PsyMedLewd. A merge of two of my favourite models for scifi stories:
Currently testing more merge iterations of these two models.
Max Ctx. Tokens - 4096
Amount to Gen. - 512
First iteration. More to come..
#########################################################################################################
This is a merge of pre-trained language models created using
mergekit.
This model was merged using the SLERP merge method.
1slices:
2 - sources:
3 - model: Undi95/PsyMedRP-v1-20B
4 layer_range: [0, 62] # PsyMedRP has 62 layers
5 - model: Undi95/MXLewd-L2-20B
6 layer_range: [0, 62] # MXLewd has 62 layers
7merge_method: slerp # Or use another method like weight_average if needed
8base_model: Undi95/MXLewd-L2-20B # Can use either as the base model
9parameters:
10 t:
11 - filter: self_attn
12 value: [0, 0.5, 0.3, 0.7, 1] # Tune these for desired effect
13 - filter: mlp
14 value: [1, 0.5, 0.7, 0.3, 0]
15 - value: 0.5 # Default averaging weight
16dtype: bfloat16 # Use preferred dtype, like fp16 or float32 if needed
