MatrixC7/Meidebenne-120b-v1.0-wbcal-2.65bpw-h6-exl2

• The calibration dataset is from VatsaDev/worldbuild.
• The measurement file is attached in the branch measurement.
• Perplexity:
  • calibration: Module quantized, calibration perplexity (quant): 14.6905
  • wikitext-103-v1: Evaluation perplexity: 12.7281

---

Meidebenne-120b-v1.0

This is a Frankenstein merge of sophosympatheia/Midnight-Rose-70B-v1.0 created using mergekit.

Hence, the name of this merged model, 'Meidebenne', is inspired by a variety of the deeper Midnight-Rose species.

For detailed instructions, it is recommended to refer to the original page of the sophosympatheia/Midnight-Rose-70B-v1.0 repository.

Merge Details

Merge Method

This model was merged using the passthrough merge method, which is exactly copied from nsfwthrowitaway69/Venus-120b-v1.2 / cognitivecomputations/MegaDolphin-120b.

Models Merged

The following models were included in the merge:

• sophosympatheia/Midnight-Rose-70B-v1.0

Configuration

The following YAML configuration was used to produce this model:

dtype: float16
merge_method: passthrough
slices:
- sources:
  - layer_range: [0, 20]
    model: "sophosympatheia/Midnight-Rose-70B-v1.0"
- sources:
  - layer_range: [10, 30]
    model: "sophosympatheia/Midnight-Rose-70B-v1.0"
- sources:
  - layer_range: [20, 40]
    model: "sophosympatheia/Midnight-Rose-70B-v1.0"
- sources:
  - layer_range: [30, 50]
    model: "sophosympatheia/Midnight-Rose-70B-v1.0"
- sources:
  - layer_range: [40, 60]
    model: "sophosympatheia/Midnight-Rose-70B-v1.0"
- sources:
  - layer_range: [50, 70]
    model: "sophosympatheia/Midnight-Rose-70B-v1.0"
- sources:
  - layer_range: [60, 80]
    model: "sophosympatheia/Midnight-Rose-70B-v1.0"