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.

i use:

export CLEARML_AGENT_EXTRA_DOCKER_ARGS=" -v /srv:/root/srv  -v /srv:/srv  -v /var/run/docker.sock:/var/run/docker.sock -v /home/user/mlbox/clearml.conf:/root/clearml.conf  --network=host  --ipc host  --hostname=\$(hostname)"

i added the required dependencies but don't feel like doing /srv/project:/root/.clearml/venvs-builds/3.7/code

although okay, if it helps to start working, then you can do it like that ...

cp does't work

cp: target '/root/.clearml/venvs-builds/3.7/code' is not a directory

i found 🙂

CLEARML_AGENT_SKIP_PYTHON_ENV_INSTALL

sounds great)

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

CostlyOstrich36 ^

all code

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

@<1523701087100473344:profile|SuccessfulKoala55>
I'm talking about something like OPTUNA

wow, that's interesting, please let me know. Are there screenshots or a demo video somewhere where you can see how the enumeration parameters are set.

i use the local free version of clearml

don't think. next week I'll try to change the code of the example, maybe it will work with the channel ID

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

maybe I can share individual modules?)

to guide you faster)

did not help(

I found a solution - I did not specify the address of the service (

here is a working code example:

basket = s3-infra.loc/
path = "s3-artifacts-test/proj_path"
task = Task.init(
        project_name=project_name,
        task_name=task_name,
        reuse_last_task_id=False,
        output_uri='s3://{basket}{path}',
    )

this is the experiment that was useful, but we stopped it, because. convergence has happened before. than we expected

And why can it be that the displayed time is zero?
although the experiments were considered for several days

Does this only work for the completed status?
and does not take into account failed and ABORTED experiments?

we just started logging metrics in clearml

Perhaps this will help

yes, I think so too

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

TimelyPenguin76
As I remember, the last one is 1.1.4
What exactly do you mean about task logger?

GorgeousSeagull44 Did it help?

basket name all ok?

AgitatedDove14 ^

CostlyOstrich36 ^