Skip to content

unitorch.cli.models.vit¤

ViTProcessor¤

Tip

core/process/vit is the section for configuration of ViTProcessor.

Bases: ViTProcessor

Vision Transformer (ViT) Processor for handling image processing tasks.

Initialize ViTProcessor.

Parameters:

Name Type Description Default
vision_config_path str

The path to the vision config file.

 required
Source code in src/unitorch/cli/models/vit/processing.py 
26
27
28
29
30
31
32
33
34
35
36
37
38
def __init__(
    self,
    vision_config_path: str,
):
    """
    Initialize ViTProcessor.

    Args:
        vision_config_path (str): The path to the vision config file.
    """
    super().__init__(
        vision_config_path=vision_config_path,
    )

from_core_configure classmethod ¤

from_core_configure(config, **kwargs)

Create an instance of ViTProcessor from a core configuration.

Parameters:

Name Type Description Default
config

The core configuration.

 required
**kwargs

Additional keyword arguments.

 {}

Returns:

Name Type Description
dict

The processed arguments for initializing the processor.
Source code in src/unitorch/cli/models/vit/processing.py 
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
@classmethod
@add_default_section_for_init("core/process/vit")
def from_core_configure(cls, config, **kwargs):
    """
    Create an instance of ViTProcessor from a core configuration.

    Args:
        config: The core configuration.
        **kwargs: Additional keyword arguments.

    Returns:
        dict: The processed arguments for initializing the processor.
    """
    config.set_default_section("core/process/vit")
    pretrained_name = config.getoption(
        "pretrained_name", "vit-base-patch16-224-in21k"
    )
    vision_config_path = config.getoption("vision_config_path", None)
    vision_config_path = pop_value(
        vision_config_path,
        nested_dict_value(pretrained_vit_infos, pretrained_name, "vision_config"),
    )

    vision_config_path = cached_path(vision_config_path)

    return {
        "vision_config_path": vision_config_path,
    }

ViTForImageClassification¤

Tip

core/model/classification/vit is the section for configuration of ViTForImageClassification.

Bases: ViTForImageClassification

Vision Transformer (ViT) for Image Classification model.

Initialize ViTForImageClassification.

Parameters:

Name Type Description Default
config_path str

The path to the model's configuration file.

 required
num_classes Optional[int]

The number of classes for image classification. Defaults to 1.

 1
Source code in src/unitorch/cli/models/vit/modeling.py 
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
def __init__(
    self,
    config_path: str,
    num_classes: Optional[int] = 1,
):
    """
    Initialize ViTForImageClassification.

    Args:
        config_path (str): The path to the model's configuration file.
        num_classes (Optional[int]): The number of classes for image classification.
            Defaults to 1.
    """
    super().__init__(
        config_path=config_path,
        num_classes=num_classes,
    )

forward ¤

forward(pixel_values: Tensor)

Forward pass of the ViTForImageClassification model.

Parameters:

Name Type Description Default
pixel_values Tensor

The input pixel values of the image.

 required

Returns:

Name Type Description
ClassificationOutputs

The output logits of the model.
Source code in src/unitorch/cli/models/vit/modeling.py 
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
@autocast(device_type=("cuda" if torch.cuda.is_available() else "cpu"))
def forward(
    self,
    pixel_values: torch.Tensor,
):
    """
    Forward pass of the ViTForImageClassification model.

    Args:
        pixel_values (torch.Tensor): The input pixel values of the image.

    Returns:
        ClassificationOutputs: The output logits of the model.
    """
    outputs = super().forward(pixel_values=pixel_values)
    return ClassificationOutputs(outputs=outputs)

from_core_configure classmethod ¤

from_core_configure(config, **kwargs)

Create an instance of ViTForImageClassification from a core configuration.

Parameters:

Name Type Description Default
config

The core configuration.

 required
**kwargs

Additional keyword arguments.

 {}

Returns:

Name Type Description
ViTForImageClassification

The initialized ViTForImageClassification instance.
Source code in src/unitorch/cli/models/vit/modeling.py 
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
@classmethod
@add_default_section_for_init("core/model/classification/vit")
def from_core_configure(cls, config, **kwargs):
    """
    Create an instance of ViTForImageClassification from a core configuration.

    Args:
        config: The core configuration.
        **kwargs: Additional keyword arguments.

    Returns:
        ViTForImageClassification: The initialized ViTForImageClassification instance.
    """
    config.set_default_section("core/model/classification/vit")
    pretrained_name = config.getoption(
        "pretrained_name", "vit-base-patch16-224-in21k"
    )
    config_path = config.getoption("config_path", None)
    config_path = pop_value(
        config_path,
        nested_dict_value(pretrained_vit_infos, pretrained_name, "config"),
    )

    config_path = cached_path(config_path)
    num_classes = config.getoption("num_classes", 1)

    inst = cls(
        config_path=config_path,
        num_classes=num_classes,
    )
    pretrained_weight_path = config.getoption("pretrained_weight_path", None)
    weight_path = pop_value(
        pretrained_weight_path,
        nested_dict_value(pretrained_vit_infos, pretrained_name, "weight"),
        check_none=False,
    )
    if weight_path is not None:
        inst.from_pretrained(weight_path)

    return inst