Can't install pytorch

It looks like everything is running successfully until I get error:
AutoModelForSeq2SeqLM requires the PyTorch library but it was not found in your environment. Checkout the instructions on the
installation page: Start Locally | PyTorch and follow the ones that match your environment.
Please note that you may need to restart your runtime after installation

I tried restarting kernel, unstalling different version of pytorch - nothing works. It looks like the memory is not enoough to install torch (2 vCPU + 4 GiB). But I don’t have permission to change the instance to a larger one.

Time is running out - what do I do?

Eventually fixed with:
!pip --no-cache-dir install torch==1.13.1

Tried all other solutions - this is the only one that worked after 90 minutes.

1 Like