// META: title=test WebNN API sigmoid operation
// META: global=window,dedicatedworker
// META: variant=?cpu
// META: variant=?gpu
// META: variant=?npu
// META: script=../resources/utils.js
// META: timeout=long
'use strict';
// https://www.w3.org/TR/webnn/#api-mlgraphbuilder-sigmoid-method
// Compute the sigmoid function of the input tensor. The calculation follows the
// expression 1 / (exp(-x) + 1).
//
// MLOperand sigmoid(MLOperand input);
const getSigmoidPrecisionTolerance = (graphResources) => {
// float32 (leaving a few ULP for roundoff)
const toleranceValueDict = {float32: 32 + 2, float16: 3};
const expectedDataType =
getExpectedDataTypeOfSingleOutput(graphResources.expectedOutputs);
return {metricType: 'ULP', value: toleranceValueDict[expectedDataType]};
};
const sigmoidTests = [
{
'name': 'sigmoid float32 1D constant tensor',
'graph': {
'inputs': {
'sigmoidInput': {
'data': [
-0.37699514627456665, -0.6848450899124146, -5.988872051239014,
4.431885719299316, -0.93868488073349, 4.591195583343506,
-2.5067026615142822, 1.5669522285461426, -2.596473217010498,
-3.64729380607605, 2.6785237789154053, -3.1051602363586426,
2.2585017681121826, -0.2865157723426819, 4.64043664932251,
1.0606156587600708, -3.536252498626709, 0.4410440921783447,
4.791460037231445, 2.0745489597320557, 0.8354471325874329,
-5.433595657348633, -4.184835910797119, -2.484982490539551
],
'descriptor': {'dimensions': [24], 'dataType': 'float32'},
'constant': true
}
},
'operators': [{
'name': 'sigmoid',
'arguments': [{'input': 'sigmoidInput'}],
'outputs': 'sigmoidOutput'
}],
'expectedOutputs': {
'sigmoidOutput': {
'data': [
0.4068518280982971, 0.33518078923225403, 0.0025002227630466223,
0.9882476925849915, 0.28116607666015625, 0.9899610877037048,
0.07538963109254837, 0.8273487091064453, 0.0693657398223877,
0.02539960853755474, 0.9357474446296692, 0.04289489984512329,
0.9053813815116882, 0.42885708808898926, 0.9904388189315796,
0.7428081631660461, 0.0282981526106596, 0.6085078120231628,
0.9917680025100708, 0.8884047269821167, 0.6975054740905762,
0.004348373040556908, 0.014996387995779514, 0.07691769301891327
],
'descriptor': {'dimensions': [24], 'dataType': 'float32'}
}
}
}
},
{
'name': 'sigmoid float32 0D tensor',
'graph': {
'inputs': {
'sigmoidInput': {
'data': [-0.37699514627456665],
'descriptor': {'dimensions': [], 'dataType': 'float32'}
}
},
'operators': [{
'name': 'sigmoid',
'arguments': [{'input': 'sigmoidInput'}],
'outputs': 'sigmoidOutput'
}],
'expectedOutputs': {
'sigmoidOutput': {
'data': [0.4068518280982971],
'descriptor': {'dimensions': [], 'dataType': 'float32'}
}
}
}
},
{
'name': 'sigmoid float32 1D tensor',
'graph': {
'inputs': {
'sigmoidInput': {
'data': [
-0.37699514627456665, -0.6848450899124146, -5.988872051239014,
4.431885719299316, -0.93868488073349, 4.591195583343506,
-2.5067026615142822, 1.5669522285461426, -2.596473217010498,
-3.64729380607605, 2.6785237789154053, -3.1051602363586426,
2.2585017681121826, -0.2865157723426819, 4.64043664932251,
1.0606156587600708, -3.536252498626709, 0.4410440921783447,
4.791460037231445, 2.0745489597320557, 0.8354471325874329,
-5.433595657348633, -4.184835910797119, -2.484982490539551
],
'descriptor': {'dimensions': [24], 'dataType': 'float32'}
