Never Train from Scratch: Fair Comparison of Long-Sequence Models Requires Data-Driven Priors (Extended Abstract)

Amos et al. "Never Train from Scratch: Fair Comparison of Long-Sequence Models Requires Data-Driven Priors (Extended Abstract)." International Joint Conference on Artificial Intelligence, 2025. doi:10.24963/IJCAI.2025/1203

[Amos et al. "Never Train from Scratch: Fair Comparison of Long-Sequence Models Requires Data-Driven Priors (Extended Abstract)." International Joint Conference on Artificial Intelligence, 2025.](https://mlanthology.org/ijcai/2025/amos2025ijcai-never/) doi:10.24963/IJCAI.2025/1203

@inproceedings{amos2025ijcai-never,
  title     = {{Never Train from Scratch: Fair Comparison of Long-Sequence Models Requires Data-Driven Priors (Extended Abstract)}},
  author    = {Amos, Ido and Berant, Jonathan and Gupta, Ankit},
  booktitle = {International Joint Conference on Artificial Intelligence},
  year      = {2025},
  pages     = {10846-10851},
  doi       = {10.24963/IJCAI.2025/1203},
  url       = {https://mlanthology.org/ijcai/2025/amos2025ijcai-never/}
}