phynexis.ml

Machine learning module. Provides neural network training and inference (via PyTorch backend), plus ML-based shape representations using signed distance functions (SDF) and Laplace fields.

Import

import phynexis

# Access submodule directly
net = phynexis.ml.NeuralNet()

# Or bind locally
from phynexis import ml
net = ml.NeuralNet()

Module Overview

Class	Description
NeuralNet	Unified neural network supporting classification, regression, and gradient computation
TrainingConfig	Training hyperparameters (step size, batch size, epochs, etc.)
LayerType	Layer type enumeration (Linear, ReLU, Tanh, etc.)
PyTorchUtils	Utility functions for PyTorch interoperability
NetField	Base class for neural-network field representations
NetSDF	Neural network-based signed distance function
NetLaplace	Neural network-based Laplace field
CodedNetSDF	SDF with latent code conditioning
CodedNetLaplace	Laplace field with latent code conditioning
LaplaceConfig	Configuration for Laplace field training and extraction
FieldType	Field type enumeration (SDF / Laplace)

C++ Namespace

phynexis::ml

pybind Module

pyml (lazy-loaded via phynexis.ml)