DoppelReflEx/MN-12B-Kakigori

TEXT GENERATIONPricing:Input $1.2 / Output $4.8Concurrent Unit Cost:1Model Size:12.2BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jan 29, 2025License:cc-by-nc-4.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

DoppelReflEx/MN-12B-Kakigori is a 12.2 billion parameter merged language model developed by DoppelReflEx, built upon cgato/Nemo-12b-Humanize-KTO-Experimental-Latest and crestf411/MN-Slush. This model is specifically noted for its decent performance in roleplay (RP) and erotic roleplay (ERP) scenarios, offering a context length of 32768 tokens. It aims to provide a balanced experience for interactive narrative generation, with reported evaluation scores surpassing its predecessor, WolfFrame.

Loading preview...

Overview

DoppelReflEx/MN-12B-Kakigori is a 12.2 billion parameter language model created by DoppelReflEx through a merge of two base models: cgato/Nemo-12b-Humanize-KTO-Experimental-Latest and crestf411/MN-Slush. The merge utilized a slerp method with specific parameter weighting. This model offers a substantial context length of 32768 tokens.

Key Capabilities

  • Roleplay (RP) and Erotic Roleplay (ERP): The model is designed and noted for its "good enough" performance in both general and erotic roleplay contexts.
  • Improved Evaluation Scores: It reportedly achieves better evaluation scores compared to its predecessor, DoppelReflEx/MN-12B-WolFrame.

Good For

  • Developers and users looking for a model specifically tuned for interactive narrative generation, particularly in roleplaying scenarios.
  • Experimentation with merged models that aim to combine characteristics from different base models for specific conversational styles.
  • Applications requiring a model with a decent balance of performance for creative text generation tasks.

Finetunes
2 models