allura-forge/l3.0-upscale-1

output

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:

• meta-llama/Meta-Llama-3-8B

Configuration

The following YAML configuration was used to produce this model:

dtype: bfloat16
merge_method: passthrough

slices:
  # untouched intro
  - sources:
      - layer_range: [0, 8]
        model: meta-llama/Meta-Llama-3-8B

  - sources:
      - layer_range: [8, 10]
        model: meta-llama/Meta-Llama-3-8B
  # 8–16 baseline
  - sources:
      - layer_range: [8, 12]
        model: meta-llama/Meta-Llama-3-8B
  # 8–16 duplicate with projections nulled
  - sources:
      - layer_range: [8, 16]
        model: meta-llama/Meta-Llama-3-8B
        parameters:
          scale:
            - filter: o_proj
              value: 0.0
            - filter: down_proj
              value: 0.0
            - value: 1.0

  # 16–24 duplicate
  - sources:
      - layer_range: [16, 20]
        model: meta-llama/Meta-Llama-3-8B
        parameters:
          scale:
            - filter: o_proj
              value: 0.0
            - filter: down_proj
              value: 0.0
            - value: 1.0
  # 16–24 baseline
  - sources:
      - layer_range: [16, 24]
        model: meta-llama/Meta-Llama-3-8B
  # 16–24 duplicate
  - sources:
      - layer_range: [16, 20]
        model: meta-llama/Meta-Llama-3-8B
        parameters:
          scale:
            - filter: o_proj
              value: 0.0
            - filter: down_proj
              value: 0.0
            - value: 1.0

  # untouched tail
  - sources:
      - layer_range: [24, 32]
        model: meta-llama/Meta-Llama-3-8B