@tensorflow/tfjs-layers - npm Package Compare versions

Comparing version 0.7.4 to 0.7.5

dist/backend/common.js

 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-var _epsilon = 1e-7;
+var tfjs_core_1 = require("@tensorflow/tfjs-core");
+var _epsilon = tfjs_core_1.ENV.get('EPSILON');
 function epsilon() {
@@ -5,0 +6,0 @@ return _epsilon;

dist/engine/container.d.ts

 import { Scalar, serialization, Tensor } from '@tensorflow/tfjs-core';
 import { JsonDict, Kwargs, NamedTensorMap, Shape } from '../types';
 import { LayerVariable } from '../variables';
-import { Layer, Node, SymbolicTensor, DisposeResult } from './topology';
+import { DisposeResult, Layer, Node, SymbolicTensor } from './topology';
 export declare function loadWeightsFromJson(weightsJSON: JsonDict, layers: Layer[], skipMismatch?: boolean): void;
@@ -55,2 +55,3 @@ export declare function loadWeightsFromNamedTensorMap(weights: NamedTensorMap, layers: Layer[], strict?: boolean): void;
     readonly stateful: boolean;
+    resetStates(): void;
 }

dist/engine/container.js

@@ -397,6 +397,3 @@ "use strict";
         this.assertNotDisposed();
-        var result = {
-            refCountAfterDispose: null,
-            numDisposedVariables: 0
-        };
+        var result = { refCountAfterDispose: null, numDisposedVariables: 0 };
         if (--this._refCount === 0) {
@@ -923,2 +920,12 @@ for (var _i = 0, _a = this.layers; _i < _a.length; _i++) {
     });
+    Container.prototype.resetStates = function () {
+        var _this = this;
+        tfjs_core_1.tidy(function () {
+            _this.layers.forEach(function (layer) {
+                if (layer.stateful) {
+                    layer.resetStates();
+                }
+            });
+        });
+    };
     return Container;
@@ -925,0 +932,0 @@ }(topology_1.Layer));

dist/engine/topology.d.ts

@@ -125,2 +125,3 @@ import { DataType, Scalar, serialization, Tensor } from '@tensorflow/tfjs-core';
     readonly stateful: boolean;
+    resetStates(): void;
     protected assertInputCompatibility(inputs: Tensor | Tensor[] | SymbolicTensor | SymbolicTensor[]): void;
@@ -127,0 +128,0 @@ call(inputs: Tensor | Tensor[], kwargs: Kwargs): Tensor | Tensor[];

dist/engine/topology.js

@@ -285,2 +285,8 @@ "use strict";
     });
+    Layer.prototype.resetStates = function () {
+        if (!this.stateful) {
+            throw new Error('Cannot call the resetStates() method of a non-stateful Layer ' +
+                'object.');
+        }
+    };
     Layer.prototype.assertInputCompatibility = function (inputs) {
@@ -287,0 +293,0 @@ inputs = generic_utils.toList(inputs);

dist/engine/training.js

@@ -346,2 +346,5 @@ "use strict";
 })(ModelLoggingVerbosity = exports.ModelLoggingVerbosity || (exports.ModelLoggingVerbosity = {}));
+function checkBatchSize(batchSize) {
+    tfc.util.assert(batchSize > 0 && Number.isInteger(batchSize), "batchSize is required to be a positive integer, but got " + batchSize);
+}
 var Model = (function (_super) {
@@ -536,2 +539,3 @@ __extends(Model, _super);
         var batchSize = config.batchSize == null ? 32 : config.batchSize;
+        checkBatchSize(batchSize);
         var standardizedOuts = this.standardizeUserData(x, y, true, batchSize);
@@ -688,2 +692,3 @@ var ins = standardizedOuts[0].concat(standardizedOuts[1]);
         var batchSize = config.batchSize == null ? 32 : config.batchSize;
+        checkBatchSize(batchSize);
         return this.predictLoop(x, batchSize);
@@ -776,3 +781,5 @@ };
                             initialEpoch: initialEpoch,
+                            samples: numTrainSamples,
                             steps: stepsPerEpoch,
+                            batchSize: batchSize,
                             verbose: verbose,
@@ -1002,3 +1009,7 @@ doValidation: doValidation,
                         this.isTraining = true;
+                        _a.label = 1;
+                    case 1:
+                        _a.trys.push([1, , 3, 4]);
                         batchSize = config.batchSize == null ? 32 : config.batchSize;
+                        checkBatchSize(batchSize);
                         standardizedOuts = this.standardizeUserData(x, y, false, batchSize);
@@ -1008,2 +1019,5 @@ inputs = standardizedOuts[0];
                         doValidation = false;
+                        valX = void 0;
+                        valY = void 0;
+                        valIns = void 0;
                         needValidationDisposal = false;
@@ -1092,2 +1106,4 @@ if (config.validationData != null && config.validationData.length > 0) {
                         outLabels = this.getDedupedMetricsNames();
+                        valFunction = void 0;
+                        callbackMetrics = void 0;
                         if (doValidation) {
@@ -1106,3 +1122,3 @@ this.makeTestFunction();
                         return [4, this.fitLoop(trainFunction, ins, outLabels, batchSize, config.epochs, config.verbose, callbacks, valFunction, valIns, config.shuffle, callbackMetrics, config.initialEpoch, null, null, config.yieldEvery)];
-                    case 1:
+                    case 2:
                         out = _a.sent();
@@ -1116,2 +1132,6 @@ if (needValidationDisposal) {
                         return [2, out];
+                    case 3:
+                        this.isTraining = false;
+                        return [7];
+                    case 4: return [2];
                 }
@@ -1118,0 +1138,0 @@ });

