[webnn] Add tests for WebNN API constant (fillSequence) (#43801)
* [webnn] Add tests for WebNN API constant (fillSequence)
* [webnn] Update test data with negative start
* [webnn] Update test data for integer test cases
* [webnn] Add a test with negative integer step
diff --git a/webnn/constant.https.any.js b/webnn/constant.https.any.js
new file mode 100644
index 0000000..afb020d
--- /dev/null
+++ b/webnn/constant.https.any.js
@@ -0,0 +1,10 @@
+// META: title=test WebNN API constant
+// META: global=window,dedicatedworker
+// META: script=./resources/utils.js
+// META: timeout=long
+
+'use strict';
+
+// https://webmachinelearning.github.io/webnn/#api-mlgraphbuilder-constant-range
+
+testWebNNOperation('constant', buildConstantRange);
\ No newline at end of file
diff --git a/webnn/gpu/constant.https.any.js b/webnn/gpu/constant.https.any.js
new file mode 100644
index 0000000..7e0933f
--- /dev/null
+++ b/webnn/gpu/constant.https.any.js
@@ -0,0 +1,10 @@
+// META: title=test WebNN API constant
+// META: global=window,dedicatedworker
+// META: script=../resources/utils.js
+// META: timeout=long
+
+'use strict';
+
+// https://webmachinelearning.github.io/webnn/#api-mlgraphbuilder-constant-range
+
+testWebNNOperation('constant', buildConstantRange, 'gpu');
\ No newline at end of file
diff --git a/webnn/resources/test_data/constant.json b/webnn/resources/test_data/constant.json
new file mode 100644
index 0000000..06fe0a7
--- /dev/null
+++ b/webnn/resources/test_data/constant.json
@@ -0,0 +1,754 @@
+{
+ "tests": [
+ {
+ "name": "constant float32 0D tensor of default float32 type",
+ "inputs": {
+ "start": {
+ "data": 0.22992068529129028,
+ "type": "float32"
+ },
+ "step": {
+ "data": 0.7537541389465332,
+ "type": "float32"
+ }
+ },
+ "outputShape": [],
+ "expected": {
+ "name": "output",
+ "shape": [],
+ "data": [
+ 0.22992068529129028
+ ],
+ "type": "float32"
+ }
+ },
+ {
+ "name": "constant float32 1D tensor of default float32 type",
+ "inputs": {
+ "start": {
+ "data": 0.22992068529129028,
+ "type": "float32"
+ },
+ "step": {
+ "data": 0.7537541389465332,
+ "type": "float32"
+ }
+ },
+ "outputShape": [24],
+ "expected": {
+ "name": "output",
+ "shape": [24],
+ "data": [
+ 0.22992068529129028,
+ 0.9836748242378235,
+ 1.737428903579712,
+ 2.491183042526245,
+ 3.2449371814727783,
+ 3.9986913204193115,
+ 4.752445697784424,
+ 5.506199836730957,
+ 6.25995397567749,
+ 7.013708114624023,
+ 7.767462253570557,
+ 8.52121639251709,
+ 9.274970054626465,
+ 10.028724670410156,
+ 10.782478332519531,
+ 11.536232948303223,
+ 12.289986610412598,
+ 13.043741226196289,
+ 13.797494888305664,
+ 14.551249504089355,
+ 15.30500316619873,
+ 16.058757781982422,
+ 16.812511444091797,
+ 17.566265106201172
+ ],
+ "type": "float32"
+ }
+ },
+ {
+ "name": "constant float32 2D tensor of default float32 type",
+ "inputs": {
+ "start": {
