datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
LongLaMP
LongLaMP Dataset
Dataset Summary
LongLaMP is a comprehensive benchmark for personalized long-form text generation. The dataset is designed to evaluate and improve the performance of language models in generating extended, personalized content across various domains and tasks.
Our dataset consists of multiple tasks focusing on different aspects of long-form text generation, including:
Personalized Email Completion
Personalized Abstract Generation
Personalized Review… See the full description on the dataset page: https://huggingface.co/datasets/LongLaMP/LongLaMP.longlamp-ag-stream
LongLaMP abstract-generation — per-user update streams
A re-curation of LongLaMP/LongLaMP, config abstract_generation_temporal
into per-user update streams: each author's whole publication history, in
chronological order, packed into disjoint chunks, where every item is both a
condition item and a possible target. Built for training a recurrence across
adapter updates, θ_t = f(chunk_t, θ_{t-1}), rather than a single
condition→output pass.
Splits are by author, 22,421 train / 200… See the full description on the dataset page: https://huggingface.co/datasets/LafouCC/longlamp-ag-stream.LongLaMP-2LongLaMP-3LongLaMP-4longlamp-ai-detection
author-ai-text-detect/longlamp-ai-detection
A fork of LongLaMP/LongLaMP (user-setting configs) with model
completions attached. Every row is a LongLaMP sample with its original fields intact — the
author field, input, output and the full profile — plus a completions list holding
the model-written texts for that same task input.
The human reference is output; each entry of completions is machine-written. There is no
label column because the nesting already says which is which.… See the full description on the dataset page: https://huggingface.co/datasets/author-ai-text-detect/longlamp-ai-detection.