Mass-Producing Failures of Multimodal Systems with Language Models

arXiv:2306.12105v2 Announce Type: replace
Abstract: Deployed multimodal systems can fail in ways that evaluators did not anticipate. In order to find these failures before deployment, we introduce MultiMon, a system that automatically identifies systematic failures -- generalizable, natural-language descriptions of patterns of model failures. To uncover systematic failures, MultiMon scrapes a corpus for examples of erroneous agreement: inputs that produce the same output, but should not. It then prompts a language model (e.g., GPT-4) to find systematic patterns of failure …

arxiv cs.cl cs.lg cs.se language language models multimodal multimodal systems systems type