}
},
'operators': [{
'name': 'sigmoid',
'arguments': [{'input': 'sigmoidInput'}],
'outputs': 'sigmoidOutput'
}],
'expectedOutputs': {
'sigmoidOutput': {
'data': [
0.4068518280982971, 0.33518078923225403, 0.0025002227630466223,
0.9882476925849915, 0.28116607666015625, 0.9899610877037048,
0.07538963109254837, 0.8273487091064453, 0.0693657398223877,
0.02539960853755474, 0.9357474446296692, 0.04289489984512329,
0.9053813815116882, 0.42885708808898926, 0.9904388189315796,
0.7428081631660461, 0.0282981526106596, 0.6085078120231628,
0.9917680025100708, 0.8884047269821167, 0.6975054740905762,
0.004348373040556908, 0.014996387995779514, 0.07691769301891327
],
'descriptor': {'dimensions': [24], 'dataType': 'float32'}
}
}
}
},
{
'name': 'sigmoid float32 2D tensor',
'graph': {
'inputs': {
'sigmoidInput': {
'data': [
-0.37699514627456665, -0.6848450899124146, -5.988872051239014,
4.431885719299316, -0.93868488073349, 4.591195583343506,
-2.5067026615142822, 1.5669522285461426, -2.596473217010498,
-3.64729380607605, 2.6785237789154053, -3.1051602363586426,
2.2585017681121826, -0.2865157723426819, 4.64043664932251,
1.0606156587600708, -3.536252498626709, 0.4410440921783447,
4.791460037231445, 2.0745489597320557, 0.8354471325874329,
-5.433595657348633, -4.184835910797119, -2.484982490539551
],
'descriptor': {'dimensions': [4, 6], 'dataType': 'float32'}
}
},
'operators': [{
'name': 'sigmoid',
'arguments': [{'input': 'sigmoidInput'}],
'outputs': 'sigmoidOutput'
}],
'expectedOutputs': {
'sigmoidOutput': {
'data': [
0.4068518280982971, 0.33518078923225403, 0.0025002227630466223,
0.9882476925849915, 0.28116607666015625, 0.9899610877037048,
0.07538963109254837, 0.8273487091064453, 0.0693657398223877,
0.02539960853755474, 0.9357474446296692, 0.04289489984512329,
0.9053813815116882, 0.42885708808898926, 0.9904388189315796,
0.7428081631660461, 0.0282981526106596, 0.6085078120231628,
0.9917680025100708, 0.8884047269821167, 0.6975054740905762,
0.004348373040556908, 0.014996387995779514, 0.07691769301891327
],
'descriptor': {'dimensions': [4, 6], 'dataType': 'float32'}
}
}
}
},
{
'name': 'sigmoid float32 3D tensor',
'graph': {
'inputs': {
'sigmoidInput': {
'data': [
-0.37699514627456665, -0.6848450899124146, -5.988872051239014,
4.431885719299316, -0.93868488073349, 4.591195583343506,
-2.5067026615142822, 1.5669522285461426, -2.596473217010498,
-3.64729380607605, 2.6785237789154053, -3.1051602363586426,
2.2585017681121826, -0.2865157723426819, 4.64043664932251,
1.0606156587600708, -3.536252498626709, 0.4410440921783447,
4.791460037231445, 2.0745489597320557, 0.8354471325874329,
-5.433595657348633, -4.184835910797119, -2.484982490539551
],
'descriptor': {'dimensions': [2, 3, 4], 'dataType': 'float32'}
}
},
'operators': [{
'name': 'sigmoid',
'arguments': [{'input': 'sigmoidInput'}],
'outputs': 'sigmoidOutput'
}],
'expectedOutputs': {
'sigmoidOutput': {
'data': [
0.4068518280982971, 0.33518078923225403, 0.0025002227630466223,
0.9882476925849915, 0.28116607666015625, 0.9899610877037048,
0.07538963109254837, 0.8273487091064453, 0.0693657398223877,
0.02539960853755474, 0.9357474446296692, 0.04289489984512329,
0.9053813815116882, 0.42885708808898926, 0.9904388189315796,
