--- tags: - not-for-all-audiences - merge - mergekit base_model: - Sao10K/Fimbulvetr-11B-v2 - KoboldAI/Mistral-7B-Holodeck-1 - KoboldAI/Mistral-7B-Erebus-v3 - senseable/WestLake-7B-v2 - fblgit/UNA-TheBeagle-7b-v1 - SanjiWatsuki/Kunoichi-DPO-v2-7B - SanjiWatsuki/Silicon-Maid-7B --- Refer to the original models for best usage. - [Sao10K/Fimbulvetr-11B-v2](https://huggingface.co/Sao10K/Fimbulvetr-11B-v2) - [KoboldAI/Mistral-7B-Holodeck-1](https://huggingface.co/KoboldAI/Mistral-7B-Holodeck-1) - [KoboldAI/Mistral-7B-Erebus-v3](https://huggingface.co/KoboldAI/Mistral-7B-Erebus-v3) - [senseable/WestLake-7B-v2](https://huggingface.co/senseable/WestLake-7B-v2) - [fblgit/UNA-TheBeagle-7b-v1](https://huggingface.co/fblgit/UNA-TheBeagle-7b-v1) - [SanjiWatsuki/Kunoichi-DPO-v2-7B](https://huggingface.co/SanjiWatsuki/Kunoichi-DPO-v2-7B) - [SanjiWatsuki/Silicon-Maid-7B](https://huggingface.co/SanjiWatsuki/Silicon-Maid-7B) --- # Mergekit Recipe ```yaml # Mistral Merging merge_method: linear dtype: float16 models: - model: KoboldAI/Mistral-7B-Holodeck-1 parameters: weight: 0.45 - model: KoboldAI/Mistral-7B-Erebus-v3 parameters: weight: 0.45 - model: senseable/WestLake-7B-v2 parameters: weight: 0.025 - model: fblgit/UNA-TheBeagle-7b-v1 parameters: weight: 0.025 - model: SanjiWatsuki/Kunoichi-DPO-v2-7B parameters: weight: 0.025 - model: SanjiWatsuki/Silicon-Maid-7B parameters: weight: 0.025 name: MV01-WestHolobuSiliKunoichiBeagle-7B --- # Mistral Stacking merge_method: passthrough dtype: float16 slices: - sources: - model: MV01-WestHolobuSiliKunoichiBeagle-7B layer_range: [0, 24] - sources: - model: MV01-WestHolobuSiliKunoichiBeagle-7B layer_range: [8, 32] name: MV01-Stacked-WestHolobuSiliKunoichiBeagle-Unhealed-10.7B --- # Mistral Stack Healing merge_method: task_arithmetic dtype: float16 base_model: Undi95/Mistral-11B-v0.1 models: - model: Sao10K/Fimbulvetr-11B-v2 parameters: weight: 1.0 - model: MV01-Stacked-WestHolobuSiliKunoichiBeagle-Unhealed-10.7B parameters: weight: 1.0 name: FimbulWestHolobuSiliKunoichiBeagle-UnfixedConfig-10.7B --- # Healed Stack Config Fixing merge_method: task_arithmetic dtype: float16 base_model: Sao10K/Fimbulvetr-11B-v2 models: - model: FimbulWestHolobuSiliKunoichiBeagle-UnfixedConfig-10.7B parameters: weight: 1.0 ```