knifeayumu/Cydonia-v4-MS3.2-Magnum-Diamond-24B

Hugging Face
TEXT GENERATIONConcurrency Cost:2Model Size:24BQuant:FP8Ctx Length:32kPublished:Jul 23, 2025License:apache-2.0Architecture:Transformer0.0K Open Weights Warm

Cydonia-v4-MS3.2-Magnum-Diamond-24B is a 24 billion parameter language model developed by knifeayumu, created through a SLERP merge of TheDrummer/Cydonia-24B-v4 and Doctor-Shotgun/MS3.2-24B-Magnum-Diamond. This model aims to refine the behavior of its constituents, specifically addressing verbosity and 'horniness' observed in MS3.2-24B-Magnum-Diamond. It is designed for general language generation tasks, offering a balanced output by combining characteristics of its merged components.

Loading preview...

Overview

Cydonia-v4-MS3.2-Magnum-Diamond-24B is a 24 billion parameter language model developed by knifeayumu. It is a merged model, combining the strengths of two pre-trained language models: TheDrummer/Cydonia-24B-v4 and Doctor-Shotgun/MS3.2-24B-Magnum-Diamond. The merge was performed using the SLERP (Spherical Linear Interpolation) method, a technique often used to combine model weights smoothly.

Key Characteristics

  • Parameter Count: 24 billion parameters.
  • Merge Method: Utilizes the SLERP merge method for combining model weights.
  • Base Model: TheDrummer/Cydonia-24B-v4 served as the base model for the merge.
  • Refinement Focus: The primary motivation for this merge was to mitigate the "horny and verbose" tendencies observed in the Doctor-Shotgun/MS3.2-24B-Magnum-Diamond component, aiming for more balanced and controlled output.

Intended Use Cases

This model is suitable for general language generation tasks where a refined and less verbose output is desired compared to its more uninhibited constituent models. It aims to provide a more tempered response profile, making it potentially useful for applications requiring more neutral or controlled text generation.

Popular Sampler Settings

Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.

temperature
top_p
top_k
frequency_penalty
presence_penalty
repetition_penalty
min_p