Sorihon/Transformed-Journey-24B

TEXT GENERATIONConcurrent Unit Cost:2Model Size:24BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jul 4, 2026Architecture:Transformer Featherless Exclusive Cold

Sorihon/Transformed-Journey-24B is a 24 billion parameter language model created by Sorihon, merged using the SCE method from Sorihon/Amended-Journey-24B, Sorihon/Experimental-Resonance-24B, and Sorihon/Magistry-Painted-Cydonia-24B. This model leverages a 32768 token context length and is designed as a composite model, combining the strengths of its constituent base models for broad language understanding and generation tasks. Its unique merge configuration aims to enhance overall performance by blending different pre-trained model characteristics.

Loading preview...

Overview

Sorihon/Transformed-Journey-24B is a 24 billion parameter language model developed by Sorihon. It was created using the SCE (Selective Channel Ensemble) merge method, a technique designed to combine the capabilities of multiple pre-trained language models. The base model for this merge was Sorihon/Amended-Journey-24B, with additional contributions from Sorihon/Experimental-Resonance-24B and Sorihon/Magistry-Painted-Cydonia-24B.

Merge Details

This model's architecture is a result of a specific merge configuration, where different weights were assigned to the contributing models:

  • Sorihon/Magistry-Painted-Cydonia-24B: weight 1.0
  • Sorihon/Experimental-Resonance-24B: weight 0.6
    The merge process utilized mergekit and was configured with dtype: bfloat16, normalize: true, and select_topk: 0.5 parameters, indicating a focus on optimizing the blend of features from the source models. The tokenizer was set to union to ensure comprehensive vocabulary coverage from the merged components.

Key Characteristics

  • Parameter Count: 24 billion parameters.
  • Context Length: Supports a substantial context window of 32768 tokens.
  • Merge Method: Employs the SCE method, which is detailed in the arXiv paper, suggesting a sophisticated approach to model combination.

Good For

This model is suitable for users looking for a robust language model that integrates the strengths of several specialized or general-purpose models. Its merged nature implies a broad range of potential applications, from complex reasoning to creative text generation, benefiting from the combined knowledge and capabilities of its constituent parts.

Finetunes
1 models