This repo contains a merge of pre-trained Gemma 2 9B Instruct language models created using
mergekit.
None of the components of this merge were trained for roleplay nor intended for it. Despite this, the resulting model can be used effectively for that function. The virtue of this model lies in its coherence, as opposed to textual richness.
This project utilizes HODACHI/EZO-Common-9B-gemma-2-it, a model based on gemma-2 and fine-tuned by Axcxept co., ltd. Its primary goal was to perform well in Japanese language tasks. Model training leveraged context-based synthesized instruction pre-training data for supervised multitask pre-training
(abstract).
We also used princeton-nlp/gemma-2-9b-it-SimPO, a demonstration of Simple Preference Optimization
(abstract).
This model was merged using the SLERP merge method.
1slices:
2- sources:
3 - model: princeton-nlp/gemma-2-9b-it-SimPO
4 layer_range: [0, 42]
5 - model: HODACHI/EZO-Common-9B-gemma-2-it
6 layer_range: [0, 42]
7merge_method: slerp
8base_model: HODACHI/EZO-Common-9B-gemma-2-it
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
17
