This is a dataset primarily for the detection of machine-authored, hallucinated, or adversarial Latin-like text, however, it also contains human authored text, mined from Acervus, and annotated by multiple LLMs with:

5 different Latin span relations
2 different categorical descriptions
2 different machine written modifications

Relations

NLI Logical Entailment (premise-hypothesis)
Question Answering (question-answer)
Quote Attribution (quote-attribution)
Dialogue e.g. chat (utterance-reply)
Prompt-Response, e.g. temporal succession, narrative sequence, or thematic association (prompt-response)

Categorical

Text Genre (ecclesiastical, historiography, law, treatise, poetry, rhetoric, general, satire)
Adversarial Highly-Latin-Like Text (Pseudo-Latin, Lorem Ipsum, & Hallucinated Latin that evaded aggressive lexical filtering)

Machine Modification

Human Original vs. Machine Authored Back Translation (human-machine)
Original vs. Machine Authored Translation (lat-eng)

Counts

models:

gpt-6.1-sol: 17544
claude-opus-5-5: 36591

relations:

machine-authorship: 18627
machine-translation: 13342
quote-attribution: 5120
prompt-response: 4122
premise-hypothesis: 5671
question-answer: 1157
dialogue: 630

Citation

If you use this, please cite:

@misc{mccarthy2026Simulacrum,
  author       = {McCarthy, Aidan},
  title        = {{Simulacrum}: Machine Authored and Annotated Latin},
  year         = {2026},
  howpublished = {\url{https://huggingface.co/datasets/aimgo/Simulacrum}},
  note         = {Dataset}
}

Downloads last month: 113

Total file size:

26.9 MB

You need to agree to share your Hugging Face information to access this dataset

Si machina nullo moderatore aberrasset...

About