tsne-js | distributed stochastic neighbor | Machine Learning library

Firstly, make sure you are copying the same package-lock.json file that you use to install deps locally, to make sure you have the same dependency tree in your container as you do locally.

COPY package.json package-lock.json /app/

Then, make sure that you are matching the same node/npm version as you run locally (replace 12 with the major version you are running, be it 10, 12, 14 or whatever):

FROM node:12

Each node version is bundled with a specific npm version (latest 12 version comes with npm 6.14.6), you can find the bundled NPM version int he changelogs, https://github.com/nodejs/node/tree/master/doc/changelogs

Additionally, instead of running npm install, you might wanna run npm ci in the container. The latter skips any checks for discrepancy between the lock file and your package.json file and just installs the locked dependency tree, which is not only faster, but will also match your local dep tree exactly.

EDIT:

In addition, this line:

COPY . /app

Would also overwrite node_modules unless you have ignored it via .dockerignore or similar.

Easiest would probably be to add a .dockerignore file to the same folder as your Dockerfile and add lines stating: