Struct CLIP 

pub struct CLIP { /* private fields */ }

CLIP Model

Implementations

impl CLIP

pub fn new(config: CLIPConfig) -> Self

Create new CLIP model

pub fn encode_image(&self, images: &Tensor) -> Result<Tensor, String>

Encode images

pub fn encode_text(&self, input_ids: &Tensor) -> Result<Tensor, String>

Encode text

pub fn forward( &self, images: &Tensor, input_ids: &Tensor, ) -> Result<Tensor, String>

Forward pass (compute similarity matrix)

pub fn zero_shot_classify( &self, images: &Tensor, text_prompts: &Tensor, ) -> Result<Vec<usize>, String>

Zero-shot classification

pub fn image_to_text_retrieval( &self, images: &Tensor, texts: &Tensor, ) -> Result<Vec<usize>, String>

Image-text retrieval (find best matching text for each image)

pub fn text_to_image_retrieval( &self, images: &Tensor, texts: &Tensor, ) -> Result<Vec<usize>, String>

Text-image retrieval (find best matching image for each text)