Thanks to @Epiculous for the dope model/ help with llm backends and support overall.
Id like to also thank @kalomaze for the dope sampler additions to ST.
ST users can find the TextGenPreset in the folder labeled so.
Quants:Thank you @bartowski, @jeiku, @konz00.
1slices:
2 - sources:
3 - model: SanjiWatsuki/Kunoichi-DPO-v2-7B
4 layer_range: [0, 32]
5 - model: Epiculous/Fett-uccine-7B
6 layer_range: [0, 32]
7merge_method: slerp
8base_model: SanjiWatsuki/Kunoichi-DPO-v2-7B
9parameters:
10 t:
11 - filter: self_attn
12 value: [0, 0.5, 0.3, 0.7, 1]
13 - filter: mlp
14 value: [1, 0.5, 0.7, 0.3, 0]
15 - value: 0.5
16dtype: bfloat16
Detailed results can be found
here
