@<1523701070390366208:profile|CostlyOstrich36> any idea what could be going wrong here?
OR do you have any example of this type of setup?

no, my server is on aws and agent is local

my code to init task:

task.set_base_docker(
    docker_image="clearml",
    docker_arguments=(
        "-d "
        f"-v {PWD}data:/home/data "
        f"-v {PWD}.passwd-s3fs:/home/.passwd-s3fs "
        f"-v {PWD}.logger_credentials:/home/.logger_credentials "
        "-v /home/$USER/.ssh/:/root/.ssh/ "
        "--rm "
        "--net=host "
        "--privileged "
        "--gpus all "
    )
)

task.set_script(
    entry_point="scripts/evaluate.py"
)

Thanks for this, but somehow didn't seem to work. What I did instead was to have a docker image added to my agent and mounted my data/cred directories through the setup code while using --docker for clearml-agent daemon.
The outstanding issue is, I'm unable to run a file by default after starting a container. I'm using the "New Experiment" pop-up window in clearml dashboard to start an experiment. But that only sets up my container with correct mounting, but doesn't run the script that I ...

fair, thanks!

Yeah I did. But found the mistake, it was in the relative path provided. Couldn't catch it earlier because clearml console didn't give any error.

Thanks for the help, CLEARML_AGENT_FORCE_CODE_DIR - this is working for me now