arXiv subjects

Eric Cullhed

Publications and source records attributed to Eric Cullhed.

3 recordsLinked to original sources

arXiv2026

Stoicheia: Character-Level Masked Diffusion for Ancient Greek Textual Restoration, Parsing, and Metrical Scansion

We introduce Stoicheia, a 405M-parameter character-level masked-diffusion encoder for Ancient Greek whose input factors into five aligned, independently maskable planes: letters, word and sentence boundaries, diacritics, capitalization, and punctuation. A single backbone can therefore restore lacunae, re-segment, accentuate, and punctuate unspaced text without task-specific retokenization. We pretrain it on an open, revision-pinned corpus of 380M words and release eleven checkpoints: ten rotated, decontaminated folds, guaranteeing that for any given literary passage at least one released model has never seen its text, and one with no exposure to documentary texts. Three experiments - reconstruction of damaged inscriptions and papyri, morphosyntactic tagging and dependency parsing, and macronization with metrical scansion - each carry a matched random-initialization control, isolating what character-level diffusion pretraining contributes: 5.6 CER points on inscription reconstruction, 12.9 LAS on parsing, and 6.0 points of balanced accuracy on macronization. On Ithaca's own test split, with identical frozen samples and strict scoring, Stoicheia reduces character error relative to both prior state-of-the-art systems, from 24.6 (Ithaca) and 23.5 (its 2025 Aeneas-framework successor) to 15.5, and raises top-1 accuracy from 63.0 and 64.0 to 74.5.

cs.CL↗

arXiv2026

Automatic Annotation of Ancient Greek Vowel Length

Prior work in Ancient Greek NLP relies on corpora that do not disambiguate the phonemic vowel length of alpha, iota, and ypsilon, together known as the dichrona. Depending on lexeme, morphology, sandhi, syntax, and conventions of period, genre, and verse form, each of these letters can represent either a long or a short vowel. Deciding and marking the correct length is known as "macronizing", a long-tail problem given the sheer mass of word forms and the context dependency of individual instances. No macronized corpus of Ancient Greek is publicly available at scale, so a stand-alone macronizer is needed. While previous work has shown how to build a static, corpus-bespoke vowel-length dictionary, the present paper constructs the first general-purpose macronizer for arbitrary Ancient Greek input. Given input carrying lemma, part-of-speech, and morphological annotation in the standard CoNLL-U format, a set of recursive modules lets less common word forms inherit markup from more common forms of the same lexical word. The macronizer's chief application is generating training data for machine learning: we show that a small character-level transformer trained on the macronizer's own output learns to generalize past the cases the rule-based system leaves unmarked, matching or exceeding its accuracy on a gold-standard, manually annotated benchmark of verse and prose. We also show that macronization can improve downstream prosodical NLP tasks like verse scansion.

cs.CL↗

arXiv2024

Instruct-Tuning Pretrained Causal Language Models for Ancient Greek Papyrology and Epigraphy

This article presents an experiment in fine-tuning a pretrained causal language model (Meta's Llama 3.1 8B Instruct) to assist with restoring missing or illegible characters in ancient Greek inscriptions and documentary papyri. Utilizing a straightforward instruction-based approach and a 95%/5% train/test split, the papyrus restoration model achieved a character error rate (CER) of 14.9%, a top-1 accuracy of 73.5%, and a top-20 accuracy of 86.0% for sequences up to 10 characters. A model was also fine-tuned for geographic attribution, reaching a top-1 accuracy of 66.4% and a top-3 accuracy of 79.9%. In chronological attribution, it demonstrated an average deviation of 21.7 years from the actual terminus post/ante quem, with a median deviation of 0 years. For inscriptions, the restoration model achieved a CER of 20.5%, a top-1 accuracy of 63.7%, and a top-20 accuracy of 83.0% for sequences up to 10 characters. In geographic attribution, it attained a top-1 accuracy of 75.0% and a top-3 accuracy of 83.7%, while in dating, it had an average deviation of 37.1 years and a median deviation of 3 years from the actual date range. Benchmarked against the state-of-the-art model (Ithaca) on a shared test set and on recently edited inscriptions, the instruction-tuned models excelled in text restoration, while also offering the practical advantage of ignoring spaces during reconstruction, which aligns with the scriptio continua of ancient textual artifacts. However, their performance in geographic and chronological attribution was lower than Ithaca's. To evaluate the approach in a more even setup, the instruction model was retrained with an 80%/10%/10% train-validation-test split, and still outperformed Ithaca in text restoration. The results suggest that fine-tuning larger pretrained causal language models using instruction templates for emendations and conjectures to ancient texts holds promise.

cs.CL↗