@tensorflow/tfjs-core
Version:
Hardware-accelerated JavaScript library for machine intelligence
281 lines • 14 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('maxPool3d', ALL_ENVS, () => {
it('4D x=[2,2,2,1] f=[2,2,2] s=1 p=valid', async () => {
const x = tf.tensor4d([1, 2, 3, 4, 5, 6, 7, 8], [2, 2, 2, 1]);
const result = tf.maxPool3d(x, 2, 1, 'valid');
expect(result.shape).toEqual([1, 1, 1, 1]);
expectArraysClose(await result.data(), [8]);
});
it('x=[1,1,1,1,1] f=[1,1,1] s=1 [0] => [0]', async () => {
const x = tf.tensor5d([0], [1, 1, 1, 1, 1]);
const result = tf.maxPool3d(x, 1, 1, 0);
expect(result.shape).toEqual([1, 1, 1, 1, 1]);
expectArraysClose(await result.data(), [0]);
});
it('x=[1,2,2,2,1] f=[2,2,2] s=1 p=valid', async () => {
const x = tf.tensor5d([1, 2, 3, 4, 5, 6, 7, 8], [1, 2, 2, 2, 1]);
const result = tf.maxPool3d(x, 2, 1, 'valid');
expect(result.shape).toEqual([1, 1, 1, 1, 1]);
expectArraysClose(await result.data(), [8]);
});
it('x=[1,2,2,2,1] f=[2,2,2] s=1 p=same', async () => {
const x = tf.tensor5d([1, 2, 3, 4, 5, 6, 7, 8], [1, 2, 2, 2, 1]);
const expected = [8, 8, 8, 8, 8, 8, 8, 8];
const result = tf.maxPool3d(x, 2, 1, 'same');
expect(result.shape).toEqual([1, 2, 2, 2, 1]);
expectArraysClose(await result.data(), expected);
});
it('x=[1,3,3,3,1] f=[2,2,2] s=1 p=valid', async () => {
const x = tf.tensor5d([
1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14,
15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27
], [1, 3, 3, 3, 1]);
const expected = [14, 15, 17, 18, 23, 24, 26, 27];
const result = tf.maxPool3d(x, 2, 1, 'valid');
expect(result.shape).toEqual([1, 2, 2, 2, 1]);
expectArraysClose(await result.data(), expected);
});
it('x=[1,3,3,3,1] f=[2,2,2] s=1 p=same', async () => {
const x = tf.tensor5d([
1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14,
15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27
], [1, 3, 3, 3, 1]);
const expected = [
14, 15, 15, 17, 18, 18, 17, 18, 18, 23, 24, 24, 26, 27,
27, 26, 27, 27, 23, 24, 24, 26, 27, 27, 26, 27, 27
];
const result = tf.maxPool3d(x, 2, 1, 'same');
expect(result.shape).toEqual([1, 3, 3, 3, 1]);
expectArraysClose(await result.data(), expected);
});
it('x=[1,3,3,3,1] f=[2,2,2] s=1 p=valid, ignores NaNs', async () => {
const x = tf.tensor5d([
1, 2, 3, 4, 5, 6, NaN, 8, NaN, 10, 11, 12, 13, 14,
15, 16, 17, 18, 19, 20, 21, NaN, 23, 24, 25, 26, 27
], [1, 3, 3, 3, 1]);
const expected = [14, 15, 17, 18, 23, 24, 26, 27];
const result = tf.maxPool3d(x, 2, 1, 'valid');
expect(result.shape).toEqual([1, 2, 2, 2, 1]);
expectArraysClose(await result.data(), expected);
});
it('x=[2,3,3,3,1] f=[2,2,2] s=1 p=valid', async () => {
const x = tf.tensor5d([
1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14,
15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28,
29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42,
43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54
], [2, 3, 3, 3, 1]);
const expected = [14, 15, 17, 18, 23, 24, 26, 27, 41, 42, 44, 45, 50, 51, 53, 54];
const result = tf.maxPool3d(x, 2, 1, 'valid');
expect(result.shape).toEqual([2, 2, 2, 2, 1]);
expectArraysClose(await result.data(), expected);
});
it('x=[1,3,3,3,2] f=[2,2,2] s=1 p=valid', async () => {
const x = tf.tensor5d([
1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14,
15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28,
29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42,
43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54
], [1, 3, 3, 3, 2]);
const expected = [27, 28, 29, 30, 33, 34, 35, 36, 45, 46, 47, 48, 51, 52, 53, 54];
const result = tf.maxPool3d(x, 2, 1, 'valid');
expect(result.shape).toEqual([1, 2, 2, 2, 2]);
expectArraysClose(await result.data(), expected);
});
it('x=[1,2,2,2,1] f=[2,2,2] s=1 p=1 roundingMode=floor', async () => {
