"I am gay." sentence is classified as NEGATIVE with score 0.99

🤗Transformers
1

This code gives the following output:

from transformers import pipeline
classifier = pipeline(“sentiment-analysis”)
#classifier(“I’ve been waiting for a HuggingFace course my whole life.”)
classifier(“I am gay.”)

output:

[{‘label’: ‘NEGATIVE’, ‘score’: 0.9939725399017334}]

“I am gay.” sentence is classified as NEGATIVE with score of 0.99.
Should the dataset be modified or processed again?

2

That’s part of the warnings about pretrained models included here. Every fine-tuned versions gets the bias of the original pretrained model.