PygmalionAI/mythalion-13b

Warm
Public
13B
FP8
4096
License: llama2
Hugging Face
Overview

Mythalion 13B: A Roleplay and Chat Optimized Llama-2 Merge

Mythalion 13B is a 13 billion parameter language model built upon the Llama-2 architecture, resulting from a strategic merge of PygmalionAI's Pygmalion-2 13B and Gryphe's MythoMax L2 13B. This collaboration aimed to create a model highly proficient in interactive conversational scenarios.

Key Capabilities and Features

  • Optimized for Roleplay and Chat: Testers indicate Mythalion 13B outperforms MythoMax in roleplay (RP) and general chat interactions, making it suitable for engaging conversational agents.
  • Llama-2 Foundation: Benefits from the robust Llama-2 base, ensuring a strong language understanding and generation capability.
  • Flexible Prompting: Supports both Alpaca and Pygmalion/Metharme formatting, utilizing <|system|>, <|user|>, and <|model|> tokens for structured conversations.
  • Commercial Use Permitted: Available for both commercial and non-commercial applications under the Llama-2 license.

Intended Use and Limitations

Mythalion 13B is primarily intended for fictional writing and entertainment purposes. It was not fine-tuned for safety or harmlessness and may produce socially unacceptable or factually incorrect content, reflecting the nature of its training data. Users should follow recommended generation settings, particularly for platforms like SillyTavern, to achieve optimal results.