SReLU

Input

  • x:
    • Required: yes
    • Type: float32
    • Description: input tensor
    • Restriction: NC1HWC0
  • tr:
    • Required: yes
    • Type: float32
    • Description: filter weight. The operator actually uses the value in the weight file. For single-operator testing purposes, it is used as an input.
    • Restriction: NC1HWC0. Has the same C dimension as the input x, with shape [1, c, 1, 1].
  • ar:
    • Required: yes
    • Type: float32
    • Description: filter weight. The operator actually uses the value in the weight file. For single-operator testing purposes, it is used as an input.
    • Restriction: NC1HWC0. Has the same C dimension as the input x, with shape [1, c, 1, 1].
  • tl:
    • Required: yes
    • Type: float32
    • Description: filter weight. The operator actually uses the value in the weight file. For single-operator testing purposes, it is used as an input.
    • Restriction: NC1HWC0. Has the same C dimension as the input x, with shape [1, c, 1, 1].
  • al:
    • Required: yes
    • Type: float32
    • Description: filter weight. The operator actually uses the value in the weight file. For single-operator testing purposes, it is used as an input.
    • Restriction: NC1HWC0. Has the same C dimension as the input x, with shape [1, c, 1, 1].

Attribute

channel_shared:
  • Required: yes
  • Type: bool
  • Description: whether to share the same weight across channels
  • Restriction: Must be false.

Output

y:
  • Required: yes
  • Type: float32
  • Description: output tensor
  • Restriction: NC1HWC0