dumping-grounds/llama-3.1-minitron-8b-width-base

merge

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

Merge Details

Merge Method

This model was merged using the passthrough merge method.

Models Merged

The following models were included in the merge:

• nvidia/Llama-3.1-Minitron-4B-Width-Base

Configuration

The following YAML configuration was used to produce this model:

slices:
  - sources:
      - model: nvidia/Llama-3.1-Minitron-4B-Width-Base
        layer_range: [0, 24]
  - sources: # add middle layers with residuals scaled to zero
      - model: nvidia/Llama-3.1-Minitron-4B-Width-Base
        layer_range: [8, 24]
        parameters:
          scale:
            - filter: o_proj
              value: 0.0
            - filter: down_proj
              value: 0.0
            - value: 1.0
  - sources: # add middle layers with residuals scaled to zero
      - model: nvidia/Llama-3.1-Minitron-4B-Width-Base
        layer_range: [8, 24]
        parameters:
          scale:
            - filter: o_proj
              value: 0.0
            - filter: down_proj
              value: 0.0
            - value: 1.0
  - sources:
      - model: nvidia/Llama-3.1-Minitron-4B-Width-Base
        layer_range: [24, 32]
merge_method: passthrough
dtype: bfloat16