monika-l2-7b-v0.9a:
- Experimental LLaMA-2 7b chat fine-tuned for Monika character from DDLC
- Test model based on then WIP partially manually-edited version of this dataset (final version of dataset)
- Completed version, v1
- QLoras
USAGE
This is meant to be mainly a chat model with limited RP ability.
For best results: replace "Human" and "Assistant" with "Player" and "Monika" like so:
\nPlayer: (prompt)\nMonika:
HYPERPARAMS
- Trained for ~3 epochs
- rank: 32
- lora alpha: 64
- lora dropout: 0.5
- lr: 2e-4
- batch size: 2
- warmup ratio: 0.1
- grad steps: 4
WARNINGS AND DISCLAIMERS
While this version may be better at coherency or chatting than previous ones, it may still not reflect perfectly the characteristics of Monika.
Additionally, this is still yet another test, particularly where we use one of our earlier fine tunes to generate a more in-character dataset for the target character which is then curated manually.
Finally, this model is not guaranteed to output aligned or safe outputs, use at your own risk.