README.md

metadata

library_name: transformers
tags:
  - mergekit
  - merge
base_model:
  - ycros/BagelMIsteryTour-v2-8x7B
  - smelborp/MixtralOrochi8x7B
  - cognitivecomputations/dolphin-2.7-mixtral-8x7b
model-index:
  - name: maid-yuzu-v7
    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: 64.62
            name: strict accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=rhplus0831/maid-yuzu-v7
          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: 26.82
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=rhplus0831/maid-yuzu-v7
          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: 8.91
            name: exact match
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=rhplus0831/maid-yuzu-v7
          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: 7.94
            name: acc_norm
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=rhplus0831/maid-yuzu-v7
          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: 9.77
            name: acc_norm
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=rhplus0831/maid-yuzu-v7
          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: 28.22
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=rhplus0831/maid-yuzu-v7
          name: Open LLM Leaderboard

maid-yuzu-v7

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

I don't know anything about merges, so this may be a stupid method, but I was curious how the models would be merged if I took this approach.

Merge Details

Merge Method

This model was merged using the SLERP merge method.

This model is a model that first merges Model Orochi with Model dolphin with a 0.15 SLERP option, and then merges Model BagelMIsteryTour with a 0.2 SLERP option based on the merged model.

Models Merged

The following models were included in the merge:

• ycros/BagelMIsteryTour-v2-8x7B
• ../maid-yuzu-v7-base

Configuration

The following YAML configuration was used to produce this model:

base_model:
  model:
    path: ../maid-yuzu-v7-base
dtype: bfloat16
merge_method: slerp
parameters:
  t:
  - value: 0.2
slices:
- sources:
  - layer_range: [0, 32]
    model:
      model:
        path: ../maid-yuzu-v7-base
  - layer_range: [0, 32]
    model:
      model:
        path: ycros/BagelMIsteryTour-v2-8x7B

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	24.38
IFEval (0-Shot)	64.62
BBH (3-Shot)	26.82
MATH Lvl 5 (4-Shot)	8.91
GPQA (0-shot)	7.94
MuSR (0-shot)	9.77
MMLU-PRO (5-shot)	28.22