RicardoEstep/RPBizkit-v9-12B

"Experimental Mistral Nemo Mix" using... "DARE TIES + Model Stock merges with Mergekit".

This is a mix of some known "RP Uncensored" models. I'm not responsible of the use of this.

I did it all for the nookie, so you can take that cookie, and stick it up your...

¡This merge is based in 5 parts!

My last version was "good" overall, but was not... "Spicy". This version actually "amazed" me in terms of "PURE Stability". I had made a "better selection" choosing the models, and force a "pure mix" between them. While testing it (more than V7 and V8, sorry of that) i think it has "a kind of a bug". It has "a kind of humanity", and sometimes does NOT act like a completly "uncensored" model.

Part One: "Simpleness".

Models Merged:

• natong19/Mistral-Nemo-Instruct-2407-abliterated # Base!
• LatitudeGames/Muse-12B
• TheDrummer/UnslopNemo-12B-v4.1
• anthracite-org/magnum-v2-12b

Configuration Used:

merge_method: dare_ties
base_model: natong19/Mistral-Nemo-Instruct-2407-abliterated
tokenizer_source: base
parameters:
  normalize: true
  int8_mask: false
  random_seed: 5318008
models:
  - model: anthracite-org/magnum-v2-12b
    parameters:
      weight: 0.3333
      density: 0.5
  - model: TheDrummer/UnslopNemo-12B-v4.1
    parameters:
      weight: 0.3333
      density: 0.5
  - model: LatitudeGames/Muse-12B
    parameters:
      weight: 0.3333
      density: 0.5
dtype: bfloat16

Part Two: "Setting".

Models Merged:

• natong19/Mistral-Nemo-Instruct-2407-abliterated # Base!
• ReadyArt/Forgotten-Safeword-12B-v4.0
• Gryphe/Pantheon-RP-1.6.1-12b-Nemo
• inflatebot/MN-12B-Mag-Mell-R1

Configuration Used:

merge_method: dare_ties
base_model: natong19/Mistral-Nemo-Instruct-2407-abliterated
tokenizer_source: base
parameters:
  normalize: true
  int8_mask: false
  random_seed: 5318008
models:
  - model: ReadyArt/Forgotten-Safeword-12B-v4.0
    parameters:
      weight: 0.3333
      density: 0.5
  - model: inflatebot/MN-12B-Mag-Mell-R1
    parameters:
      weight: 0.3333
      density: 0.5
  - model: Gryphe/Pantheon-RP-1.6.1-12b-Nemo
    parameters:
      weight: 0.3333
      density: 0.5
dtype: bfloat16

Part Three: "Phantasy".

Models Merged:

• natong19/Mistral-Nemo-Instruct-2407-abliterated # Base!
• FallenMerick/MN-Violet-Lotus-12B
• Elizezen/Himeyuri-v0.1-12B
• ChaoticNeutrals/Nera_Noctis-12B

Configuration Used:

merge_method: dare_ties
base_model: natong19/Mistral-Nemo-Instruct-2407-abliterated
tokenizer_source: base
parameters:
  normalize: true
  int8_mask: false
  random_seed: 5318008
models:
  - model: ChaoticNeutrals/Nera_Noctis-12B
    parameters:
      weight: 0.3333
      density: 0.5
  - model: FallenMerick/MN-Violet-Lotus-12B
    parameters:
      weight: 0.3333
      density: 0.5
  - model: Elizezen/Himeyuri-v0.1-12B
    parameters:
      weight: 0.3333
      density: 0.5
dtype: bfloat16

Part Four: "Darkness".

Models Merged:

• natong19/Mistral-Nemo-Instruct-2407-abliterated # Base!
• SicariusSicariiStuff/Impish_Bloodmoon_12B
• MuXodious/Wayfarer-2-12B-absolute-heresy
• ArliAI/Mistral-Nemo-12B-ArliAI-RPMax-v1.2

Configuration Used:

merge_method: dare_ties
base_model: natong19/Mistral-Nemo-Instruct-2407-abliterated
tokenizer_source: base
parameters:
  normalize: true
  int8_mask: false
  random_seed: 5318008
models:
  - model: MuXodious/Wayfarer-2-12B-absolute-heresy
    parameters:
      weight: 0.3333
      density: 0.5
  - model: ArliAI/Mistral-Nemo-12B-ArliAI-RPMax-v1.2
    parameters:
      weight: 0.3333
      density: 0.5
  - model: SicariusSicariiStuff/Impish_Bloodmoon_12B
    parameters:
      weight: 0.3333
      density: 0.5
dtype: bfloat16

Part Five: "The Final Mix".

Models Merged:

• natong19/Mistral-Nemo-Instruct-2407-abliterated # Base!
• ./output1 #Simpleness.
• ./output2 #Setting.
• ./output3 #Phantasy.
• ./output4 #Darkness.

Configuration Used:

merge_method: model_stock
base_model: natong19/Mistral-Nemo-Instruct-2407-abliterated
tokenizer_source: base
parameters:
  normalize: true
  int8_mask: false
  filter_wise: true
models:
  - model: ./output1  #Simpleness.
  - model: ./output2  #Setting.
  - model: ./output3  #Phantasy.
  - model: ./output4  #Darkness. 
dtype: bfloat16

Final Notes;

The Tokenizer & Embedding Sizes (131072) are clean + It's "supposed to" support now a full "128K Context Size".

The recommended chat template to use is "Alpaca (with "RAW" inputs)". The configuration files are already "Tweaked", to don't use any "Chat Template". But, now, it's "supposed to" full support too, the "optional_chat_template.7z (Mistral V3)" if is decompressed.

• Quantized Model, Here RicardoEstep/RPBizkit-v9-12B-GGUF.