pg19-subsample

# Jet-Long Evaluation Datasets This repository contains the evaluation data used in the paper [Jet-Long: Efficient Long-Context Extension with Dynamic Bifocal RoPE](https://huggingface.co/papers/2607.07740). - **Code:** [GitHub Repository](https://github.com/jet-ai-projects/jet-long) ## Dataset Description These datasets are processed versions of standard benchmarks used to evaluate the long-context capabilities of Jet-Long: - **RULER-500**: A dataset for evaluating long-context understanding and retrieval up to 128K context. - **PG-19-subsample**: A subsampled version of the PG-19 dataset used for evaluating long-context language modeling and perplexity. ## Citation If you find Jet-Long or these datasets useful, please cite: ```bibtex @misc{jetlong2026, title={Jet-Long: Efficient Long-Context Extension with Dynamic Bifocal RoPE}, author={Haozhan Tang and Zerui Wang and Yuxian Gu and Song Han and Han Cai}, year={2026}, eprint={2607.07740}, archivePrefix={arXiv}, primaryClass={cs.LG}, url={https://arxiv.org/abs/2607.07740}, } ```