/**
|
* @license
|
* Copyright 2018 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 { ENGINE } from './engine';
|
import { Tensor, Variable } from './tensor';
|
import { convertToTensor, convertToTensorArray } from './tensor_util_env';
|
import * as util from './util';
|
/**
|
* Provided `f(x)`, returns another function `g(x, dy?)`, which gives the
|
* gradient of `f(x)` with respect to `x`.
|
*
|
* If `dy` is provided, the gradient of `f(x).mul(dy).sum()` with respect to
|
* `x` is computed instead. `f(x)` must take a single tensor `x` and return a
|
* single tensor `y`. If `f()` takes multiple inputs, use `tf.grads` instead.
|
*
|
* ```js
|
* // f(x) = x ^ 2
|
* const f = x => x.square();
|
* // f'(x) = 2x
|
* const g = tf.grad(f);
|
*
|
* const x = tf.tensor1d([2, 3]);
|
* g(x).print();
|
* ```
|
*
|
* ```js
|
* // f(x) = x ^ 3
|
* const f = x => x.pow(tf.scalar(3, 'int32'));
|
* // f'(x) = 3x ^ 2
|
* const g = tf.grad(f);
|
* // f''(x) = 6x
|
* const gg = tf.grad(g);
|
*
|
* const x = tf.tensor1d([2, 3]);
|
* gg(x).print();
|
* ```
|
*
|
* @param f The function f(x), to compute gradient for.
|
*
|
* @doc {heading: 'Training', subheading: 'Gradients'}
|
*/
|
function grad(f) {
|
util.assert(util.isFunction(f), () => 'The f passed in grad(f) must be a function');
|
return (x, dy) => {
|
// x can be of any dtype, thus null as the last argument.
|
const $x = convertToTensor(x, 'x', 'tf.grad', 'string_or_numeric');
|
const $dy = (dy != null) ? convertToTensor(dy, 'dy', 'tf.grad') : null;
|
return ENGINE.tidy(() => {
|
const { value, grads } = ENGINE.gradients(() => f($x), [$x], $dy);
|
if ($dy != null) {
|
util.assertShapesMatch(value.shape, $dy.shape, 'The shape of dy passed in grad(f)(x, dy) must match the shape ' +
|
'returned by f(x)');
|
}
|
checkGrads(grads);
|
return grads[0];
|
});
|
};
|
}
|
/**
|
* Provided `f(x1, x2,...)`, returns another function `g([x1, x2,...], dy?)`,
|
* which gives an array of gradients of `f()` with respect to each input
|
* [`x1`,`x2`,...].
|
*
|
* If `dy` is passed when calling `g()`, the gradient of
|
* `f(x1,...).mul(dy).sum()` with respect to each input is computed instead.
|
* The provided `f` must take one or more tensors and return a single tensor
|
* `y`. If `f()` takes a single input, we recommend using `tf.grad` instead.
|
*
|
* ```js
|
* // f(a, b) = a * b
|
* const f = (a, b) => a.mul(b);
|
* // df / da = b, df / db = a
|
* const g = tf.grads(f);
|
*
|
* const a = tf.tensor1d([2, 3]);
|
* const b = tf.tensor1d([-2, -3]);
|
* const [da, db] = g([a, b]);
|
* console.log('da');
|
* da.print();
|
* console.log('db');
|
* db.print();
|
* ```
|
*
|
* @param f The function `f(x1, x2,...)` to compute gradients for.
|
*
|
* @doc {heading: 'Training', subheading: 'Gradients'}
|
*/
|
function grads(f) {
|
util.assert(util.isFunction(f), () => 'The f passed in grads(f) must be a function');
|
return (args, dy) => {
|
util.assert(Array.isArray(args), () => 'The args passed in grads(f)(args) must be an array ' +
|
'of `Tensor`s or `TensorLike`s');
|
// args can be of any dtype, thus null as the last argument.
|
const $args = convertToTensorArray(args, 'args', 'tf.grads', 'string_or_numeric');
|
const $dy = (dy != null) ? convertToTensor(dy, 'dy', 'tf.grads') : null;
|
return ENGINE.tidy(() => {
|
const { value, grads } = ENGINE.gradients(() => f(...$args), $args, $dy);
|
if ($dy != null) {
|
util.assertShapesMatch(value.shape, $dy.shape, 'The shape of dy passed in grads(f)([x1,...], dy) must ' +
|
'match the shape returned by f([x1,...])');
|
}
|
checkGrads(grads);
|
return grads;
|
});
|
};
|
}
|
/**
|
* Like `tf.grad`, but also returns the value of `f()`. Useful when `f()`
|
* returns a metric you want to show.
