IkariDev/Athena-v4

TEXT GENERATIONConcurrency Cost:1Model Size:13BQuant:FP8Ctx Length:4kPublished:Oct 7, 2023License:cc-by-nc-4.0Architecture:Transformer0.0K Open Weights Cold

IkariDev/Athena-v4 is a 13 billion parameter experimental language model developed by IkariDev and Undi95, built upon a complex merge of several 13B models including Athena-v3, Xwin-LM, PsyMedRP, Thespis, and Airoboros. This model is specifically fine-tuned for roleplay (RP), erotic roleplay (ERP), and general conversational tasks, utilizing the Alpaca prompt format. Its unique merging recipe aims to enhance its capabilities in interactive and creative text generation.

Loading preview...

Athena-v4: An Experimental Merged Language Model

Athena-v4 is a 13 billion parameter experimental language model developed by IkariDev and Undi95. This model is the result of a sophisticated merging process, combining several distinct 13B models to create a unique conversational agent. It is designed to be highly versatile, particularly excelling in interactive and creative text generation scenarios.

Key Capabilities & Features

  • Complex Merging Architecture: Athena-v4 is built from a multi-stage merge of prominent 13B models, including Athena-v3, Xwin-LM/Xwin-LM-13B-V0.1, Undi95/PsyMedRP-v1-13B, cgato/Thespis-13b-v0.2, and jondurbin/airoboros-l2-13b-3.0. This intricate recipe aims to leverage the strengths of its constituent models.
  • Alpaca Prompt Format: The model is optimized to work with the Alpaca prompt template, ensuring straightforward integration and consistent response generation for instruction-based tasks.
  • Community-Rated Performance: The model's performance is supported by user ratings, indicating its effectiveness in its target applications.

Good For

  • Roleplay (RP): Optimized for engaging in detailed and dynamic roleplaying scenarios.
  • Erotic Roleplay (ERP): Specifically designed to handle and generate content for erotic roleplay.
  • General Conversational Tasks: Capable of general chat and instruction-following, making it suitable for a wide range of interactive applications.