Deep Automodulators

Authors: Ari Heljakka, Yuxin Hou, Juho Kannala, Arno Solin

NeurIPS 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental demonstrated on four image data sets. Besides style-mixing, we show state-of-the-art results in autoencoder comparison, and visual image quality nearly indistinguishable from state-of-the-art GANs. We demonstrate promising qualitative and quantitative performance and applications on FFHQ, CELEBA-HQ, and LSUN Bedrooms and Cars data sets.
Researcher Affiliation Collaboration Ari Heljakka1,2 Yuxin Hou1 Juho Kannala1 Arno Solin1 1Aalto University 2Gen Mind {ari.heljakka, yuxin.hou, juho.kannala, arno.solin}@aalto.fi
Pseudocode No The paper does not contain any explicitly labeled 'Pseudocode' or 'Algorithm' blocks.
Open Source Code Yes Our source code is available at https://github.com/AaltoVision/automodulator.
Open Datasets Yes We demonstrate promising qualitative and quantitative performance and applications on FFHQ, CELEBA-HQ, and LSUN Bedrooms and Cars data sets.
Dataset Splits No The paper mentions training on 'the actual 60k training set of FFHQ only' and 'CELEBA-HQ at 256x256 with 40M seen samples (50k FID batch)', but does not provide specific percentages or counts for training, validation, and test splits across all datasets, nor does it explicitly reference predefined standard splits.
Hardware Specification No The paper only states that 'The authors wish to acknowledge the Aalto Science-IT project and CSC IT Center for Science, Finland, for computational resources,' without providing specific hardware details like GPU or CPU models.
Software Dependencies No The paper does not mention any specific software dependencies or their version numbers, such as Python, PyTorch, or TensorFlow versions.
Experiment Setup Yes Margin Mgap = 0.5, except for CELEBAHQ and Bedrooms 128 128 (Mgap = 0.2) and CELEBAHQ 256 256 (Mgap = 0.4). We train the model variants on CELEBA-HQ [24] data set to convergence (40M seen samples) and choose the best of three restarts with different random seeds.