Add webnn softmax tests

webnn/resources/test_data/softmax.json

@@ -0,0 +1,70 @@
+{
+  "tests": [
+    {
+      "name": "softmax float32 2D tensor",
+      "inputs": {
+        "x": {
+          "shape": [4, 6],
+          "data": [
+            -9.602943250149636,
+            5.353947088147631,
+            4.982506021190005,
+            -1.8709681620943286,
+            2.275207739606868,
+            -1.9403444022539098,
+            1.8985638253724968,
+            1.7877445512108103,
+            -7.243864108734894,
+            7.827657028853274,
+            -2.4196847673132726,
+            -8.018804523844905,
+            -3.1493873901132474,
+            7.674345195428224,
+            -5.183967275728079,
+            -7.301743057626533,
+            -7.121610082852072,
+            -8.29701166999875,
+            -5.77890866334104,
+            -9.76213424697228,
+            7.332167409724249,
+            3.9077721241449215,
+            -5.2970783047100545,
+            5.0440509491607965
+          ],
+          "type": "float32"
+        }
+      },
+      "expected": {
+        "name": "output",
+        "shape": [4, 6],
+        "data": [
+          1.8385014755040174e-7,
+          0.575650691986084,
+          0.3970489799976349,
+          0.0004191970219835639,
+          0.026489850133657455,
+          0.0003911005624104291,
+          0.00264744833111763,
+          0.00236973213031888,
+          2.833488679243601e-7,
+          0.9949471354484558,
+          0.00003527247827150859,
+          1.3054766156983533e-7,
+          0.000019920609702239744,
+          0.9999766945838928,
+          0.0000026043292109534377,
+          3.1329790317613515e-7,
+          3.7513549955292547e-7,
+          1.1580249292819644e-7,
+          0.0000017836379129221314,
+          3.322107744452296e-8,
+          0.8818095922470093,
+          0.028719622641801834,
+          0.000002887773234760971,
+          0.08946608752012253
+        ],
+        "type": "float32"
+      }
+    }
+  ]
+}

webnn/resources/utils.js

@@ -60,13 +60,26 @@ const getExpectedData = (resources, outputName) => {
   return data;
 };
 
+/**
+ * Get ULP tolerance of softmax operation.
+ * @param {Object} resources - Resources used for building a graph
+ * @returns {Number} A tolerance number
+ */
+const getSoftmaxPrecisionTolerance = (resources) => {
+  // Compute the softmax values of the 2-D input tensor along axis 1.
+  const inputShape = resources.inputs[Object.keys(resources.inputs)[0]].shape;
+  const tolerance = inputShape[1] * 3 + 3;
+  return tolerance;
+};
+
 // Refer to precision metrics on https://github.com/webmachinelearning/webnn/issues/265#issuecomment-1256242643
 const PrecisionMetrics = {
   clamp: {ULP: {float32: 0, float16: 0}},
   concat: {ULP: {float32: 0, float16: 0}},
   relu: {ULP: {float32: 0, float16: 0}},
   reshape: {ULP: {float32: 0, float16: 0}},
   slice: {ULP: {float32: 0, float16: 0}},
+  softmax: {ULP: {float32: getSoftmaxPrecisionTolerance, float16: getSoftmaxPrecisionTolerance}},
 };
 
 /**
@@ -82,7 +95,7 @@ const getPrecisonTolerance = (operationName, metricType, resources) => {
   let tolerance = PrecisionMetrics[operationName][metricType][precisionType];
   // If the tolerance is dynamic, then evaluate the function to get the value.
   if (tolerance instanceof Function) {
-    tolerance = tolerance(resources, operationName);
+    tolerance = tolerance(resources);
   }
   return tolerance;
 };

webnn/softmax.https.any.js

@@ -0,0 +1,10 @@
+// META: title=test WebNN API softmax operation
+// META: global=window,dedicatedworker
+// META: script=./resources/utils.js
+// META: timeout=long
+
+'use strict';
+
+// https://webmachinelearning.github.io/webnn/#api-mlgraphbuilder-softmax
+
+testWebNNOperation('softmax', '/webnn/resources/test_data/softmax.json', buildOperationWithSingleInput);