bamec66557/MISCHIEVOUS-12B-Mix_III_IV_V

merge

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

Merge Details

Merge Method

This model was merged using the SLERP merge method.

Models Merged

The following models were included in the merge:

• bamec66557/MISCHIEVOUS-12B-Mix_III_ex_V
• bamec66557/MISCHIEVOUS-12B-Mix_0.4v

Configuration

The following YAML configuration was used to produce this model:

slices:
  - sources:
      - model: bamec66557/MISCHIEVOUS-12B-Mix_0.4v
        layer_range: [0, 15]
      - model: bamec66557/MISCHIEVOUS-12B-Mix_III_ex_V
        layer_range: [0, 15]
    parameters:
      t:
        - value: 0.8
  - sources:
      - model: bamec66557/MISCHIEVOUS-12B-Mix_0.4v
        layer_range: [15, 30]
      - model: bamec66557/MISCHIEVOUS-12B-Mix_III_ex_V
        layer_range: [15, 30]
    parameters:
      t:
        - value: [0.8, 1.0, 0.8]
  - sources:
      - model: bamec66557/MISCHIEVOUS-12B-Mix_0.4v
        layer_range: [30, 40]
      - model: bamec66557/MISCHIEVOUS-12B-Mix_III_ex_V
        layer_range: [30, 40]
    parameters:
      t:
        - value: 1.0
        - filter: self_attn
          value: [0.9, 1.0, 1.1]
merge_method: slerp
base_model: bamec66557/MISCHIEVOUS-12B-Mix_0.4v
dtype: bfloat16
regularization:
  - method: weight_clipping
    clip_range: [-0.1, 0.1]
postprocessing:
  - operation: gaussian_smoothing
    sigma: 1.2
  - operation: normalize

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	23.22
IFEval (0-Shot)	40.31
BBH (3-Shot)	34.85
MATH Lvl 5 (4-Shot)	12.84
GPQA (0-shot)	9.40
MuSR (0-shot)	12.34
MMLU-PRO (5-shot)	29.60