LLM Security(@llm_sec):Does your LLM truly unlearn? An embarrassingly simple approach to recover unlearned knowledge "This paper reveals that applying quantization to models that have undergone unlearning can restore the "forgotten" information." "for unlearning methods with utility constraints, the unlearned model retains an average of 21% of the intended forgotten knowledge in full precision, which significantly increases to 83% after 4-bit quantization" code: https://t.co/Gn3AOAn48D paper: https://t.co/5nuo2ClNQh (not peer reviewed)

LLM Security

@llm_sec

Research, papers, jobs, and news on large language model security. Got something relevant? DM / tag @llm_sec

가입 April 2023

292 팔로잉 중 9.8K 팬

LLM Security@llm_sec

2024.11.04 20:22

Does your LLM truly unlearn? An embarrassingly simple approach to recover unlearned knowledge "This paper reveals that applying quantization to models that have undergone unlearning can restore the "forgotten" information." "for unlearning methods with utility constraints, the unlearned model retains an average of 21% of the intended forgotten knowledge in full precision, which significantly increases to 83% after 4-bit quantization" code: paper: (not peer reviewed)