@tensorflow/tfjs-core
Version:
Hardware-accelerated JavaScript library for machine intelligence
214 lines • 10.8 kB
JavaScript
;
/**
* @license
* Copyright 2018 Google Inc. All Rights Reserved.
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
* =============================================================================
*/
var __extends = (this && this.__extends) || (function () {
var extendStatics = function (d, b) {
extendStatics = Object.setPrototypeOf ||
({ __proto__: [] } instanceof Array && function (d, b) { d.__proto__ = b; }) ||
function (d, b) { for (var p in b) if (b.hasOwnProperty(p)) d[p] = b[p]; };
return extendStatics(d, b);
};
return function (d, b) {
extendStatics(d, b);
function __() { this.constructor = d; }
d.prototype = b === null ? Object.create(b) : (__.prototype = b.prototype, new __());
};
})();
var __awaiter = (this && this.__awaiter) || function (thisArg, _arguments, P, generator) {
return new (P || (P = Promise))(function (resolve, reject) {
function fulfilled(value) { try { step(generator.next(value)); } catch (e) { reject(e); } }
function rejected(value) { try { step(generator["throw"](value)); } catch (e) { reject(e); } }
function step(result) { result.done ? resolve(result.value) : new P(function (resolve) { resolve(result.value); }).then(fulfilled, rejected); }
step((generator = generator.apply(thisArg, _arguments || [])).next());
});
};
var __generator = (this && this.__generator) || function (thisArg, body) {
var _ = { label: 0, sent: function() { if (t[0] & 1) throw t[1]; return t[1]; }, trys: [], ops: [] }, f, y, t, g;
return g = { next: verb(0), "throw": verb(1), "return": verb(2) }, typeof Symbol === "function" && (g[Symbol.iterator] = function() { return this; }), g;
function verb(n) { return function (v) { return step([n, v]); }; }
function step(op) {
if (f) throw new TypeError("Generator is already executing.");
while (_) try {
if (f = 1, y && (t = op[0] & 2 ? y["return"] : op[0] ? y["throw"] || ((t = y["return"]) && t.call(y), 0) : y.next) && !(t = t.call(y, op[1])).done) return t;
if (y = 0, t) op = [op[0] & 2, t.value];
switch (op[0]) {
case 0: case 1: t = op; break;
case 4: _.label++; return { value: op[1], done: false };
case 5: _.label++; y = op[1]; op = [0]; continue;
case 7: op = _.ops.pop(); _.trys.pop(); continue;
default:
if (!(t = _.trys, t = t.length > 0 && t[t.length - 1]) && (op[0] === 6 || op[0] === 2)) { _ = 0; continue; }
if (op[0] === 3 && (!t || (op[1] > t[0] && op[1] < t[3]))) { _.label = op[1]; break; }
if (op[0] === 6 && _.label < t[1]) { _.label = t[1]; t = op; break; }
if (t && _.label < t[2]) { _.label = t[2]; _.ops.push(op); break; }
if (t[2]) _.ops.pop();
_.trys.pop(); continue;
}
op = body.call(thisArg, _);
} catch (e) { op = [6, e]; y = 0; } finally { f = t = 0; }
if (op[0] & 5) throw op[1]; return { value: op[0] ? op[1] : void 0, done: true };
}
};
Object.defineProperty(exports, "__esModule", { value: true });
var engine_1 = require("../engine");
var globals_1 = require("../globals");
var ops_1 = require("../ops/ops");
var serialization_1 = require("../serialization");
var optimizer_1 = require("./optimizer");
var AdamOptimizer = /** @class */ (function (_super) {
__extends(AdamOptimizer, _super);
function AdamOptimizer(learningRate, beta1, beta2, epsilon) {
if (epsilon === void 0) { epsilon = null; }
var _this = _super.call(this) || this;
_this.learningRate = learningRate;
_this.beta1 = beta1;
_this.beta2 = beta2;
_this.epsilon = epsilon;
_this.accumulatedFirstMoment = [];
_this.accumulatedSecondMoment = [];
globals_1.tidy(function () {
// accB* will be updated by batch.
