GPT-4 Turbo (preview)

[teetone]

Tested with MedQA and NarrativeQA with just 10 instances. It seems to generate reasonable completions.

Information about the model: https://help.openai.com/en/articles/8555510-gpt-4-turbo

[JosselinSomervilleRoberts]

@teetone the test that is failing is in test_model_properties. This test was used for the refactorization and I think we should remove it. I will send a PR for this. In the meantime, you can simply add the corresponding ModelDeployment to the test and everything will pass.

[JosselinSomervilleRoberts]

LGTM once you fix the test

[teetone]

@teetone the test that is failing is in test_model_properties. This test was used for the refactorization and I think we should remove it. I will send a PR for this. In the meantime, you can simply add the corresponding ModelDeployment to the test and everything will pass.

@JosselinSomervilleRoberts Actually, your test is good. It made me realize that I should add a separate window service for the 128K context length :)