Skip to content

Incorrect Toxic/NonToxic Prediction Value #10106

Closed
@lzinga

Description

@lzinga

After going through the documents for the text sentiment ML I wasn't getting the expected results as stated in the docs. After a bit more testing I downloaded the actual files and came out with the same result.

Sentiment: This is a very rude movie | Prediction: Toxic | Probability: 0.5297049
Sentiment: He is the best, and the article should say that. | Prediction: Toxic | Probability: 0.9918675

These two lines are both predicting that the sentiments are "Toxic" however based on the docs and what
should naturally be expected the second line should be considered "Non-Toxic" which is being shown as "Toxic" instead.

The below result is what the expectation would be if working correctly.

Sentiment: This is a very rude movie | Prediction: Toxic | Probability: 0.4585565
Sentiment: He is the best, and the article should say that. | Prediction: Not Toxic | Probability: 0.9924279


Document Details

Do not edit this section. It is required for docs.microsoft.com ➟ GitHub issue linking.

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions