steps.clip.model.predictor

predictor

Functions:

Name	Description
predict	Inference step for CLIP on an image-only dataset.

predict(model, dataset)

Inference step for CLIP on an image-only dataset.

This step extracts image embeddings using the provided CLIP model and dataset.

Parameters:

Name	Type	Description	Default
model	CLIPModel	The CLIP model instance with loaded weights and processor.	required
dataset	CocoDataset	A COCO-style dataset containing image assets.	required

Returns:

Type	Description
list[PicselliaCLIPEmbeddingPrediction]	A list of predictions with image embeddings for each asset in the dataset.