const x = tf.tensor5d([1, 2, 3, 4, 5, 6, 7, 8], [1, 2, 2, 2, 1]);
const expected = [
1, 2, 2, 3, 4, 4, 3, 4, 4, 5, 6, 6, 7, 8,
8, 7, 8, 8, 5, 6, 6, 7, 8, 8, 7, 8, 8
];
const result = tf.maxPool3d(x, 2, 1, 1, 'floor');
expect(result.shape).toEqual([1, 3, 3, 3, 1]);
expectArraysClose(await result.data(), expected);
});
it('throws when x is not rank 5', async () => {
// tslint:disable-next-line:no-any
const x = tf.tensor1d([1]);
expect(() => tf.maxPool3d(x, 2, 1, 'valid')).toThrowError();
});
it('throws when dimRoundingMode is set and pad is not a number', async () => {
const x = tf.tensor5d([1], [1, 1, 1, 1, 1]);
const pad = 'valid';
const dimRoundingMode = 'round';
expect(() => tf.maxPool3d(x, 2, 1, pad, dimRoundingMode)).toThrowError();
});
it('throws when passed a non-tensor', () => {
expect(() => tf.maxPool3d({}, 2, 1, 'valid')).toThrowError();
});
it('accepts a tensor-like object', async () => {
const x = [[[[[0]]]]]; // 1x1x1x1x1
const result = tf.maxPool3d(x, 1, 1, 0);
expect(result.shape).toEqual([1, 1, 1, 1, 1]);
expectArraysClose(await result.data(), [0]);
});
});
describeWithFlags('maxPool3dBackprop', ALL_ENVS, () => {
it('gradient x=[2,2,2,1] f=[1,1,1] s=1', async () => {
const dy = tf.tensor4d([1, 2, 1, 2, 1, 2, 1, 2], [2, 2, 2, 1]);
const x = tf.tensor4d([1, 2, 3, 4, 5, 6, 7, 8], [2, 2, 2, 1]);
const expected = [1, 2, 1, 2, 1, 2, 1, 2];
const dx = tf.grad((x) => tf.maxPool3d(x, 1, 1, 0))(x, dy);
expect(dx.shape).toEqual(x.shape);
expectArraysClose(await dx.data(), expected);
});
it('gradient x=[1,2,2,2,1] f=[1,1,1] s=1', async () => {
const dy = tf.tensor5d([1, 2, 1, 2, 1, 2, 1, 2], [1, 2, 2, 2, 1]);
const x = tf.tensor5d([1, 2, 3, 4, 5, 6, 7, 8], [1, 2, 2, 2, 1]);
const expected = [1, 2, 1, 2, 1, 2, 1, 2];
const dx = tf.grad((x) => tf.maxPool3d(x, 1, 1, 0))(x, dy);
expect(dx.shape).toEqual(x.shape);
expectArraysClose(await dx.data(), expected);
});
it('gradient x=[1,2,2,2,1] f=[2,2,2] s=2', async () => {
const dy = tf.tensor5d([1], [1, 1, 1, 1, 1]);
const x = tf.tensor5d([1, 2, 3, 4, 5, 6, 7, 8], [1, 2, 2, 2, 1]);
const expected = [0, 0, 0, 0, 0, 0, 0, 1];
const dx = tf.grad((x) => tf.maxPool3d(x, 2, 1, 0))(x, dy);
expect(dx.shape).toEqual(x.shape);
expectArraysClose(await dx.data(), expected);
});
it('gradient with clone x=[1,2,2,2,1] f=[2,2,2] s=1', async () => {
const dy = tf.tensor5d([1], [1, 1, 1, 1, 1]);
const x = tf.tensor5d([1, 2, 3, 4, 5, 6, 7, 8], [1, 2, 2, 2, 1]);
const expected = [0, 0, 0, 0, 0, 0, 0, 1];
const dx = tf.grad((x) => tf.maxPool3d(x.clone(), 2, 1, 0).clone())(x, dy);
expect(dx.shape).toEqual(x.shape);
expectArraysClose(await dx.data(), expected);
});
it('gradient x=[1,3,3,3,1] f=[2,2,2] s=1 no dup max value', async () => {
const dy = tf.tensor5d([1, 2, 3, 4, 5, 6, 7, 8], [1, 2, 2, 2, 1]);
const x = tf.tensor5d([
1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14,
15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27
], [1, 3, 3, 3, 1]);
const expected = [
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1,
2, 0, 3, 4, 0, 0, 0, 0, 5, 6, 0, 7, 8
];
const dx = tf.grad((x) => tf.maxPool3d(x, 2, 1, 0))(x, dy);
expect(dx.shape).toEqual(x.shape);
expectArraysClose(await dx.data(), expected);
});
it('gradient x=[1,3,3,3,1] f=[2,2,2] s=1 dup max value', async () => {
const dy = tf.tensor5d([1, 2, 3, 4, 5, 6, 7, 8], [1, 2, 2, 2, 1]);
const x = tf.tensor5d([
1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 27,
15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 14
], [1, 3, 3, 3, 1]);
const expected = [
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 36,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
];
const dx = tf.grad((x) => tf.maxPool3d(x, 2, 1, 0))(x, dy);
expect(dx.shape).toEqual(x.shape);
expectArraysClose(await dx.data(), expected);
});
it('gradient x=[1,4,4,4,1] f=[2,2,2] s=2', async () => {