+ "data": 0.22992068529129028,
+ "type": "float32"
+ },
+ "step": {
+ "data": 0.7537541389465332,
+ "type": "float32"
+ }
+ },
+ "outputShape": [4, 6],
+ "expected": {
+ "name": "output",
+ "shape": [4, 6],
+ "data": [
+ 0.22992068529129028,
+ 0.9836748242378235,
+ 1.737428903579712,
+ 2.491183042526245,
+ 3.2449371814727783,
+ 3.9986913204193115,
+ 4.752445697784424,
+ 5.506199836730957,
+ 6.25995397567749,
+ 7.013708114624023,
+ 7.767462253570557,
+ 8.52121639251709,
+ 9.274970054626465,
+ 10.028724670410156,
+ 10.782478332519531,
+ 11.536232948303223,
+ 12.289986610412598,
+ 13.043741226196289,
+ 13.797494888305664,
+ 14.551249504089355,
+ 15.30500316619873,
+ 16.058757781982422,
+ 16.812511444091797,
+ 17.566265106201172
+ ],
+ "type": "float32"
+ }
+ },
+ {
+ "name": "constant float32 3D tensor of default float32 type",
+ "inputs": {
+ "start": {
+ "data": 0.22992068529129028,
+ "type": "float32"
+ },
+ "step": {
+ "data": 0.7537541389465332,
+ "type": "float32"
+ }
+ },
+ "outputShape": [2, 3, 4],
+ "expected": {
+ "name": "output",
+ "shape": [2, 3, 4],
+ "data": [
+ 0.22992068529129028,
+ 0.9836748242378235,
+ 1.737428903579712,
+ 2.491183042526245,
+ 3.2449371814727783,
+ 3.9986913204193115,
+ 4.752445697784424,
+ 5.506199836730957,
+ 6.25995397567749,
+ 7.013708114624023,
+ 7.767462253570557,
+ 8.52121639251709,
+ 9.274970054626465,
+ 10.028724670410156,
+ 10.782478332519531,
+ 11.536232948303223,
+ 12.289986610412598,
+ 13.043741226196289,
+ 13.797494888305664,
+ 14.551249504089355,
+ 15.30500316619873,
+ 16.058757781982422,
+ 16.812511444091797,
+ 17.566265106201172
+ ],
+ "type": "float32"
+ }
+ },
+ {
+ "name": "constant float32 4D tensor of default float32 type",
+ "inputs": {
+ "start": {
+ "data": 0.22992068529129028,
+ "type": "float32"
+ },
+ "step": {
+ "data": 0.7537541389465332,
+ "type": "float32"
+ }
+ },
+ "outputShape": [2, 2, 2, 3],
+ "expected": {
+ "name": "output",
+ "shape": [2, 2, 2, 3],
+ "data": [
+ 0.22992068529129028,
+ 0.9836748242378235,
+ 1.737428903579712,
+ 2.491183042526245,
+ 3.2449371814727783,
+ 3.9986913204193115,
+ 4.752445697784424,
+ 5.506199836730957,
+ 6.25995397567749,
+ 7.013708114624023,
+ 7.767462253570557,
+ 8.52121639251709,
+ 9.274970054626465,
+ 10.028724670410156,
+ 10.782478332519531,
+ 11.536232948303223,
+ 12.289986610412598,
+ 13.043741226196289,
+ 13.797494888305664,
+ 14.551249504089355,
+ 15.30500316619873,
+ 16.058757781982422,
+ 16.812511444091797,
+ 17.566265106201172
+ ],
+ "type": "float32"
+ }
+ },
+ {
+ "name": "constant float32 4D tensor of default float32 type step > 0",
+ "inputs": {
+ "start": {
+ "data": 0.22992068529129028,
+ "type": "float32"
+ },
+ "step": {
+ "data": 0.7615746259689331,
+ "type": "float32"
+ }
+ },
+ "outputShape": [2, 2, 2, 3],
+ "expected": {
+ "name": "output",
+ "shape": [2, 2, 2, 3],
+ "data": [
+ 0.22992068529129028,
