.

License

AGPL-3.0-only.