I combined like 6 different Gemma 3 4B models together to see what happens. Answer: nothing really. It's like the most average model ever.
This is a merge of pre-trained language models created using
mergekit.
This model was merged using the
SLERP merge method.
1# config_redo.yaml
2models:
3 - model: ./merged_final
4 - model: ./other_gemma
5merge_method: slerp
6base_model: ./merged_final
7parameters:
8 t: 0.5
9dtype: bfloat16
