mistral 7b instruct v0.2
#4
by
cognitivetech
- opened
Wondering if we'll ever see Samantha paired with Mistral 7b Instruct v0.2. Its quite a lot better than 0.1, and for my purposes, still beats any other 7b.
Would love to see it with more warmth of personality.
I do not train on instruct models.
I train on base models.
I do not train on instruct models.
cognitivecomputations/samantha-1.1-westlake-7b
well, hey! there's a first for everything.
thanks for your work.
maybe I can just satisfy my cravings with this merge
https://huggingface.co/MaziyarPanahi/samantha-1.2-mistral-7b-Mistral-7B-Instruct-v0.2-slerp-GGUF (mistral template)
cognitivetech
changed discussion status to
closed
actually, that's one of the best merges I've tried. looking forward to working with Sam's EQ alongside a powerful instruct model