amylynn/Prima-Llama-8B

merged_model

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 meta-llama/Meta-Llama-3.1-8B as a base.

Models Merged

The following models were included in the merge:

• theo77186/Llama-3.2-8B-Instruct + amylynn/llama-3.1-8b-lora-deepseek-r1-distill-llama-8b
• SicariusSicariiStuff/Llama-3.3-8B-Instruct-128K_Abliterated + amylynn/llama-3.1-8b-lora-deephermes-3-llama-3-8b-preview
• DavidAU/L3.1-RP-Hero-Dirty_Harry-8B

Configuration

The following YAML configuration was used to produce this model:

merge_method: dare_ties
base_model: meta-llama/Meta-Llama-3.1-8B

models:
  # 1. Your main adult RP creative base (heavier weight to preserve erotic style & prose)
  - model: DavidAU/L3.1-RP-Hero-Dirty_Harry-8B
    parameters:
      weight: 0.82
      density: 0.75     # Keep most of the creative/erotic deltas

  # 2. DeepSeek-R1 reasoning adapter (strong analytical boost)
  - model: theo77186/Llama-3.2-8B-Instruct+amylynn/llama-3.1-8b-lora-deepseek-r1-distill-llama-8b
    parameters:
      weight: 0.58
      density: 0.65

  # 3. DeepHermes-3 adapter (judgment, thinking mode, coherence)
  - model: SicariusSicariiStuff/Llama-3.3-8B-Instruct-128K_Abliterated+amylynn/llama-3.1-8b-lora-deephermes-3-llama-3-8b-preview
    parameters:
      weight: 0.55
      density: 0.60

parameters:
  normalize: true
  int8_mask: true

dtype: bfloat16