LoRA Fine-tuning Efficiently Undoes Safety Training in Llama 2-Chat 70B.
Simon LermenCharlie Rogers-SmithJeffrey LadishPublished in: CoRR (2023)
Keyphrases
- fine tuning
- fine tuned
- agent technology
- learning perl
- viable alternative
- fine tune
- training process
- training set
- test set
- training algorithm
- database
- electronic commerce
- online learning
- supervised learning
- international conference
- content analysis
- training examples
- training phase
- programming language
- instant messaging
- general purpose
- decision making
- machine learning