Explanation and Description
This model is an experiment. Originally, I had in mind making a remix of the excellent Bumoch/Forgotten_Magic_24B with some updated (and more-Heretic'd) donor models, but I got impatient waiting to see if I'd be granted access to some gated repos, so I cooked this up instead.
The hubris in question is my own. I'm a complete newbie. Who am I to think I could make anything good? And in fact, it might not be. I've tested that it works and does indeed produce pretty unhinged output if asked to. I don't really do roleplaying, so I'm not sure how I would test that. Or much of anything else. Caveat emptor.
This model is also not really intended to be the final product. My long-term plan is to upscale it and turn it into something suitable for long-form creative fiction writing, but I'm still trying to put together a decent dataset for retraining purposes. That might be a while. In the meantime, though, I see no value in keeping this hidden away. Someone might even like it. I can hope.
Any suggestions or criticisms are always welcome. I can't promise I'll have the vaguest idea how to act on them, but they're welcome nevertheless.
Deluxe thanks to team mradermacher for graciously quantizing this!
Merge Details
Merge Method
This is a merge of pre-trained language models created using mergekit. This model was merged using the Karcher Mean merge method.
Models Merged
The following models were included in the merge:
Configuration
The following YAML configuration was used to produce this model:
merge_method: karcher
models:
- model: sophosympatheia/Magistry-24B-v1.1
- model: llmfan46/Forgotten-Transgression-24B-v4.1-ultra-uncensored-heretic
- model: yvvki/Erotophobia-24B-v1.1
- model: yamatazen/EsotericKnowledge-24B
- model: MuXodious/Cydonia-24B-v4.3-absolute-heresy
- model: TheDrummer/Precog-24B-v1
- model: DarkArtsForge/Asmodeus-24B-v3
- model: grayarea/Magidonia-24B-v4.3-heretic-v1.2
tokenizer:
source: union
chat_template: auto
dtype: float32
out_dtype: bfloat16