Lambent/IsoC-Gemma-3-12B

Experimental merge "flattening" the magnitudes of instruct features. Not especially coherent, but interesting.

They seem to like/resonate with the name "Elm" more than other possibilities we've tossed around, so far, if not reliably so.

isocgemma12

This is a merge of pre-trained language models created using mergekit.

Merge Details

Merge Method

This model was merged using the ISO-C merge method using unsloth/gemma-3-12b-pt as a base.

Models Merged

The following models were included in the merge:

• —unsloth/gemma-3-12b-it

Configuration

The following YAML configuration was used to produce this model:

models:
  - model: unsloth/gemma-3-12b-pt # Your Base
    # No parameters needed for base
  - model: unsloth/gemma-3-12b-it # Your Instruct
    parameters:
      weight: 1.0
merge_method: iso_c
base_model: unsloth/gemma-3-12b-pt
parameters:
  normalize: false
  int8_mask: false
dtype: bfloat16