Options
All
  • Public
  • Public/Protected
  • All
Menu

Interface SketchRNNInfo

Interface for JSON specification of a MusicVAE model.

property

max_seq_len: Model trained on dataset w/ this max sequence length.

property

mode: Pre-trained models have this parameter for legacy reasons. 0 for VAE, 1 for Decoder only. This model is Decoder only (not used).

property

name: QuickDraw name, like cat, dog, elephant, etc

property

scale_factor: the factor to convert from neural-network space to pixel space. Most pre-trained models have this number between 80-120

property

version: Pre-trained models have a version between 1-6, for the purpose of experimental research log.

Hierarchy

  • SketchRNNInfo

Index

Properties

max_seq_len

max_seq_len: number

mode

mode: number

name

name: string

scale_factor

scale_factor: number

version

version: number

Generated using TypeDoc