• search
    search
  • reviewers
    reviewers
  • feeds
    feeds
  • assignments
    assignments
  • settings
  • logout

Bypassing the Safety Training of Open-Source LLMs with Priming Attacks.

Jason VegaIsha ChaudharyChangming XuGagandeep Singh
Published in: CoRR (2023)
Keyphrases
  • open source
  • open source software
  • source code
  • test set
  • training process
  • case study
  • training algorithm
  • neural network
  • training set
  • online learning
  • intrusion detection
  • training examples
  • training phase
  • dos attacks