onnx_diagnostic.tasks.zero_shot_image_classification

onnx_diagnostic.tasks.zero_shot_image_classification.get_inputs(model: Module, config: Any | None, dummy_max_token_id: int, batch_size: int = 2, sequence_length: int = 30, input_width: int = 224, input_height: int = 224, input_channels: int = 3, batch_size_image=3, add_second_input: int = 1, **kwargs)

Generates inputs for task zero-short-image-classification.

Parameters:

• model – model to get the missing information

• config – configuration used to generate the model

• dummy_max_token_id – vocabulary size

• batch_size – batch size

• sequence_length – sequence length

• batch_size_image – number of images

• input_channels – input channel

• input_width – input width

• input_height – input height

Returns:

dictionary

# input_ids:T7s2x7 # attention_mask:T7s2x7 # pixel_values:T1s2x3x224x224

onnx_diagnostic.tasks.zero_shot_image_classification.random_input_kwargs(config: Any) → Tuple[Dict[str, Any], Callable]

Inputs kwargs.

If the configuration is None, the function selects typical dimensions.

onnx_diagnostic.tasks.zero_shot_image_classification.reduce_model_config(config: Any) → Dict[str, Any]

Reduces a model size.