Mistral's Shieldstral: 3B open-weights model for multimodal moderation
Shieldstral introduces a 3B open-weights multimodal safety classifier that outperforms models up to 7x its size.
Summary
Mistral’s new Shieldstral model is a multimodal safety classifier that boasts impressive performance, exceeding the capabilities of larger models by up to seven times. This advancement could significantly enhance moderation processes across various platforms.
Comments (133 of ) ⌛