Open-MMUnlearning: Unifying Methods and Evaluation for MLLM Unlearning

Key points

• As multimodal large language models (MLLMs) become more capable and widely deployed, concerns about privacy and safety have become increasingly pressing.1
• Machine unlearning offers one approach to addressing these concerns by removing designated information from trained models while preserving unrelated capabilities.1
• Using a common evaluation protocol, we compare ten representative unlearning methods.1
• We further introduce a metric meta-evaluation protocol that tests faithfulness using models with controlled exposure to target knowledge and robustness under quantization and relearning.1

Sources (1)

• [1]Open-MMUnlearning: Unifying Methods and Evaluation for MLLM Unlearning
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 7, 04:30 PM

  We introduce Open-MMUnlearning, an open-source, extensible framework that integrates target-model preparation, multimodal data processing, unlearning, and evaluation through shared interfaces and structured configurations.

  As multimodal large language models (MLLMs) become more capable and widely deployed, concerns about privacy and safety have become increasingly pressing.

Extractive summary: sentences quoted from the sources.