23 lines
636 B
YAML
23 lines
636 B
YAML
|
base_model: Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v8
|
||
|
|
merge_method: slerp
|
||
|
|
tokenizer_source: base
|
||
|
|
dtype: bfloat16
|
||
|
|
parameters:
|
||
|
|
t:
|
||
|
|
- filter: self_attn
|
||
|
|
value: [0.0, 0.5, 0.3, 0.7, 1.0]
|
||
|
|
- filter: mlp
|
||
|
|
value: [1.0, 0.5, 0.7, 0.3, 0.0]
|
||
|
|
- value: 0.5
|
||
|
|
slices:
|
||
|
|
- sources:
|
||
|
|
- model: Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v8
|
||
|
|
layer_range: [0, 24]
|
||
|
|
- model: Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v5
|
||
|
|
layer_range: [0, 24]
|
||
|
|
- sources:
|
||
|
|
- model: Lunzima/NQLSG-Qwen2.5-14B-MegaFusion-v8
|
||
|
|
layer_range: [24, 48]
|
||
|
|
- model: Lunzima/NQLSG-Qwen2.5-14B-OriginalFusion
|
||
|
|
layer_range: [24, 48]
|