This version uses a diffrent config setup, with the actual base model of the two merges as the "base_model". Test both for yourself and see which is better at coding. Benchmarks coming soon.

Config can be found bellow:

yaml

models:
  - model: deepseek-ai_deepseek-coder-6.7b-instruct
    parameters:
      weight: 1
  - model: ise-uiuc_Magicoder-S-DS-6.7B
    parameters:
      weight: 1
merge_method: task_arithmetic
base_model: deepseek-ai_deepseek-coder-6.7b-base
parameters:
  normalize: true
  int8_mask: true
dtype: float16

RichardErkhov/rombodawg_-_DeepMagic-Coder-7b-Alt-8bits

Original model description:

license: other license_name: deepseek license_link: https://github.com/deepseek-ai/DeepSeek-Coder/blob/main/LICENSE-MODEL

RichardErkhov/
rombodawg_-_DeepMagic-Coder-7b-Alt-8bits