elvanalabs/

sarcasm-statements-90

	Dataset Name : 90 Labelled Sarcasm Statements Dataset




	
	Dataset Description

This dataset contains 70 short English-language text statements, each manually labeled as either sarcastic (True) or non-sarcastic (False). The goal is to help researchers and developers train and evaluate models for sarcasm detection, especially in low-resource or few-shot settings. The dataset captures a mix of:

Dry sarcasm

Ironic praise

Overstated negativity or positivity

Context-independent… See the full description on the dataset page: https://huggingface.co/datasets/elvanalabs/sarcasm-statements-90.