NVIDIA multilingual response safety dataset
Over half a million examples in 12 languages for training LLM safety filters (guard models).
- Text classification
- Text generation
- English
- Arabic
- German
- +9
Hugging Face datasets with plain-language descriptions: what's inside and what models to train on them. 4,788 in the catalog so far, with new ones added every hour.
Answered by Gemini 4 · the chat is kept only in this tab
Over half a million examples in 12 languages for training LLM safety filters (guard models).
English quotes from Goodreads for multi-label classification and text generation.
100k to 1M pairs of Chinese sentences with labels for training logical inference and deduplication models.
A corpus of English financial tweets labeled across 20 thematic categories.
Spoken version of the MASSIVE text corpus for ASR, intent detection, and slot filling tasks.
Synthetic financial documents in six languages with personal data markup for NER training.
90k text fragments with language labels for classification.
English dataset for evaluating language model skills in cognitive behavioral therapy (CBT).
A crowdsourced corpus of branching social stories in English for training and evaluating ethical reasoning.
Task collections where a state and question require a typed decision: yes/no, class, number.
About 500k GPT-4 answers filtered by FLAN annotations for training instruction models.
Annotated customer emails and support responses for support automation.
Tabular and text data on pushes, pull requests, releases, and discussions from public repositories.
English medical texts for sentence classification and relation extraction between drugs, doses, and adverse effects.
1 million to 10 million English examples with chains of thought for generation and classification.
A classification dataset of English tweets annotated for hate speech, offensive language, and neutral text.
Metadata of 2.99 million scientific publications from 1991 to 2026 in parquet format for search and analytics.
Multilingual set for training LLM safety filters: jailbreaks and injections.
A balanced text set for training malicious prompt classifiers in English.
Real English AI agent trajectories for evaluating the safety of their actions.
A large corpus of 18+ Chinese novels for language modeling and classification tasks.
Text corpus for comparing human and ChatGPT answers; suitable for classification and similarity evaluation.
10,331 community-rated prompts for training and evaluating prompt ranking.
1,535 climate claims from the internet and 7,675 annotated evidence sentences from English Wikipedia.