Source code
Revision control
Copy as Markdown
Other Tools
Test Info: Warnings
- This test has a WPT meta file that expects 1 subtest issues.
- This WPT test may be referenced by the following Test IDs:
- /webnn/conformance_tests/split.https.any.html?cpu - WPT Dashboard Interop Dashboard
- /webnn/conformance_tests/split.https.any.html?gpu - WPT Dashboard Interop Dashboard
- /webnn/conformance_tests/split.https.any.html?npu - WPT Dashboard Interop Dashboard
- /webnn/conformance_tests/split.https.any.worker.html?cpu - WPT Dashboard Interop Dashboard
- /webnn/conformance_tests/split.https.any.worker.html?gpu - WPT Dashboard Interop Dashboard
- /webnn/conformance_tests/split.https.any.worker.html?npu - WPT Dashboard Interop Dashboard
// META: title=test WebNN API split operation
// META: global=window,dedicatedworker
// META: variant=?cpu
// META: variant=?gpu
// META: variant=?npu
// META: script=../resources/utils.js
// META: timeout=long
'use strict';
// Split the input tensor into a number of sub tensors along the given axis.
//
// dictionary MLSplitOptions {
// [EnforceRange] unsigned long axis = 0;
// };
//
// sequence<MLOperand> split(
// MLOperand input,
// ([EnforceRange] unsigned long or sequence<[EnforceRange] unsigned long>)
// splits, optional MLSplitOptions options = {});
const getSplitPrecisionTolerance = (graphResources) => {
const toleranceValueDict = {float32: 0, float16: 0};
const dataType =
graphResources
.expectedOutputs[Object.keys(graphResources.expectedOutputs)[0]]
.descriptor.dataType;
return {metricType: 'ULP', value: toleranceValueDict[dataType]};
};
const splitTests = [
{
'name': 'split float32 1D constant tensor number splits default options',
'graph': {
'inputs': {
'splitInput': {
'data': [
-64.52056884765625, -84.60513305664062, -67.99282836914062,
-23.446075439453125, -85.64382934570312, 46.87752151489258,
-68.11224365234375, 75.99607849121094, -61.05668640136719,
-90.92643737792969, 53.916622161865234, 84.16268920898438,
-95.57494354248047, -52.40757751464844, -29.007186889648438,
71.65496063232422, 50.66357421875, 21.364582061767578,
-27.127241134643555, 65.1489486694336, -30.40681266784668,
-6.818390369415283, 46.673622131347656, -21.12453842163086
],
'descriptor': {shape: [24], dataType: 'float32'},
'constant': true
}
},
'operators': [{
'name': 'split',
'arguments': [{'input': 'splitInput'}, {'splits': 3}],
'outputs': ['splitOutput1', 'splitOutput2', 'splitOutput3']
}],
'expectedOutputs': {
'splitOutput1': {
'data': [
-64.52056884765625, -84.60513305664062, -67.99282836914062,
-23.446075439453125, -85.64382934570312, 46.87752151489258,
-68.11224365234375, 75.99607849121094
],
'descriptor': {shape: [8], dataType: 'float32'}
},
'splitOutput2': {
'data': [
-61.05668640136719, -90.92643737792969, 53.916622161865234,
84.16268920898438, -95.57494354248047, -52.40757751464844,
-29.007186889648438, 71.65496063232422
],
'descriptor': {shape: [8], dataType: 'float32'}
},
'splitOutput3': {
'data': [
50.66357421875, 21.364582061767578, -27.127241134643555,
65.1489486694336, -30.40681266784668, -6.818390369415283,
46.673622131347656, -21.12453842163086
],
'descriptor': {shape: [8], dataType: 'float32'}
}
}
}
},
{
'name': 'split float32 1D tensor number splits default options',
'graph': {
'inputs': {
'splitInput': {
'data': [
-64.52056884765625, -84.60513305664062, -67.99282836914062,
-23.446075439453125, -85.64382934570312, 46.87752151489258,
-68.11224365234375, 75.99607849121094, -61.05668640136719,
