This repository was archived by the owner on Sep 30, 2024. It is now read-only.
feat(cody-gateway): Add FLAGGED_MODEL_NAMES check#63013
Merged
Conversation
sqs
approved these changes
May 31, 2024
Co-authored-by: Quinn Slack <quinn@slack.org>
chenkc805
approved these changes
May 31, 2024
|
|
||
| func TestIsFlaggedRequest(t *testing.T) { | ||
| validPreamble := "You are cody-gateway." | ||
| flaggedModelNames := []string{"dangerous-llm-model"} |
Contributor
There was a problem hiding this comment.
maybe "expensive-llm-model" instead?
Contributor
There was a problem hiding this comment.
actually this is just a test file so ignore
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
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
3 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Cody Gateway may "flag" a request as a way to either identify clearly abusive traffic, and reject it before it reaches an LLM, or to allow for further analysis.
This PR adds the ability to specify a set of model names, and Cody Gateway will in-turn flag all requests for that specific model. This adds another knob for being able to identify abusive behavior and/or confirm that any sudden spikes in a particular LLM model are for legitimate reasons.
With this PR no "default" is provided. And a follow-up PR will need to be made in the
infrastructurerepo to add model names as applicable.model_usedreason won't erroneously ban the user. (Since in this case, the "flag" isn't a sign of anything nefarious.)Test plan
Updated existing unit tests.
Changelog
Add a FLAGGED_MODEL_NAMES env var for Cody Gateway. Any requests using a supplied model will be flagged.