Starting April 29, 2025, Gemini 1.5 Pro and Gemini 1.5 Flash models are not available in projects that have no prior usage of these models, including new projects. For details, see Model versions and lifecycle.
When enabled, a linear path from the maximally blurred image to the input
image is created. Using a blurred baseline instead of zero (black image) is
motivated by the BlurIG approach explained here:
https://arxiv.org/abs/2004.03383
When enabled, a linear path from the maximally blurred image to the input
image is created. Using a blurred baseline instead of zero (black image) is
motivated by the BlurIG approach explained here:
https://arxiv.org/abs/2004.03383
When enabled, the gradients are approximated by averaging the gradients
from noisy samples in the vicinity of the inputs. Adding
noise can help improve the computed gradients. Refer to this paper for more
details: https://arxiv.org/pdf/1706.03825.pdf
When enabled, the gradients are approximated by averaging the gradients
from noisy samples in the vicinity of the inputs. Adding
noise can help improve the computed gradients. Refer to this paper for more
details: https://arxiv.org/pdf/1706.03825.pdf
Required. The number of steps for approximating the path integral.
A good value to start is 50 and gradually increase until the
sum to diff property is within the desired error range.
Valid range of its value is [1, 100], inclusively.
When enabled, a linear path from the maximally blurred image to the input
image is created. Using a blurred baseline instead of zero (black image) is
motivated by the BlurIG approach explained here:
https://arxiv.org/abs/2004.03383
When enabled, the gradients are approximated by averaging the gradients
from noisy samples in the vicinity of the inputs. Adding
noise can help improve the computed gradients. Refer to this paper for more
details: https://arxiv.org/pdf/1706.03825.pdf
[[["Easy to understand","easyToUnderstand","thumb-up"],["Solved my problem","solvedMyProblem","thumb-up"],["Other","otherUp","thumb-up"]],[["Hard to understand","hardToUnderstand","thumb-down"],["Incorrect information or sample code","incorrectInformationOrSampleCode","thumb-down"],["Missing the information/samples I need","missingTheInformationSamplesINeed","thumb-down"],["Other","otherDown","thumb-down"]],["Last updated 2025-08-16 UTC."],[],[],null,["# Interface IntegratedGradientsAttributionOrBuilder (1.32.0)\n\n public interface IntegratedGradientsAttributionOrBuilder extends MessageOrBuilder\n\nImplements\n----------\n\n[MessageOrBuilder](https://cloud.google.com/java/docs/reference/protobuf/latest/com.google.protobuf.MessageOrBuilder.html)\n\nMethods\n-------\n\n### getBlurBaselineConfig()\n\n public abstract BlurBaselineConfig getBlurBaselineConfig()\n\nConfig for IG with blur baseline.\n\nWhen enabled, a linear path from the maximally blurred image to the input\nimage is created. Using a blurred baseline instead of zero (black image) is\nmotivated by the BlurIG approach explained here:\n\u003chttps://arxiv.org/abs/2004.03383\u003e\n\n`.google.cloud.vertexai.v1.BlurBaselineConfig blur_baseline_config = 3;`\n\n### getBlurBaselineConfigOrBuilder()\n\n public abstract BlurBaselineConfigOrBuilder getBlurBaselineConfigOrBuilder()\n\nConfig for IG with blur baseline.\n\nWhen enabled, a linear path from the maximally blurred image to the input\nimage is created. Using a blurred baseline instead of zero (black image) is\nmotivated by the BlurIG approach explained here:\n\u003chttps://arxiv.org/abs/2004.03383\u003e\n\n`.google.cloud.vertexai.v1.BlurBaselineConfig blur_baseline_config = 3;`\n\n### getSmoothGradConfig()\n\n public abstract SmoothGradConfig getSmoothGradConfig()\n\nConfig for SmoothGrad approximation of gradients.\n\nWhen enabled, the gradients are approximated by averaging the gradients\nfrom noisy samples in the vicinity of the inputs. Adding\nnoise can help improve the computed gradients. Refer to this paper for more\ndetails: \u003chttps://arxiv.org/pdf/1706.03825.pdf\u003e\n\n`.google.cloud.vertexai.v1.SmoothGradConfig smooth_grad_config = 2;`\n\n### getSmoothGradConfigOrBuilder()\n\n public abstract SmoothGradConfigOrBuilder getSmoothGradConfigOrBuilder()\n\nConfig for SmoothGrad approximation of gradients.\n\nWhen enabled, the gradients are approximated by averaging the gradients\nfrom noisy samples in the vicinity of the inputs. Adding\nnoise can help improve the computed gradients. Refer to this paper for more\ndetails: \u003chttps://arxiv.org/pdf/1706.03825.pdf\u003e\n\n`.google.cloud.vertexai.v1.SmoothGradConfig smooth_grad_config = 2;`\n\n### getStepCount()\n\n public abstract int getStepCount()\n\nRequired. The number of steps for approximating the path integral.\nA good value to start is 50 and gradually increase until the\nsum to diff property is within the desired error range.\n\nValid range of its value is \\[1, 100\\], inclusively.\n\n`int32 step_count = 1 [(.google.api.field_behavior) = REQUIRED];`\n\n### hasBlurBaselineConfig()\n\n public abstract boolean hasBlurBaselineConfig()\n\nConfig for IG with blur baseline.\n\nWhen enabled, a linear path from the maximally blurred image to the input\nimage is created. Using a blurred baseline instead of zero (black image) is\nmotivated by the BlurIG approach explained here:\n\u003chttps://arxiv.org/abs/2004.03383\u003e\n\n`.google.cloud.vertexai.v1.BlurBaselineConfig blur_baseline_config = 3;`\n\n### hasSmoothGradConfig()\n\n public abstract boolean hasSmoothGradConfig()\n\nConfig for SmoothGrad approximation of gradients.\n\nWhen enabled, the gradients are approximated by averaging the gradients\nfrom noisy samples in the vicinity of the inputs. Adding\nnoise can help improve the computed gradients. Refer to this paper for more\ndetails: \u003chttps://arxiv.org/pdf/1706.03825.pdf\u003e\n\n`.google.cloud.vertexai.v1.SmoothGradConfig smooth_grad_config = 2;`"]]