correct template for Nemo model?

[alist]

Hi @guinmoon, thanks a ton for building this App. I am looking to run a 12B q4 GGUF Mistral-Nemo model on an 8Gb iPad (I have a 16gb on its way..), inference occurs but it seems like the decoding is failing since the tokens are not as expected....

Any ideas for settings or things I could be doing wrong? I really recommend this model, would love to get it working. Thank you.