CultriX/Qwen2.5-14B-Brocav4

merge

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

Merge Details

Merge Method

This model was merged using the della_linear merge method using CultriX/Qwen2.5-14B-Wernickev3 as a base.

Models Merged

The following models were included in the merge:

• allknowingroger/QwenSlerp6-14B
• qingy2019/Qwen2.5-Math-14B-Instruct
• CultriX/Qwen2.5-14B-Broca
• sometimesanotion/Qwen2.5-14B-Vimarckoso
• djuna/Q2.5-Veltha-14B-0.5
• CultriX/Qwenfinity-2.5-14B
• CultriX/SeQwence-14Bv1

Configuration

The following YAML configuration was used to produce this model:

# Refined configuration using della_linear for optimized logical and multitask reasoning.
base_model: CultriX/Qwen2.5-14B-Wernickev3
merge_method: della_linear
parameters:
  epsilon: 0.012        # Ultra-fine parameter scaling for precision.
  lambda: 1.4           # Emphasis on significant model contributions.
  normalize: true       # Ensures balanced parameter integration.
  smoothing_factor: 0.1 # Precisely tuned for smooth task-specific blending.

gradient_clipping:
  CultriX/Qwen2.5-14B-Wernickev3: 0.86  # Stabilized multitask performance.
  CultriX/Qwenfinity-2.5-14B: 0.83      # Multitask integration refinement.
  djuna/Q2.5-Veltha-14B-0.5: 0.91       # Strengthened advanced reasoning.
  CultriX/Qwen2.5-14B-Broca: 0.85       # Logical and contextual reasoning stability.
  qingy2019/Qwen2.5-Math-14B-Instruct: 0.93 # Mathematical reasoning prioritization.
  CultriX/SeQwence-14Bv1: 0.88          # Generalist multitask contributions.
  sometimesanotion/Qwen2.5-14B-Vimarckoso: 0.89 # Multi-step reasoning enhancement.
  allknowingroger/QwenSlerp6-14B: 0.87  # Contextual reasoning integration.

models:
  - model: CultriX/Qwen2.5-14B-Wernickev3
    parameters:
      weight: 0.26       # Core backbone for multitask reasoning.
      density: 0.7       # Prioritizes critical reasoning parameters.
  - model: CultriX/Qwenfinity-2.5-14B
    parameters:
      weight: 0.23       # Comprehensive multitask contributor.
      density: 0.65
  - model: djuna/Q2.5-Veltha-14B-0.5
    parameters:
      weight: 0.22       # Advanced reasoning support for GPQA and MUSR.
      density: 0.72
  - model: CultriX/Qwen2.5-14B-Broca
    parameters:
      weight: 0.15       # Logical reasoning and factual QA enhancements.
      density: 0.65
  - model: qingy2019/Qwen2.5-Math-14B-Instruct
    parameters:
      weight: 0.18       # Mathematical reasoning priority.
      density: 0.73
  - model: CultriX/SeQwence-14Bv1
    parameters:
      weight: 0.14       # Generalist multitask backbone.
      density: 0.63
  - model: sometimesanotion/Qwen2.5-14B-Vimarckoso
    parameters:
      weight: 0.12       # Multi-step reasoning tasks contributor.
      density: 0.6
  - model: allknowingroger/QwenSlerp6-14B
    parameters:
      weight: 0.1        # Contextual reasoning improvements.
      density: 0.62

adaptive_merge_parameters:
  task_weights:
    tinyArc: 1.6           # Logical reasoning improvements.
    tinyHellaswag: 1.5     # Contextual consistency.
    tinyMMLU: 1.65         # Domain knowledge enhancement.
    tinyTruthfulQA: 1.9    # Accurate factual reasoning.
    tinyTruthfulQA_mc1: 1.7 # Multiple-choice reasoning focus.
    tinyWinogrande: 1.75   # Advanced reasoning boost.
    IFEval: 1.9            # Instruction-following tasks prioritized.
    BBH: 1.7               # Complex reasoning support.
    MATH: 2.1              # Mathematical excellence emphasized.
    GPQA: 1.8              # Graduate-level QA enhanced.
    MUSR: 1.9              # Multi-step reasoning strengthened.
    MMLU-PRO: 1.8          # Domain multitask performance maximized.

tokenizer_source: CultriX/Qwen2.5-14B-Wernickev3  # Tokenizer aligned with backbone.