vihsd
Datasets
All datasets matching “vihsd”vihsd
Dataset Card for Dataset Name
ViHSD - Vietnamese Hate Speech Detection Dataset
Dataset Details
Dataset Description
The dataset contains about 33K annotated comments from social networks. Each has one of three labels: HATE, OFFENSIVE, CLEAN
Uses
Use directly from the Hugging face dataset loader
Direct Use
from datasets import load_dataset
train = load_dataset("sonlam1102/vihsd", split="train")
dev = load_dataset("sonlam1102/vihsd"… See the full description on the dataset page: https://huggingface.co/datasets/uitnlp/vihsd.ViHSD
Dataset Card for ViHSD
1. Dataset Summary
ViHSD (Vietnamese Hate Speech Detection) is a unified CSV‐based dataset for hate speech detection in Vietnamese social media texts. It consolidates train/dev/test splits into a single file and uses a type field to distinguish:
Train: ~26K examples
Dev: ~3.3K examples
Test: ~3.3K examples
Each example is a short user comment labelled as one of:
HATE
OFFENSIVE
CLEAN
This version is adapted from the original sonlam1102/vihsd… See the full description on the dataset page: https://huggingface.co/datasets/visolex/ViHSD.vihsd-explainable-dpo
vihsd-explainable-dpo
DPO preference dataset derived from vominhmanh/vihsd-explainable for Direct Preference Optimization (DPO).
Each example is a preference pair (chosen vs rejected) for the same prompt.
Schema (per example):
prompt (string): the original SFT prompt for Vietnamese moderation (user instruction).
chosen (string): JSON string with keys explanation, evidence, label — preferred (longer/more informative) explanation.
rejected (string): JSON string with keys explanation… See the full description on the dataset page: https://huggingface.co/datasets/vominhmanh/vihsd-explainable-dpo.vihsd-explainable-dpo
vihsd-explainable-dpo
DPO preference dataset derived from vominhmanh/vihsd-explainable for Direct Preference Optimization (DPO).
Each example is a preference pair (chosen vs rejected) for the same prompt.
Schema (per example):
prompt (string): the original SFT prompt for Vietnamese moderation (user instruction).
chosen (string): JSON string with keys explanation, evidence, label — preferred (longer/more informative) explanation.
rejected (string): JSON string with keys… See the full description on the dataset page: https://huggingface.co/datasets/vlinhd11/vihsd-explainable-dpo.UIT-ViHSD
UIT-ViHSD Dataset
This is a copy instance of the original dataset provided by UIT. Please visit https://nlp.uit.edu.vn/datasets to obtain a usage permission before using this dataset.
ViHSD