AgitatedDove14
I suspect that it will be difficult to assign the old IP - I will consult with the admin guys.

Architecturally it sounds right to work with the base, though I don't have that much experience. I correctly understand, What then it is necessary to replace the address in a line of a way for a file?

CostlyOstrich36
we have a render server and a file server - one machine, unfortunately I'm not so familiar with clearml yet to set it all up separately.

AgitatedDove14
yes, IP-based access, with DNS for some reason took 5 times longer and we abandoned it (MD is about our DNS and its settings).

GorgeousSeagull44
Cool!
Please tell me how to run it, I have never run JS before.

CostlyOstrich36
usability of the pytorch_lightning logger

we log the average reward of each action for the RL agent.
If the agent you did this action on the current episode, then his average reward will be nan , not 0. for obvious reasons. And we would like it to be visualized in the same way as in the tensorboard, for informational content.

AgitatedDove14
if I had to choose between logging or not logging, I would choose logging
If you choose between logging as 0 or as nan, then I would choose as nan
If you choose between skipping or logging like nan, then here I find it difficult, it seems that it is better to log than skip, but you need to think.

to a greater extent, we are used to the tensorboard, where nan is logged in a special way, and this behavior seems to be natural.

CostlyOstrich36
*If the agent did not perform a certain action, then its average reward per episode for this action will be nan , not 0

class LitMNIST(LightningModule): ... self.log('test/test_nan', np.nan, prog_bar=False, logger=True, on_step=True, on_epoch=False) ...

CostlyOstrich36
Will wait)

not nice that this logging is misleading

import numpy as np np.nan

About migration - we saved the data archives and copied them to a new server, extracting them to the appropriate folders and setting the necessary rights, and rebuilding the docker image and launching the container

Before all this, we migrated to the new version according to the instructions and everything went well, all the data after the restart was displayed correctly.

And only after that we began the process of switching to new hardware - with a large disk.

when the old server is up - all the pictures in the new server are also opened from the old server, if you click on open at the link address

AgitatedDove14 yes, that's right, he's changed

AgitatedDove14 ^

GorgeousSeagull44 did you restart docker compose?

[ctrl + C] [ctrl + V]
https://github.com/allegroai/clearml/issues/604

AgitatedDove14

CostlyOstrich36 ^

AgitatedDove14
cool

in theory, a calm launch is possible at 1.17.1-2?

in documentation link https://docs.docker.com/compose/install/

My question is, which version do you need docker compose?

https://github.com/docker/compose

Yes, of course, there is a lot of code...

maybe I can share individual modules?)

to guide you faster)

ah, I get it, I use (pytorch) lightning, and that's where it all comes from.

@<1523701070390366208:profile|CostlyOstrich36> @<1523701205467926528:profile|AgitatedDove14> maybe there is already a fix in some version?

Is that what the thing was called a bucket?

I figured it out, I had to click on the model to push, and only after that it would appear as available.

all code

I found what the problem is, I had port 8091 specified, and the file server was raised to 8081

🙂