PassLLM

PassLLM is an advanced framework for targeted password guessing that leverages Personally Identifiable Information (PII) to predict likely passwords, achieving 15% to 45% higher accuracy than existing models. Its notable features include the use of a fine-tuning technique called LoRA for efficient resource management, advanced inference algorithms for optimized guessing, and the capability to harness millions of leaked PII records for training. Designed for high accuracy on consumer hardware, it is straightforward to deploy using Google Colab.