|
*
|
* The result is a rich object with the following properties:
|
* - grad: The gradient of `f(x)` w.r.t. `x` (result of `tf.grad`).
|
* - value: The value returned by `f(x)`.
|
*
|
* ```js
|
* // f(x) = x ^ 2
|
* const f = x => x.square();
|
* // f'(x) = 2x
|
* const g = tf.valueAndGrad(f);
|
*
|
* const x = tf.tensor1d([2, 3]);
|
* const {value, grad} = g(x);
|
*
|
* console.log('value');
|
* value.print();
|
* console.log('grad');
|
* grad.print();
|
* ```
|
*
|
* @doc {heading: 'Training', subheading: 'Gradients'}
|
*/
|
function valueAndGrad(f) {
|
util.assert(util.isFunction(f), () => 'The f passed in valueAndGrad(f) must be a function');
|
return (x, dy) => {
|
util.assert(x instanceof Tensor, () => 'The x passed in valueAndGrad(f)(x) must be a tensor');
|
util.assert(dy == null || dy instanceof Tensor, () => 'The dy passed in valueAndGrad(f)(x, dy) must be a tensor');
|
const { grads, value } = ENGINE.gradients(() => f(x), [x], dy);
|
checkGrads(grads);
|
return { grad: grads[0], value };
|
};
|
}
|
/**
|
* Like `tf.grads`, but returns also the value of `f()`. Useful when `f()`
|
* returns a metric you want to show.
|
*
|
* The result is a rich object with the following properties:
|
* - grads: The gradients of `f()` w.r.t. each input (result of `tf.grads`).
|
* - value: The value returned by `f(x)`.
|
*
|
* ```js
|
* // f(a, b) = a * b
|
* const f = (a, b) => a.mul(b);
|
* // df/da = b, df/db = a
|
* const g = tf.valueAndGrads(f);
|
*
|
* const a = tf.tensor1d([2, 3]);
|
* const b = tf.tensor1d([-2, -3]);
|
* const {value, grads} = g([a, b]);
|
*
|
* const [da, db] = grads;
|
*
|
* console.log('value');
|
* value.print();
|
*
|
* console.log('da');
|
* da.print();
|
* console.log('db');
|
* db.print();
|
* ```
|
*
|
* @doc {heading: 'Training', subheading: 'Gradients'}
|
*/
|
function valueAndGrads(f) {
|
util.assert(util.isFunction(f), () => 'The f passed in valueAndGrads(f) must be a function');
|
return (args, dy) => {
|
util.assert(Array.isArray(args) && args.every(arg => arg instanceof Tensor), () => 'The args passed in valueAndGrads(f)(args) must be array of ' +
|
'tensors');
|
util.assert(dy == null || dy instanceof Tensor, () => 'The dy passed in valueAndGrads(f)(args, dy) must be a tensor');
|
const res = ENGINE.gradients(() => f(...args), args, dy);
|
if (dy != null) {
|
util.assertShapesMatch(res.value.shape, dy.shape, 'The shape of dy passed in valueAndGrads(f)([x1,...], dy) must ' +
|
'match the shape returned by f([x1,...])');
|
}
|
checkGrads(res.grads);
|
return res;
|
};
|
}
|
/**
|
* Computes and returns the gradient of f(x) with respect to the list of
|
* trainable variables provided by `varList`. If no list is provided, it
|
* defaults to all trainable variables.
|
*
|
* ```js
|
* const a = tf.variable(tf.tensor1d([3, 4]));
|
* const b = tf.variable(tf.tensor1d([5, 6]));
|
* const x = tf.tensor1d([1, 2]);
|
*
|
* // f(a, b) = a * x ^ 2 + b * x
|
* const f = () => a.mul(x.square()).add(b.mul(x)).sum();
|
* // df/da = x ^ 2, df/db = x
|
* const {value, grads} = tf.variableGrads(f);
|
*
|
* Object.keys(grads).forEach(varName => grads[varName].print());
|
* ```
|
*
|
* @param f The function to execute. f() should return a scalar.
|
* @param varList The list of variables to compute the gradients with respect
|
* to. Defaults to all trainable variables.
|
* @returns An object with the following keys and values:
|
* - `value`: The value of the function `f`.
|
* - `grads`: A map from the names of the variables to the gradients.
|
* If the `varList` argument is provided explicitly and contains a subset of
|
* non-trainable variables, this map in the return value will contain keys
|
* that map the names of the non-trainable variables to `null`.
