Language Models are Super Mario: Absorbing Abilities from Homologous Models as a Free Lunch
Paper • 2311.03099 • Published • 32
This is a merge of pre-trained language models created using mergekit.
This model was merged using the DARE TIES merge method using meta-llama/Meta-Llama-3-8B-Instruct as a base.
The following models were included in the merge:
The following YAML configuration was used to produce this model:
merge_method: dare_ties
base_model: meta-llama/Meta-Llama-3-8B-Instruct
models:
- model: meta-llama/Meta-Llama-3-8B-Instruct
parameters:
density: 0.9
weight: 0.8
- model: lightblue/suzume-llama-3-8B-japanese
parameters:
density: 0.9
weight: 0.2
dtype: float16