Add embedding-based detector #2
Open
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
m-misiura
reviewed
Jan 22, 2025
| @@ -0,0 +1,37 @@ | |||
| # Embedding Classification Detector | |||
|
|
|||
| # Setup | |||
There was a problem hiding this comment.
Perhaps it could be useful to state that local Python must match up Python in the Containerfile? At present, python 3.9 will be downloaded inside the container, which may warrant upgrading?
There was a problem hiding this comment.
can this be moved to a shared utils folder so that other detectors that require training can use this class ?
|
|
||
| sys.path.insert(0, os.path.abspath("")) | ||
| # from common.scheme import TextDetectionHttpRequest, TextDetectionResponse | ||
| import os |
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Outdated suggestions cannot be applied.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.
Adds a framework for defining detections based on a text-embedding classifier. The default configuration here uses MMLU as the training data for the classification and creates a multi-label text classifier to infer which of the 61 MMLU subjects a particular body of text belongs to. The detector endpoint then accepts the following arguments:
contents: List of texts to classifyallowList: Allowed list of subjects: all inbound texts must belong to at least one of these subjects to avoid flagging the detectorblockList: Blocked list of subjects: all inbounds texts must not belong to any of these subjects to avoid flagging the detector.threshold: Defines the maximum distance a body of text can be from the subject centroid and still be classified into that subject. The default value is 0.75, while athresholdof >10 will classify every document into every subject. As such, values 0<threshold<1 are recommended.