|
--- |
|
base_model: |
|
- Undi95/PsyMedRP-v1-20B |
|
- Undi95/MXLewd-L2-20B |
|
library_name: transformers |
|
tags: |
|
- mergekit |
|
- merge |
|
|
|
--- |
|
# merge |
|
|
|
This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit). |
|
|
|
## Merge Details |
|
### Merge Method |
|
|
|
This model was merged using the SLERP merge method. |
|
|
|
### Models Merged |
|
|
|
The following models were included in the merge: |
|
* [Undi95/PsyMedRP-v1-20B](https://huggingface.co/Undi95/PsyMedRP-v1-20B) |
|
* [Undi95/MXLewd-L2-20B](https://huggingface.co/Undi95/MXLewd-L2-20B) |
|
|
|
### Configuration |
|
|
|
The following YAML configuration was used to produce this model: |
|
|
|
```yaml |
|
slices: |
|
- sources: |
|
- model: Undi95/PsyMedRP-v1-20B |
|
layer_range: [0, 62] # PsyMedRP has 62 layers |
|
- model: Undi95/MXLewd-L2-20B |
|
layer_range: [0, 62] # MXLewd has 62 layers |
|
merge_method: slerp |
|
base_model: Undi95/PsyMedRP-v1-20B |
|
parameters: |
|
t: |
|
- filter: self_attn |
|
value: [0.7, 0.6, 0.8, 0.9, 1] # Boost PsyMedRP's role in attention, especially deeper layers |
|
- filter: mlp |
|
value: [0.6, 0.7, 0.8, 0.4, 0.5] # Adjust MLP for more abstraction but also balance |
|
- value: 0.65 # Favor PsyMedRP overall in the blending |
|
dtype: bfloat16 # Preferred dtype, or adjust to fp16 for performance |
|
``` |
|
|
