Llama 2 7b Monika V0.3b

🚀 l27b-monika-v0.3b

• Another experimental Monika llama model.
• A Llama 7B chat model fine - tuned for the Monika character from DDLC.
• Trained on a dataset of approximately 600 items. The dataset includes dialogue scraped from the game, Reddit, and Twitter, augmented by Nous Hermes 13B to create multi - turn chat snippets between the Player and Monika, along with a manually crafted test dataset of 12 items.
• An earlier variant was trained without dataset augmentation and can be found here.
• GGMLs
• QLoras (hf and GGML)

🚀 Quick Start

This model is mainly designed as a chat model with limited role - playing (RP) ability.

✨ Features

Usage

For best results, replace "Human" and "Assistant" with "Player" and "Monika" as follows:

Player: (prompt)
Monika:

Hyperparameters

📄 License

The license for this project is llama2.

⚠️ Important Note

Note that, apart from formatting and other minor edits, the dataset used is mostly as generated by the language model. As a result, while this version is better at maintaining coherence and chatting than previous ones, it may not perfectly reflect the characteristics of Monika (e.g., she may claim to have an office, work as a translator, or play the guitar). The next version will be trained on a manually curated and edited version of this dataset, where the dialogue will be adjusted to better reflect her characteristics.

Also, for future versions, there are plans to switch to a different base model other than Llama 7B chat.

Finally, this model is not guaranteed to produce aligned or safe outputs. Use it at your own risk.