Reputation
Badges 1
16 × Eureka!@<1523701087100473344:profile|SuccessfulKoala55> any insight on this? still blocked
@<1523701087100473344:profile|SuccessfulKoala55> any insight on this? still blocked
I am running the job by:
- making a git commit
- running my jupyter notebook locally to create the “base” task
- cloning that and changing ‘configuration’
- enqueueing
the agent is running on the separate worker machine running OSX,
being launched byclearml-agent daemon --detached
local development is fine, but local-running-first I am just unsure about
the problem seems to come and go … really not sure if anything I am changing is resolving it,
e.g. restarting the daemon,
changing to bash shell as the default login for osx
I suspect that this is something to do with system python vs conda python … but, very opaque
conda has 3.11.6 installed in the base env,
and the base shell appears to not have python installed at all
And the clearml-agent
is installed in the “base” conda env
@<1523701087100473344:profile|SuccessfulKoala55> well, afaik, the base Xcode python 3.9.6 is installed with the Xcode tools, in /usr/bin
, and then anaconda installed as the user running the clearml-agent
,
conda init is in both the .bashrc
and .bash_profile
And the task seems to not be installing from requirements.txt, because one package ends up missing
🤔 could one create empty draft experiment placeholders, finalize all their code, commit it, and then and manually connect the impl as it exists in a file in a commit in a repo to that draft, once I’ve decided that all development is done?
🤔 does a job/experiment always reproduce from git + commit hash + specific file?
(again, haven’t read through all the docs yet, a lot to take in)
but I’ll try out what you said - parameterize it all, and run a short version locally, then scale up in the UI
huh, ok … the need to run stuff locally first is a little odd, ngl… just have to get a sense of what the workflow here is
but it looks like execute_remotely
is indeed what I want, and I should just standardize on using that,
either in-situ, or in an outer runner script that just calls that first, then calls down to my actual implementation