Static quantization tutorial missing a step

[rfejgin]

tutorials/advanced_source/static_quantization_tutorial.py

Line 514 in 7bd0315

per_channel_quantized_model.qconfig = torch.quantization.get_default_qconfig('fbgemm')

If I understand correctly, one needs to also set
torch.backends.quantized.engine = "fbgemm"

I tried to quantize a model without the missing step and got strange errors about certain operations not being supported on the FBGEMM backend. They go away with this step added.

cc @jerryzh168 @jianyuh @sekyondaMeta @svekars @carljparker @NicolasHug @kit1980 @subramen

[Samsonboadi]

/assigntome

[svekars]

Can you please close one of the PRs for this issue?

[Samsonboadi]

Done

[svekars]

This issue has been unassigned due to inactivity. If you are still planning to work on this, you can still send a PR referencing this issue.

[Viditagarwal7479]

In the current version of tutorials this python file has been replaced with a .rst file with fbgemm backend changed to x86. So shouldn't this issue be closed?

[krishnakalyan3]

/assigntome

[svekars]

You can still add to the .rst file but with the updated backend, like torch.backend.quantized.engine = 'x86'

[krishnakalyan3]

@svekars this change has already been made to the rst
/~https://github.com/pytorch/tutorials/blob/main/advanced_source/static_quantization_tutorial.rst?plain=1#L462 and link to the doc reflecting the same https://pytorch.org/tutorials/advanced/static_quantization_tutorial.html.

This issue is pointing to a stale branch.