SIM-CoT performance on larger models

Hi

Thanks for releasing such an insightful paper and clean repository!

I am working on reproducing your method and applying it to other model architectures.

Besides LLaMA 8B, have you experimented with larger LLMs? If so, how does the performance hold up? I'd love to hear your thoughts or expectations on scaling this up.

Thanks again!

The issue names no files, tests, or entry points. Start by reviewing the repository's existing SIM-CoT implementation and evaluation setup, then determine whether larger-model experiments are documented; done would require reported scaling results or a documented maintainer response.

Written by the indexing model from the issue text.