OddTheGreat/Comet_12B_V.4

merge

This is a merge of pre-trained language models

Goal of this merge was to create good all purpose uncensored model, without too much positive bias.

Model has no problems with any themes, tested both as assistant and in rp on 400 answers, vision works too, use default gemma 3 mmproj.

Model performs good, is adequately smart for 12B, okay context attention, good at instruction following. I like how it replies.

Also, tested on RU, and it works fine, prose can be really nice.

Im not used to gemma slop yet, so i can't judge that field yet.