AuriAetherwiing/MN-12B-Starcannon-v3

TEXT GENERATIONConcurrency Cost:1Model Size:12BQuant:FP8Ctx Length:32kPublished:Aug 6, 2024Architecture:Transformer0.0K Cold

AuriAetherwiing/MN-12B-Starcannon-v3 is a 12 billion parameter language model merge created by AuriAetherwiing using the TIES method. Based on nothingiisreal/MN-12B-Celeste-V1.9, it integrates anthracite-org/magnum-12b-v2 to enhance its capabilities. This model is designed for general text generation tasks, with both constituent models now ChatML trained, offering improved conversational performance. It provides a versatile foundation for various natural language processing applications.

Loading preview...

Model Overview

AuriAetherwiing/MN-12B-Starcannon-v3 is a 12 billion parameter language model resulting from a merge of pre-trained models. Developed by AuriAetherwiing, this model utilizes the TIES merge method, building upon nothingiisreal/MN-12B-Celeste-V1.9 as its base.

Key Capabilities

  • Merged Architecture: Combines anthracite-org/magnum-12b-v2 and nothingiisreal/MN-12B-Celeste-V1.9 to leverage their respective strengths.
  • ChatML Training: Both constituent models have undergone ChatML training, suggesting improved performance in conversational and instruction-following tasks.
  • Mergekit Integration: Created using mergekit, indicating a structured and reproducible merging process.
  • Parameter Count: Features 12 billion parameters, offering a balance between performance and computational requirements.

Good For

  • General Text Generation: Suitable for a wide range of natural language processing tasks.
  • Conversational AI: Enhanced by ChatML training, it can be applied to chatbots and interactive applications.
  • Experimentation: As a merged model, it provides a foundation for developers looking to explore combined model capabilities.