Locutusque/Llama-3-Yggdrasil-2.0-8B

merge

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 NousResearch/Meta-Llama-3-8B as a base.

Models Merged

The following models were included in the merge:

• Locutusque/Llama-3-NeuralHercules-5.0-8B
• NousResearch/Hermes-2-Theta-Llama-3-8B
• Locutusque/llama-3-neural-chat-v2.2-8b

Configuration

The following YAML configuration was used to produce this model:

models:
  - model: NousResearch/Meta-Llama-3-8B
    # No parameters necessary for base model
  - model: NousResearch/Hermes-2-Theta-Llama-3-8B
    parameters:
      density: 0.6
      weight: 0.55
  - model: Locutusque/llama-3-neural-chat-v2.2-8b
    parameters:
      density: 0.55
      weight: 0.4
  - model: Locutusque/Llama-3-NeuralHercules-5.0-8B
    parameters:
      density: 0.65
      weight: 0.6
    
merge_method: dare_ties
base_model: NousResearch/Meta-Llama-3-8B
parameters:
  int8_mask: true
dtype: bfloat16

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	20.22
IFEval (0-Shot)	53.71
BBH (3-Shot)	26.92
MATH Lvl 5 (4-Shot)	6.87
GPQA (0-shot)	1.68
MuSR (0-shot)	8.07
MMLU-PRO (5-shot)	24.07