+ 0.9914953112602234,
+ 1.7530698776245117,
+ 2.5146446228027344,
+ 3.276219129562378,
+ 4.0377936363220215,
+ 4.799368381500244,
+ 5.560943126678467,
+ 6.3225178718566895,
+ 7.084092140197754,
+ 7.845666885375977,
+ 8.6072416305542,
+ 9.368816375732422,
+ 10.130391120910645,
+ 10.891965866088867,
+ 11.653539657592773,
+ 12.415114402770996,
+ 13.176689147949219,
+ 13.938263893127441,
+ 14.699838638305664,
+ 15.461413383483887,
+ 16.22298812866211,
+ 16.984561920166016,
+ 17.746137619018555
+ ],
+ "type": "float32"
+ }
+ },
+ {
+ "name": "constant float32 4D tensor of default float32 type step = 0",
+ "inputs": {
+ "start": {
+ "data": 0.22992068529129028,
+ "type": "float32"
+ },
+ "step": {
+ "data": 0,
+ "type": "float32"
+ }
+ },
+ "outputShape": [2, 2, 2, 3],
+ "expected": {
+ "name": "output",
+ "shape": [2, 2, 2, 3],
+ "data": [
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028,
+ 0.22992068529129028
+ ],
+ "type": "float32"
+ }
+ },
+ {
+ "name": "constant float32 4D tensor of default float32 type step < 0",
+ "inputs": {
+ "start": {
+ "data": 0.22992068529129028,
+ "type": "float32"
+ },
+ "step": {
+ "data": -0.6248052716255188,
+ "type": "float32"
+ }
+ },
+ "outputShape": [2, 2, 2, 3],
+ "expected": {
+ "name": "output",
+ "shape": [2, 2, 2, 3],
+ "data": [
+ 0.22992068529129028,
+ -0.3948845863342285,
+ -1.0196897983551025,
+ -1.6444951295852661,
+ -2.2693004608154297,
+ -2.8941056728363037,
+ -3.5189108848571777,
+ -4.143716335296631,
+ -4.768521308898926,
+ -5.393326759338379,
+ -6.018132209777832,
+ -6.642937183380127,
+ -7.26774263381958,
+ -7.892547607421875,
+ -8.517353057861328,
+ -9.142158508300781,
+ -9.766963958740234,
+ -10.391769409179688,
+ -11.016573905944824,
+ -11.641379356384277,
+ -12.26618480682373,
+ -12.890990257263184,
+ -13.515795707702637,
+ -14.140600204467773
+ ],
+ "type": "float32"
+ }
+ },
+ {
+ "name": "constant float32 5D tensor of default float32 type",
+ "inputs": {
+ "start": {
+ "data": 0.22992068529129028,
+ "type": "float32"
+ },
+ "step": {
+ "data": 0.7537541389465332,
+ "type": "float32"
+ }
+ },
+ "outputShape": [2, 1, 4, 1, 3],
+ "expected": {
+ "name": "output",
+ "shape": [2, 1, 4, 1, 3],
+ "data": [
+ 0.22992068529129028,
+ 0.9836748242378235,
+ 1.737428903579712,
+ 2.491183042526245,
+ 3.2449371814727783,
+ 3.9986913204193115,
+ 4.752445697784424,
+ 5.506199836730957,
+ 6.25995397567749,
+ 7.013708114624023,
+ 7.767462253570557,
+ 8.52121639251709,
+ 9.274970054626465,
+ 10.028724670410156,
+ 10.782478332519531,
+ 11.536232948303223,
+ 12.289986610412598,
+ 13.043741226196289,
+ 13.797494888305664,
+ 14.551249504089355,
+ 15.30500316619873,
+ 16.058757781982422,
+ 16.812511444091797,
+ 17.566265106201172
+ ],
+ "type": "float32"
+ }
+ },
+ {
+ "name": "constant float32 4D tensor of explict float32 type",
+ "inputs": {
+ "start": {
