@<1554638160548335616:profile|AverageSealion33> looks like hydra pulls the config relative to the scripts directory, and not the current working directory. The pipeline controller actually creates a temp file in /tmp
when it pulls the step, so the script's directory will be /tmp
and when searching for ../data
, hydra will search in /
. The .git
likely caused your repository to be pulled, so your repo structure was created in /tmp
, which caused the step to run correctly.
What you could do is this to make hydra search via cwd:
from clearml import Task
import os
import hydra
@hydra.main(config_path=os.path.join(os.getcwd(), "../configs/preprocessing/"), config_name="config", version_base=None)
def main(cfg):
Task.init("hydra", "hydra")
print(cfg)
if __name__ == "__main__":
main()
@<1554638160548335616:profile|AverageSealion33> Can you run the script with HYDRA_FULL_ERROR=1
. Also, what if you run the script without clearml? Do you get the same error?
@<1523701435869433856:profile|SmugDolphin23> I'm having trouble understanding your words. I don't use the agent here.
@<1523701435869433856:profile|SmugDolphin23> @<1523701070390366208:profile|CostlyOstrich36> Maybe you could look at my code ( None ) to get a better view of the case. (I really doubt that the look at the code will help to solve it since it is related to presence of git). Theres no clearml agent or any complex settings.
Hi @<1554638160548335616:profile|AverageSealion33> , can you add a log with and without git? What is the use case - why do you want to remove git? And just making sure - this is run through agent?
Hi @<1554638160548335616:profile|AverageSealion33> ! We pull git repos to copy the directory your task is running in. Because you deleted .git
, we can't do that anymore. I think that, to fix this, you could just run the agent in the directory .git
previously existed.
- Attached verbose logs.
- There's no point to run script with pipeline without clearml since the script is based on clearml feature (pipeline itself). All other scripts that can be run without clearml works fine.
Hydra config file can't be read while running preprocess.py via pipeline, but it works good running it directly. (no git case)
@<1523701070390366208:profile|CostlyOstrich36> 1) I attached logs in text file. 2) I want to develop my code in docker container and theres no place for git (as far as I understand the purpose of using docker at all) - regardless of my use case, this type of simple pipeline should work with no git dependency, right? 3) I don't use the agent in this case