The accompanying files are select GGUF quants of a merge of pre-trained language models created using mergekit.

A coherent Llama 3 model was merged at low weight into a Llama 3.1 model. No fine-tuning was performed afterward. The resulting model is mostly coherent for roleplay text generation, retaining long context capability of 3.1, despite none of the merge components having been specifically for roleplay nor intended for it. The coherence of this model may be surprising given the difference in tokenizers between 3 and 3.1. It is plausible that 3.1 was trained on a base which originated with 3, hence an internal mapping or feature existing that can translate internally between the two tokenizers.

Testing has been performed out to 16K context, using temperature 1 and minP 0.01. This model leans toward being creative, so adjust temperature upward or downward as desired.

There remain initial format consistency issues with the merged model, but this can be mitigated in an Instruct prompt. Additionally, promptsteering was employed to vary the text generation output to avoid some of the common failings observed during text generation with Llama 3/3.1 8B models. The complete Instruct prompt used during testing is available below.

Built with Llama.

This model was merged using the SLERP merge method.

The following models were included in the merge:

The following YAML configuration was used to produce this model: