Jayveersinh-Raj
/

PolyGuard

Text Classification

Cross-lingual-nlp

zero-shot-transfer

toxicity-analysis

abuse-detection

multilinguality

Model card Files Files and versions

Jayveersinh-Raj commited on May 7, 2023

Commit

3dccfc7

·

1 Parent(s): c4700ad

Update README.md

Files changed (1) hide show

README.md +1 -1

README.md CHANGED Viewed

@@ -188,7 +188,7 @@ Moreover, our model flags mostly severe toxicity, since toxicity is a subjective
 ## Bias, Risks, and Limitations
 <!-- This section is meant to convey both technical and sociotechnical limitations. -->
-Toxicity is a subjective issue, however the model is very well balanced to flag mostly severe toxicity. The model has never flagged non toxic sentence as toxic. Its performance on non toxicity is 100%, making it a very good choice for the purpose of flagging or blocking users. In addition, if the language is very low resource, then the model might misclassify, but the performance is still good.
 ### Recommendations

 ## Bias, Risks, and Limitations
 <!-- This section is meant to convey both technical and sociotechnical limitations. -->
+Toxicity is a subjective issue, however the model is very well balanced to flag mostly severe toxicity. The model has never flagged non toxic sentence as toxic. Its performance on non toxicity is 100%, making it a very good choice for the purpose of flagging or blocking users. In addition, if the language is very low resource, and/or distant from English, then the model might misclassify, but the performance is still good.
 ### Recommendations