LRN

Description

Performs normalization on the local input region of the input tensor. The local region is defined across channels. For a tensor x[n, c, d1, ..., dk] with shape (N x C x D1 x D2, ..., Dk), the region is defined as follows.

Input

x: input tensor of type float16 or float.

Attribute

• alpha: float, indicating the scaling factor.
• beta: float, indicating the exponent.
• bias: float, specifically, float32. This parameter is optional. It indicates the offset. Generally, it is greater than 0 to avoid dividing by 0.
• size: int, indicating the number of channels to be summed. Only an odd number is supported.

Output

y: output tensor of the identical data type and shape to the input.

Constraints

None.

ONNX Opset Support

Opset v8/v9/v10/v11/v12/v13/v14/v15/v16/v17/v18