This is a straight SLERP merge of two great models - the llama-2 version of Nous Capybara 7b and Undi95's impressive experiment putting an Mlewd LORA on the xwin 7b 0.2 base.

I don't think this is quite as good as the Mistral version of Cabybara in my breif tests, but it does seem close, bringing some story awareness and logic from both xwin and capybara and with pretty decent prose (both base models pretty good there). A good alternative I think to the newer Nous Capybara Mistral model/beta for RP if you find Mistrals repetition issues frustrating. The merge is more about doubling up on synergy than it is cancelling any weaknesses.

Full credit to Nous, Undi95 and the seemingly dozens of people that put work into models for his mega-merge.