-
Notifications
You must be signed in to change notification settings - Fork 1.2k
feature: upgrade Neo MxNet to 1.7 #1934
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
This reverts commit b3f8abc.
AWS CodeBuild CI Report
Powered by github-codebuild-logs, available on the AWS Serverless Application Repository |
AWS CodeBuild CI Report
Powered by github-codebuild-logs, available on the AWS Serverless Application Repository |
AWS CodeBuild CI Report
Powered by github-codebuild-logs, available on the AWS Serverless Application Repository |
AWS CodeBuild CI Report
Powered by github-codebuild-logs, available on the AWS Serverless Application Repository |
"1.4": "1.5", | ||
"1.4.0": "1.5", | ||
"1.4.1": "1.5" | ||
"0.12.1": "1.7", |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Now all versions will map to 1.7?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
yeah. the older version has different repo now. If you compile a model with neo service now. The only container you can use to serve the model is the 1.7 one.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
🤕
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thats right, Neo doesnt support compiling with different versions of any Framework (MX, TF, or PT) yet. Whatever version of the framework used to train the model, only the one version in Neo (i.e MX-1.7, PT-1.4, TF-1.15) will be used to load that model. backward compatibility is assumed
@@ -165,7 +159,7 @@ def test_inferentia_deploy_model( | |||
|
|||
model.compile( | |||
target_instance_family=inf_instance_family, | |||
input_shape={"data": [1, 1, 28, 28]}, | |||
input_shape={"data": [1, 1, 28, 28], "softmax_label": [1]}, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This isnt needed for Inferentia compilation flow. But if it works thats ok, it doesnt hurt anything.
@@ -175,7 +175,7 @@ def _create_compilation_job(input_shape, output_location): | |||
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Are there similar tests for pytorch?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
No, we only have unit tests for pytorch. It's debatable if we have a gap there since the mxnet test is already testing the part of the logic that calls the neo service to compile model. We are planning to revamp our build system to handle more parallel integ testing. Once the system is more scalable we will be in a better position to add integ tests.
AWS CodeBuild CI Report
Powered by github-codebuild-logs, available on the AWS Serverless Application Repository |
AWS CodeBuild CI Report
Powered by github-codebuild-logs, available on the AWS Serverless Application Repository |
AWS CodeBuild CI Report
Powered by github-codebuild-logs, available on the AWS Serverless Application Repository |
AWS CodeBuild CI Report
Powered by github-codebuild-logs, available on the AWS Serverless Application Repository |
This reverts commit b3f8abc. Co-authored-by: Rui Wang Napieralski <[email protected]>
This reverts commit b3f8abc.
Issue #, if available:
Description of changes:
Testing done:
Merge Checklist
Put an
x
in the boxes that apply. You can also fill these out after creating the PR. If you're unsure about any of them, don't hesitate to ask. We're here to help! This is simply a reminder of what we are going to look for before merging your pull request.General
Tests
unique_name_from_base
to create resource names in integ tests (if appropriate)By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.