| Qlora finetuned from mlabonne/Llama-3.1-70B-Instruct-lorablated | |
| Trained on long form story content, instruction-following creative writing, roleplay, and adherence to system prompts. I tried to use a variety of prompting methods, but a markdown-like style for sorting information works best I think. | |
| My goal with this finetune is to have a fully uncensored model capable of a variety of styles and genres, and most importantly avoids the GPT-isms common in many creative models. | |
| Uses the Llama-3 chat template. No specific system prompts or keywords are necessary. | |
| This model is fully uncensored, you are responsible for what you choose to do with it. |