DreadPoor/Jingle_Jangle_Jingle-8B-BREADCRUMBS

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 bunnycore/Tulu-3.1-8B-SuperNova as a base.

Models Merged

The following models were included in the merge:

• johnsutor/Llama-3-8B-Instruct_dare_ties-density-0.9
• refuelai/Llama-3-Refueled
• DreadPoor/Aspire-8B-model_stock

Configuration

The following YAML configuration was used to produce this model:

base_model: bunnycore/Tulu-3.1-8B-SuperNova
dtype: bfloat16
merge_method: breadcrumbs
parameters:
  normalize: 1.0
slices:
- sources:
  - layer_range: [0, 32]
    model: johnsutor/Llama-3-8B-Instruct_dare_ties-density-0.9
    parameters:
      density: 0.3
      weight: 1
      gamma: 0.01
  - layer_range: [0, 32]
    model: refuelai/Llama-3-Refueled
    parameters:
      density: 0.3
      weight: 1
      gamma: 0.01
  - layer_range: [0, 32]
    model: DreadPoor/Aspire-8B-model_stock
    parameters:
      density: 0.3
      weight: 1
      gamma: 0.01
  - layer_range: [0, 32]
    model: bunnycore/Tulu-3.1-8B-SuperNova
    parameters:
      density: 0.5
      weight: 1
      gamma: 0.01

Open LLM Leaderboard Evaluation Results

Detailed results can be found here! Summarized results can be found here!

Metric	Value (%)
Average	27.71
IFEval (0-Shot)	73.23
BBH (3-Shot)	35.92
MATH Lvl 5 (4-Shot)	11.71
GPQA (0-shot)	4.92
MuSR (0-shot)	11.41
MMLU-PRO (5-shot)	29.06