Llama-3-70b-Toxic-DPO-v0.1

Finetune of Llama-3-70b-Instruct using unalignment/toxic-dpo-v0.2, using MonsterAPI

Model Details

Trained for 1 epoch, use convert-lora-to-ggml.py in this repo to merge with the Llama-3-70b GGUF.

• Developed by: Meta
• Shared by: leafspark
• Model type: Large Language Model
• Language(s) (NLP): English
• License: Llama-3 Community License

Uses

Use at your own risk; I am not responsible for what you do with this model.

Prompt Format:

Vicuna, merge with WizardLM maybe:

{system}
USER: {user} ASSISTANT: {assistant}</s>

Framework versions

• PEFT 0.10.0