merging native pytorch model locally

[sorobedio]

hello, is it possible to merge a local model I meant non huggingface model.
what should be the parameters in such cases.
what I wan is to perform some preprocessing merge the models evaluate the merge locally then save the checkpoints.

[hoooooli]

可以的，yml文件中的模型填写本地路径就可以，没有其他需要改变的

[zyp100200]

可以的，yml文件中的模型填写本地路径就可以，没有其他需要改变的

怎么写呢？
models:

• model: "model1.pth"
  parameters:
  density: 0.5
  weight: 0.5
• model: "model2.pth"
  parameters:
  density: 0.5
  weight: 0.5
  merge_method: ties
  base_model: "model1.pth"
  parameters:
  normalize: true
  int8_mask: true
  dtype: float16
  这是yaml文件，会出现 Input should be a valid list [type=list_type, input_value='float16', input_type=str]
  的错误