metadata

base_model:
  - Kaoeiri/MS_Moingooistral-2409-22B
  - InferenceIllusionist/SorcererLM-22B
  - Kaoeiri/MS-Magpantheonsel-lark-v4x1.6.2-Cydonia-vXXX-22B-5
  - Envoid/Mistral-Small-NovusKyver
  - Kaoeiri/MS_dampf-2409-22B
  - Saxo/Linkbricks-Horizon-AI-Japanese-Superb-V1-22B
  - TheDrummer/Cydonia-22B-v1.1
  - TroyDoesAI/BlackSheep-MermaidMistral-22B
  - Kaoeiri/MS-Inky-2409-22B
  - Gryphe/Pantheon-RP-Pure-1.6.2-22b-Small
  - DigitalSouls/BlackSheep-DigitalSoul-22B
  - Darkknight535/MS-Moonlight-22B-v3
  - >-
    hf-100/Mistral-Small-Spellbound-StoryWriter-22B-instruct-0.2-chkpt-200-16-bit
  - Kaoeiri/MS_fujin-2409-22B
  - Kaoeiri/MS_a-coolyte-2409-22B
  - crestf411/MS-sunfall-v0.7.0
  - unsloth/Mistral-Small-Instruct-2409
  - Kaoeiri/MS-Quadrosiac-2409-22B
  - ArliAI/Mistral-Small-22B-ArliAI-RPMax-v1.1
  - Jellywibble/MistralSmall1500CTXDummy
library_name: transformers
tags:
  - mergekit
  - merge
  - not-for-all-audiences
license: cc-by-nc-nd-4.0

merge

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

Merge Details

Merge Method

This model was merged using the DARE TIES merge method using unsloth/Mistral-Small-Instruct-2409 as a base.

Models Merged

The following models were included in the merge:

Configuration

The following YAML configuration was used to produce this model:

models:
  # Primary Fiction & Monster Core
  - model: Kaoeiri/MS_Moingooistral-2409-22B  # Monster fiction core
    parameters:
      weight: 0.23
      density: 1.21
  
  - model: Kaoeiri/MS-Magpantheonsel-lark-v4x1.6.2-Cydonia-vXXX-22B-5  # Main writing engine
    parameters:
      weight: 0.85
      density: 0.76

  # World Building & Fantasy
  - model: Kaoeiri/MS-Inky-2409-22B  # Descriptive world dynamics
    parameters:
      weight: 0.24
      density: 0.73
  
  - model: Gryphe/Pantheon-RP-Pure-1.6.2-22b-Small  # Rich interaction
    parameters:
      weight: 0.26
      density: 0.74

  # Character & Monster Development
  - model: DigitalSouls/BlackSheep-DigitalSoul-22B  # Combat and conflict
    parameters:
      weight: 0.18
      density: 0.68
  
  - model: InferenceIllusionist/SorcererLM-22B  # Magical elements
    parameters:
      weight: 0.19
      density: 0.70

  # Adventure & Environment
  - model: TheDrummer/Cydonia-22B-v1.1  # Evil aspects
    parameters:
      weight: 0.15
      density: 0.65

  # Nature & Creatures
  - model: crestf411/MS-sunfall-v0.7.0  # Nature elements
    parameters:
      weight: 0.21
      density: 0.70
  
  - model: Kaoeiri/MS_a-coolyte-2409-22B
    parameters:
      weight: 0.22
      density: 0.69

  # Future & Fantasy Elements
  - model: Kaoeiri/MS_fujin-2409-22B
    parameters:
      weight: 0.16
      density: 0.68
  
  - model: Kaoeiri/MS-Quadrosiac-2409-22B
    parameters:
      weight: 0.17
      density: 0.71
  
  - model: Kaoeiri/MS_dampf-2409-22B
    parameters:
      weight: 0.17
      density: 0.71

  # Story Enhancement
  - model: hf-100/Mistral-Small-Spellbound-StoryWriter-22B-instruct-0.2-chkpt-200-16-bit
    parameters:
      weight: 0.18
      density: 0.67
  
  - model: ArliAI/Mistral-Small-22B-ArliAI-RPMax-v1.1
    parameters:
      weight: 0.12
      density: 0.63
  
  - model: Darkknight535/MS-Moonlight-22B-v3
    parameters:
      weight: 0.11
      density: 0.63

  # Community Tuning
  - model: Jellywibble/MistralSmall1500CTXDummy
    parameters:
      weight: 0.04
      density: 0.60

  # Cultural Enhancement
  - model: Saxo/Linkbricks-Horizon-AI-Japanese-Superb-V1-22B
    parameters:
      weight: 0.20
      density: 0.58

  # Character Detail
  - model: Envoid/Mistral-Small-NovusKyver
    parameters:
      weight: 0.27
      density: 0.71

  # Fantasy Creatures
  - model: TroyDoesAI/BlackSheep-MermaidMistral-22B
    parameters:
      weight: 0.24
      density: 0.70

merge_method: dare_ties
base_model: unsloth/Mistral-Small-Instruct-2409

parameters:
  density: 0.88    # Increased for richer fictional detail
  epsilon: 0.05    # Reduced for more consistent narrative style
  lambda: 1.40     # Increased to preserve creative writing characteristics

dtype: bfloat16
tokenizer_source: union