Editing Models with Task Arithmetic
Paper
• 2212.04089 • Published
• 7
This is a merge of pre-trained language models created using mergekit.
This model was merged using the task arithmetic merge method using Delta-Vector/Holland-4B-V1 as a base.
The following models were included in the merge:
The following YAML configuration was used to produce this model:
base_model: Delta-Vector/Holland-4B-V1
dtype: bfloat16
merge_method: task_arithmetic
parameters:
normalize: true
slices:
- sources:
- layer_range: [0, 32]
model: Delta-Vector/Holland-4B-V1
- layer_range: [0, 32]
model: Delta-Vector/HollandV2R1
parameters:
weight: 0.30
- layer_range: [0, 32]
model: IntervitensInc/intv-mimi-sus
parameters:
weight: 0.30
- layer_range: [0, 32]
model: anthracite-org/magnum-v2-4b
parameters:
weight: 0.40