@tensorflow/tfjs-core
Version:
Hardware-accelerated JavaScript library for machine intelligence
133 lines • 5.77 kB
JavaScript
/**
* @license
* Copyright 2020 Google LLC. 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.
* =============================================================================
*/
import * as tf from '../index';
import { ALL_ENVS, describeWithFlags } from '../jasmine_util';
import { expectArraysClose } from '../test_util';
describeWithFlags('oneHot', ALL_ENVS, () => {
it('Depth 1 throws error', () => {
const indices = tf.tensor1d([0, 0, 0], 'int32');
expect(() => tf.oneHot(indices, 1)).toThrowError();
});
it('Depth 2, diagonal', async () => {
const indices = tf.tensor1d([0, 1], 'int32');
const res = tf.oneHot(indices, 2);
expect(res.shape).toEqual([2, 2]);
expectArraysClose(await res.data(), [1, 0, 0, 1]);
});
it('Scalar input as Tensor', async () => {
const indices = tf.scalar(2, 'int32');
const res = tf.oneHot(indices, 4);
expect(res.shape).toEqual([4]);
expectArraysClose(await res.data(), [0, 0, 1, 0]);
});
it('Scalar input as number', async () => {
const indices = 2;
const res = tf.oneHot(indices, 4);
expect(res.shape).toEqual([4]);
expectArraysClose(await res.data(), [0, 0, 1, 0]);
});
it('oneHot with chaining compiles', () => {
const indices = 2;
// Asserts that there is no compiler error.
tf.oneHot(indices, 4).toFloat();
});
it('Depth 2, transposed diagonal', async () => {
const indices = tf.tensor1d([1, 0], 'int32');
const res = tf.oneHot(indices, 2);
expect(res.shape).toEqual([2, 2]);
expectArraysClose(await res.data(), [0, 1, 1, 0]);
});
it('Depth 3, 4 events', async () => {
const indices = tf.tensor1d([2, 1, 2, 0], 'int32');
const res = tf.oneHot(indices, 3);
expect(res.shape).toEqual([4, 3]);
expectArraysClose(await res.data(), [0, 0, 1, 0, 1, 0, 0, 0, 1, 1, 0, 0]);
});
it('Out of range events do not trigger onValue', async () => {
const indices = tf.tensor1d([-1, 5, 12345], 'int32');
const res = tf.oneHot(indices, 5);
expect(res.shape).toEqual([3, 5]);
expectArraysClose(await res.data(), [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]);
});
it('Depth 2 onValue=3, offValue=-2', async () => {
const indices = tf.tensor1d([0, 1], 'int32');
const res = tf.oneHot(indices, 2, 3, -2);
expect(res.shape).toEqual([2, 2]);
expectArraysClose(await res.data(), [3, -2, -2, 3]);
});
it('indices not int32 throws error', () => {
const indices = tf.tensor1d([0, 1], 'float32');
expect(() => tf.oneHot(indices, 2)).toThrowError();
});
it('check output dtype', () => {
const expectedType = 'int32';
const indices = tf.tensor1d([0, 1], 'int32');
const res = tf.oneHot(indices, 2);
expect(res.dtype).toEqual(expectedType);
});
it('oneHot accepts a tensor-like object', async () => {
const res = tf.oneHot([0, 1], 2);
expect(res.shape).toEqual([2, 2]);
expectArraysClose(await res.data(), [1, 0, 0, 1]);
});
it('has gradient', async () => {
const a = tf.tensor1d([0, 1, 2], 'int32');
const dy = tf.ones([3, 3], 'float32');
const da = tf.grad((x) => tf.oneHot(x, 3))(a, dy);
expect(da.dtype).toBe('float32');
expect(da.shape).toEqual([3]);
expectArraysClose(await da.data(), [0, 0, 0]);
});
it('gradient with clones', async () => {
const a = tf.tensor1d([0, 1, 2], 'int32');
const dy = tf.ones([3, 3], 'float32');
const da = tf.grad((x) => tf.oneHot(x.clone(), 3).clone())(a, dy);
expect(da.dtype).toBe('float32');
expect(da.shape).toEqual([3]);
expectArraysClose(await da.data(), [0, 0, 0]);
});
it('gradient when indices is 3d', async () => {
const a = tf.tensor3d([1, 2, 3, 4], [1, 2, 2], 'int32');
const dy = tf.ones([1, 2, 2, 3], 'float32');
const depth = 3;
const da = tf.grad(x => tf.oneHot(x, depth))(a, dy);
expect(da.dtype).toBe('float32');
expect(da.shape).toEqual(a.shape);
expectArraysClose(await da.data(), [0, 0, 0, 0]);
});
it('oneHot with indices as 2d', async () => {
const indices = tf.tensor2d([[1, 3], [2, 3]], [2, 2], 'int32');
const depth = 4;
const res = tf.oneHot(indices, depth);
expect(res.shape).toEqual([2, 2, depth]);
expectArraysClose(await res.data(), [0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 1]);
});
it('Supports chaining', async () => {
const indices = tf.tensor2d([[1, 2, 3], [2, 3, 1], [4, 5, 6]], [3, 3], 'int32');
const depth = 6;
const onValue = 3;
const offValue = 7;
const res = indices.oneHot(depth, onValue, offValue);
expect(res.shape).toEqual([3, 3, 6]);
expectArraysClose(await res.data(), [
7, 3, 7, 7, 7, 7, 7, 7, 3, 7, 7, 7, 7, 7, 7, 3, 7, 7,
7, 7, 3, 7, 7, 7, 7, 7, 7, 3, 7, 7, 7, 3, 7, 7, 7, 7,
7, 7, 7, 7, 3, 7, 7, 7, 7, 7, 7, 3, 7, 7, 7, 7, 7, 7
]);
});
});
//# sourceMappingURL=one_hot_test.js.map