README.md

metadata

base_model:
  - CultriX/Qwen2.5-14B-MegaMerge-pt1
  - CultriX/Qwen2.5-14B-Wernicke
  - CultriX/Qwen2.5-14B-MergeStock
library_name: transformers
tags:
  - mergekit
  - merge
license: apache-2.0
language:
  - en
model-index:
  - name: Qwen2.5-14B-MegaMerge-pt2
    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: 52.35
            name: strict accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=CultriX/Qwen2.5-14B-MegaMerge-pt2
          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: 50.64
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=CultriX/Qwen2.5-14B-MegaMerge-pt2
          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: 30.06
            name: exact match
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=CultriX/Qwen2.5-14B-MegaMerge-pt2
          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: 19.13
            name: acc_norm
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=CultriX/Qwen2.5-14B-MegaMerge-pt2
          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: 18.25
            name: acc_norm
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=CultriX/Qwen2.5-14B-MegaMerge-pt2
          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: 49.15
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/open-llm-leaderboard/open_llm_leaderboard?query=CultriX/Qwen2.5-14B-MegaMerge-pt2
          name: Open LLM Leaderboard
metrics:
  - accuracy

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 CultriX/Qwen2.5-14B-MegaMerge-pt1 as a base.

Models Merged

The following models were included in the merge:

• CultriX/Qwen2.5-14B-Wernicke
• CultriX/Qwen2.5-14B-MergeStock

Configuration

The following YAML configuration was used to produce this model:

# final_dare_ties_merge.yaml

models:
  - model: CultriX/Qwen2.5-14B-MergeStock
    parameters:
      density: 0.5  # Retain 50% of the most significant parameters
      weight: 0.6    # Emphasize MergeStock's contributions
  - model: CultriX/Qwen2.5-14B-Wernicke
    parameters:
      density: 0.5  # Retain 50% of the most significant parameters
      weight: 0.4    # Incorporate Wernicke's contributions
merge_method: dare_ties
base_model: CultriX/Qwen2.5-14B-MegaMerge-pt1
parameters:
  normalize: true
  int8_mask: true
dtype: bfloat16
tokenizer_source: Qwen/Qwen2.5-14B-Instruct

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

Metric	Value
Avg.	36.69
IFEval (0-Shot)	56.83
BBH (3-Shot)	50.91
MATH Lvl 5 (4-Shot)	27.34
GPQA (0-shot)	17.23
MuSR (0-shot)	18.74
MMLU-PRO (5-shot)	49.12