Zero-shot sentiment with a fixed label set

The prompt

Classify the text into neutral, negative, or positive
Text: I think the food was okay.
Sentiment:

Expected result

"Neutral" — a single-word label with no extra commentary, matching the exact three-way scheme given in the instruction.

Why it works

The baseline zero-shot classification template: naming the exact label set up front keeps the model's output constrained and easy to parse, rather than it inventing its own wording.