[D] Intuition about "Discrete Latents" in paper about VQ-VAEs

• chainer-VQ-VAE

  1 82 0.0 Python

  A Chainer implementation of VQ-VAE.

Code for https://arxiv.org/abs/1711.00937 found: https://github.com/dhgrs/chainer-VQ-VAE

• deep-vector-quantization

  2 463 0.0 Jupyter Notebook

  VQVAEs, GumbelSoftmaxes and friends

How the codebook is initialised makes a big difference, if more codes are used at the start then they often stay in use but it can be finicky to figure that out. I've found initialising codes normally with small standard deviation (e.g. 0.01) to help (so codes lie on a hypersphere). Failing that, here is an example with k-means initialisation https://github.com/karpathy/deep-vector-quantization .

• InfluxDB

  www.influxdata.com featured

  Power Real-Time Data Analytics at Scale. Get real-time insights from all types of time series data with InfluxDB. Ingest, query, and analyze billions of data points in real-time with unbounded cardinality.

  

Suggest a related project