README.md

metadata

license: llama3
library_name: transformers
tags:
  - mergekit
  - merge
base_model:
  - flammenai/Mahou-1.2-llama3-8B
  - lemon07r/Llama-3-RedMagic2-8B
  - lemon07r/Lllama-3-RedElixir-8B
  - NousResearch/Meta-Llama-3-8B
  - nbeerbower/llama-3-spicy-abliterated-stella-8B
model-index:
  - name: Llama-3-RedMagic4-8B
    results:
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: IFEval (0-Shot)
          type: HuggingFaceH4/ifeval
          args:
            num_few_shot: 0
        metrics:
          - type: inst_level_strict_acc and prompt_level_strict_acc
            value: 48.64
            name: strict accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=lemon07r/Llama-3-RedMagic4-8B
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: BBH (3-Shot)
          type: BBH
          args:
            num_few_shot: 3
        metrics:
          - type: acc_norm
            value: 19.48
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=lemon07r/Llama-3-RedMagic4-8B
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: MATH Lvl 5 (4-Shot)
          type: hendrycks/competition_math
          args:
            num_few_shot: 4
        metrics:
          - type: exact_match
            value: 8.31
            name: exact match
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=lemon07r/Llama-3-RedMagic4-8B
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: GPQA (0-shot)
          type: Idavidrein/gpqa
          args:
            num_few_shot: 0
        metrics:
          - type: acc_norm
            value: 5.37
            name: acc_norm
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=lemon07r/Llama-3-RedMagic4-8B
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: MuSR (0-shot)
          type: TAUR-Lab/MuSR
          args:
            num_few_shot: 0
        metrics:
          - type: acc_norm
            value: 4.38
            name: acc_norm
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=lemon07r/Llama-3-RedMagic4-8B
          name: Open LLM Leaderboard
      - task:
          type: text-generation
          name: Text Generation
        dataset:
          name: MMLU-PRO (5-shot)
          type: TIGER-Lab/MMLU-Pro
          config: main
          split: test
          args:
            num_few_shot: 5
        metrics:
          - type: acc
            value: 29.73
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=lemon07r/Llama-3-RedMagic4-8B
          name: Open LLM Leaderboard

Llama-3-RedMagic4-8B

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

Merge Details

Merge Method

This model was merged using the Model Stock merge method using NousResearch/Meta-Llama-3-8B as a base.

Models Merged

The following models were included in the merge:

• flammenai/Mahou-1.2-llama3-8B
• lemon07r/Llama-3-RedMagic2-8B
• lemon07r/Lllama-3-RedElixir-8B
• nbeerbower/llama-3-spicy-abliterated-stella-8B

Configuration

The following YAML configuration was used to produce this model:

base_model: NousResearch/Meta-Llama-3-8B
dtype: bfloat16
merge_method: model_stock
slices:
- sources:
  - layer_range: [0, 32]
    model: lemon07r/Llama-3-RedMagic2-8B
  - layer_range: [0, 32]
    model: lemon07r/Lllama-3-RedElixir-8B
  - layer_range: [0, 32]
    model: nbeerbower/llama-3-spicy-abliterated-stella-8B
  - layer_range: [0, 32]
    model: flammenai/Mahou-1.2-llama3-8B
  - layer_range: [0, 32]
    model: NousResearch/Meta-Llama-3-8B

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	19.32
IFEval (0-Shot)	48.64
BBH (3-Shot)	19.48
MATH Lvl 5 (4-Shot)	8.31
GPQA (0-shot)	5.37
MuSR (0-shot)	4.38
MMLU-PRO (5-shot)	29.73