mergekit-community/MergeStage3

merge

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

Merge Details

Merge Method

This model was merged using the TIES merge method using CultriX/MergeStage2 as a base.

Models Merged

The following models were included in the merge:

• CultriX/Qwen2.5-14B_Virtuoso-small-v2-LoRA_r128
• arcee-ai/Virtuoso-Small-v2

Configuration

The following YAML configuration was used to produce this model:

# Stage 3: Ties Merge (Optional - Targeted Fine-tuning)
name: Qwen2.5-14B-Stage3
merge_method: ties
base_model: CultriX/MergeStage2
tokenizer_source: base  # Verify and update if needed
dtype: bfloat16
parameters:
  normalize: true
  rescale: false
  int8_mask: true
  density: 0.9  # Adjusted
  weight: 1.0
models:
  - model: arcee-ai/Virtuoso-Small-v2  # IFEval focus
    parameters:
      density: 0.8  # Adjusted
      weight: 0.7  # Adjusted
  - model: CultriX/Qwen2.5-14B_Virtuoso-small-v2-LoRA_r128  # IFEval LoRA
    parameters:
      density: 0.7  # Adjusted
      weight: 0.5  # Adjusted