/BadT2I

[MM'23 Oral] "Text-to-image diffusion models can be easily backdoored through multimodal data poisoning"

Primary LanguagePythonMIT LicenseMIT

Issues