PRelu

• Domain: ai.onnx

• Since version: 16

PRelu takes input data (Tensor ) and slope tensor as input, and produces one output data (Tensor ) where the function f(x) = slope * x for x  = 0., is applied to the data tensor elementwise.

Inputs

• X (T): Input tensor

• slope (T): Slope tensor. The shape of slope can be smaller than first input X; if so, its shape must be unidirectional broadcastable to X

Outputs

• Y (T): Output tensor (same size as X)

Type Constraints

• T: Constrain input and output types to float/int tensors. Allowed types: tensor(bfloat16), tensor(double), tensor(float), tensor(float16), tensor(int32), tensor(int64), tensor(uint32), tensor(uint64).

Examples

test_cc_prelu

Node:
  PRelu(x, slope) -> (y)

Inputs:
  x: shape=(2, 3), dtype=float32
    [[-3., -1.,  0.],
     [ 1.,  2.,  3.]]
  slope: shape=(2, 3), dtype=float32
    [[0.25, 0.5 , 0.75],
     [0.1 , 0.2 , 0.3 ]]

Outputs:
  y: shape=(2, 3), dtype=float32
    [[-0.75, -0.5 ,  0.  ],
     [ 1.  ,  2.  ,  3.  ]]

test_cc_prelu_bcast

Node:
  PRelu(x, slope) -> (y)

Inputs:
  x: shape=(2, 3), dtype=float32
    [[-1., -2., -3.],
     [ 1.,  2.,  3.]]
  slope: shape=(3,), dtype=float32
    [0.1, 0.2, 0.3]

Outputs:
  y: shape=(2, 3), dtype=float32
    [[-0.1       , -0.4       , -0.90000004],
     [ 1.        ,  2.        ,  3.        ]]

test_cc_prelu_empty_shape_scalars

Node:
  PRelu(x, slope) -> (y)

Inputs:
  x: shape=(), dtype=float32
    2.5
  slope: shape=(), dtype=float32
    3.5

Outputs:
  y: shape=(), dtype=float32
    2.5

test_cc_prelu_empty_shape_zero_dim

Node:
  PRelu(x, slope) -> (y)

Inputs:
  x: shape=(0,), dtype=float32
    []
  slope: shape=(0,), dtype=float32
    []

Outputs:
  y: shape=(0,), dtype=float32
    []

test_cc_prelu_empty_shape_zero_dim_2d

Node:
  PRelu(x, slope) -> (y)

Inputs:
  x: shape=(0, 3), dtype=float32
    []
  slope: shape=(0, 3), dtype=float32
    []

Outputs:
  y: shape=(0, 3), dtype=float32
    []

test_cc_prelu_inf

Node:
  PRelu(x, slope) -> (y)

Inputs:
  x: shape=(4,), dtype=float32
    [     inf,     -inf,  5.0e+30, -2.5e+00]
  slope: shape=(4,), dtype=float32
    [0.25, 0.5 , 0.25, 0.25]

Outputs:
  y: shape=(4,), dtype=float32
    [      inf,      -inf,  5.00e+30, -6.25e-01]

test_prelu_broadcast

Node:
  PRelu(x, slope) -> (y)

Inputs:
  x: shape=(3, 4, 5), dtype=float32
    [[[-0.62216586,  0.23071946, -0.44037476,  1.1018444 ,  0.33993453],
      [ 0.6673572 ,  0.23079044, -0.33776435, -0.7145624 ,  0.23379254],
      [-0.87260103, -0.03636372,  0.3375118 , -0.56752586,  1.3069218 ],
      [ 0.23701055, -1.4519347 ,  0.5864804 , -1.6773233 ,  0.667711  ]],

     [[ 0.59248346,  0.25854757,  0.9993237 , -0.2519604 ,  0.4396883 ],
      [-0.12366938,  1.5436682 , -1.5088953 , -0.02319983, -1.4253234 ],
      [-1.2474204 , -1.5215695 ,  1.252772  , -1.7240764 ,  0.42119765],
      [-0.7997555 , -0.5946146 ,  0.34001034,  0.25650212,  0.39954498]],

     [[ 0.30716214, -0.5623004 ,  1.3781636 ,  0.5957701 , -0.3376929 ],
      [ 0.11759786, -1.637436  ,  0.37568974,  2.3201282 , -0.38706338],
      [ 1.8695779 ,  1.0533305 , -0.699427  , -1.2625815 , -0.61388   ],
      [ 0.7215071 ,  0.21707447,  0.07948666,  0.6312519 , -0.6393902 ]]]
  slope: shape=(5,), dtype=float32
    [ 0.9154805, -0.5188539,  1.3467076, -1.0085899,  1.2497014]

Outputs:
  y: shape=(3, 4, 5), dtype=float32
    [[[-0.56958073,  0.23071946, -0.593056  ,  1.1018444 ,  0.33993453],
      [ 0.6673572 ,  0.23079044, -0.4548698 ,  0.7207004 ,  0.23379254],
      [-0.7988492 ,  0.01886746,  0.3375118 ,  0.5724008 ,  1.3069218 ],
      [ 0.23701055,  0.753342  ,  0.5864804 ,  1.6917313 ,  0.667711  ]],

