datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
ML-ESG-3-TrainML-ESG-3-Testesgard-esrs-datapoints
Esgard — ESRS Datapoint Library (open)
Curated CSRD/ESRS datapoints derived from Commission Delegated Regulation (EU) 2023/2772 (v0.1: ~70 datapoints across ESRS 2, E1, S1, G1).
Used by the Esgard compliance copilot for document-to-datapoint mapping. Fields: id, standard, disclosure_requirement, name, datapoint_type, description, search_terms, unit_hints, regulation_version, phase_in, mandatory.
Licensing: derived from EU public legal texts; see the Esgard repository (MIT).
esg_report_data_10k_jsonesgf-dataset-v0.26ESG_ratingsES-Geoentities-XLDataset consisting of 158649 html-cleaned samples taken from the Dataset 10 of the Epstein Files.
Samples WITH Location Entity: 52883
Samples WITHOUT Location Entity: 105766
The dataset was labeled using a weak labeling approach using qwen3 70b.
It is BIO-tagged, B-ADDR and I-ADDR mark location entities.
esgdata