* Convert text string to integer indices. * * @param {string} text Input text. * @returns {number[]} Indices of the characters of `text`.

* Perform training on a randomly sampled batch from the replay buffer. * * @param {number} batchSize Batch size. * @param {number} gamma Rewa

* Update the cart-pole system using an action. * @param {number} action Only the sign of `action` matters. * A value > 0 leads to a rightward

Generate some fake MNIST data for testing.

Calculate the current Q-values and the best action.

* Draw confusion matrix.

* Draw confusion matrix.

* Check the status of locally saved model. * * @returns If the locally saved model exists, the model info as a JSON * object. Else, `undefi

* Load a local csv file and prepare the data for training. Data source: * https://archive.ics.uci.edu/ml/datasets/Abalone * * @param {string} csvPa

Center and scale input image so the pixel values fall into [0, 255].

Simulate weight de-quantization. Args: w: (a numpy.ndarray) The weight to be quantized. bits: (int) number of bits used for the quantiza

* Coordinates are expected to be in the [0, 1] range.

* Display result results (from backend or frontend). * * @param {object} results Search result object. Assumed to have the following * fields: *

* Generate an evenly spaced 2d latent space.

Refresh examples of real MNIST images.

* Unzip file * * @param {string} sourcePath Source zip file path. * @param {string} destPath destination path.

* Takes the state of one complete game and returns features suitable for * training. Returns an object containing features = player1's hand represen

* Generate a number (i.e., batch) of examples. * * @param {number} batchSize Number of example image in the batch. * @param {number} numCircl

* Generate values for the latent vector and show them with the sliders. * * @param {bool} fixedLatent Whether to use fixed value for the latent *

* Use `textGenerator` to generate random text, show the characters on the * screen as they are generated one by one.

* Get the assets URL path, e.g. `http://assets-url-path/` if the metadata URL * is `http://assets-url-path/metadata.json`.

* Captures a frame from the webcam and normalizes it between -1 and 1. * Returns a batched image (1-element batch) of shape [1, w, h, c].

* Get policy-network logits and the action based on state-tensor inputs. * * @param {tf.Tensor} inputs A tf.Tensor instance of shape `[batchSize

* Return a tf.Model to tag tokens in an input sequence. * * @param {Object} opts * @return {tf.Model} the model instance

* Get current state as a tf.Tensor of shape [1, 4].

* Get all test data as a data tensor and a labels tensor. * * @param {number} numExamples Optional number of examples to get. If not * pr

* Get the test images URL path, e.g. `http://test-images-url-path/` if the * test images index is `http://test-images-url-path/index.json`.

Get word index. Args: reverse: Reverse the index, so that the returned index is from index values to words. Returns: The word inde

* Handles an input image as data URL. Displays it in the query element, and * sends it for inference to the right handler depending on the model ty

Display the indeterminate progress bar.

* Initializes the MNIST Transfer CNN demo.

* Initializes the Sentiment demo.

* Is the current environment Node.js? * * This logic is specific to Electron, because it checks * `process.type`.

Loads training and test data.

* Load IMDB data features from a local file. * * @param {string} filePath Data file on local filesystem. * @param {string} numWords Number of words