A time-dilation schedule makes the mode-probability learning phase survive in high dimension, and the learned flow autoencoder recovers the mixture's p and σ² in two separate phases.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Phase-aware Training Schedule Simplifies Learning in Flow-Based Generative Models
A time-dilation schedule makes the mode-probability learning phase survive in high dimension, and the learned flow autoencoder recovers the mixture's p and σ² in two separate phases.