Calme-Ties-78B

Calme-Ties-78B

Calme-Ties-78B is a 78-billion-parameter model merged using the TIES methodology, based on the Qwen2 architecture. It integrates two sub-base models: calme-3.2-instruct-78B by MaziyarPanahi and CalmeRys-78B-Orpo-v0.1 by dfurman, which serves as the base model. The merging process assigns equal weight and density to both models, with additional parameters enabling normalization and int8 masking. The model operates using the bfloat16 data type.

Merged Models

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

Merge Method

This model was merged using the TIES merge method using dfurman/CalmeRys-78B-Orpo-v0.1 as a base.

Models Merged

The following models were included in the merge:

• MaziyarPanahi/calme-3.2-instruct-78b

Configuration

The following YAML configuration was used to produce this model:

models:
  - model: MaziyarPanahi/calme-3.2-instruct-78b
    parameters:
      weight: 1
      density: 1
merge_method: ties
base_model: dfurman/CalmeRys-78B-Orpo-v0.1
parameters:
  weight: 1
  density: 1
  normalize: true
  int8_mask: true
dtype: bfloat16