DevilSthenoAnjirSoup-L3-8B

An experimental roleplaying model.

First, these three models were merged together using the Model Stock method:

• Sao10K/L3-8B-Stheno-v3.1 (for its creativity)
• Hastagaras/Anjir-8B-L3 (for its spiciness)
• mlabonne/NeuralDaredevil-8B-abliterated (for its intelligence)

Then, aaditya/Llama3-OpenBioLLM-8B was SLERP merged into the result at 0.1 weight to (maybe?) give the model better anatomical knowledge.

Uses the Llama 3 Instruct prompt format.

Recommended sampler settings:

• Temperature: 0.7
• Min-P: 0.08

This is a merge created using mergekit.

Merge Details

Merge Method

This model was merged using the SLERP merge method.

Models Merged

The following models were included in the merge:

• aaditya/Llama3-OpenBioLLM-8B
• Revile/SthenoSoup

Configuration

The following YAML configuration was used to produce this model:

slices:
  - sources:
      - model: Revile/SthenoSoup
        layer_range: [0, 32]
      - model: aaditya/Llama3-OpenBioLLM-8B
        layer_range: [0, 32]
merge_method: slerp
base_model: Revile/SthenoSoup
parameters:
  t:
    - value: 0.1
dtype: bfloat16