This is a merge of pre-trained language models created using
mergekit.
This model was merged using the
SCE merge method using D:\mergekit\yamls\IPOplectic as a base.
1merge_method: sce
2select_topk: 0.25
3
4models:
5
6
7 - model: "D:\\mergekit\\yamls\\IPOplectic"
8 - model: "D:\\mergekit\\yamls\\sce_smoot_v2"
9 - model: schonsense/70B_lampblack_writers_room
10 - model: LatitudeGames/Nova-70B-Llama-3.3
11 - model: LatitudeGames/Wayfarer-Large-70B-Llama-3.3
12
13
14base_model: "D:\\mergekit\\yamls\\IPOplectic"
15
16parameters:
17 normalize: false
18 int8_mask: true
19
20dtype: float32
21out_dtype: bfloat16
22
23tokenizer:
24 source: union
25 pad_to_multiple_of: 8
26
