A merge between an ol' fav: Undi95/MXLewd-L2-20B
and:
An extended XwinLM: Elfrino/XwinXtended-20B
*NOTE: FIRST Merge of XwinXtended and another 20B that finally works and isn't drunk. And it's rather pretty good and coherent, still testing...
######################################################################################################################
merge
This is a merge of pre-trained language models created using mergekit.
Merge Details
Merge Method
This model was merged using the SLERP merge method.
Models Merged
The following models were included in the merge:
Configuration
The following YAML configuration was used to produce this model:
slices:
- sources:
- model: Undi95/MXLewd-L2-20B
layer_range: [0, 62]
- model: Elfrino/XwinXtended-20B
layer_range: [0, 62]
merge_method: slerp
base_model: Undi95/MXLewd-L2-20B
parameters:
t:
- filter: self_attn
value: [0.8, 0.8, 0.9, 0.7, .8]
- filter: mlp
value: [.8, 0.8, 0.9, 0.8, .7]
- value: 0.2
dtype: bfloat16