dist/exports_layers.d.ts

 import { InputLayerConfig } from './engine/input_layer';
 import { Layer, LayerConfig } from './engine/topology';
 import { input } from './exports';
-import { ELULayerConfig, LeakyReLULayerConfig, SoftmaxLayerConfig, ThresholdedReLULayerConfig } from './layers/advanced_activations';
+import { ELULayerConfig, LeakyReLULayerConfig, ReLULayerConfig, SoftmaxLayerConfig, ThresholdedReLULayerConfig } from './layers/advanced_activations';
 import { ConvLayerConfig, Cropping2DLayerConfig, SeparableConvLayerConfig, UpSampling2DLayerConfig } from './layers/convolutional';
@@ -17,2 +17,3 @@ import { DepthwiseConv2DLayerConfig } from './layers/convolutional_depthwise';
 export declare function elu(config?: ELULayerConfig): Layer;
+export declare function reLU(config?: ReLULayerConfig): Layer;
 export declare function leakyReLU(config?: LeakyReLULayerConfig): Layer;
@@ -19,0 +20,0 @@ export declare function softmax(config?: SoftmaxLayerConfig): Layer;

dist/exports_layers.js

@@ -29,2 +29,6 @@ "use strict";
 exports.elu = elu;
+function reLU(config) {
+    return new advanced_activations_1.ReLU(config);
+}
+exports.reLU = reLU;
 function leakyReLU(config) {
@@ -31,0 +35,0 @@ return new advanced_activations_1.LeakyReLU(config);

dist/layers/advanced_activations.d.ts

 import { serialization, Tensor } from '@tensorflow/tfjs-core';
 import { Layer, LayerConfig } from '../engine/topology';
 import { Kwargs, Shape } from '../types';
+export interface ReLULayerConfig extends LayerConfig {
+    maxValue?: number;
+}
+export declare class ReLU extends Layer {
+    static className: string;
+    maxValue: number;
+    constructor(config?: ReLULayerConfig);
+    call(inputs: Tensor | Tensor[], kwargs: Kwargs): Tensor | Tensor[];
+    computeOutputShape(inputShape: Shape | Shape[]): Shape | Shape[];
+    getConfig(): serialization.ConfigDict;
+}
 export interface LeakyReLULayerConfig extends LayerConfig {
@@ -5,0 +16,0 @@ alpha?: number;

dist/layers/advanced_activations.js

@@ -20,2 +20,34 @@ "use strict";
 var types_utils_1 = require("../utils/types_utils");
+var ReLU = (function (_super) {
+    __extends(ReLU, _super);
+    function ReLU(config) {
+        var _this = _super.call(this, config == null ? {} : config) || this;
+        _this.supportsMasking = true;
+        if (config != null) {
+            _this.maxValue = config.maxValue;
+        }
+        return _this;
+    }
+    ReLU.prototype.call = function (inputs, kwargs) {
+        inputs = types_utils_1.getExactlyOneTensor(inputs);
+        var output = tfjs_core_1.relu(inputs);
+        if (this.maxValue != null) {
+            output = tfjs_core_1.clipByValue(output, 0, this.maxValue);
+        }
+        return output;
+    };
+    ReLU.prototype.computeOutputShape = function (inputShape) {
+        return inputShape;
+    };
+    ReLU.prototype.getConfig = function () {
+        var config = { maxValue: this.maxValue };
+        var baseConfig = _super.prototype.getConfig.call(this);
+        Object.assign(config, baseConfig);
+        return config;
+    };
+    ReLU.className = 'ReLU';
+    return ReLU;
+}(topology_1.Layer));
+exports.ReLU = ReLU;
+tfjs_core_1.serialization.SerializationMap.register(ReLU);
 var LeakyReLU = (function (_super) {
@@ -22,0 +54,0 @@ __extends(LeakyReLU, _super);

