NeverSleep/Noromaid-13b-v0.2

TEXT GENERATIONConcurrency Cost:1Model Size:13BQuant:FP8Ctx Length:4kPublished:Dec 10, 2023License:cc-by-nc-4.0Architecture:Transformer0.0K Open Weights Cold

NeverSleep/Noromaid-13b-v0.2 is a 13 billion parameter experimental language model developed by IkariDev and Undi. This model is a merge of existing models and incorporates new, private datasets, including 'no_robots' for human-like behavior and 'Aesir Private RP dataset' for fresh roleplay data. It is designed for roleplay (RP), erotic roleplay (ERP), and general conversational tasks, offering a distinct alternative to other LLMs in its class.

Loading preview...

Overview

NeverSleep/Noromaid-13b-v0.2 is an experimental 13 billion parameter language model, a collaborative effort by IkariDev and Undi. This version is a merge of existing models and integrates new, private datasets to enhance its capabilities, particularly in conversational and roleplay scenarios. Users are advised that this is a highly experimental release, with a previous stable version (Noromaid 0.1.1) available for those seeking more predictable performance.

Key Capabilities

  • Roleplay (RP) and Erotic Roleplay (ERP): Specifically designed and tested for engaging in detailed roleplay and erotic roleplay interactions.
  • General Conversational Tasks: Capable of handling a variety of general conversational prompts.
  • Enhanced Human-like Behavior: Utilizes the no_robots dataset to promote more natural and human-like responses.
  • Fresh Data Integration: Incorporates new, private datasets, including the Aesir Private RP dataset, to provide novel and diverse content, avoiding common data saturation issues.

Prompting and Usage

Noromaid-13b-v0.2 supports both a custom prompting format, with downloadable SillyTavern config files for context and instruct modes, and the standard Alpaca prompting format. This flexibility allows users to choose the method best suited for their application. The model is currently undergoing testing, and the developers encourage community feedback for optimal settings and future improvements.