Minhyun/ainize-dart100-reproduction-20260911

DART task datasets for Ainize reproduction

This bundle preserves 100 distinct task configurations from the existing DART-derived datasets. Each JSONL file is byte-identical to its authenticated Ainize canonical download. The manifest links configuration names, row counts, SHA256 values and Ainize dataset IDs.

Source: Financial Supervisory Service Open DART, https://opendart.fss.or.kr/. These are historical disclosure-derived answers, not a promise of current company information or financial advice. Source reuse terms and publisher rights must be confirmed before public upload; no new license is asserted by this packaging step.

The train split contains the question/answer records used by Ainize. alt_prompt, where present, is an alternate question for the same fact and is not an independent dataset. A dataset ID is not a trained model ID. Uploading this repository does not demonstrate 100 successful teach jobs, inference support, a model deployment or a public Ainize listing.

After upload, verify the Hub commit and every configuration by downloading at that exact revision. Do not treat this README or the local manifest as proof of remote publication.

Format reference: https://huggingface.co/docs/datasets/en/repository_structure

The HF configurations read typed Parquet files so date-like answers and leading zeros remain strings. The original canonical JSONL files are also retained byte-for-byte; the manifest binds both formats.