An application for training deep learning models in your browser

Jordan Anaya (of Pizzagate fame) writes:

For the last year I [Anaya] have been working on a web application that implements how I trained deep learning models at Johns Hopkins entirely in the browser. It’s available at https://aleaaxis.net/

I’m not sure if you have any experience with building deep learning models, but there’s a data generation component which you might find interesting. The data generation wasn’t meant to be very complex but I’d be interested if you have suggestions for improvement.

I think it’s going to be really useful for introducing students to deep learning. There’s a video at the learn page if you want to quickly see some examples. If you run into any issues let me know (I haven’t tested on different hardware / operating systems).

I don’t know enough about deep learning to be able to evaluate this, but it could be of interest to some of you. I like the idea that it can simulate from a generative model.

4 thoughts on “An application for training deep learning models in your browser”

  1. Well I tried. I usually like such things so I watched the intro video and then tried uploading a data set to test it out. After 15 different tries, I give up. My files were rejected for file type, browser/file incompatibility, missing data (non allowed), unallowed syntax in column names (spaces?), and so on. I might eventually find a file that works, but I’ve run out of energy. LLMs work much better. So, I appreciate the effort, I can’t say anything good about the implementation. Maybe data generation works better, but I didn’t try that.

    • It has to be a text file, not an excel file. I have seen Firefox have issues with .csv extensions however, and since I don’t have a mac I’m not sure about safari. What type of files were they?

      Yes, if your data entries contain columns or spaces then my parser will detect a variable number of columns. If you have coding experience you can use the JSON upload to the left where you have already parsed your data into inputs and targets.

      If there are common datasets you’d like to try I could add them to the “Get Data” tab.

      • I did submit .csv files. When it did not work in Firefox I switched to Edge. Still no luck. I switched to a data set with no missing data (the one with missing data did not work) and still no luck – something about extra spaces – in the column titles I presume. Anyway, it was more work than I wanted since I have other ways to use deep learning algorithms.

        • Okay, sorry about that. I tried to make the data upload flexible, it will split your data on commas or spaces, but maybe I should let people select if they have comma or whitespace separated data, which would allow them to have entries that contain the delimiters.

Leave a Reply

Your email address will not be published. Required fields are marked *