AML-6 model manifest, tamper detection implementation #7
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
zeryx
commented
Sep 28, 2021
lemonez
reviewed
Sep 28, 2021
lemonez
reviewed
Sep 28, 2021
lemonez
reviewed
Sep 28, 2021
lemonez
reviewed
Sep 28, 2021
lemonez
suggested changes
Sep 28, 2021
…unction does not have a parameter; don't try to pass it one.
lemonez
approved these changes
Sep 28, 2021
kennydaniel
reviewed
Oct 6, 2021
aslisabanci
reviewed
Oct 8, 2021
lemonez
approved these changes
Oct 13, 2021
aslisabanci
reviewed
Oct 14, 2021
|
Would also love to run the tests and even add more to them but VSCode is being a pain in the head about my unittest framework and test discovery right now. So I'll conclude without that, but hope to resolve my local setup soon for future maintenance. |
aslisabanci
approved these changes
Oct 15, 2021
kennydaniel
reviewed
Oct 15, 2021
| @@ -0,0 +1 @@ | |||
| {"required_files": [{"name": "squeezenet", "source_uri": "data://AlgorithmiaSE/image_cassification_demo/squeezenet1_1-f364aa15.pth", "fail_on_tamper": true, "metadata": {"dataset_md5_checksum": "46a44d32d2c5c07f7f66324bef4c7266"}, "md5_checksum": "46a44d32d2c5c07f7f66324bef4c7266"}, {"name": "labels", "source_uri": "data://AlgorithmiaSE/image_cassification_demo/imagenet_class_index.json", "fail_on_tamper": true, "metadata": {"dataset_md5_checksum": "46a44d32d2c5c07f7f66324bef4c7266"}, "md5_checksum": "c2c37ea517e94d9795004a39431a14cb"}], "optional_files": [{"name": "mobilenet", "source_uri": "data://AlgorithmiaSE/image_cassification_demo/mobilenet_v2-b0353104.pth", "fail_on_tamper": false, "metadata": {"dataset_md5_checksum": "46a44d32d2c5c07f7f66324bef4c7266"}}], "timestamp": "1633450866.985464", "lock_checksum": "24f5eca888d87661ca6fc08042e40cb7"} | |||
There was a problem hiding this comment.
Can we pretty print the freeze files instead of one long json line?
kennydaniel
approved these changes
Oct 15, 2021
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This is the first implementation of model file tamper detection, and model manifest integration into the developer environment.
This optional enhancement can be utilized by defining a
model_manifest.json, which while utilizing the Python CLI; will freeze that manifest file into amanifest_file.json.freezewhich contains the md5 hashes of all model files that are currently defined in the manifest, along with a timestamp.This system allows for model files to automatically be downloaded safely and verified for security before they are even accessible to the algorithm developer; which they are available as a filehandle that they can then do whatever prep necessary.
A number of our customers have asked for this, and our implementation will improve over time.
To show an example of what an algorithm implementation might look like; take a look here:
https://gist.github.com/zeryx/d64140c385a9bda7aecfe42d3ce9bbff
the model_manifest.json would exist in the
rootdirectory, alongside thealgorithmia.conffile. It would compiled into alockfile using thealgo compilefunction defined in the following client PR (algorithmiaio/algorithmia-python#113), which is then to be checked in to the git repo.