|
*
|
* @doc {heading: 'Training', subheading: 'Gradients'}
|
*/
|
function variableGrads(f, varList) {
|
util.assert(util.isFunction(f), () => 'The f passed in variableGrads(f) must be a function');
|
util.assert(varList == null ||
|
Array.isArray(varList) && varList.every(v => v instanceof Variable), () => 'The varList passed in variableGrads(f, varList) must be an array ' +
|
'of variables');
|
const specifiedVarList = varList != null;
|
if (!specifiedVarList) {
|
// Get all of the trainable variables.
|
varList = [];
|
for (const varName in ENGINE.registeredVariables) {
|
varList.push(ENGINE.registeredVariables[varName]);
|
}
|
}
|
const specifiedNonTrainable = specifiedVarList ? varList.filter(variable => !variable.trainable) : null;
|
// Prune non-trainable variables.
|
const originalVarCount = varList.length;
|
varList = varList.filter(variable => variable.trainable);
|
util.assert(varList.length > 0, () => `variableGrads() expects at least one of the input variables to ` +
|
`be trainable, but none of the ${originalVarCount} variables is ` +
|
`trainable.`);
|
const allowNoGradients = true;
|
const { value, grads } = ENGINE.gradients(f, varList, null, allowNoGradients);
|
util.assert(grads.some(g => g != null), () => 'Cannot find a connection between any variable and the result of ' +
|
'the loss function y=f(x). Please make sure the operations that ' +
|
'use variables are inside the function f passed to minimize().');
|
util.assert(value.rank === 0, () => `The f passed in variableGrads(f) must return a scalar, but it ` +
|
`returned a rank-${value.rank} tensor`);
|
const namedGrads = {};
|
varList.forEach((v, i) => {
|
if (grads[i] != null) {
|
namedGrads[v.name] = grads[i];
|
}
|
});
|
if (specifiedNonTrainable != null) {
|
// If varList is explicitly provided and contains non-trainable values,
|
// add them to the returned gradients with `null` values.
|
specifiedNonTrainable.forEach(v => namedGrads[v.name] = null);
|
}
|
return { value, grads: namedGrads };
|
}
|
/**
|
* Overrides the gradient computation of a function `f`.
|
*
|
* Takes a function
|
* `f(...inputs, save) => {value: Tensor, gradFunc: (dy, saved) => Tensor[]}`
|
* and returns another function `g(...inputs)` which takes the same inputs as
|
* `f`. When called, `g` returns `f().value`. In backward mode, custom gradients
|
* with respect to each input of `f` are computed using `f().gradFunc`.
|
*
|
* The `save` function passed to `f` should be used for saving tensors needed
|
* in the gradient. And the `saved` passed to the `gradFunc` is a
|
* `NamedTensorMap`, which contains those saved tensors.
|
*
|
* ```js
|
* const customOp = tf.customGrad((x, save) => {
|
* // Save x to make sure it's available later for the gradient.
|
* save([x]);
|
* // Override gradient of our custom x ^ 2 op to be dy * abs(x);
|
* return {
|
* value: x.square(),
|
* // Note `saved.x` which points to the `x` we saved earlier.
|
* gradFunc: (dy, saved) => [dy.mul(saved[0].abs())]
|
* };
|
* });
|
*
|
* const x = tf.tensor1d([-1, -2, 3]);
|
* const dx = tf.grad(x => customOp(x));
|
*
|
* console.log(`f(x):`);
|
* customOp(x).print();
|
* console.log(`f'(x):`);
|
* dx(x).print();
|
* ```
|
*
|
* @param f The function to evaluate in forward mode, which should return
|
* `{value: Tensor, gradFunc: (dy, saved) => Tensor[]}`, where `gradFunc`
|
* returns the custom gradients of `f` with respect to its inputs.
|
*
|
* @doc {heading: 'Training', subheading: 'Gradients'}
|
*/
|
function customGrad(f) {
|
return ENGINE.customGrad(f);
|
}
|
function checkGrads(grads) {
|
const numNullGradients = grads.filter(g => g == null).length;
|
if (numNullGradients > 0) {
|
throw new Error(`Cannot compute gradient of y=f(x) with respect to x. Make sure that
|
the f you passed encloses all operations that lead from x to y.`);
|
}
|
}
|
export { customGrad, variableGrads, valueAndGrad, valueAndGrads, grad, grads, };
|
//# sourceMappingURL=data:application/json;base64,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
|
