DipeshChaudhary/nepali-gector-style-token-level-tag

尼泊尔语语法错误校正序列标注数据集，适用于训练GEC-ToR风格的模型，包含词级别校正标签，并包括正确和错误的句子，用于模型稳定性。数据集经过内容感知对齐算法处理，支持复杂的交换操作。分为训练集、验证集和校准集。

Nepali grammatical error correction sequence tagging dataset designed for training GEC-ToR-style models, containing word-level correction tags, including both correct and incorrect sentences for model stability. The dataset has been processed with a content-aware alignment algorithm to support complex swap operations. It is split into training, validation, and calibration sets.