-90.92643737792969, 53.916622161865234, 84.16268920898438,
-95.57494354248047, -52.40757751464844, -29.007186889648438,
71.65496063232422, 50.66357421875, 21.364582061767578,
-27.127241134643555, 65.1489486694336, -30.40681266784668,
-6.818390369415283, 46.673622131347656, -21.12453842163086
],
'descriptor': {shape: [24], dataType: 'float32'}
}
},
'operators': [{
'name': 'split',
'arguments': [{'input': 'splitInput'}, {'splits': 3}],
'outputs': ['splitOutput1', 'splitOutput2', 'splitOutput3']
}],
'expectedOutputs': {
'splitOutput1': {
'data': [
-64.52056884765625, -84.60513305664062, -67.99282836914062,
-23.446075439453125, -85.64382934570312, 46.87752151489258,
-68.11224365234375, 75.99607849121094
],
'descriptor': {shape: [8], dataType: 'float32'}
},
'splitOutput2': {
'data': [
-61.05668640136719, -90.92643737792969, 53.916622161865234,
84.16268920898438, -95.57494354248047, -52.40757751464844,
-29.007186889648438, 71.65496063232422
],
'descriptor': {shape: [8], dataType: 'float32'}
},
'splitOutput3': {
'data': [
50.66357421875, 21.364582061767578, -27.127241134643555,
65.1489486694336, -30.40681266784668, -6.818390369415283,
46.673622131347656, -21.12453842163086
],
'descriptor': {shape: [8], dataType: 'float32'}
}
}
}
},
{
'name': 'split float32 2D tensor number splits default options',
'graph': {
'inputs': {
'splitInput': {
'data': [
-64.52056884765625, -84.60513305664062, -67.99282836914062,
-23.446075439453125, -85.64382934570312, 46.87752151489258,
-68.11224365234375, 75.99607849121094, -61.05668640136719,
-90.92643737792969, 53.916622161865234, 84.16268920898438,
-95.57494354248047, -52.40757751464844, -29.007186889648438,
71.65496063232422, 50.66357421875, 21.364582061767578,
-27.127241134643555, 65.1489486694336, -30.40681266784668,
-6.818390369415283, 46.673622131347656, -21.12453842163086
],
'descriptor': {shape: [8, 3], dataType: 'float32'}
}
},
'operators': [{
'name': 'split',
'arguments': [{'input': 'splitInput'}, {'splits': 2}],
'outputs': ['splitOutput1', 'splitOutput2']
}],
'expectedOutputs': {
'splitOutput1': {
'data': [
-64.52056884765625, -84.60513305664062, -67.99282836914062,
-23.446075439453125, -85.64382934570312, 46.87752151489258,
-68.11224365234375, 75.99607849121094, -61.05668640136719,
-90.92643737792969, 53.916622161865234, 84.16268920898438
],
'descriptor': {shape: [4, 3], dataType: 'float32'}
},
'splitOutput2': {
'data': [
-95.57494354248047, -52.40757751464844, -29.007186889648438,
71.65496063232422, 50.66357421875, 21.364582061767578,
-27.127241134643555, 65.1489486694336, -30.40681266784668,
-6.818390369415283, 46.673622131347656, -21.12453842163086
],
'descriptor': {shape: [4, 3], dataType: 'float32'}
}
}
}
},
{
'name': 'split float32 3D tensor number splits default options',
'graph': {
'inputs': {
'splitInput': {
'data': [
-64.52056884765625, -84.60513305664062, -67.99282836914062,
-23.446075439453125, -85.64382934570312, 46.87752151489258,
-68.11224365234375, 75.99607849121094, -61.05668640136719,
-90.92643737792969, 53.916622161865234, 84.16268920898438,
-95.57494354248047, -52.40757751464844, -29.007186889648438,
71.65496063232422, 50.66357421875, 21.364582061767578,
-27.127241134643555, 65.1489486694336, -30.40681266784668,
-6.818390369415283, 46.673622131347656, -21.12453842163086
],
'descriptor': {shape: [4, 3, 2], dataType: 'float32'}
}
},
'operators': [{
'name': 'split',
'arguments': [{'input': 'splitInput'}, {'splits': 2}],