+ "data": 0.22992068529129028,
+ "type": "float32"
+ },
+ "step": {
+ "data": 0.7537541389465332,
+ "type": "float32"
+ }
+ },
+ "outputShape": [2, 2, 2, 3],
+ "type": "float32",
+ "expected": {
+ "name": "output",
+ "shape": [2, 2, 2, 3],
+ "data": [
+ 0.22992068529129028,
+ 0.9836748242378235,
+ 1.737428903579712,
+ 2.491183042526245,
+ 3.2449371814727783,
+ 3.9986913204193115,
+ 4.752445697784424,
+ 5.506199836730957,
+ 6.25995397567749,
+ 7.013708114624023,
+ 7.767462253570557,
+ 8.52121639251709,
+ 9.274970054626465,
+ 10.028724670410156,
+ 10.782478332519531,
+ 11.536232948303223,
+ 12.289986610412598,
+ 13.043741226196289,
+ 13.797494888305664,
+ 14.551249504089355,
+ 15.30500316619873,
+ 16.058757781982422,
+ 16.812511444091797,
+ 17.566265106201172
+ ],
+ "type": "float32"
+ }
+ },
+ {
+ "name": "constant float32 4D tensor of float16 type",
+ "inputs": {
+ "start": {
+ "data": 0.22992068529129028,
+ "type": "float32"
+ },
+ "step": {
+ "data": 0.7537541389465332,
+ "type": "float32"
+ }
+ },
+ "outputShape": [2, 2, 2, 3],
+ "type": "float16",
+ "expected": {
+ "name": "output",
+ "shape": [2, 2, 2, 3],
+ "data": [
+ 0.22998046875,
+ 0.98388671875,
+ 1.7373046875,
+ 2.490234375,
+ 3.244140625,
+ 3.998046875,
+ 4.75390625,
+ 5.5078125,
+ 6.26171875,
+ 7.015625,
+ 7.765625,
+ 8.5234375,
+ 9.2734375,
+ 10.03125,
+ 10.78125,
+ 11.5390625,
+ 12.2890625,
+ 13.046875,
+ 13.796875,
+ 14.5546875,
+ 15.3046875,
+ 16.0625,
+ 16.8125,
+ 17.5625
+ ],
+ "type": "float16"
+ }
+ },
+ {
+ "name": "constant float32 4D tensor of int32 type",
+ "inputs": {
+ "start": {
+ "data": 7,
+ "type": "float32"
+ },
+ "step": {
+ "data": 1,
+ "type": "float32"
+ }
+ },
+ "outputShape": [2, 2, 2, 3],
+ "type": "int32",
+ "expected": {
+ "name": "output",
+ "shape": [2, 2, 2, 3],
+ "data": [
+ 7,
+ 8,
+ 9,
+ 10,
+ 11,
+ 12,
+ 13,
+ 14,
+ 15,
+ 16,
+ 17,
+ 18,
+ 19,
+ 20,
+ 21,
+ 22,
+ 23,
+ 24,
+ 25,
+ 26,
+ 27,
+ 28,
+ 29,
+ 30
+ ],
+ "type": "int32"
+ }
+ },
+ {
+ "name": "constant float32 4D tensor of uint32 type",
+ "inputs": {
+ "start": {
+ "data": 7,
+ "type": "float32"
+ },
+ "step": {
+ "data": 1,
+ "type": "float32"
+ }
+ },
+ "outputShape": [2, 2, 2, 3],
+ "type": "uint32",
+ "expected": {
+ "name": "output",
+ "shape": [2, 2, 2, 3],
+ "data": [
+ 7,
+ 8,
+ 9,
+ 10,
+ 11,
+ 12,
+ 13,
+ 14,
+ 15,
+ 16,
+ 17,
+ 18,
+ 19,
+ 20,
+ 21,
+ 22,
+ 23,
+ 24,
+ 25,
+ 26,
+ 27,
+ 28,
+ 29,
+ 30
+ ],
+ "type": "uint32"
+ }
+ },
+ {
+ "name": "constant float32 4D tensor of int64 type",
+ "inputs": {
+ "start": {
+ "data": 7,
+ "type": "float32"
+ },
+ "step": {
+ "data": 1,
+ "type": "float32"
+ }
+ },
+ "outputShape": [2, 2, 2, 3],
+ "type": "int64",
+ "expected": {
+ "name": "output",
+ "shape": [2, 2, 2, 3],
+ "data": [
+ "7",
+ "8",
+ "9",
+ "10",
+ "11",
+ "12",
+ "13",
+ "14",
+ "15",
+ "16",
+ "17",
+ "18",
+ "19",
+ "20",
+ "21",
+ "22",
