Clément T.
Clément T.
Thanks for your answers, unfortunately I don't think it fixes it entirely, if I just add the two lines in the for loop: ```import gc from collections import Counter import...
> there're jax objects but they might not store data. Could you double check the memory? The memory does increase (it's already in the output, going from 301.4 mb to...
> maybe numpyro also caches something. I'll check later. Thanks, FYI I think this is similar to https://forum.pyro.ai/t/memory-leak-for-hierarchical-mixture-model/3807 which was not solved 😞
> Did you get OOM with the above solution? The increase in memory seems small to me. The issue is that I want to serve this model using [ml-server ](https://mlserver.readthedocs.io/en/stable/)...
Hey, sorry for the late reply: > Do you think that memory will increase forever? There are some ways you can serve a jax program with dynamic shapes. One of...
@kishankedia, I finally gave up on training the network from this repo and focuses on [this repo ](https://github.com/janivanecky/Depth-Estimation/tree/master/dataset) (the weights to perform inference are provided by the author and explanations...
@kishankedia I think [this paper](https://arxiv.org/pdf/1704.02157.pdf) is pretty much state of the art, however their weight file is missing. The repo I sent you a link in my last comment([Ivanecky](https://github.com/janivanecky/Depth-Estimation/)) only...
Hi, I think you are right gedeck