'outputs': ['splitOutput1', 'splitOutput2']
}],
'expectedOutputs': {
'splitOutput1': {
'data': [
-64.52056884765625, -84.60513305664062, -67.99282836914062,
-23.446075439453125, -85.64382934570312, 46.87752151489258,
-68.11224365234375, 75.99607849121094, -61.05668640136719,
-90.92643737792969, 53.916622161865234, 84.16268920898438
],
'descriptor': {shape: [2, 3, 2], dataType: 'float32'}
},
'splitOutput2': {
'data': [
-95.57494354248047, -52.40757751464844, -29.007186889648438,
71.65496063232422, 50.66357421875, 21.364582061767578,
-27.127241134643555, 65.1489486694336, -30.40681266784668,
-6.818390369415283, 46.673622131347656, -21.12453842163086
],
'descriptor': {shape: [2, 3, 2], dataType: 'float32'}
}
}
}
},
{
'name': 'split float32 4D tensor number splits default options',
'graph': {
'inputs': {
'splitInput': {
'data': [
-64.52056884765625, -84.60513305664062, -67.99282836914062,
-23.446075439453125, -85.64382934570312, 46.87752151489258,
-68.11224365234375, 75.99607849121094, -61.05668640136719,
-90.92643737792969, 53.916622161865234, 84.16268920898438,
-95.57494354248047, -52.40757751464844, -29.007186889648438,
71.65496063232422, 50.66357421875, 21.364582061767578,
-27.127241134643555, 65.1489486694336, -30.40681266784668,
-6.818390369415283, 46.673622131347656, -21.12453842163086
],
'descriptor': {shape: [12, 1, 1, 2], dataType: 'float32'}
}
},
'operators': [{
'name': 'split',
'arguments': [{'input': 'splitInput'}, {'splits': 4}],
'outputs':
['splitOutput1', 'splitOutput2', 'splitOutput3', 'splitOutput4']
}],
'expectedOutputs': {
'splitOutput1': {
'data': [
-64.52056884765625, -84.60513305664062, -67.99282836914062,
-23.446075439453125, -85.64382934570312, 46.87752151489258
],
'descriptor': {shape: [3, 1, 1, 2], dataType: 'float32'}
},
'splitOutput2': {
'data': [
-68.11224365234375, 75.99607849121094, -61.05668640136719,
-90.92643737792969, 53.916622161865234, 84.16268920898438
],
'descriptor': {shape: [3, 1, 1, 2], dataType: 'float32'}
},
'splitOutput3': {
'data': [
-95.57494354248047, -52.40757751464844, -29.007186889648438,
71.65496063232422, 50.66357421875, 21.364582061767578
],
'descriptor': {shape: [3, 1, 1, 2], dataType: 'float32'}
},
'splitOutput4': {
'data': [
-27.127241134643555, 65.1489486694336, -30.40681266784668,
-6.818390369415283, 46.673622131347656, -21.12453842163086
],
'descriptor': {shape: [3, 1, 1, 2], dataType: 'float32'}
}
}
}
},
{
'name': 'split float32 5D tensor number splits default options',
'graph': {
'inputs': {
'splitInput': {
'data': [
-64.52056884765625, -84.60513305664062, -67.99282836914062,
-23.446075439453125, -85.64382934570312, 46.87752151489258,
-68.11224365234375, 75.99607849121094, -61.05668640136719,
-90.92643737792969, 53.916622161865234, 84.16268920898438,
-95.57494354248047, -52.40757751464844, -29.007186889648438,
71.65496063232422, 50.66357421875, 21.364582061767578,
-27.127241134643555, 65.1489486694336, -30.40681266784668,
-6.818390369415283, 46.673622131347656, -21.12453842163086
],
'descriptor': {shape: [6, 1, 1, 2, 2], dataType: 'float32'}
}
},
'operators': [{
'name': 'split',
'arguments': [{'input': 'splitInput'}, {'splits': 2}],
'outputs': ['splitOutput1', 'splitOutput2']
}],
'expectedOutputs': {
'splitOutput1': {
'data': [
-64.52056884765625, -84.60513305664062, -67.99282836914062,
-23.446075439453125, -85.64382934570312, 46.87752151489258,
-68.11224365234375, 75.99607849121094, -61.05668640136719,
-90.92643737792969, 53.916622161865234, 84.16268920898438
],
