Incorrect sentiment analysis from twitter-roberta-base-sentiment-model

Hi, Brand new to Hugging Face and a non-coder. :slight_smile: I did follow the directions:

and successfully created a google sheet ( yay!! :blush: ) that collects and analyzes tweets on my chosen topic “offshorewind”. The sentiment analysis model used is

However, when looking at the sentiments assigned, I see that they are not correct, despite a high confidence score.

So far all 4 tweets analyzed have been categorized as “negative” however I would say that only row 4 is negative, while row 5 should be positive and rows 2 and 3 are neutral.

Any suggestions on how to resolve/improve analysis? Thanks!