My main goal with this one was to merge the smartness of the base Instruct Nemo with the better prose from the different roleplaying fine-tunes. This is version v0.1, still to be tested. Weights shamelessly stolen from @ParasiticRogue (thank you, friend). All credits and thanks go to Intervitens, Mistralai, NeverSleep and ShuttleAI for providing amazing models used in the merge.

Both Mistral Instruct and ChatML should work.

Or...

<|im_start|>system
{system}<|im_end|>
<|im_start|>user
{user}<|im_end|>
<|im_start|>assistant
{assistant}<|im_end|>

V1: https://huggingface.co/MarinaraSpaghetti/Nemomix-v1.0-12B

V2: https://huggingface.co/MarinaraSpaghetti/Nemomix-v2.0-12B

V3: https://huggingface.co/MarinaraSpaghetti/Nemomix-v3.0-12B

Lower Temperature of 0.35 recommended, although I had luck with Temperatures above one (1.0-1.2) if you crank up the Min P (0.01-0.1). Run with base DRY of 0.8/1.75/2/0 and you're good to go.

This is a merge of pre-trained language models created using mergekit.

This model was merged using the DARE TIES merge method using F:\mergekit\mistralaiMistral-Nemo-Base-2407 as a base.

The following models were included in the merge:

The following YAML configuration was used to produce this model: