This is a merge of pre-trained language models created using
mergekit.
This model was merged using the
DARE TIES merge method using
Gryphe/Pantheon-RP-1.0-8b-Llama-3 as a base.
1slices:
2- sources:
3 - layer_range: [0, 16]
4 model: Sao10K/L3-8B-Stheno-v3.2
5 parameters:
6 density: 0.5
7 weight: 1.0
8 - layer_range: [0, 16]
9 model: Gryphe/Pantheon-RP-1.0-8b-Llama-3
10 parameters:
11 density: 0.5
12 weight: 0.9
13- sources:
14 - layer_range: [16, 32]
15 model: Sao10K/L3-8B-Stheno-v3.2
16 parameters:
17 density: 0.5
18 weight: 0.9
19 - layer_range: [16, 32]
20 model: Gryphe/Pantheon-RP-1.0-8b-Llama-3
21 parameters:
22 density: 0.5
23 weight: 1.0
24merge_method: dare_ties
25tokenizer_source: base
26base_model: Gryphe/Pantheon-RP-1.0-8b-Llama-3
27parameters:
28 int8_mask: true
29dtype: bfloat16