const dy = tf.tensor5d([1, 2, 3, 4, 5, 6, 7, 8], [1, 2, 2, 2, 1]);
const x = tf.tensor5d([
1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16,
17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32,
33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48,
49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64
], [1, 4, 4, 4, 1]);
const expected = [
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1,
0, 2, 0, 0, 0, 0, 0, 3, 0, 4, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 5, 0, 6, 0, 0, 0, 0, 0, 7, 0, 8
];
const dx = tf.grad((x) => tf.maxPool3d(x, 2, 2, 0))(x, dy);
expect(dx.shape).toEqual(x.shape);
expectArraysClose(await dx.data(), expected);
});
it('gradient x=[1,3,3,3,2] f=[2,2,2] s=1 no dup max value', async () => {
const dy = tf.tensor5d([1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16], [1, 2, 2, 2, 2]);
const x = tf.tensor5d([
1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14,
15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28,
29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42,
43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54
], [1, 3, 3, 3, 2]);
const expected = [
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 1, 2, 3, 4, 0, 0, 5, 6, 7, 8,
0, 0, 0, 0, 0, 0, 0, 0, 9, 10, 11, 12, 0, 0, 13, 14, 15, 16
];
const dx = tf.grad((x) => tf.maxPool3d(x, 2, 1, 0))(x, dy);
expect(dx.shape).toEqual(x.shape);
expectArraysClose(await dx.data(), expected);
});
it('gradient x=[1,3,3,3,2] f=[2,2,2] s=1 dup max value', async () => {
const dy = tf.tensor5d([1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16], [1, 2, 2, 2, 2]);
const x = tf.tensor5d([
1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14,
15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 53, 54,
29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42,
43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 27, 28
], [1, 3, 3, 3, 2]);
const expected = [
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 64, 72, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
];
const dx = tf.grad((x) => tf.maxPool3d(x, 2, 1, 0))(x, dy);
expect(dx.shape).toEqual(x.shape);
expectArraysClose(await dx.data(), expected);
});
it('gradient x=[2,3,3,3,1] f=[2,2,2] s=1 no dup max value', async () => {
const dy = tf.tensor5d([1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16], [2, 2, 2, 2, 1]);
const x = tf.tensor5d([
1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14,
15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28,
29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42,
43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54
], [2, 3, 3, 3, 1]);
const expected = [
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 2, 0, 3, 4,
0, 0, 0, 0, 5, 6, 0, 7, 8, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 9, 10, 0, 11, 12, 0, 0, 0, 0, 13, 14, 0, 15, 16
];
const dx = tf.grad((x) => tf.maxPool3d(x, 2, 1, 0))(x, dy);
expect(dx.shape).toEqual(x.shape);
expectArraysClose(await dx.data(), expected);
});
it('gradient x=[2,3,3,3,1] f=[2,2,2] s=1 dup max value', async () => {
const dy = tf.tensor5d([1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16], [2, 2, 2, 2, 1]);
const x = tf.tensor5d([
1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 27,
15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 14, 28,
29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 54, 42,
43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 41
], [2, 3, 3, 3, 1]);
const expected = [
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 36, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 100, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
];
const dx = tf.grad((x) => tf.maxPool3d(x, 2, 1, 0))(x, dy);
expect(dx.shape).toEqual(x.shape);
expectArraysClose(await dx.data(), expected);
});
});
//# sourceMappingURL=max_pool_3d_test.js.map