Merge pull request #238 from BruceDai/support_latest_spec

Support latest spec
webmachinelearning · Mar 26, 2024 · d20c9f4 · d20c9f4
2 parents 866d72b + 2444bb1
commit d20c9f4
Show file tree

Hide file tree

Showing 257 changed files with 3,217 additions and 5,108 deletions.
diff --git a/package.json b/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@webmachinelearning/webnn-polyfill",
-  "version": "0.1.9",
+  "version": "0.1.10",
   "description": "WebNN API polyfill",
   "main": "dist/webnn-polyfill.js",
   "jsdelivr": "dist/webnn-polyfill.js",

diff --git a/src/ml.ts b/src/ml.ts
@@ -1,8 +1,7 @@
 import {MLContext, MLContextOptions} from './nn/context';
-import * as utils from './nn/utils';
 
 /**
- * [spec](https://webmachinelearning.github.io/webnn/#ml)
+ * [spec](https://webmachinelearning.github.io/webnn/#api-ml)
  */
 export class ML {
   /** @ignore */
@@ -20,11 +19,4 @@ export class ML {
       resolve(context);
     });
   }
-
-  createContextSync(options: MLContextOptions = {}): MLContext {
-    utils.assert(
-        typeof window === 'undefined' && typeof importScripts === 'function',
-        'createContextSync() should only be allowed in dedicated worker.');
-    return new MLContext(options);
-  }
 }
diff --git a/src/nn/context.ts b/src/nn/context.ts
@@ -44,7 +44,7 @@ export interface MLContextOptions {
 }
 
 /**
- * [API spec](https://webmachinelearning.github.io/webnn/#mlcontext)
+ * [API spec](https://webmachinelearning.github.io/webnn/#api-mlcontext)
  */
 export class MLContext {
   private options_: MLContextOptions;
@@ -79,7 +79,7 @@ export class MLContext {
   }
 
   /**
-   * [spec](https://webmachinelearning.github.io/webnn/#dom-mlcontext-compute)
+   * [spec](https://webmachinelearning.github.io/webnn/#api-mlcontext-compute)
    */
   async compute(
       graph: MLGraph,
@@ -89,19 +89,6 @@ export class MLContext {
     return result;
   }
 
-  /**
-   * [spec](https://webmachinelearning.github.io/webnn/#dom-mlcontext-computesync)
-   */
-  computeSync(
-      graph: MLGraph,
-      inputs: MLNamedArrayBufferViews,
-      outputs: MLNamedArrayBufferViews): void {
-      utils.assert(
-          typeof window === 'undefined' && typeof importScripts === 'function',
-          'computeSync() should only be allowed in dedicated worker.');
-      graph.computeSync(inputs, outputs);
-  }
-
   /** @internal */
   // Expose tf interfance for setting backend.
   get tf(): unknown {

diff --git a/src/nn/graph.ts b/src/nn/graph.ts
@@ -146,7 +146,8 @@ export class MLGraph {
       utils.assert(
           utils.isTypedArray(resource),
           'Only resource of ArrayBufferView type is supported.');
-      utils.validateTypedArray(resource, inputOperand.desc.type, dimensions);
+      utils.validateTypedArray(
+          resource, inputOperand.desc.dataType, dimensions);
     }
   }
 
@@ -178,7 +179,7 @@ export class MLGraph {
       for (const inputName of this.inputs_.keys()) {
         const inputOperand = this.inputs_.get(inputName);
         const typedArrayConstructor =
-            utils.getTypedArray(inputOperand.desc.type);
+            utils.getTypedArray(inputOperand.desc.dataType);
         const inputBuffer = new typedArrayConstructor(
             utils.sizeFromDimensions(inputOperand.desc.dimensions));
         inputs[inputName] = inputBuffer;
@@ -208,30 +209,13 @@ export class MLGraph {
       const tensor = outputTensors[outputName] as tf.Tensor;
       const desc = utils.createOperandDescriptorFromTensor(tensor);
       const resource = outputs[outputName] ;
-      utils.validateTypedArray(resource, desc.type, desc.dimensions);
+      utils.validateTypedArray(resource, desc.dataType, desc.dimensions);
       resource.set(await tensor.data());
       tf.dispose(tensor);
     }
     return {inputs, outputs};
   }
 
-  /** @internal */
-  computeSync(
-    inputs: MLNamedArrayBufferViews,
-    outputs: MLNamedArrayBufferViews): void {
-    const outputTensors: tf.TensorContainerObject =
-        this.computeOutputTensors(inputs, outputs);
-    // Setup the outputs.
-    for (const outputName of Object.keys(outputTensors)) {
-      const tensor = outputTensors[outputName] as tf.Tensor;
-      const desc = utils.createOperandDescriptorFromTensor(tensor);
-      const resource = outputs[outputName] ;
-      utils.validateTypedArray(resource, desc.type, desc.dimensions);
-      resource.set(tensor.dataSync());
-      tf.dispose(tensor);
-    }
-  }
-
   /** @ignore */
   constructor(outputs?: MLNamedOperands) {
     utils.assert(outputs !== undefined, 'Invalid argument');
@@ -252,14 +236,6 @@ export class MLGraph {
     return graph;
   }
 
-  /** @internal */
-  static buildAndCompileSync(outputs?: MLNamedOperands): MLGraph {
-    const graph = new MLGraph(outputs);
-    graph.build();
-    graph.compileSync();
-    return graph;
-  }
-
   private build(): void {
     const visitedOps: Set<Operation> = new Set();
     for (const output of this.outputs_.values()) {
@@ -306,11 +282,6 @@ export class MLGraph {
     await this.computeOnce();
   }
 
-  private compileSync(): void {
-    this.allocateConstants();
-    this.computeOnceSync();
-  }
-
   private allocateConstants(): void {
     for (const constant of this.constants_) {
       this.constantTensors_.set(
@@ -327,15 +298,6 @@ export class MLGraph {
     }
   }
 
-  private computeOnceSync(): void {
-    const outputTensors = this.computeOutputTensors();
-    for (const outputName of Object.keys(outputTensors)) {
-      const tensor = outputTensors[outputName] as tf.Tensor;
-      tensor.dataSync();
-      tf.dispose(tensor);
-    }
-  }
-
   /** @ignore */
   // For memory leak testing.
   dispose(): void {