daichira/

structeval-t-sft-v2-xml

This dataset is the full, refined, format-specific subset for StructEval-T, focusing exclusively on strictly validated XML transformations.

Total Samples: 4,503 Verified Quality: 100% strictly validated using AST/parsers (e.g. json.loads, xml.etree.ElementTree, yaml.safe_load). Only samples that successfully parse as valid XML without errors are included. Source: This is a split from the unified structeval-t-sft-v2 dataset.… See the full description on the dataset page: https://huggingface.co/datasets/daichira/structeval-t-sft-v2-xml.