Language Models are Super Mario: Absorbing Abilities from Homologous Models as a Free Lunch
Paper
• 2311.03099 • Published
• 30
This is a merge of pre-trained language models created using mergekit.
This model was merged using the DARE TIES merge method using w4r10ck/SOLAR-10.7B-Instruct-v1.0-uncensored as a base.
The following models were included in the merge:
The following YAML configuration was used to produce this model:
merge_method: dare_ties
base_model: w4r10ck/SOLAR-10.7B-Instruct-v1.0-uncensored
parameters:
normalize: true
models:
- model: w4r10ck/SOLAR-10.7B-Instruct-v1.0-uncensored+jeiku/Luna_LoRA_SOLAR
parameters:
weight: 0.65
- model: w4r10ck/SOLAR-10.7B-Instruct-v1.0-uncensored+typeof/openhermes-2.5-mistral-lora
parameters:
weight: 1
- model: w4r10ck/SOLAR-10.7B-Instruct-v1.0-uncensored+Technonia/mistral-7b-dolly5k
parameters:
weight: 0.8
- model: w4r10ck/SOLAR-10.7B-Instruct-v1.0-uncensored+Anarchist/mistral_7b_lora_smol_pippa
parameters:
weight: 0.55
dtype: float16