MN-12B-Paskuda

merge

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

Merge Details

Merge Method

This model was merged using the Model Breadcrumbs merge method using OddTheGreat/Sinner_12B_V.1 as a base.

Models Merged

The following models were included in the merge:

• BeaverAI/MN-2407-DSK-QwQify-v0.1-12B
• Nitral-AI/Mag-Mell-Reasoner-12B

Configuration

The following YAML configuration was used to produce this model:

models:
  - model: OddTheGreat/Sinner_12B_V.1
    parameters:
      weight: [0.9, 1, 0.8, 0.65, 0.2, 0.05, 0.1, 0.35, 0.65, 0.8]
  - model: BeaverAI/MN-2407-DSK-QwQify-v0.1-12B
    parameters:
      weight: [0.1, 10e-5, 10e-5, 0.1, 0.5, 0.5, 0.5, 0.35, 0.1, 10e-5]
      density: [0.35, 1, 1, 0.35, 0.35, 0.5, 0.5, 0.6, 0.9, 1]
      gamma: [5e-4, 5e-3, 5e-3, 5e-4, 5e-4, 0.01, 0.01, 0.01, 5e-3, 5e-4]
  - model: Nitral-AI/Mag-Mell-Reasoner-12B
    parameters:
      weight: [10e-5, 10e-5, 0.2, 0.25, 0.3, 0.45, 0.4, 0.3, 0.25, 0.2]
      density: [1, 1, 0.9, 0.9, 0.85, 0.85, 0.8, 0.85, 0.9, 0.9]
      gamma: 0.03
merge_method: breadcrumbs
base_model: OddTheGreat/Sinner_12B_V.1
parameters:
  int8_mask: false
  rescale: true
  normalize: false
dtype: float32
out_dtype: bfloat16
tokenizer_source: BeaverAI/MN-2407-DSK-QwQify-v0.1-12B