version requirement for using "requirement.txt" when serving a model #1221

billdoors · 2020-01-09T20:34:36Z

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

I have read the CONTRIBUTING doc
I used the commit message format described in CONTRIBUTING
I have used the regional endpoint when creating S3 and/or STS clients (if appropriate)
I have updated any necessary documentation, including READMEs and API docs (if appropriate)

Tests

I have added tests that prove my fix is effective or that my feature works (if appropriate)
I have checked that my tests are not configured for a specific region or account (if appropriate)
I have used 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.

…folder, packages in "requirements.txt" will fail to install.

…erving. Please check: aws/sagemaker-pytorch-inference-toolkit#28 for more information.

sagemaker-bot · 2020-01-09T20:35:31Z

AWS CodeBuild CI Report

Result: SUCCEEDED
Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

sagemaker-bot · 2020-01-09T20:35:43Z

AWS CodeBuild CI Report

Result: SUCCEEDED
Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

sagemaker-bot · 2020-01-09T20:35:48Z

AWS CodeBuild CI Report

Result: SUCCEEDED
Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

sagemaker-bot · 2020-01-09T20:42:18Z

AWS CodeBuild CI Report

Result: SUCCEEDED
Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

laurenyu

the reality is a bit all over the place, unfortunately:

for training, it should be supported for all versions
for inference, requirements.txt should be in the source_dir for v0.4.0-1.1.0 and in the model tarball for v1.3.1 under a directory named code/

billdoors · 2020-01-10T16:32:33Z

the reality is a bit all over the place, unfortunately:

for training, it should be supported for all versions

for inference, requirements.txt should be in the source_dir for v0.4.0-1.1.0 and in the model tarball for v1.3.1 under a directory named code/

I re-write it as below:
If there are other packages you want to use with your script, you can include a requirements.txt file in the same directory as your training script to install other dependencies at runtime. Both requirements.txt and your training script should be put in the same folder. You must specify this folder in source_dir argument when creating PyTorch estimator. A requirements.txt file is a text file that contains a list of items that are installed by using pip install. You can also specify the version of an item to install.

The function of installing packages using requirements.txt is supported for all PyTorch versions during training. When serving a PyTorch model, this function is supported but how to use it varies with PyTorch Versions. For PyTorch 1.3.1 or newer, requirements.txt must be in source_dir. For PyTorch 1.2.0, requirements.txt must be under directory code. For PyTorch 0.4.0 to 1.1.0, requirements.txt must be in source_dir and also in model tarball.

For information about the format of a requirements.txt file, see Requirements Files <https://pip.pypa.io/en/stable/user_guide/#requirements-files>__ in the pip documentation.

Question:
This part is unclear to me: "For PyTorch 0.4.0 to 1.1.0, requirements.txt must be in source_dir and also in model tarball"

sagemaker-bot · 2020-01-10T20:14:31Z

AWS CodeBuild CI Report

Result: SUCCEEDED
Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

sagemaker-bot · 2020-01-10T20:14:48Z

AWS CodeBuild CI Report

Result: SUCCEEDED
Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

sagemaker-bot · 2020-01-10T20:14:49Z

AWS CodeBuild CI Report

Result: SUCCEEDED
Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

sagemaker-bot · 2020-01-10T20:20:40Z

AWS CodeBuild CI Report

Result: SUCCEEDED
Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

laurenyu · 2020-01-11T01:47:08Z

Question:
This part is unclear to me: "For PyTorch 0.4.0 to 1.1.0, requirements.txt must be in source_dir and also in model tarball"

what I meant was: requirements.txt should be (in the source_dir for v0.4.0-1.1.0) && (in the model tarball for v1.3.1 under a directory named code/). I can see how that ended up being ambiguous though 😂

If there are other packages you want to use with your script, you can include a requirements.txt file in the same directory as your training script to install other dependencies at runtime. Both requirements.txt and your training script should be put in the same folder. You must specify this folder in source_dir argument when creating PyTorch estimator. A requirements.txt file is a text file that contains a list of items that are installed by using pip install. You can also specify the version of an item to install.

I would move "A requirements.txt file is a text file that contains a list of items that are installed by using pip install. You can also specify the version of an item to install." to go with the last paragraph ("For information about the format of a requirements.txt file, see Requirements Files in the pip documentation.")

The function of installing packages using requirements.txt is supported for all PyTorch versions during training. When serving a PyTorch model, this function is supported but how to use it varies with PyTorch Versions. For PyTorch 1.3.1 or newer, requirements.txt must be in source_dir. For PyTorch 1.2.0, requirements.txt must be under directory code. For PyTorch 0.4.0 to 1.1.0, requirements.txt must be in source_dir and also in model tarball.