_this.accBeta1 = ops_1.scalar(beta1).variable();
_this.accBeta2 = ops_1.scalar(beta2).variable();
});
if (epsilon == null) {
_this.epsilon = engine_1.ENGINE.backend.epsilon();
}
return _this;
}
AdamOptimizer.prototype.applyGradients = function (variableGradients) {
var _this = this;
var varNames = Array.isArray(variableGradients) ?
variableGradients.map(function (v) { return v.name; }) :
Object.keys(variableGradients);
globals_1.tidy(function () {
var oneMinusAccBeta1 = ops_1.sub(1, _this.accBeta1);
var oneMinusAccBeta2 = ops_1.sub(1, _this.accBeta2);
varNames.forEach(function (name, i) {
var value = engine_1.ENGINE.registeredVariables[name];
var trainable = false;
if (_this.accumulatedFirstMoment[i] == null) {
_this.accumulatedFirstMoment[i] = {
originalName: name + "/m",
variable: globals_1.tidy(function () { return ops_1.zerosLike(value).variable(trainable); })
};
}
if (_this.accumulatedSecondMoment[i] == null) {
_this.accumulatedSecondMoment[i] = {
originalName: name + "/v",
variable: globals_1.tidy(function () { return ops_1.zerosLike(value).variable(trainable); })
};
}
var gradient = Array.isArray(variableGradients) ?
variableGradients[i].tensor :
variableGradients[name];
if (gradient == null) {
return;
}
var firstMoment = _this.accumulatedFirstMoment[i].variable;
var secondMoment = _this.accumulatedSecondMoment[i].variable;
var newFirstMoment = firstMoment.mul(_this.beta1).add(gradient.mul(1 - _this.beta1));
var newSecondMoment = secondMoment.mul(_this.beta2)
.add(gradient.square().mul(1 - _this.beta2));
var biasCorrectedFirstMoment = newFirstMoment.div(oneMinusAccBeta1);
var biasCorrectedSecondMoment = newSecondMoment.div(oneMinusAccBeta2);
firstMoment.assign(newFirstMoment);
secondMoment.assign(newSecondMoment);
var newValue = biasCorrectedFirstMoment
.div(biasCorrectedSecondMoment.sqrt().add(_this.epsilon))
.mul(-_this.learningRate)
.add(value);
value.assign(newValue);
});
_this.accBeta1.assign(_this.accBeta1.mul(_this.beta1));
_this.accBeta2.assign(_this.accBeta2.mul(_this.beta2));
});
this.incrementIterations();
};
AdamOptimizer.prototype.dispose = function () {
this.accBeta1.dispose();
this.accBeta2.dispose();
if (this.accumulatedFirstMoment != null) {
globals_1.dispose(this.accumulatedFirstMoment.map(function (v) { return v.variable; }));
}
if (this.accumulatedSecondMoment != null) {
globals_1.dispose(this.accumulatedSecondMoment.map(function (v) { return v.variable; }));
}
};
AdamOptimizer.prototype.getWeights = function () {
return __awaiter(this, void 0, void 0, function () {
var variables;
return __generator(this, function (_a) {
switch (_a.label) {
case 0:
variables = this.accumulatedFirstMoment.concat(this.accumulatedSecondMoment);
return [4 /*yield*/, this.saveIterations()];
case 1: return [2 /*return*/, [_a.sent()].concat(variables.map(function (v) { return ({ name: v.originalName, tensor: v.variable }); }))];
}
});
});
};
AdamOptimizer.prototype.setWeights = function (weightValues) {
return __awaiter(this, void 0, void 0, function () {
var variableCount, trainable;
var _this = this;
return __generator(this, function (_a) {
switch (_a.label) {
case 0: return [4 /*yield*/, this.extractIterations(weightValues)];
case 1:
weightValues = _a.sent();
globals_1.tidy(function () {
_this.accBeta1.assign(ops_1.pow(_this.beta1, _this.iterations_ + 1));
_this.accBeta2.assign(ops_1.pow(_this.beta2, _this.iterations_ + 1));
});
variableCount = weightValues.length / 2;
trainable = false;
this.accumulatedFirstMoment =
weightValues.slice(0, variableCount).map(function (v) { return ({
originalName: v.name,
variable: v.tensor.variable(trainable)
}); });
this.accumulatedSecondMoment =
weightValues.slice(variableCount, variableCount * 2)
.map(function (v) { return ({
originalName: v.name,
variable: v.tensor.variable(trainable)
}); });
return [2 /*return*/];
}
});
});
};
AdamOptimizer.prototype.getConfig = function () {
return {
'learningRate': this.learningRate,
'beta1': this.beta1,
'beta2': this.beta2,
'epsilon': this.epsilon,
};
};
/** @nocollapse */
AdamOptimizer.fromConfig = function (cls, config) {
return new cls(config['learningRate'], config['beta1'], config['beta2'], config['epsilon']);
};
/** @nocollapse */
AdamOptimizer.className = 'Adam'; // Note: Name matters for Python compatibility.
return AdamOptimizer;
}(optimizer_1.Optimizer));
exports.AdamOptimizer = AdamOptimizer;
serialization_1.registerClass(AdamOptimizer);
//# sourceMappingURL=adam_optimizer.js.map