Github Action gets killed on each invocation, would like to know why

I have a github action in a public repo that runs some python code. Within this code, it downloads a dataset and starts training a simple linear regression model. The dataset is ~ 400 MB and the model is very basic. My guess is this is well within the disk usage and memory bounds.

However, the action gets killed on each invocation and all I see is the error code 137. I have logged in to the VM running the action using the tmate action. I see that the dataset is successfully downloaded and there is sufficient disk space.

How can I find out what is causing the action to get killed?

Here is a link to the latest failure of the action: https://github.com/sanzgiri/action-numerai/actions/runs/116260289

I would appreciate any help.

Thanks,
Ashutosh

Hi @sanzgiri,
Based on my test, I found that the issue was caused by this line in train.py:

data = nx.download('numerai_dataset.zip')

It will first download the Numerai dataset and then load it. I think there is something wrong when load it.
I would encourage you add load=False .

nx.download('numerai_dataset.zip', load=False)

Thanks @yanjingzhu. I was able to get the reason for the error after using a ubuntu-16.04 runner

MemoryError: Unable to allocate 2.40 GiB for an array with shape (313, 2056947) and data type float32.

Is there a way I can bump up the memory available to the runner? According to https://help.github.com/en/actions/reference/virtual-environments-for-github-hosted-runners, the VM has 7 GB of RAM. Is this accurate?

Sorry to tell you that you could not increase the RAM memory of hosted runner. The workaround is to use a self-hosted runner.
But there are some security issue of using self-hosted runners with public repositories. Forks of your public repository can potentially run dangerous code on your self-hosted runner machine by creating a pull request that executes the code in a workflow.