Long-sequence benchmark results

[Jeanna-AI]

Thank you for your interesting work and for sharing the code.

I'm currently attempting to reproduce your results for the long-sequence benchmark. However, I couldn't find any examples, code or documentation on how to run these experiments.

Could you please assist me with the steps or provide guidance on how to initiate the runs to reproduce your long-sequence results?

[WoodScene]

Thank you for your interest in our work!

The benchmarks and training code for the long-sequence experiments are included in the repository. You can follow the provided instructions in the documentation to execute the experiments. Everything required to reproduce the results is available, and the setup steps are clearly outlined.