Modelpublic

Xiaojian9992024/Phi-3.5-Small-Instruct-UNOFFICAL

sourceHugging Faceupdated 2y agoView on Hugging Face

0likes38downloads

Model Card

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 using microsoft/Phi-3.5-mini-instruct as a base.

Models Merged

The following models were included in the merge:

Configuration

The following YAML configuration was used to produce this model:

yaml

base_model: microsoft/Phi-3.5-mini-instruct  # Base model to expand
merge_method: passthrough  # Use passthrough to add layers
slices:
  - sources:
      - model: microsoft/Phi-3.5-mini-instruct  # Source model for the base layers
        layer_range: [0, 32]  # Use all existing layers (assuming 38 layers in the base model)
  - sources:
      - model: microsoft/Phi-3.5-mini-instruct  # Source model for the additional layers
        layer_range: [16, 32]  # Add 8 new layers initialized from the base model
  - sources:
      - model: microsoft/Phi-3.5-mini-instruct  # Source model for the base layers
        layer_range: [16, 32]
  - sources:
      - model: microsoft/Phi-3.5-mini-instruct  # Source model for the base layers
        layer_range: [4, 32]
tokenizer_source: microsoft/Phi-3.5-mini-instruct  # Use the tokenizer from the base model
dtype: float16  # Data type for the merged model