This is the benchmark we adopt in our TMLR2025 paper Long-context LLMs Struggle with Long In-context Learning. Check out our leaderboard at https://huggingface.co/spaces/TIGER-Lab/LongICL-Leaderboard.

Cite our work by

@misc{li2024longcontext,
      title={Long-context LLMs Struggle with Long In-context Learning}, 
      author={Tianle Li and Ge Zhang and Quy Duc Do and Xiang Yue and Wenhu Chen},
      year={2024},
      eprint={2404.02060},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}