V0.25 #152
Merged
V0.25 #152
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
Contributor
hverlind
commented
Feb 15, 2024
hverlind
commented
- Support for latest GPT-3.5 Turbo and GPT-4 Turbo Preview models (0125)
- Support for new OpenAI embedding models (text-embedding-3-small, text-embedding-3-large)
- Increased prompts per project limit from 100 to 250
This is too fragile if we e.g. also need to map 'gpt-3.5-turbo' which is a prefix of 'gpt-3.5-turbo-16k'.
Latest announcement of OpenAI is that both gpt-3.5-turbo and gpt-3.5-turbo-16k will start pointing to gpt-3.5-turbo-0125 on Feb 16. That means there is no real point in exposing two separate models in PlayFetch anymore. But rather than migrating all existing versions, we can keep the gpt-3.5-turbo pointing to the same fixed model as before (at least until it gets discontinued in June), and only have the gpt-3.5-turbo-16k alias actually point to the latest model.
This would happen automatically (if we didn't remap) on Feb 16, but we need to update metadata anyway (otherwise the changed pricing wouldn't be reflected in PlayFetch).
This is easy because we didn't hardcode the embedding model even though there was only a single option at the time. (https://openai.com/blog/new-embedding-models-and-api-updates)
Goal is to co-locate more semantically. Components folder now contains only the generic components and its subfolders have now become sibling folders within the client folder.
Now that we have thematic component folders within the client folder, we can move some helper files that were already there into them.
This logic does not need to live in common either.
Might run into some performance issues but this would only be the case for teams who actually need this many prompts in a single project, so it feels pretty safe to increase this limit without risking to break anything else, plus it's easy to isolate the change to only prompt queries by simply making the default limit entity dependent (rather than having to start auditing all call sites).
There is also a new gpt-4-turbo-preview alias available that could be used to do this automatically, but that would require migrating our models which doesn't feel worth it right now.
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.