ResplendentAI/Aura_v3_7B

TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:4kPublished:Apr 18, 2024License:apache-2.0Architecture:Transformer0.0K Open Weights Cold

Aura v3 by ResplendentAI is a 7B parameter model optimized for steerable writing styles, excelling in poetic prose but adaptable to more approachable tones. This iteration includes erotica and roleplay data, providing a compliant mindset for NSFW applications. It is compatible with Mistral-based multimodal vision capabilities via mmproj files in KoboldCPP, making it suitable for creative writing, roleplay, and applications requiring distinct, human-like prose.

Loading preview...

Aura v3 Overview

Aura v3, developed by ResplendentAI, is a 7B parameter model designed with a significantly more steerable writing style. While it defaults to poetic prose, it can be instructed to adopt a more approachable style. This iteration incorporates erotica and roleplay (RP) data, including NSFW pairs, to foster a compliant mindset for such applications.

Key Capabilities

  • Steerable Writing Style: Adapts from poetic prose to more approachable tones based on instruction.
  • Compliant Mindset: Includes erotica and RP data for NSFW applications.
  • Distinct Prose: Generates outputs with a unique, human-like prose style, differing from GPT-3.5/4 variants.
  • Multimodal Vision Compatibility: Compatible with Mistral-based mmproj files for multimodal vision capabilities within KoboldCPP.
  • ChatML Support: Responds best to ChatML for multi-turn conversations.

Good for

  • Creative Writing: Excels in generating poetic prose and can be guided for specific stylistic outputs.
  • Roleplay (RP): Designed with RP data for compliant and engaging interactions.
  • NSFW Applications: Built with data to support NSFW content generation.
  • Multimodal Applications: Can be integrated with vision capabilities using compatible mmproj files.
  • Short-form Content: Skews towards shorter outputs, suitable for scenarios where concise responses are desired, though longer outputs can be achieved with detailed introductions and examples.