word2vec app with Algorithmia

Back in 2013 (time really flies) just after Google released word2vec I wrote a small wrapper library around it and wrote a blogpost + small app to query a pretrained model deployed in YHat. Since YHat was adquired by Alteryx in 2017 this app stopped working.

So I decided to rewrite this app this time using Algorithmia which is an amazing service to deploy machine learning models and expose them as a REST API.

Algorithmia integrates with a Data Science workflow in the right places using Git and language specific clients. Every algorithm is a Git repo and you can clone it to work on it locally or edit in their editor UI. Creating an endpoint is as eassy as writting an apply function in basically any language. It automatically picks dependencies and builds the models, handles versioning and you can upload pretrained models or connect to external sources. It’s quite an amazing service.

The endpoint is deployed in algorithmia as danielfrg/word2vec and you can find docs there and even make sample queries from their UI (I love that!).

After that I decided to rewrite the app using Vue.JS and the Algorithmia JS client. It’s just an static page I deployed using Netlify. Code for the app is on GitHub: danielfrg/word2vec-app.

You can find the app at word2vec.danielfrg.com. It allows you to do some simple queries for word similariy and analogies and it looks like this:

I am quite amazed by Algorithmia and can’t wait to deploy more models. Services like Algorithmia that combined with amazing free static hosting from Netlify is a great way to deploy Machine Learning applications for free :). Special shout to to Vue.JS for being what I always wanted in a JS framework.