Creative and coherent, more tests needed...
A SLERP merge of two nice models:
and
#############################################################
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/PsyMedRP-v1-20B
layer_range: [0, 62] # PsyMedRP has 62 layers
- model: tavtav/Rose-20B
layer_range: [0, 62] # Rose-20B has 62 layers
merge_method: slerp # Or use another method like weight_average if needed
base_model: Undi95/PsyMedRP-v1-20B # PsyMedRP as the base for reasoning dominance
parameters:
t:
- filter: self_attn
value: [0.3, 0.7, 1, 0.7, 0.4] # Boost creativity in pattern recognition
- filter: mlp
value: [0.2, 0.6, 1, 0.8, 0.5] # Emphasize creativity in decision-making and connections
- value: 0.639 # Slight leaning toward PsyMedRP's base characteristics
dtype: bfloat16