'descriptor': {shape: [3, 1, 1, 2, 2], dataType: 'float32'}
},
'splitOutput2': {
'data': [
-95.57494354248047, -52.40757751464844, -29.007186889648438,
71.65496063232422, 50.66357421875, 21.364582061767578,
-27.127241134643555, 65.1489486694336, -30.40681266784668,
-6.818390369415283, 46.673622131347656, -21.12453842163086
],
'descriptor': {shape: [3, 1, 1, 2, 2], dataType: 'float32'}
}
}
}
},
{
'name': 'split float32 4D tensor array splits default options',
'graph': {
'inputs': {
'splitInput': {
'data': [
-64.52056884765625, -84.60513305664062, -67.99282836914062,
-23.446075439453125, -85.64382934570312, 46.87752151489258,
-68.11224365234375, 75.99607849121094, -61.05668640136719,
-90.92643737792969, 53.916622161865234, 84.16268920898438,
-95.57494354248047, -52.40757751464844, -29.007186889648438,
71.65496063232422, 50.66357421875, 21.364582061767578,
-27.127241134643555, 65.1489486694336, -30.40681266784668,
-6.818390369415283, 46.673622131347656, -21.12453842163086
],
'descriptor': {shape: [12, 1, 1, 2], dataType: 'float32'}
}
},
'operators': [{
'name': 'split',
'arguments': [{'input': 'splitInput'}, {'splits': [3, 3, 3, 3]}],
'outputs':
['splitOutput1', 'splitOutput2', 'splitOutput3', 'splitOutput4'],
}],
'expectedOutputs': {
'splitOutput1': {
'data': [
-64.52056884765625, -84.60513305664062, -67.99282836914062,
-23.446075439453125, -85.64382934570312, 46.87752151489258
],
'descriptor': {shape: [3, 1, 1, 2], dataType: 'float32'}
},
'splitOutput2': {
'data': [
-68.11224365234375, 75.99607849121094, -61.05668640136719,
-90.92643737792969, 53.916622161865234, 84.16268920898438
],
'descriptor': {shape: [3, 1, 1, 2], dataType: 'float32'}
},
'splitOutput3': {
'data': [
-95.57494354248047, -52.40757751464844, -29.007186889648438,
71.65496063232422, 50.66357421875, 21.364582061767578
],
'descriptor': {shape: [3, 1, 1, 2], dataType: 'float32'}
},
'splitOutput4': {
'data': [
-27.127241134643555, 65.1489486694336, -30.40681266784668,
-6.818390369415283, 46.673622131347656, -21.12453842163086
],
'descriptor': {shape: [3, 1, 1, 2], dataType: 'float32'}
}
}
}
},
{
'name': 'split float32 4D tensor number splits options.axis',
'graph': {
'inputs': {
'splitInput': {
'data': [
-64.52056884765625, -84.60513305664062, -67.99282836914062,
-23.446075439453125, -85.64382934570312, 46.87752151489258,
-68.11224365234375, 75.99607849121094, -61.05668640136719,
-90.92643737792969, 53.916622161865234, 84.16268920898438,
-95.57494354248047, -52.40757751464844, -29.007186889648438,
71.65496063232422, 50.66357421875, 21.364582061767578,
-27.127241134643555, 65.1489486694336, -30.40681266784668,
-6.818390369415283, 46.673622131347656, -21.12453842163086
],
'descriptor': {shape: [12, 1, 1, 2], dataType: 'float32'}
}
},
'operators': [{
'name': 'split',
'arguments':
[{'input': 'splitInput'}, {'splits': 3}, {'options': {'axis': 0}}],
'outputs': ['splitOutput1', 'splitOutput2', 'splitOutput3'],
}],
'expectedOutputs': {
'splitOutput1': {
'data': [
-64.52056884765625, -84.60513305664062, -67.99282836914062,
-23.446075439453125, -85.64382934570312, 46.87752151489258,
-68.11224365234375, 75.99607849121094
],
'descriptor': {shape: [4, 1, 1, 2], dataType: 'float32'}
},
'splitOutput2': {
'data': [
-61.05668640136719, -90.92643737792969, 53.916622161865234,
84.16268920898438, -95.57494354248047, -52.40757751464844,
-29.007186889648438, 71.65496063232422
],
'descriptor': {shape: [4, 1, 1, 2], dataType: 'float32'}
},
'splitOutput3': {
'data': [