     [[ 0.59248346,  0.25854757,  0.9993237 ,  0.2541247 ,  0.4396883 ],
      [-0.11321691,  1.5436682 , -2.0320406 ,  0.02339911, -1.7812285 ],
      [-1.1419891 ,  0.7894723 ,  1.252772  ,  1.738886  ,  0.42119765],
      [-0.73216057,  0.3085181 ,  0.34001034,  0.25650212,  0.39954498]],

     [[ 0.30716214,  0.29175174,  1.3781636 ,  0.5957701 , -0.42201528],
      [ 0.11759786,  0.84959006,  0.37568974,  2.3201282 , -0.48371366],
      [ 1.8695779 ,  1.0533305 , -0.9419237 ,  1.2734269 , -0.7671667 ],
      [ 0.7215071 ,  0.21707447,  0.07948666,  0.6312519 , -0.79904675]]]

test_prelu_example

Node:
  PRelu(x, slope) -> (y)

Inputs:
  x: shape=(3, 4, 5), dtype=float32
    [[[-0.7513602 ,  0.51713336,  0.7695295 , -0.22347532,  0.18149193],
      [ 0.24913743,  1.994417  , -0.07321609,  0.1076815 ,  2.7155633 ],
      [-0.47996232,  1.631689  ,  0.49418923, -2.1702976 ,  2.58957   ],
      [ 0.14372788,  0.06502817,  1.441035  , -0.6060344 ,  1.2044882 ]],

     [[ 0.73592454, -1.4978608 , -1.3068415 ,  2.285458  ,  0.94369376],
      [ 0.24891663, -1.1944486 , -0.20609704, -2.257049  , -1.3082974 ],
      [ 1.7855092 , -0.24503152,  0.6468487 ,  0.05505385,  0.0182795 ],
      [-0.18279839,  0.14583768, -1.6932459 ,  2.7747114 , -0.18945445]],

     [[-0.8139422 , -0.83692306, -1.5865561 , -1.0409492 , -0.04504394],
      [-0.3157502 ,  0.20416091,  0.6947444 , -0.93289083,  0.02533321],
      [-0.7196668 , -2.7953355 ,  1.6615764 , -1.6629553 , -1.2022487 ],
      [ 1.0108496 , -0.98562527,  0.70152247,  0.48280644, -0.1113701 ]]]
  slope: shape=(3, 4, 5), dtype=float32
    [[[-1.0029005 ,  1.0619555 , -0.30622792, -0.5332439 ,  1.4375478 ],
      [ 1.2071373 , -0.6458864 ,  0.8300795 ,  0.21273202, -0.7806308 ],
      [-0.40428525,  0.6376805 ,  1.1788383 , -0.4927185 ,  2.5766451 ],
      [-1.7393267 , -1.0279831 , -0.6508921 , -2.7996702 ,  0.2743459 ]],

     [[ 0.78074443,  1.0220166 , -0.2684288 ,  0.6771564 , -0.21049254],
      [ 0.6646813 , -1.1061122 ,  1.5222495 , -2.4748578 ,  0.76601636],
      [ 0.05163119,  1.052501  , -0.36132503, -0.23450962,  0.11239426],
      [ 2.8382437 ,  1.5054117 , -0.7077736 , -0.50181675,  2.6686907 ]],

     [[ 0.13030747, -0.88599855, -0.22711682,  0.37078258, -0.25641453],
      [-1.1376325 ,  0.06863393,  0.33454004,  0.8406287 , -1.4702488 ],
      [ 0.20493318, -1.3650485 ,  0.04658019,  0.7189073 , -1.0817056 ],
      [-1.1260862 , -2.0177746 , -0.2981477 ,  0.33546326,  0.69503844]]]

Outputs:
  y: shape=(3, 4, 5), dtype=float32
    [[[ 0.7535395 ,  0.51713336,  0.7695295 ,  0.11916685,  0.18149193],
      [ 0.24913743,  1.994417  , -0.06077517,  0.1076815 ,  2.7155633 ],
      [ 0.19404168,  1.631689  ,  0.49418923,  1.0693457 ,  2.58957   ],
      [ 0.14372788,  0.06502817,  1.441035  ,  1.6966964 ,  1.2044882 ]],

     [[ 0.73592454, -1.5308386 ,  0.3507939 ,  2.285458  ,  0.94369376],
      [ 0.24891663,  1.3211942 , -0.3137311 ,  5.5858755 , -1.0021772 ],
      [ 1.7855092 , -0.25789592,  0.6468487 ,  0.05505385,  0.0182795 ],
      [-0.51882637,  0.14583768,  1.1984348 ,  2.7747114 , -0.5055953 ]],

     [[-0.10606275,  0.7415126 ,  0.36033356, -0.38596585,  0.01154992],
      [ 0.3592077 ,  0.20416091,  0.6947444 , -0.7842148 ,  0.02533321],
      [-0.1474836 ,  3.8157687 ,  1.6615764 , -1.1955107 ,  1.300479  ],
      [ 1.0108496 ,  1.9887697 ,  0.70152247,  0.48280644, -0.0774065 ]]]

Differences with previous version (9)

SchemaDiff: PRelu (domain 'ai.onnx')

• old version: 9

• new version: 16

• breaking: no

Type constraints:

• changed ‘T’: added types: [‘tensor(bfloat16)’]

Version History

• Version 9

• Version 7

• Version 6

• Version 1