Almost. For 1.3.1+, requirements.txt must be under code/ within model.tar.gz. 1.2.0 doesn't have requirements.txt support at all. For 0.4.0-1.1.0, you can remove "and also in the model tarball."

sagemaker-bot · 2020-01-13T14:45:13Z

AWS CodeBuild CI Report

Result: SUCCEEDED
Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

sagemaker-bot · 2020-01-13T14:45:28Z

AWS CodeBuild CI Report

Result: SUCCEEDED
Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

sagemaker-bot · 2020-01-13T14:45:31Z

AWS CodeBuild CI Report

Result: SUCCEEDED
Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

sagemaker-bot · 2020-01-13T14:48:26Z

AWS CodeBuild CI Report

Result: SUCCEEDED
Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

sagemaker-bot · 2020-01-13T14:48:49Z

AWS CodeBuild CI Report

Result: SUCCEEDED
Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

sagemaker-bot · 2020-01-13T14:49:06Z

AWS CodeBuild CI Report

Result: SUCCEEDED
Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

sagemaker-bot · 2020-01-13T14:51:07Z

AWS CodeBuild CI Report

Result: SUCCEEDED
Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

sagemaker-bot · 2020-01-13T14:55:10Z

AWS CodeBuild CI Report

Result: SUCCEEDED
Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

laurenyu · 2020-01-13T17:04:43Z

thank you!! :)

sagemaker-bot · 2020-01-13T17:05:17Z

AWS CodeBuild CI Report

Result: SUCCEEDED
Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

sagemaker-bot · 2020-01-13T17:05:30Z

AWS CodeBuild CI Report

Result: SUCCEEDED
Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

sagemaker-bot · 2020-01-13T17:05:33Z

AWS CodeBuild CI Report

Result: SUCCEEDED
Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

sagemaker-bot · 2020-01-13T17:11:08Z

AWS CodeBuild CI Report

Result: SUCCEEDED
Build Logs (available for 30 days)

Powered by github-codebuild-logs, available on the AWS Serverless Application Repository

Qingwei Li added 3 commits January 7, 2020 22:44

fix typo

4b4d4ba

Without specifying "source_dir" and put requirements.txt in that …

346998c

…folder, packages in "requirements.txt" will fail to install.

Add PyTorch version requirement to allow installing pacakges during s…

61852dd

…erving. Please check: aws/sagemaker-pytorch-inference-toolkit#28 for more information.

laurenyu reviewed Jan 9, 2020

View reviewed changes

revised doc

d1b389c

Doc improvement"

ab8988f

Doc improvement"

cef44c1

laurenyu previously approved these changes Jan 13, 2020

View reviewed changes

Merge branch 'master' into master

2df8fb6

laurenyu dismissed their stale review via 2df8fb6 January 13, 2020 17:04

laurenyu approved these changes Jan 13, 2020

View reviewed changes

laurenyu merged commit 335fd60 into aws:master Jan 13, 2020

laurenyu mentioned this pull request Jan 14, 2020

doc: add version requirement for using "requirements.txt" when serving an MXNet model #1232

Merged

7 tasks

version requirement for using "requirement.txt" when serving a model #1221

version requirement for using "requirement.txt" when serving a model #1221

Uh oh!

Conversation

billdoors commented Jan 9, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Merge Checklist

General

Tests

Uh oh!

sagemaker-bot commented Jan 9, 2020

AWS CodeBuild CI Report

Uh oh!

sagemaker-bot commented Jan 9, 2020

AWS CodeBuild CI Report

Uh oh!

sagemaker-bot commented Jan 9, 2020

AWS CodeBuild CI Report

Uh oh!

sagemaker-bot commented Jan 9, 2020

AWS CodeBuild CI Report

Uh oh!

laurenyu left a comment

Choose a reason for hiding this comment

Uh oh!

billdoors commented Jan 10, 2020

Uh oh!

sagemaker-bot commented Jan 10, 2020

AWS CodeBuild CI Report

Uh oh!

sagemaker-bot commented Jan 10, 2020

AWS CodeBuild CI Report

Uh oh!

sagemaker-bot commented Jan 10, 2020

AWS CodeBuild CI Report

Uh oh!

sagemaker-bot commented Jan 10, 2020

AWS CodeBuild CI Report

Uh oh!

laurenyu commented Jan 11, 2020

Uh oh!

sagemaker-bot commented Jan 13, 2020

AWS CodeBuild CI Report

Uh oh!

sagemaker-bot commented Jan 13, 2020

AWS CodeBuild CI Report

Uh oh!

sagemaker-bot commented Jan 13, 2020

AWS CodeBuild CI Report

Uh oh!

sagemaker-bot commented Jan 13, 2020

AWS CodeBuild CI Report

Uh oh!

sagemaker-bot commented Jan 13, 2020

AWS CodeBuild CI Report

Uh oh!

sagemaker-bot commented Jan 13, 2020

AWS CodeBuild CI Report

Uh oh!

sagemaker-bot commented Jan 13, 2020

AWS CodeBuild CI Report

Uh oh!

sagemaker-bot commented Jan 13, 2020

AWS CodeBuild CI Report

Uh oh!

laurenyu commented Jan 13, 2020

Uh oh!

sagemaker-bot commented Jan 13, 2020

AWS CodeBuild CI Report

Uh oh!

sagemaker-bot commented Jan 13, 2020

AWS CodeBuild CI Report

Uh oh!

sagemaker-bot commented Jan 13, 2020

AWS CodeBuild CI Report

Uh oh!

sagemaker-bot commented Jan 13, 2020

AWS CodeBuild CI Report

Uh oh!

Uh oh!

billdoors commented Jan 9, 2020 •

edited

Loading