datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
ar_sarcasm
Dataset Card for ArSarcasm
Dataset Summary
ArSarcasm is a new Arabic sarcasm detection dataset.
The dataset was created using previously available Arabic sentiment analysis
datasets (SemEval 2017
and ASTD) and adds sarcasm and
dialect labels to them.
The dataset contains 10,547 tweets, 1,682 (16%) of which are sarcastic.
For more details, please check the paper
From Arabic Sentiment Analysis to Sarcasm Detection: The ArSarcasm Dataset
Supported Tasks and… See the full description on the dataset page: https://huggingface.co/datasets/iabufarha/ar_sarcasm.Sarcasm_News_HeadlinePast studies in Sarcasm Detection mostly make use of Twitter datasets collected using hashtag based supervision but such datasets are noisy in terms of labels and language. Furthermore, many tweets are replies to other tweets and detecting sarcasm in these requires the availability of contextual tweets.
To overcome the limitations related to noise in Twitter datasets, this Headlines dataset for Sarcasm Detection is collected from two news website. TheOnion aims at producing sarcastic versions… See the full description on the dataset page: https://huggingface.co/datasets/raquiba/Sarcasm_News_Headline.sarcasm-on-redditCopied from: Sarcasm on Reddit. https://www.kaggle.com/datasets/danofer/sarcasm
Which in turn came from:
@unpublished{SARC,
authors={Mikhail Khodak and Nikunj Saunshi and Kiran Vodrahalli},
title={A Large Self-Annotated Corpus for Sarcasm},
url={https://arxiv.org/abs/1704.05579},
year=2017
}
license: mit
language:
- en
figlang2020-sarcasmhttps://github.com/EducationalTestingService/sarcasm
@inproceedings{ghosh-etal-2020-report,
title = "A Report on the 2020 Sarcasm Detection Shared Task",
author = "Ghosh, Debanjan and
Vajpayee, Avijit and
Muresan, Smaranda",
booktitle = "Proceedings of the Second Workshop on Figurative Language Processing",
month = jul,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
url =… See the full description on the dataset page: https://huggingface.co/datasets/tasksource/figlang2020-sarcasm.kor_sarcasm
Dataset Card for Korean Sarcasm Detection
Dataset Summary
The Korean Sarcasm Dataset was created to detect sarcasm in text, which can significantly alter the original meaning of a sentence. 9319 tweets were collected from Twitter and labeled for sarcasm or not_sarcasm. These tweets were gathered by querying for: 역설, 아무말, 운수좋은날, 笑, 뭐래 아닙니다, 그럴리없다, 어그로, irony sarcastic, and sarcasm. The dataset was pre-processed by removing the keyword hashtag, urls and mentions of the user… See the full description on the dataset page: https://huggingface.co/datasets/SpellOnYou/kor_sarcasm.Metaphor_Driven_Sarcasm_dataset_source
Metaphor_Driven_Sarcasm_dataset_source
Dataset Description
This is a Chinese metaphor-driven sarcasm detection dataset, containing 235,909 texts with a four‑layer annotation structure. The dataset is converted from the BSD‑MM‑V3 dataset and is specifically designed for studying the role of metaphor in sarcastic expressions.
Dataset Summary
Volume: 235,909 texts
Language: Chinese
Annotation Levels: 4 (sarcasm category, sarcasm style, metaphor… See the full description on the dataset page: https://huggingface.co/datasets/Arrebol-yzq/Metaphor_Driven_Sarcasm_dataset_source.multimodal_sarcasm_detectioncopy of data-of-multimodal-sarcasm-detection
# usage
from datasets import load_dataset
from transformers import CLIPImageProcessor, CLIPTokenizer
from torch.utils.data import DataLoader
image_processor = CLIPImageProcessor.from_pretrained(clip_path)
tokenizer = CLIPTokenizer.from_pretrained(clip_path)
def tokenization(example):
text_inputs = tokenizer(example["text"], truncation=True, padding=True, return_tensors="pt")
image_inputs = image_processor(example["image"], return_tensors="pt")… See the full description on the dataset page: https://huggingface.co/datasets/quaeast/multimodal_sarcasm_detection.SARC_Sarcasm
SARC_Sarcasm
Dataset Summary
A large corpus for sarcasm research and for training and evaluating systems for sarcasm detection is presented. The corpus comprises 1.3 million sarcastic statements, a quantity that is tenfold more substantial than any preceding dataset, and includes many more instances of non-sarcastic statements. This allows for learning in both balanced and unbalanced label regimes. Each statement is self-annotated; that is to say, sarcasm is labeled by… See the full description on the dataset page: https://huggingface.co/datasets/CreativeLang/SARC_Sarcasm.conversational-sarcasm-benchmark
Conversational Sarcasm Benchmark — Audio-Grounded, Metadata-Only
A benchmark of 1,168 conversational sarcasm units drawn from 64 English-language
YouTube videos (predominantly stand-up comedy and comedic conversation). Every unit
pairs a short target utterance with the preceding context that makes its
figurative reading available, and carries a categorical label plus a free-text rationale.
This repository contains no audio. It ships annotations, transcriptions, and the
source… See the full description on the dataset page: https://huggingface.co/datasets/darksyntax0/conversational-sarcasm-benchmark.task1489_sarcasmdetection_tweet_classification
Dataset Card for Natural Instructions (https://github.com/allenai/natural-instructions) Task: task1489_sarcasmdetection_tweet_classification
Additional Information
Citation Information
The following paper introduces the corpus in detail. If you use the corpus in published work, please cite it:
@misc{wang2022supernaturalinstructionsgeneralizationdeclarativeinstructions,
title={Super-NaturalInstructions: Generalization via Declarative Instructions on 1600+… See the full description on the dataset page: https://huggingface.co/datasets/Lots-of-LoRAs/task1489_sarcasmdetection_tweet_classification.beyond_sarcasm_detection_sftreddit-sarcasmsarcasm
Dataset Card for Sarcasm Detection Dataset
Dataset Details
Dataset Description
The Sarcasm Detection Dataset is designed for identifying instances of sarcasm in text. The dataset aims to address difficulties in sarcasm detection due to the subjective and contextual nature of language.
