TensorflowJs extract model.predict() result in a Javascript variable

DanielFebruary 8, 2023February 24, 2023

After we configure and train our TensorflowJs model, we can run the predict() function to get the results of that model:

//1. create and configure the TensorflowJs model

//2. train the model

//3. use the model
model.predict(inputs)

If we want to store the results of model.predict() in a separate variable so that we can later refer to it we can do that using the arraySync() function:

const result = model.predict(tf.tensor1d([input_val_here]));
const predictedVal = resultTensor.arraySync()[0];

Keep in mind that, for example, if we use a 2d tensor we will need to update the index as needed:

const result = model.predict(tf.tensor2d([input_val_here], [1, 1]));
const predictedVal = resultTensor.arraySync()[0][0];

Also, you can use the print() function to directly log the result:

model.predict(tf.tensor1d([input_val_here])).print()