Sorry for the bombarding with errors.. but here comes another one 🙂
I have torch installed locally (through the transformers library) and when sending it to a queue, I getfrom simpletransformers.classification.classification_model import ClassificationModel File "/root/.trains/venvs-builds/3.6/lib/python3.6/site-packages/simpletransformers/classification/classification_model.py", line 20, in import torch ModuleNotFoundError: No module named 'torch'pip freeze | grep torch shows it is indeed installed locally (output torch==1.6.0 ), logs on the agent shows torch not even installed (while other libs did) - what could be the reason for that?
WackyRabbit7 the auto detection will only import direct packages you import (so that we do not end up with bloated venvs)
It seems that the transformers library does not have it as a requirements, otherwise it would have pulled it...
In your code you can always do either:import torchorTask.add_requirements('torch')
Cool, now I understand the auto detection better
