datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
NFLWR2025CLEANED
NFLWR2025CLEANED
NFLWR2025CLEANED is a cleaned ChatML instruction dataset built from SebastianAndreu/24679_NFL_WR_Dataset_2025. It is designed for supervised fine-tuning of models that answer questions about 2025 NFL wide receiver statistics.
The dataset transforms structured receiver statistics into conversational examples covering lookup, comparison, efficiency, usage, quarter splits, player totals, and concise scouting-style explanations.
Dataset Summary
Source… See the full description on the dataset page: https://huggingface.co/datasets/clarkkitchen22/NFLWR2025CLEANED.nfl-play-normalization-2019-2022
NFL play normalization, 2019-2022
This dataset contains 188452 real raw nflverse play descriptions from 2019 through 2022. Each record pairs the raw description with a canonical JSON expected record for NFL play normalization.
License and attribution
The source nflverse releases are CC BY 4.0. This derived dataset retains source season, game ID, play ID, source URLs, and SHA-256 checksums in metadata/training-manifest.json. Attribute nflverse when using this… See the full description on the dataset page: https://huggingface.co/datasets/AdamRoch/nfl-play-normalization-2019-2022.nfl-play-normalization-v2
NFL Play Normalization V2 Data-Efficiency Curve
This dataset contains four nested training sets with 3,125, 6,250, 12,500,
and 25,000 unchanged real nflverse play descriptions from the 2019 through
2022 seasons. The v2 selection targets laterals, repeated fumbles, accepted
penalties after turnovers, and descriptions with several yardage clauses.
The curve manifest pins every dataset and point-manifest checksum. The review
file explains the deterministic selection and the… See the full description on the dataset page: https://huggingface.co/datasets/AdamRoch/nfl-play-normalization-v2.