Ancient Greek LLM Stylometry: a Resource and Evaluation

A chunk-level corpus and evaluation of state-of-the-art LLM-generated Ancient Greek (GPT-5.5, Claude 4.8; with GPT-4o and Claude 3.5 as a longitudinal baseline) compared stylometrically and linguistically to human Attic orators. Includes code, prompts, feature-extraction and embedding pipelines, and analyses. Code is released under MIT; the released data/corpus is released under CC BY-SA 4.0 (the human Greek texts derive from Perseus Digital Library editions and inherit the ShareAlike term). Prepared as a language resource for Language Resources and Evaluation (Springer).