Move model parameter above input/messages/file to make downstream clients more efficient for OpenAI API implementors #284
Conversation
This file contains 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
for CreateChatCompletionRequest, CreateEmbeddingRequest, CreateTranscriptionRequest, CreateTranslationRequest
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Outdated suggestions cannot be applied.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.
At $WORK we offer an OpenAI compatible API and route based on the model parameter.
Some clients, like the openai-python client, are generated from this API spec, so the python client always sends the input before the model.
This leads to us buffering the whole input, before being able to parse the model and route to the correct machine hosting that model. I filed an issue with the python client, but it would be hard to change there and it may not be the only project relying on the order of the API spec.
Someone would have to confirm, that this change actually changes the order in requests of openai-python. I tried but did not get far, since they use stainless in a github action.
I think it would be good to change the order of the parameters in the API spec to an improved order, so that downstream clients adopt it. Then, providers of the OpenAI API will be able to route larger requests and with less overall latency.
I wrote some more details in the openai/openai-python#1525 issue.
I changed the spec how I am imagining it currently only for CreateChatCompletionRequest, CreateEmbeddingRequest, CreateTranscriptionRequest, and CreateTranslationRequest. Those are the ones that I have looked into, but the issue may apply to other requests as well, and to other "meta" parameters apart from model.