0.7428081631660461, 0.0282981526106596, 0.6085078120231628,
0.9917680025100708, 0.8884047269821167, 0.6975054740905762,
0.004348373040556908, 0.014996387995779514, 0.07691769301891327
],
'descriptor': {'dimensions': [2, 3, 4], 'dataType': 'float32'}
}
}
}
},
{
'name': 'sigmoid float32 4D tensor',
'graph': {
'inputs': {
'sigmoidInput': {
'data': [
-0.37699514627456665, -0.6848450899124146, -5.988872051239014,
4.431885719299316, -0.93868488073349, 4.591195583343506,
-2.5067026615142822, 1.5669522285461426, -2.596473217010498,
-3.64729380607605, 2.6785237789154053, -3.1051602363586426,
2.2585017681121826, -0.2865157723426819, 4.64043664932251,
1.0606156587600708, -3.536252498626709, 0.4410440921783447,
4.791460037231445, 2.0745489597320557, 0.8354471325874329,
-5.433595657348633, -4.184835910797119, -2.484982490539551
],
'descriptor': {'dimensions': [2, 2, 2, 3], 'dataType': 'float32'}
}
},
'operators': [{
'name': 'sigmoid',
'arguments': [{'input': 'sigmoidInput'}],
'outputs': 'sigmoidOutput'
}],
'expectedOutputs': {
'sigmoidOutput': {
'data': [
0.4068518280982971, 0.33518078923225403, 0.0025002227630466223,
0.9882476925849915, 0.28116607666015625, 0.9899610877037048,
0.07538963109254837, 0.8273487091064453, 0.0693657398223877,
0.02539960853755474, 0.9357474446296692, 0.04289489984512329,
0.9053813815116882, 0.42885708808898926, 0.9904388189315796,
0.7428081631660461, 0.0282981526106596, 0.6085078120231628,
0.9917680025100708, 0.8884047269821167, 0.6975054740905762,
0.004348373040556908, 0.014996387995779514, 0.07691769301891327
],
'descriptor': {'dimensions': [2, 2, 2, 3], 'dataType': 'float32'}
}
}
}
},
{
'name': 'sigmoid float32 5D tensor',
'graph': {
'inputs': {
'sigmoidInput': {
'data': [
-0.37699514627456665, -0.6848450899124146, -5.988872051239014,
4.431885719299316, -0.93868488073349, 4.591195583343506,
-2.5067026615142822, 1.5669522285461426, -2.596473217010498,
-3.64729380607605, 2.6785237789154053, -3.1051602363586426,
2.2585017681121826, -0.2865157723426819, 4.64043664932251,
1.0606156587600708, -3.536252498626709, 0.4410440921783447,
4.791460037231445, 2.0745489597320557, 0.8354471325874329,
-5.433595657348633, -4.184835910797119, -2.484982490539551
],
'descriptor': {'dimensions': [2, 1, 4, 1, 3], 'dataType': 'float32'}
}
},
'operators': [{
'name': 'sigmoid',
'arguments': [{'input': 'sigmoidInput'}],
'outputs': 'sigmoidOutput'
}],
'expectedOutputs': {
'sigmoidOutput': {
'data': [
0.4068518280982971, 0.33518078923225403, 0.0025002227630466223,
0.9882476925849915, 0.28116607666015625, 0.9899610877037048,
0.07538963109254837, 0.8273487091064453, 0.0693657398223877,
0.02539960853755474, 0.9357474446296692, 0.04289489984512329,
0.9053813815116882, 0.42885708808898926, 0.9904388189315796,
0.7428081631660461, 0.0282981526106596, 0.6085078120231628,
0.9917680025100708, 0.8884047269821167, 0.6975054740905762,
0.004348373040556908, 0.014996387995779514, 0.07691769301891327
],
'descriptor': {'dimensions': [2, 1, 4, 1, 3], 'dataType': 'float32'}
}
}
}
}
];
if (navigator.ml) {
sigmoidTests.forEach((test) => {
webnn_conformance_test(
buildGraphAndCompute, getSigmoidPrecisionTolerance, test);
});
} else {
test(() => assert_implements(navigator.ml, 'missing navigator.ml'));
}
Codebase Browser
chromium