🌟 Overview

StarlightMoon-Foxfire-12B is a merge built on top of Impish_Bloodmoon_12B by SicariusSicariiStuff.

The model uses the model_stock method

🧩 Models Merged

This model results from a merge between:

• SicariusSicariiStuff/Impish_Bloodmoon_12B (Base)
• SicariusSicariiStuff/Impish_Bloodmoon_12B_Abliterated
• DreadPoor/Famino-12B-Model_Stock
• PygmalionAI/Eleusis-12B
• PygmalionAI/Pygmalion-3-12B
• MrRikyz/Foxfire_Bloom
• Vortex5/Azure-Starlight-12B

🛠️ Merge Details

Method: MODEL_STOCK

The merge was performed using mergekit with the following parameters:

• Base Model: Impish_Bloodmoon_12B_Abliterated
• Dtype: Float32
• Out_Dtype BFloat16
• lambda: 0.82

base_model: SicariusSicariiStuff/Impish_Bloodmoon_12B
dtype: float32
merge_method: model_stock
modules:
  default:
    slices:
    - sources:
      - layer_range: [0, 40]
        model: SicariusSicariiStuff/Impish_Bloodmoon_12B_Abliterated
        parameters:
          weight: 0.54
      - layer_range: [0, 40]
        model: DreadPoor/Famino-12B-Model_Stock
        parameters:
          weight: 0.38
      - layer_range: [0, 40]
        model: PygmalionAI/Eleusis-12B
        parameters:
          weight: 0.12
      - layer_range: [0, 40]
        model: PygmalionAI/Pygmalion-3-12B
        parameters:
          weight: 0.12
      - layer_range: [0, 40]
        model: MrRikyz/Foxfire_Bloom
        parameters:
          weight: 0.22
      - layer_range: [0, 40]
        model: Vortex5/Azure-Starlight-12B
        parameters:
          weight: 0.1
      - layer_range: [0, 40]
        model: SicariusSicariiStuff/Impish_Bloodmoon_12B
out_dtype: bfloat16
parameters:
  lambda: 0.82
  normalize: 1.0
tokenizer:
  source: MrRikyz/Foxfire_Bloom

✨ Acknowledgements

Thanks to the authors of the original models for their incredible work:

• SicariusSicariiStuff for Impish_Bloodmoon_12B and Impish_Bloodmoon_12B_Abliterated
• DreadPoor for Famino-12B-Model_Stock
• PygmalionAI for Eleusis-12B and PygmalionAI/Pygmalion-3-12B
• Vortex5 for Azure-Starlight-12B