RicardoEstep/RPBizkit-v10-12B

"Experimental Mistral Nemo Mix" using... "Model Stock merge 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...

From the "errors" without head, to "the simpler excellence". This model is "quality". "Version V6" leaked stability, "Version V9" leaked creativity. This model has "pruned weights" full of "pure stable creativity", from original "base" models "tunned for dark RP". No more "mixes, from mixed mixes", without the "fix_mistral_regex" patch problem. All done with patience in mind... Have fun! 🫡

Models Merged:

• TheDrummer/UnslopNemo-12B-v4.1 as base.
• NeverSleep/Lumimaid-v0.2-12B
• Lambent/arsenic-nemo-unleashed-12B
• allura-org/MN-12b-RP-Ink
• LatitudeGames/Muse-12B
• SicariusSicariiStuff/Impish_Bloodmoon_12B
• ReadyArt/Omega-Darker_The-Final-Directive-12B
• HumanLLMs/Human-Like-Mistral-Nemo-Instruct-2407
• anthracite-org/magnum-v2-12b
• Gryphe/Pantheon-RP-1.6.1-12b-Nemo
• ArliAI/Mistral-Nemo-12B-ArliAI-RPMax-v1.2
• MuXodious/Wayfarer-2-12B-absolute-heresy

Configuration Used:

merge_method: model_stock
base_model: TheDrummer/UnslopNemo-12B-v4.1
tokenizer_source: base
parameters:
  normalize: true
  int8_mask: false
  filter_wise: true
models:
  - model: anthracite-org/magnum-v2-12b
  - model: LatitudeGames/Muse-12B
  - model: Gryphe/Pantheon-RP-1.6.1-12b-Nemo
  - model: allura-org/MN-12b-RP-Ink
  - model: MuXodious/Wayfarer-2-12B-absolute-heresy
  - model: SicariusSicariiStuff/Impish_Bloodmoon_12B
  - model: NeverSleep/Lumimaid-v0.2-12B
  - model: Lambent/arsenic-nemo-unleashed-12B
  - model: ReadyArt/Omega-Darker_The-Final-Directive-12B
  - model: HumanLLMs/Human-Like-Mistral-Nemo-Instruct-2407
  - model: ArliAI/Mistral-Nemo-12B-ArliAI-RPMax-v1.2
dtype: bfloat16

Final Notes;

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

I recommend to use "Alpaca (with "RAW" inputs)" as chat template. The configuration files "as is" are "Tweaked", to not use any "Chat Template". Still, it's "supposed to" full support "optional_chat_template.jinja.7z (Mistral V3)", if is decompressed.

Rapid Benchmarking - "64-Shot + 5318008 Seed" Evaluation.

These are the results of running "lm-evaluation-harness", using the "Q8_0 Model", with an "8K Context Window". Executed on a free "Google Collab - CUDA-accelerated backend (T4)" instance.

• Quantized Model, Here RicardoEstep/RPBizkit-v10-12B-GGUF.