datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
MUCH-trash-only-for-reproducibility
[Trash-only dataset] MUCH: A Multilingual Claim Hallucination Benchmark
Jérémie Dentan1, Alexi Canesse1, Davide Buscaldi1, 2, Aymen Shabou3, Sonia Vanier1
1LIX (École Polytechnique, IP Paris, CNSR), 2LIPN (Université Sorbonne Paris Nord), 3Crédit Agricole SA
Important Notice: Trash-Only
This dataset contains only the samples that were filtered out from the MUCH benchmark due to insufficient quality, referred to as “trash.” It is provided for research reproducibility… See the full description on the dataset page: https://huggingface.co/datasets/orailix/MUCH-trash-only-for-reproducibility.trash-mult-dpotrash250-max-tokens-trash-15mask3k-trash-ttt-022225-step2-mastertrash-mult-dpo-23k-trash-ttt-022225-step13k-trash-ttt-022225-step1-master3k-trash-ttt-022225-step3-master3k-max-tokens-trash-15maskvi-error-correction-super-small-trash