RP-Spectrum-24B

RP-Spectrum-24B

<img src="https://huggingface.co/Casual-Autopsy/RP-Spectrum-24B/resolve/main/RP-Spectrum.png">

A merge of some of the best RP finetunes the community has to offer.

Note: This model can be very creative, due to this I wouldn't recommend a temp over 0.7.

Merge Details

Merge Method

This model was merged using the DELLA merge method using mistralai/Mistral-Small-3.2-24B-Instruct-2506 as a base.

Models Merged

The following models were included in the merge:

• mistralai/Mistral-Small-3.2-24B-Instruct-2506
• zerofata/MS3.2-PaintedFantasy-v3-24B
• TheDrummer/Cydonia-24B-v4
• ReadyArt/Dark-Nexus-24B-v2.0
• Gryphe/Codex-24B-Small-3.2
• Doctor-Shotgun/MS3.2-24B-Magnum-Diamond

Configuration

The following YAML configuration was used to produce this model:

models:
  - model: mistralai/Mistral-Small-3.2-24B-Instruct-2506

  - model: zerofata/MS3.2-PaintedFantasy-v3-24B # Pink/Red
    parameters:
      weight: 0.3
      density: 1.0
      epsilon: 0.0
  - model: TheDrummer/Cydonia-24B-v4 # Black
    parameters:
      weight: 0.3
      density: 0.4
      epsilon: 0.25
  - model: CrucibleLab/M3.2-24B-Loki-V1.3 # Green
    parameters:
      weight: 0.3
      density: 0.4
      epsilon: 0.25
  - model: Gryphe/Codex-24B-Small-3.2 # Blue (Yes, blue. Fuck you.)
    parameters:
      weight: 0.3
      density: 0.4
      epsilon: 0.25
  - model: ReadyArt/Dark-Nexus-24B-v2.0 # Orange
    parameters:
      weight: 0.3
      density: 0.4
      epsilon: 0.25
  - model: Doctor-Shotgun/MS3.2-24B-Magnum-Diamond # White
    parameters:
      weight: 0.3
      density: 0.4
      epsilon: 0.25

# Seed: 199803
merge_method: della
parameters:
  lambda: 0.9
  normalize: true
  int8_mask: false
base_model: mistralai/Mistral-Small-3.2-24B-Instruct-2506
dtype: float32
out_dtype: bfloat16

tokenizer:
  source: zerofata/MS3.2-PaintedFantasy-v3-24B