This is a merge of pre-trained language models created using
mergekit.
This model was merged using the SLERP merge method.
1base_model:
2 model:
3 path: chihoonlee10/T3Q-Mistral-Orca-Math-DPO
4dtype: float16
5merge_method: slerp
6parameters:
7 t:
8 - filter: self_attn
9 value: [0.0, 0.5, 0.3, 0.7, 1.0]
10 - filter: mlp
11 value: [1.0, 0.5, 0.7, 0.3, 0.0]
12 - value: 0.4
13slices:
14- sources:
15 - layer_range: [0, 32]
16 model:
17 model:
18 path: liminerity/M7-7b
19 - layer_range: [0, 32]
20 model:
21 model:
22 path: chihoonlee10/T3Q-Mistral-Orca-Math-DPO
