ockerman0/MN-12B-Starcannon-v4-unofficial
The ockerman0/MN-12B-Starcannon-v4-unofficial is a 12 billion parameter language model with a 32768 token context length, created by ockerman0 as an unofficial continuation of the "Starcannon" series. This model is a merge of nothingiisreal/MN-12B-Celeste-V1.9 and anthracite-org/magnum-12b-v2.5-kto, utilizing the TIES merge method. It is designed to combine the strengths of its constituent models, offering a versatile foundation for various generative AI tasks.
Loading preview...
Model Overview
The ockerman0/MN-12B-Starcannon-v4-unofficial is a 12 billion parameter language model, an unofficial continuation of the "Starcannon" series. It features a substantial context length of 32768 tokens, making it suitable for processing longer inputs and generating more coherent, extended outputs.
Merge Details
This model was constructed using the TIES merge method via mergekit. The base model for this merge was nothingiisreal/MN-12B-Celeste-V1.9. It was merged with anthracite-org/magnum-12b-v2.5-kto.
Configuration
The merge configuration applied specific density and weight parameters to each constituent model:
anthracite-org/magnum-12b-v2.5-kto: density 0.3, weight 0.5nothingiisreal/MN-12B-Celeste-V1.9: density 0.7, weight 0.5
The merge process also included normalize: true and int8_mask: true parameters, with the final model output in bfloat16 dtype.
Usage
This model provides a merged architecture intended to leverage the combined capabilities of its base models. Users can find static GGUF quants at mradermacher/MN-12B-Starcannon-v4-unofficial-GGUF and imatrix GGUF quants at mradermacher/MN-12B-Starcannon-v4-unofficial-i1-GGUF.
Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.