@tensorflow/tfjs-core
Version:
Hardware-accelerated JavaScript library for machine intelligence
115 lines • 4.99 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('cumsum', ALL_ENVS, () => {
it('1D standard', async () => {
const res = tf.tensor1d([1, 2, 3, 4]).cumsum();
expect(res.shape).toEqual([4]);
expectArraysClose(await res.data(), [1, 3, 6, 10]);
});
it('1D reverse', async () => {
const reverse = true;
const exclusive = false;
const res = tf.tensor1d([1, 2, 3, 4]).cumsum(0, exclusive, reverse);
expect(res.shape).toEqual([4]);
expectArraysClose(await res.data(), [10, 9, 7, 4]);
});
it('1D exclusive', async () => {
const exclusive = true;
const res = tf.tensor1d([1, 2, 3, 4]).cumsum(0, exclusive);
expect(res.shape).toEqual([4]);
expectArraysClose(await res.data(), [0, 1, 3, 6]);
});
it('1D exclusive reverse', async () => {
const reverse = true;
const exclusive = true;
const res = tf.tensor1d([1, 2, 3, 4]).cumsum(0, exclusive, reverse);
expect(res.shape).toEqual([4]);
expectArraysClose(await res.data(), [9, 7, 4, 0]);
});
it('gradient: 1D', async () => {
const a = tf.tensor1d([1, 2, 3]);
const dy = tf.tensor1d([4, 5, 6]);
const da = tf.grad(x => tf.cumsum(x))(a, dy);
expect(da.shape).toEqual([3]);
expectArraysClose(await da.data(), [15, 11, 6]);
});
it('gradient with clones', async () => {
const a = tf.tensor1d([1, 2, 3]);
const dy = tf.tensor1d([4, 5, 6]);
const da = tf.grad(x => tf.cumsum(x.clone()).clone())(a, dy);
expect(da.shape).toEqual([3]);
expectArraysClose(await da.data(), [15, 11, 6]);
});
it('2D standard', async () => {
const res = tf.tensor2d([[1, 2], [3, 4]]).cumsum(1);
expect(res.shape).toEqual([2, 2]);
expectArraysClose(await res.data(), [1, 3, 3, 7]);
});
it('2D reverse exclusive', async () => {
const reverse = true;
const exclusive = true;
const res = tf.tensor2d([[1, 2], [3, 4]]).cumsum(1, exclusive, reverse);
expect(res.shape).toEqual([2, 2]);
expectArraysClose(await res.data(), [2, 0, 4, 0]);
});
it('2D axis=0', async () => {
const res = tf.tensor2d([[1, 2], [3, 4]]).cumsum();
expect(res.shape).toEqual([2, 2]);
expectArraysClose(await res.data(), [1, 2, 4, 6]);
});
it('3D standard', async () => {
const res = tf.tensor3d([[[0, 1], [2, 3]], [[4, 5], [6, 7]]]).cumsum(2);
expect(res.shape).toEqual([2, 2, 2]);
expectArraysClose(await res.data(), [0, 1, 2, 5, 4, 9, 6, 13]);
});
it('4d axis=2', async () => {
const input = tf.ones([1, 32, 46, 4]);
const res = tf.cumsum(input, 2, false, false);
expect(res.shape).toEqual([1, 32, 46, 4]);
const earlySlice = tf.slice(res, [0, 0, 0, 0], [1, 1, 46, 1]);
const lateSlice = tf.slice(res, [0, 31, 0, 0], [1, 1, 46, 1]);
const expectedDataInEachSlice = [
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
];
expectArraysClose(await earlySlice.data(), expectedDataInEachSlice);
expectArraysClose(await lateSlice.data(), expectedDataInEachSlice);
});
it('handle permutation properly', async () => {
const res = tf.ones([1, 240, 1, 10]).cumsum(1);
expect(res.shape).toEqual([1, 240, 1, 10]);
});
it('throws when passed a non-tensor', () => {
expect(() => tf.cumsum({}))
.toThrowError(/Argument 'x' passed to 'cumsum' must be a Tensor/);
});
it('accepts a tensor-like object', async () => {
const res = tf.cumsum([1, 2, 3, 4]);
expect(res.shape).toEqual([4]);
expectArraysClose(await res.data(), [1, 3, 6, 10]);
});
it('throws error for string tensor', () => {
expect(() => tf.cumsum([
'a', 'b', 'c'
])).toThrowError(/Argument 'x' passed to 'cumsum' must be numeric tensor/);
});
});
//# sourceMappingURL=cumsum_test.js.map