CNR-ILC/gs-dataset-train

用于微调BERT模型以进行古希腊语文本填充的数据集。数据来源于多种资源，包括MAAT（机器可操作的古代文本语料库）、PDL（Perseus数字图书馆）、First1KGreek（古希腊前一千年的协作语料库）和TLG（古希腊语词典）。数据集分为训练集和开发集，训练集包含用于MLM的匿名文本块，开发集包含验证子集（包括P.Herc.块）。

Dataset in Ancient Greek used for fine-tuning BERT models for text-infilling in Ancient Greek. The texts come from a combination of resources including MAAT (Machine-Actionable Ancient Text corpus), PDL (Perseus Digital Library), First1KGreek (collaborative corpus of the first thousand years of Greek literature), and TLG (Thesaurus Linguae Graecae). The dataset is split into train (anonymized text blocks for MLM) and dev (validation subset including P.Herc. blocks).