50.66357421875, 21.364582061767578, -27.127241134643555,
65.1489486694336, -30.40681266784668, -6.818390369415283,
46.673622131347656, -21.12453842163086
],
'descriptor': {shape: [4, 1, 1, 2], dataType: 'float32'}
}
}
}
},
{
'name': 'split float32 5D tensor array splits=[3, 3] options.axis=2',
'graph': {
'inputs': {
'splitInput': {
'data': [
-64.52056884765625, -84.60513305664062, -67.99282836914062,
-23.446075439453125, -85.64382934570312, 46.87752151489258,
-68.11224365234375, 75.99607849121094, -61.05668640136719,
-90.92643737792969, 53.916622161865234, 84.16268920898438,
-95.57494354248047, -52.40757751464844, -29.007186889648438,
71.65496063232422, 50.66357421875, 21.364582061767578,
-27.127241134643555, 65.1489486694336, -30.40681266784668,
-6.818390369415283, 46.673622131347656, -21.12453842163086
],
'descriptor': {shape: [1, 1, 6, 2, 2], dataType: 'float32'}
}
},
'operators': [{
'name': 'split',
'arguments': [
{'input': 'splitInput'}, {'splits': [3, 3]}, {'options': {'axis': 2}}
],
'outputs': ['splitOutput1', 'splitOutput2']
}],
'expectedOutputs': {
'splitOutput1': {
'data': [
-64.52056884765625, -84.60513305664062, -67.99282836914062,
-23.446075439453125, -85.64382934570312, 46.87752151489258,
-68.11224365234375, 75.99607849121094, -61.05668640136719,
-90.92643737792969, 53.916622161865234, 84.16268920898438
],
'descriptor': {shape: [1, 1, 3, 2, 2], dataType: 'float32'}
},
'splitOutput2': {
'data': [
-95.57494354248047, -52.40757751464844, -29.007186889648438,
71.65496063232422, 50.66357421875, 21.364582061767578,
-27.127241134643555, 65.1489486694336, -30.40681266784668,
-6.818390369415283, 46.673622131347656, -21.12453842163086
],
'descriptor': {shape: [1, 1, 3, 2, 2], dataType: 'float32'}
}
}
}
},
{
'name': 'split float32 5D tensor array splits=[2, 4] options.axis=0',
'graph': {
'inputs': {
'splitInput': {
'data': [
-64.52056884765625, -84.60513305664062, -67.99282836914062,
-23.446075439453125, -85.64382934570312, 46.87752151489258,
-68.11224365234375, 75.99607849121094, -61.05668640136719,
-90.92643737792969, 53.916622161865234, 84.16268920898438,
-95.57494354248047, -52.40757751464844, -29.007186889648438,
71.65496063232422, 50.66357421875, 21.364582061767578,
-27.127241134643555, 65.1489486694336, -30.40681266784668,
-6.818390369415283, 46.673622131347656, -21.12453842163086
],
'descriptor': {shape: [6, 1, 1, 2, 2], dataType: 'float32'}
}
},
'operators': [{
'name': 'split',
'arguments': [
{'input': 'splitInput'}, {'splits': [2, 4]}, {'options': {'axis': 0}}
],
'outputs': ['splitOutput1', 'splitOutput2']
}],
'expectedOutputs': {
'splitOutput1': {
'data': [
-64.52056884765625, -84.60513305664062, -67.99282836914062,
-23.446075439453125, -85.64382934570312, 46.87752151489258,
-68.11224365234375, 75.99607849121094
],
'descriptor': {shape: [2, 1, 1, 2, 2], dataType: 'float32'}
},
'splitOutput2': {
'data': [
-61.05668640136719, -90.92643737792969, 53.916622161865234,
84.16268920898438, -95.57494354248047, -52.40757751464844,
-29.007186889648438, 71.65496063232422, 50.66357421875,
21.364582061767578, -27.127241134643555, 65.1489486694336,
-30.40681266784668, -6.818390369415283, 46.673622131347656,
-21.12453842163086
],
'descriptor': {shape: [4, 1, 1, 2, 2], dataType: 'float32'}
}
}
}
}
];
if (navigator.ml) {
splitTests.forEach((test) => {
webnn_conformance_test(
buildAndExecuteGraph, getSplitPrecisionTolerance, test);
});
} else {
test(() => assert_implements(navigator.ml, 'missing navigator.ml'));
}