ignore_Q2.5-test

merge

It's noticeably better than Q2.5-Qwetiapin-32B, but I'll leave it as is until arcee-ai releases a 32B R1 distill. (If that will happen though)

Merge Details

Merge Method

This model was merged using the Model Breadcrumbs with TIES merge method using arcee-ai/Virtuoso-Medium-v2 as a base.

Models Merged

The following models were included in the merge:

• crestf411/Q2.5-32B-Slush
• nbeerbower/Dumpling-Qwen2.5-32B
• maldv/Qwenstein2.5-32B-Instruct
• allura-org/Qwen2.5-32b-RP-Ink

Configuration

The following YAML configuration was used to produce this model:

models:
  - model: arcee-ai/Virtuoso-Medium-v2
  - model: maldv/Qwenstein2.5-32B-Instruct
    parameters:
        density: 0.5
        weight: 0.4
        gamma: 0.03
  - model: allura-org/Qwen2.5-32b-RP-Ink
    parameters:
        density: 0.77
        gamma: 0.007
        weight: 0.07
  - model: nbeerbower/Dumpling-Qwen2.5-32B
    parameters:
        density: 0.88
        gamma: 0.008
        weight: 0.28
  - model: crestf411/Q2.5-32B-Slush
    parameters:
        density: 0.75
        gamma: 0.01
        weight: 0.05
base_model: arcee-ai/Virtuoso-Medium-v2
merge_method: breadcrumbs_ties
parameters:
  int8_mask: true
  rescale: true
  normalize: true
dtype: bfloat16
tokenizer_source: base