search

InterDigitalInc / CompressAI

A PyTorch library and evaluation platform for end-to-end compression research
https://interdigitalinc.github.io/CompressAI/
BSD 3-Clause Clear License
1.19k stars 232 forks source link

BMSHJ2018 Parameters #2

Closed alexhepburn closed 4 years ago

alexhepburn commented 4 years ago

I was wondering if the training parameters for both 'bmshj2018_factorized' & 'bmshj2018_hyperprior' were anywhere in the documentation? E.g. the dataset, optimisation, learning rates, training epochs ect.

I am assuming that the quality parameter loads different pre-trained networks depending on the lambda parameter they were trained with in the rate-distortion loss. It would be good to what lambda corresponds to the quality parameter. Thanks!

alexhepburn commented 4 years ago

Sorry, I completely missed this in the documentation, found here: https://interdigitalinc.github.io/CompressAI/zoo.html#training