+ "23",
+ "24",
+ "25",
+ "26",
+ "27",
+ "28",
+ "29",
+ "30"
+ ],
+ "type": "int64"
+ }
+ },
+ {
+ "name": "constant float32 4D tensor of int8 type step > 0",
+ "inputs": {
+ "start": {
+ "data": -9,
+ "type": "float32"
+ },
+ "step": {
+ "data": 1,
+ "type": "float32"
+ }
+ },
+ "outputShape": [2, 2, 2, 3],
+ "type": "int8",
+ "expected": {
+ "name": "output",
+ "shape": [2, 2, 2, 3],
+ "data": [
+ -9,
+ -8,
+ -7,
+ -6,
+ -5,
+ -4,
+ -3,
+ -2,
+ -1,
+ 0,
+ 1,
+ 2,
+ 3,
+ 4,
+ 5,
+ 6,
+ 7,
+ 8,
+ 9,
+ 10,
+ 11,
+ 12,
+ 13,
+ 14
+ ],
+ "type": "int8"
+ }
+ },
+ {
+ "name": "constant float32 4D tensor of int8 type step < 0",
+ "inputs": {
+ "start": {
+ "data": 7,
+ "type": "float32"
+ },
+ "step": {
+ "data": -2,
+ "type": "float32"
+ }
+ },
+ "outputShape": [2, 2, 2, 3],
+ "type": "int8",
+ "expected": {
+ "name": "output",
+ "shape": [2, 2, 2, 3],
+ "data": [
+ 7,
+ 5,
+ 3,
+ 1,
+ -1,
+ -3,
+ -5,
+ -7,
+ -9,
+ -11,
+ -13,
+ -15,
+ -17,
+ -19,
+ -21,
+ -23,
+ -25,
+ -27,
+ -29,
+ -31,
+ -33,
+ -35,
+ -37,
+ -39
+ ],
+ "type": "int8"
+ }
+ },
+ {
+ "name": "constant float32 4D tensor of uint8 type",
+ "inputs": {
+ "start": {
+ "data": 7,
+ "type": "float32"
+ },
+ "step": {
+ "data": 1,
+ "type": "float32"
+ }
+ },
+ "outputShape": [2, 2, 2, 3],
+ "type": "uint8",
+ "expected": {
+ "name": "output",
+ "shape": [2, 2, 2, 3],
+ "data": [
+ 7,
+ 8,
+ 9,
+ 10,
+ 11,
+ 12,
+ 13,
+ 14,
+ 15,
+ 16,
+ 17,
+ 18,
+ 19,
+ 20,
+ 21,
+ 22,
+ 23,
+ 24,
+ 25,
+ 26,
+ 27,
+ 28,
+ 29,
+ 30
+ ],
+ "type": "uint8"
+ }
+ }
+ ]
+}
\ No newline at end of file
diff --git a/webnn/resources/utils.js b/webnn/resources/utils.js
index f91d662..c1fc294 100644
--- a/webnn/resources/utils.js
+++ b/webnn/resources/utils.js
@@ -292,6 +292,7 @@
cast: {ULP: {float32: 1, float16: 1, int32: 0, uint32: 0, int64: 0, int8: 0, uint8: 0}},
clamp: {ULP: {float32: 0, float16: 0}},
concat: {ULP: {float32: 0, float16: 0}},
+ constant: {ULP: {float32: 2, float16: 2, int32: 0, uint32: 0, int64: 0, int8: 0, uint8: 0}},
conv2d: {ULP: {float32: getConv2dPrecisionTolerance, float16: getConv2dPrecisionTolerance}},
convTranspose2d: {ULP: {float32: getConv2dPrecisionTolerance, float16: getConv2dPrecisionTolerance}},
// Begin Element-wise binary operations
@@ -635,6 +636,13 @@
return namedOutputOperand;
};
+const buildConstantRange = (operationName, builder, resources) => {
+ const namedOutputOperand = {};
+ // invoke builder.constant(start, step, outputShape, type)
+ namedOutputOperand[resources.expected.name] = builder[operationName](resources.inputs.start, resources.inputs.step, resources.outputShape, resources.type);
+ return namedOutputOperand;
+};
+
const buildConvTranspose2d = (operationName, builder, resources) => {
// MLOperand convTranspose2d(MLOperand input, MLOperand filter, optional MLConvTranspose2dOptions options = {});
const namedOutputOperand = {};