dist/layers/merge.js

@@ -15,5 +15,5 @@ "use strict";
 var tfjs_core_1 = require("@tensorflow/tfjs-core");
+var state_1 = require("../backend/state");
 var K = require("../backend/tfjs_backend");
 var topology_1 = require("../engine/topology");
-var state_1 = require("../backend/state");
 var errors_1 = require("../errors");
@@ -207,6 +207,5 @@ var generic_utils = require("../utils/generic_utils");
         return tfjs_core_1.tidy(function () {
-            var output = tfc.zeros(inputs[0].shape);
-            for (var _i = 0, inputs_3 = inputs; _i < inputs_3.length; _i++) {
-                var input = inputs_3[_i];
-                output = tfc.add(output, input);
+            var output = inputs[0].clone();
+            for (var i = 1; i < inputs.length; ++i) {
+                output = tfc.add(output, inputs[i]);
             }
@@ -238,6 +237,5 @@ return output;
         return tfjs_core_1.tidy(function () {
-            var output = tfc.ones(inputs[0].shape);
-            for (var _i = 0, inputs_4 = inputs; _i < inputs_4.length; _i++) {
-                var input = inputs_4[_i];
-                output = tfc.mul(output, input);
+            var output = inputs[0].clone();
+            for (var i = 1; i < inputs.length; ++i) {
+                output = tfc.mul(output, inputs[i]);
             }
@@ -269,6 +267,5 @@ return output;
         return tfjs_core_1.tidy(function () {
-            var output = tfc.zeros(inputs[0].shape);
-            for (var _i = 0, inputs_5 = inputs; _i < inputs_5.length; _i++) {
-                var input = inputs_5[_i];
-                output = tfc.add(output, input);
+            var output = inputs[0].clone();
+            for (var i = 1; i < inputs.length; ++i) {
+                output = tfc.add(output, inputs[i]);
             }
@@ -275,0 +272,0 @@ return tfc.mul(state_1.getScalar(1 / inputs.length), output);

dist/layers/recurrent.d.ts

@@ -39,2 +39,3 @@ import * as tfc from '@tensorflow/tfjs-core';
     states: Tensor[];
+    private keptStates;
     private numConstants;
@@ -41,0 +42,0 @@ constructor(config: RNNLayerConfig);

dist/utils/generic_utils.js

@@ -117,3 +117,10 @@ "use strict";
             if (fn == null) {
-                throw new errors_1.ValueError("Unknown " + printableModuleName + ": " + identifier);
+                throw new errors_1.ValueError("Unknown " + printableModuleName + ": " + identifier + ". " +
+                    "This may be due to one of the following reasons:\n" +
+                    ("1. The " + printableModuleName + " is defined in Python, in which ") +
+                    "case it needs to be ported to TensorFlow.js or your JavaScript " +
+                    "code.\n" +
+                    ("2. The custom " + printableModuleName + " is defined in JavaScript, ") +
+                    "but is not registered properly with " +
+                    "tf.serialization.registerClass().");
             }
@@ -142,3 +149,10 @@ }
         if (cls == null) {
-            throw new errors_1.ValueError("Unknown " + printableModuleName + ": " + className);
+            throw new errors_1.ValueError("Unknown " + printableModuleName + ": " + className + ". " +
+                "This may be due to one of the following reasons:\n" +
+                ("1. The " + printableModuleName + " is defined in Python, in which ") +
+                "case it needs to be ported to TensorFlow.js or your JavaScript " +
+                "code.\n" +
+                ("2. The custom " + printableModuleName + " is defined in JavaScript, ") +
+                "but is not registered properly with " +
+                "tf.serialization.registerClass().");
         }
@@ -145,0 +159,0 @@ if (fromConfig != null) {

dist/version.d.ts

@@ -1,2 +0,2 @@
-declare const version = "0.7.4";
+declare const version = "0.7.5";
 export { version };

dist/version.js

 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-var version = '0.7.4';
+var version = '0.7.5';
 exports.version = version;
 //# sourceMappingURL=version.js.map

package.json

 {
   "name": "@tensorflow/tfjs-layers",
-  "version": "0.7.4",
+  "version": "0.7.5",
   "description": "TensorFlow layers API in JavaScript",
@@ -13,3 +13,3 @@ "private": false,
   "devDependencies": {
-    "@tensorflow/tfjs-core": "~0.12.11",
+    "@tensorflow/tfjs-core": "~0.12.15",
     "@types/jasmine": "~2.5.53",
@@ -49,4 +49,4 @@ "clang-format": "~1.2.2",
   "peerDependencies": {
-    "@tensorflow/tfjs-core": "~0.12.11"
+    "@tensorflow/tfjs-core": "~0.12.15"
   }
 }

New alerts

License Policy Violation

License

This package is not allowed per your license policy. Review the package's license to ensure compliance.

Found 1 instance in 1 package

Fixed alerts

License Policy Violation

License

This package is not allowed per your license policy. Review the package's license to ensure compliance.

Found 1 instance in 1 package

Improved metrics

Total package byte prevSize

5559691

increased by0.96%

Lines of code

26860

increased by0.76%

No dependency changes