| license: apache-2.0 | |
| tags: | |
| - merge | |
| - mergekit | |
| - mlabonne/AlphaMonarch-7B | |
| - bardsai/jaskier-7b-dpo-v5.6 | |
| - macadeliccc/MBX-7B-v3-DPO | |
| # pastiche-crown-clown-7B-dare | |
| pastiche-crown-clown-7B-dare is a DARE merge of the following models using [mergekit](https://github.com/cg123/mergekit): | |
| * [mlabonne/AlphaMonarch-7B](https://huggingface.co/mlabonne/AlphaMonarch-7B) | |
| * [bardsai/jaskier-7b-dpo-v5.6](https://huggingface.co/bardsai/jaskier-7b-dpo-v5.6) | |
| * [macadeliccc/MBX-7B-v3-DPO](https://huggingface.co/macadeliccc/MBX-7B-v3-DPO) | |
| See the paper [Language Models are Super Mario: Absorbing Abilities from Homologous Models as a Free Lunch](https://arxiv.org/abs/2311.03099) for more on the method. | |
| ## 🧩 Configuration | |
| ```yaml | |
| models: | |
| - model: mlabonne/NeuralMonarch-7B | |
| # No parameters necessary for base model | |
| - model: mlabonne/AlphaMonarch-7B | |
| parameters: | |
| density: 0.53 | |
| weight: 0.4 | |
| - model: bardsai/jaskier-7b-dpo-v5.6 | |
| parameters: | |
| density: 0.53 | |
| weight: 0.3 | |
| - model: macadeliccc/MBX-7B-v3-DPO | |
| parameters: | |
| density: 0.53 | |
| weight: 0.3 | |
| merge_method: dare_ties | |
| base_model: mlabonne/NeuralMonarch-7B | |
| parameters: | |
| int8_mask: true | |
| dtype: bfloat16 | |
| ``` |