README.md

metadata

base_model:
  - cognitivecomputations/dolphin-2.2.1-mistral-7b
  - l3utterfly/mistral-7b-v0.1-layla-v4-chatml
library_name: transformers
tags:
  - mergekit
  - merge

merged

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

Merge Details

Merge Method

This model was merged using the task_swapping merge method using /content/mergekit/test as a base.

Models Merged

The following models were included in the merge:

• cognitivecomputations/dolphin-2.2.1-mistral-7b
• /content/mergekit/tri
• l3utterfly/mistral-7b-v0.1-layla-v4-chatml

Configuration

The following YAML configuration was used to produce this model:

base_model:
  model:
    path: /content/mergekit/test
dtype: bfloat16
merge_method: task_swapping
slices:
- sources:
  - layer_range: [0, 32]
    model:
      model:
        path: l3utterfly/mistral-7b-v0.1-layla-v4-chatml
    parameters:
      diagonal_offset: 4.0
      random_mask: 0.1
      random_mask_seed: 1956557.0
      weight: 0.4
  - layer_range: [0, 32]
    model:
      model:
        path: cognitivecomputations/dolphin-2.2.1-mistral-7b
    parameters:
      diagonal_offset: 4.0
      random_mask: 0.1
      random_mask_seed: 18019.0
      weight: 0.333
  - layer_range: [0, 32]
    model:
      model:
        path: /content/mergekit/tri
    parameters:
      diagonal_offset: 4.0
      random_mask: 0.05
      random_mask_seed: 666666.0
      weight: 0.5
  - layer_range: [0, 32]
    model:
      model:
        path: /content/mergekit/test