Simple and Effective VAE Training with Calibrated Decoders

Authors: Oleh Rybkin, Kostas Daniilidis, Sergey Levine

ICML 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We perform the first comprehensive comparative analysis of calibrated decoder and provide recommendations for simple and effective VAE training. Our analysis covers a range of image and video datasets and several single-image and sequential VAE models.
Researcher Affiliation Academia 1University of Pennsylvania 2UC Berkeley. Correspondence to: Oleh Rybkin <oleh@seas.upenn.edu>.
Pseudocode No The paper does not contain any clearly labeled pseudocode or algorithm blocks.
Open Source Code Yes Project website: https: //orybkin.github.io/sigma-vae/.
Open Datasets Yes We use a small convolutional VAE model on SVHN (Netzer et al., 2011), a larger hierarchical HVAE model (Maaløe et al., 2019) on the Celeb A (Liu et al., 2015) and CIFAR (Krizhevsky et al., 2009) datasets, and a sequence VAE model called SVG (Denton & Fergus, 2018) on the BAIR Pushing dataset (Finn & Levine, 2017).
Dataset Splits No The paper mentions the datasets used and image/frame sizes but does not provide specific train/validation/test splits (e.g., percentages, sample counts, or references to standard splits).
Hardware Specification No The paper does not provide specific details about the hardware (e.g., GPU/CPU models, memory) used for running the experiments.
Software Dependencies No The paper mentions using software like TensorFlow and PyTorch by citing their papers, but it does not specify exact version numbers for any software dependencies.
Experiment Setup Yes We observe that setting λmin = 6 to lower bound the standard deviation to be at least half of the distance between allowed color values works well in practice.