MaCow: Masked Convolutional Generative Flow
Authors: Xuezhe Ma, Xiang Kong, Shanghang Zhang, Eduard Hovy
NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimenting with four benchmark datasets for images, CIFAR-10, Image Net, LSUN, and Celeb A-HQ, we demonstrate the effectiveness of MACOW as a density estimator by consistently achieving significant improvements over Glow on all the three datasets. |
| Researcher Affiliation | Academia | Xuezhe Ma, Xiang Kong, Shanghang Zhang, Eduard Hovy Carnegie Mellon University Pittsburgh, PA, USA xuezhem,xiangk@cs.cmu.edu, shanghaz@andrew.cmu.edu, hovy@cmu.edu |
| Pseudocode | No | The paper describes the model architecture and components in detail within the text and figures, but it does not include any formal pseudocode or algorithm blocks. |
| Open Source Code | Yes | 3https://github.com/Xuezhe Max/macow |
| Open Datasets | Yes | Experimenting with four benchmark datasets for images, CIFAR-10, Image Net, LSUN, and Celeb A-HQ, we demonstrate the effectiveness of MACOW as a density estimator by consistently achieving significant improvements over Glow on all the three datasets. |
| Dataset Splits | No | The paper mentions training on various datasets and evaluating performance, but it does not explicitly provide specific details about the training, validation, or test dataset splits (e.g., percentages, counts) in the main text. |
| Hardware Specification | Yes | For fair comparison, we reimplemented Glow using Py Torch (Paszke et al., 2017), and all experiments are conducted on a single NVIDIA TITAN X GPU. |
| Software Dependencies | No | The paper mentions 'Py Torch (Paszke et al., 2017)' as a software used, but it does not provide specific version numbers for PyTorch or any other relevant libraries/dependencies. |
| Experiment Setup | Yes | For a step of MACOW, we use T = 2 masked convolution units, and the Glow step is the same as that described in Kingma and Dhariwal (2018) where an Act Norm is followed by an Invertible 1 1 convolution, which is followed by a coupling layer. Each coupling layer includes three convolution layers where the first and last convolutions are 3 3, while the center convolution is 1 1. For low-resolution images, we use affine coupling layers with 512 hidden channels, while for high-resolution images we use additive layers with 256 hidden channels to reduce memory cost. ELU (Clevert et al., 2015) is used as the activation function throughout the flow architecture. |