without_japanese_bilingual-gpt-neox_4b_task_arithmetic

without_japanese_bilingual-gpt-neox_4b_task_arithmetic is a merge of the following models using mergekit:

• rinna/bilingual-gpt-neox-4b
• rinna/japanese-gpt-neox-3.6b

🧩 Configuration

models:
  - model: rinna/bilingual-gpt-neox-4b
    parameters:
      weight: 0.5
  - model: rinna/japanese-gpt-neox-3.6b
    parameters:
      weight: 0.5

merge_method: task_arithmetic
base_model: rinna/bilingual-gpt-neox-4b
parameters:
  normalize: true
  int8_mask: true

dtype: bfloat16