Fix OpenAI __reduce__ methods to handle Azure deployment names. #1002
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
There was a bug in how we pickled OpenAI class instances specifically with Azure. Azure OpenAI is a bit weird in that the "deployment name" of a model can be different than the normal OpenAI name. This makes it impossible to use the standard models from the OpenAIModels Enum. Instead, we typically created separate OpenAIModel instances. The problem was that when we pickled the OpenAI object, it saved just the model name, and then tried to recreate the model option by using the Enum. This change pickles the entire OpenAIModel class instead of just the name. Also adds a test that demonstrates the bug (and fix). The other tests were passing because they called the OpenAI methods directly and didn't pickle the object. The object will be pickled in basically any Sycamore job using ray.
HenryL27
approved these changes
Nov 6, 2024
| def test_azure_pickle(azure_llm): | ||
| pickled = pickle.dumps(azure_llm) | ||
| _ = pickle.loads(pickled) | ||
| assert True |
There was a problem hiding this comment.
maybe assert that the original and dumped/loaded are pointing at the same endpoint or something?
There was a problem hiding this comment.
I guess if the point is just 'this doesn't crash' then it dont matter
There was a problem hiding this comment.
Yes, my intention here was that it doesn't raise. I think we can rely on the pickle protocol assuming that it doesn't crash.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.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.
There was a bug in how we pickled OpenAI class instances specifically with Azure. Azure OpenAI is a bit weird in that the "deployment name" of a model can be different than the normal OpenAI name. This makes it impossible to use the standard models from the OpenAIModels Enum. Instead, we typically created separate OpenAIModel instances. The problem was that when we pickled the OpenAI object, it saved just the model name, and then tried to recreate the model option by using the Enum. This change pickles the entire OpenAIModel class instead of just the name. Also adds a test that demonstrates the bug (and fix). The other tests were passing because they called the OpenAI methods directly and didn't pickle the object. The object will be pickled in basically any Sycamore job using ray.