Mythalion 13b

🚀 Mythalion 13B

A merge of Pygmalion-2 13B and MythoMax 13B, offering enhanced text generation capabilities for various applications.

🚀 Quick Start

This long-awaited new model based on Llama - 2 is now available. It's a blend of Pygmalion-2 13B and MythoMax L2 13B, created in collaboration with Gryphe.

✨ Features

• High - performance in RP/Chat: According to testers, it seems to outperform MythoMax in RP/Chat scenarios.
• Flexible Prompting: Can be prompted using both Alpaca and Pygmalion formatting.
• License - compliant: Freely available for both commercial and non - commercial use under the Llama - 2 license.

📚 Documentation

Model Details

The finer details of the merge are available in our blogpost.

⚠️ Important Note

Please make sure you follow the recommended generation settings for SillyTavern here for the best results!

Prompting

This model can be prompted using two different formats:

Alpaca formatting:

### Instruction:
<prompt>

### Response:
<leave a newline blank for model to respond>

Pygmalion/Metharme formatting:

<|system|>Enter RP mode. Pretend to be {{char}} whose persona follows:
{{persona}}

You shall reply to the user while staying in character, and generate long responses.
<|user|>Hello!<|model|>{model's response goes here}

The model has been trained on prompts using three different roles, denoted by the tokens: <|system|>, <|user|> and <|model|>. The <|system|> prompt can inject out - of - channel information, the <|user|> prompt indicates user input, and the <|model|> token indicates that the model should generate a response. These tokens can be chained to form a conversation history.

Limitations and biases

The intended use - case for this model is fictional writing for entertainment purposes. Any other usage is out of scope.

⚠️ Important Note

This model was not fine - tuned to be safe and harmless. The base model and this fine - tune have been trained on data containing profanity and offensive texts. It may produce socially unacceptable or undesirable text, and outputs might often be factually wrong or misleading.

Acknowledgements

We would like to thank SpicyChat for sponsoring the training for the Pygmalion-2 13B model.

📄 License

This model is freely available for both commercial and non - commercial use, as per the Llama - 2 license.

📦 Training Data