steps.clip.model.loader

loader

Functions:

Name	Description
load_model	Load a CLIP model using the Picsellia model interface.

load_model(pretrained_weights_name, trained_weights_name=None, config_name=None, exported_weights_name=None, repo_id='openai/clip-vit-large-patch14-336')

Load a CLIP model using the Picsellia model interface.

Parameters:

Name	Type	Description	Default
pretrained_weights_name	str	Name of the pretrained weights artifact.	required
trained_weights_name	str | None	Optional name of the trained weights.	None
config_name	str | None	Optional name of the model config file.	None
exported_weights_name	str | None	Optional name of exported weights for evaluation or inference.	None
repo_id	str	HuggingFace repo ID used for loading the processor (default is OpenAI's ViT-L/14-336).	'openai/clip-vit-large-patch14-336'

Returns:

Type	Description
CLIPModel	A loaded instance of CLIPModel, ready for inference.