Gaussian positional embedding for strokes.

Subclass torch's LayerNorm to handle fp16.

MLP with optional residual connections.

Training orchestrator for NeuralSVG models. Handles: - Training loop execution - Optimization and scheduling - Checkpoint management

Learning rate scheduler types.

Learning rate scheduler for sketch training. This class provides various learning rate scheduling functions: - Sinusoidal ramp with exponenti

Training logger for NeuralSVG.

Gaussian embedding for RGB colors.

A ResNet class that is similar to torchvision's but contains the following changes: - There are now 3 "stem" convolutions as opposed to 1, wi

r"""Allows the model to jointly attend to information from different representation subspaces. See reference: Attention Is All You Need .

Multi-output MLP for vector graphics generation.

Neural renderer for vector graphics using NeRF architecture.

Score Distillation Sampling loss.

Base model class for NeuralSVG implementations.

Configuration for training and evaluation data.

Configuration for logging and experiment tracking.

Logging configuration.

Model architecture and training configuration.

Model configuration.

Optimization configuration.

Optimization configuration for training.

Neural painter model for generating vector graphics from text.

Learning rate scheduler types.

Main training configuration.

Configuration for training steps and intervals.

Training steps configuration.