Saving Model on Node: Cannot find any save handlers

To get help from the community, check out our Google group.

TensorFlow.js version

tfjs: 0.13.0 tfjs-node: 0.1.17 node-version: v8.11.2

Browser version

Run with node, so no browser involved

Describe the problem or feature request

I’m trying to save my model like this: const save = await model.save('file:///tmp/my-model-1'); but I get the error:

UnhandledPromiseRejectionWarning: Error: Cannot find any save handlers for URL 'file:///tmp/my-model-1' at new ValueError (/Users/jonas/RestAPITest/node_modules/@tensorflow/tfjs-layers/dist/errors.js:36:28) at Sequential.<anonymous> (/Users/jonas/RestAPITest/node_modules/@tensorflow/tfjs-layers/dist/engine/training.js:1160:39) at step (/Users/jonas/RestAPITest/node_modules/@tensorflow/tfjs-layers/dist/engine/training.js:42:23) at Object.next (/Users/jonas/RestAPITest/node_modules/@tensorflow/tfjs-layers/dist/engine/training.js:23:53) at /Users/jonas/RestAPITest/node_modules/@tensorflow/tfjs-layers/dist/engine/training.js:17:71 at new Promise (<anonymous>) at __awaiter (/Users/jonas/RestAPITest/node_modules/@tensorflow/tfjs-layers/dist/engine/training.js:13:12) at Sequential.Model.save (/Users/jonas/RestAPITest/node_modules/@tensorflow/tfjs-layers/dist/engine/training.js:1152:16) at train (/Users/jonas/RestAPITest/ml.js:41:30) at <anonymous> (node:43798) UnhandledPromiseRejectionWarning: Unhandled promise rejection. This error originated either by throwing inside of an async function without a catch block, or by rejecting a promise which was not handled with .catch(). (rejection id: 3) (node:43798) [DEP0018] DeprecationWarning: Unhandled promise rejections are deprecated. In the future, promise rejectionsthat are not handled will terminate the Node.js process with a non-zero exit code.

I followed this tutorial

And this is a stackoverflow post with the same issue but there is no solution.

Code to reproduce the bug / link to feature request

Here is my full code:

const tf = require('@tensorflow/tfjs');
const tfNode = require('@tensorflow/tfjs-node');

const model = tf.sequential();

const hidden = tf.layers.dense({ units: 4, inputShape: 3, activation: 'sigmoid' });
const output = tf.layers.dense({ units: 1, activation: 'sigmoid' });

model.add(hidden);
model.add(output);

const optimzer = tf.train.sgd(0.5);
model.compile({
    optimizer: optimzer,
    loss: tf.losses.meanSquaredError
});

const dataTasks = [];
const dataSolutions = [];
for (let i = 0; i < 90; i++) {
    dataTasks[i] = Math.round(Math.random());
    if (i % 3 == 0) {
        if (dataTasks[i] == 0) {
            dataSolutions.push(0)
        } else if (dataTasks[i] == 1) {
            dataSolutions.push(1)
        }
    } 
}
const tasks = tf.tensor2d(dataTasks, [30, 3]);
const solution = tf.tensor2d(dataSolutions, [30, 1]);

train();

async function train() {
    for (let i = 0; i < 1000; i++) {
        const response = await model.fit(tasks, solution, { shuffle: true})
    }
    console.log('trained');

    const save = await model.save('file:///tmp/my-model-1');
    console.log(save);
}

[caisq]

@JonasJW This minimal code below works for me, maybe you can try it out:

package.json:

{
  "dependencies": {
    "@tensorflow/tfjs-node": "0.1.17"
  }
}

model-save.js:

const tf = require('@tensorflow/tfjs');
require('@tensorflow/tfjs-node');

const model = tf.sequential();
model.add(tf.layers.dense({units: 1, inputShape: [1]}));

model.save('file://./model-1a');

Run model-save.js with node model-save.js

[piercus]

Same problem here, I have fixed it by doing

After editing package.json, i’ve done

rm -rf node_modules
rm package-lock.json
npm install

This problem will occur in many situations :

• When using npm install --global-style
• When using npm install --legacy-bundling
• If package.json contains both @tensorflow/tfjs and @tensorflow/tfjs-node
• If package.json contains both @tensorflow/tfjs-node-gpu and @tensorflow/tfjs-node on the same project with different versions

I think it would be better to just do

const tf = require('@tensorflow/tfjs-node');

@caisq can you explain the reason behind this ‘double-require’ design ?

Thank you for this great binding