Which model to use to classify a sentence as suggestion or not?

I have a dataset containing sentences. Each sentence is a part of reviews given by customers. I want to classify each sentence as a suggestion or not.
Example-
Sentence: They should hire more dentists.
Output: Suggestion

Sentence: I am disappointed with their treatment.
Output: Not Suggestion

Sentence: I did not like the front desk staff behaviour, for they were all rude.
Output: Not suggestion

Sentence: Maybe, they should keep the hospital very clean.
Output: Suggestion

Sentence: I would not recommend this hospital to anyone.
Output: Not suggestion

Sentence: It would have been nice if they had told me earlier about my appointment cancellation so that my time would not have been wasted.
Output: Suggestion.

I have an unlabeled dataset. Dataset has only sentences which I need to classify.

ZeroShotClassificationPipeline ZeroShotClassificationPipeline can use models that have been fine-tuned on an NLI task. Models that have been fine-tuned on an NLI task

Few-shot prompting Few-shot prompting is another way to classify sentences with a large language model and a few examples.