Magnolia-v4-Gemma2-8k-9B

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

The model has been tested on narrative text generation. Instability was sometimes observed in a prior version when characters were portrayed as emotionally charged, prompting this rebuild around the original Instruct model as a core.

Merge Details

Merge Method

This model was merged using the task arithmetic merge method using google/gemma-2-9b-it as a base.

Models Merged

The following models were included in the merge:

• grimjim/Magnolia-v2-Gemma2-8k-9B
• AXCXEPT/EZO-Humanities-9B-gemma-2-it
• grimjim/lemon07r_Gemma-2-Ataraxy-v4c-9B_fixed

Configuration

The following YAML configuration was used to produce this model:

base_model: google/gemma-2-9b-it
dtype: bfloat16
merge_method: task_arithmetic
parameters:
  normalize: true
models:
  - model: google/gemma-2-9b-it
  - model: grimjim/Magnolia-v2-Gemma2-8k-9B
    parameters:
      weight: 0.5
  - model: grimjim/lemon07r_Gemma-2-Ataraxy-v4c-9B_fixed
    parameters:
      weight: 0.15
  - model: AXCXEPT/EZO-Humanities-9B-gemma-2-it
    parameters:
      weight: 0.3