Resolving Interference When Merging Models
Paper
•
2306.01708
•
Published
•
15
This is a merge of pre-trained language models created using mergekit.
This model was merged using the TIES merge method using Delta-Vector/Control-Nanuq-8B as a base.
The following models were included in the merge:
The following YAML configuration was used to produce this model:
models:
- model: Delta-Vector/Control-Nanuq-8B # Primary roleplay model
parameters:
density: 0.6
weight: 0.65
- model: lolzinventor/Meta-Llama-3.1-8B-SurviveV3 # Survival model
parameters:
density: 0.3
weight: 0.15
- model: AlexBefest/CardProjector-R1-preview-8B-v1.1 # Card creator
parameters:
density: 0.5
weight: 0.20
merge_method: ties
base_model: Delta-Vector/Control-Nanuq-8B
parameters:
normalize: true
int8_mask: true
dtype: float16