DELLA-Merging: Reducing Interference in Model Merging through Magnitude-Based Sampling
Paper
•
2406.11617
•
Published
•
10
This is a merge of pre-trained language models created using mergekit.
This model was merged using the DELLA merge method using shisa-ai/shisa-v2-mistral-nemo-12b as a base.
The following models were included in the merge:
The following YAML configuration was used to produce this model:
merge_method: della
dtype: bfloat16
out_dtype: bfloat16
base_model: shisa-ai/shisa-v2-mistral-nemo-12b
models:
- model: Elizezen/Himeyuri-v0.1-12B
parameters:
density: 0.75
weight: 0.8
- model: TheDrummer/Rocinante-12B-v1.1
parameters:
density: 0.6
weight: 0.6
parameters:
epsilon: 0.1
lambda: 1.0