Llama-MagicalGirl

MagicalGirl

No image for this model. A auditory replacement has been provided.

<audio id="player" controls> <source src="https://files.catbox.moe/7z2w06.flac" type="audio/flac" /> </audio> <script src="https://cdn.plyr.io/3.7.8/plyr.js"></script>

This is a merge of pre-trained language models created using mergekit.

Sampling Settings

I keep playing around with sampler settings more often than not due to model not being super creative or just overly verbose. Anyway, I landed on the following for this model:

Temperature: 1.4
Min P: 0.03

This applies retroactively to KaraKaraWitch/Llama-3.X-Workout-70B as well.

Notes

• Seems to fit my requirements for the most part. Not too sure how exactly others would feel but I find that this is the model I envisioned.
• Without a system prompt, the model can get quite offensive and dark. Consider writing a simple system prompt before using.

Merge Details

Merge Method

This model was merged using the SCE merge method using KaraKaraWitch/Llama-3.X-Workout-70B as a base.

Models Merged

The following models were included in the merge:

• SicariusSicariiStuff/Negative_LLAMA_70B
• TheDrummer/Nautilus-70B-v0.1
• Steelskull/L3.3-Nevoria-R1-70b
• Tarek07/Inception-LLaMa-70B

Configuration

The following YAML configuration was used to produce this model:

models:
  - model: SicariusSicariiStuff/Negative_LLAMA_70B
  - model: TheDrummer/Nautilus-70B-v0.1
  - model: Tarek07/Inception-LLaMa-70B
  - model: Steelskull/L3.3-Nevoria-R1-70b

merge_method: sce
base_model: KaraKaraWitch/Llama-3.X-Workout-70B
parameters:
  select_topk: 1.0
dtype: bfloat16