Conflicting AWS and HF docs for AWS inf instances, confused about which one to follow

[Yan Sidyakin]

I'm trying to set up inference on AWS inf for a pretrained HF transformer. I followed this HF tutorial and ran into errors, opened an issue on aws-neuron-sdk repo, was told that env setup is wrong. I'm quite confused now whether I should follow AWS tutorials, which advise to setup env according to this AWS doc, or HF tutorials which setup env with two pip install lines. Would appreciate the help!

[Philipp Schmid]

Hey @sidyakinian,

The tutorial uses the exact same setup than the AWS Doc see: https://awsdocs-neuron.readthedocs-hosted.com/en/latest/neuron-intro/pytorch-setup/pytorch-install.html#compile-on-compute-instance.

# Note: There is no DLAMI Conda environment for this framework version
#       Framework will be installed/updated inside a Python environment

# Activate PyTorch
source activate aws_neuron_pytorch_p36

# Set Pip repository  to point to the Neuron repository
pip config set global.extra-index-url https://pip.repos.neuron.amazonaws.com

#Install Neuron PyTorch
pip install torch-neuron==1.9.1.* neuron-cc[tensorflow] "protobuf<4" torchvision

[Yan Sidyakin]

@philschmid Makes sense, sorry, got a little confused with torch versions. Thanks!