Uses
Direct Use
The dataset can be used for training machine learning models to detect sarcasm in text, which has applications in sentiment analysis, social… See the full description on the dataset page: https://huggingface.co/datasets/SarcasmNet/sarcasm.hinglish-sarcasm-datasetThis repository contains the Hindi-English Code-mixed dataset used for the fine-tuning and testing experiments as mentioned in the paper https://arxiv.org/abs/2602.21933
Telugu-SarcasmDo cite the below references for using the dataset:
@article{marreddy2022resource, title={Am I a Resource-Poor Language? Data Sets, Embeddings, Models and Analysis for four different NLP tasks in Telugu Language},
author={Marreddy, Mounika and Oota, Subba Reddy and Vakada, Lakshmi Sireesha and Chinni, Venkata Charan and Mamidi, Radhika},
journal={Transactions on Asian and Low-Resource Language Information Processing}, publisher={ACM New York, NY} }
@article{marreddy2022multi… See the full description on the dataset page: https://huggingface.co/datasets/mounikaiiith/Telugu-Sarcasm.friends_chandler_bing_sarcasmukr-tweets-sentiment-toxicity-sarcasmsarcasm
Dataset Card for Dataset Name
Dataset Summary
This dataset card aims to be a base template for new datasets. It has been generated using this raw template.
Supported Tasks and Leaderboards
[More Information Needed]
Languages
[More Information Needed]
Dataset Structure
Data Instances
[More Information Needed]
Data Fields
[More Information Needed]
Data Splits
[More Information Needed]
Dataset Creation… See the full description on the dataset page: https://huggingface.co/datasets/daniel2588/sarcasm.CN-SarcasmBench📘 CN-SarcasmBench
CN-SarcasmBench is a high-quality Chinese benchmark designed to evaluate large language models on sarcasm understanding, classification, and response generation in real-world online conversations. Based on over 1,200 Bilibili comment threads, it offers rich contextual data and multi-level tasks to reveal the performance gap between current models and human-level sarcastic comprehension. Ideal for researchers and developers aiming to improve nuanced language understanding… See the full description on the dataset page: https://huggingface.co/datasets/Devon018/CN-SarcasmBench.Multi-Sarcasm
Multi-Sarcasm Dataset
Multi-Sarcasm is a curated conversational dataset containing 23,715 parent–reply pairs, combining Reddit and chatbot dialogues. Each instance includes a comment (comment), its parent comment (parent_comment), a binary label (label) for sarcasm, and a graded sarcasm level (sarcasm_level).
The dataset was re-annotated using a hybrid LLM-assisted pre-annotation followed by systematic human validation, ensuring consistency, diversity, and reliability.… See the full description on the dataset page: https://huggingface.co/datasets/Dalilame/Multi-Sarcasm.self-annotated_reddit_climate_comment
Dataset Card for Self-annotated Reddit Climate Comment
Dataset Structure
This JSON example represents an example portion of the dataset. This nested structure allows for efficient navigation and analysis of posts, comments, and replies within specific subreddit communities and individual posts.
{
"id": "1006cei",
"post_title": "Amazing Water Filter Invention",
"post_author": "User123",
"post_body": "Check out this incredible water filter!",
"post_url":… See the full description on the dataset page: https://huggingface.co/datasets/SarcasmNet/self-annotated_reddit_climate_comment.tweet_sarcasmsarcasm_headlines_multilingual
Dataset Card for Multilingual Sarcasm Detection
Dataset Summary
Dataset consists of news article headlines in Dutch, English and Italian. The news article headlines are both from actual news sources and sarcastic/satirical newspapers. The news article is determined sarcastic/non-sarcastic based on the news article source.
The sources of news articles are:
The Huffington Post (en, non-sarcastic)
The Onion (en, sarcastic)
NOS (nl, non-sarcastic)
De Speld (nl, sarcastic)
Il… See the full description on the dataset page: https://huggingface.co/datasets/helinivan/sarcasm_headlines_multilingual.embed_sarcasm_seed500_v2tweet_sarcasmSarcasm-dataset
Sarcasm Dataset
This dataset contains sarcastic sentence along with their binary label
Dataset Desciption:
Number of Rows: 5,000
Number of Columns: 2
Column Names: 'Tweet', 'Slang (yes/no)'
Description: The dataset contains tweets annotated for the use of slang. It includes a binary label ('yes' or 'no') indicating the presence of slang in each tweet.
SARCASM_VS_NON_SARCASMmuse-sarcasm-explanation
MuSe: Multimodal Sarcasm Explanation (Reformatted)
This repository provides a Hugging Face-compatible version of the MuSe (MORE) dataset.
Modifications in this version
To make the dataset easier to use with the datasets library, the following changes were made:
Unified Schema: Merged separate OCR and Non-OCR files into a single test split.
Metadata Flags: Added an is_ocr (boolean) column to distinguish between image types.
Image Integration: Converted image paths into a… See the full description on the dataset page: https://huggingface.co/datasets/alita9/muse-sarcasm-explanation.sarcasm-2.0-redditembed_sarcasm_seed500