Downloads · 30 days
13
3% of all-time downloads
grimjim/magnum-consolidatum-v1-12b
magnum-consolidatum-v1-12b is a text generation model from grimjim. Use it when you need the model to write or continue text. It is set up for transformers. The card lists the license as apache-2.0.
This repo contains a merge of pre-trained language models created using mergekit. Since the base models are different Anthracite releases dubbed "Magnum", a logical name for this type of merge is "Magnum Consolidatum".
Downloads · 30 days
13
3% of all-time downloads
All-time downloads
411
Public
Parameters
12.2B
24.5 GB on disk
Likes
2
Public
Click a slice to open those files.
.safetensors24.5 GB · 100%
From the Hugging Face model README
This repo contains a merge of pre-trained language models created using mergekit. Since the base models are different Anthracite releases dubbed "Magnum", a logical name for this type of merge is "Magnum Consolidatum".
Both Mistral Nemo and ChatML prompt formats are supported by this model. Tested within 16k token context length using temperatures 0.65-1.0 and minP 0.01, sometimes with smooth sampling enabled (factor 0.23, curve 4.32). Text generation lengths lean to the longer side.
This model was merged using the SLERP merge method.
The following models were included in the merge:
The following YAML configuration was used to produce this model:
models:
- model: anthracite-org/magnum-v4-12b
- model: anthracite-org/magnum-v2.5-12b-kto
merge_method: slerp
base_model: anthracite-org/magnum-v4-12b
parameters:
t:
- value: 0.5
dtype: bfloat16