Build better models faster
Start new thread
Gideon Mendels

Comet.ml - Automatically track machine learning code and experiments

by

Comet.ml allows machine learning teams to automatically track their machine learning code, experiments, hyperparameters and results. It works from every computer, whether it’s your laptop or a your favorite cloud provider.

Add a comment

Replies

Best
Roy Hermann Ueland

Love this product! The team has done a really good job on helping encapsulate a lot of the problems that ML engineers have. Already done very well - looks very promising for the future of machine learning. Congrats!

Pros:

- Slick Interface

- Super useful features

- Easy to integrate

Cons:

• Wish this was out sooner!

Jacqueline von Tesmar
Hey Makers! Can you tell us more about some of the highlights of what you've built here?
Gideon Mendels
Maker
@jacqvon Hey Jacqueline, We built Comet.ml after working at Machine Learning teams at both big tech companies (Google/Wix.com), startups and academia. We realized that managing machine learning workflow is really painful as there are no dedicated tools to do this. Comet works by connecting to your current workflow and then version, monitor and optimize your machine learning models. It works whether you're a student doing homework or a company with a huge cluster. Happy to answer any questions!
Mark Russell Filaroski
Can I get a demo? mark@compliancePT.com if Yes. Early Healthcare customer potential for you all. Thank you.
Gideon Mendels
Maker
@compliancept @mark_filaroski Sure Mark! I'll send you an email.
Alon E

We've been using the product for the last couple months on beta and it completely changed the way we approach and develop anything related to data.

Comet.ml team is super responsive and helped us with any support inquiry

Pros:

One liner integration with all popular python ML libraries.

Must have in every data team

Cons:

No support for R language

Amir Coren

This is such a great product! Highly recommended, as it extremely easy to use.

Pros:

Really easy to integrate.

Cons:

Limited tracking on production models

Fabio Calefato
@amir_coren Hi, do you mind elaborating more on your comment of Limited tracking on production models. Maybe give an example. Thanks for your time and help!
Tomer Aharoni

We have been using Comet for only a day now to visualize our training data for an autonomous drone hunter. It's been a game changer for us in the way we understand our data and our models. Also, it saves us a lot of time because now, instead of asking specific people questions, we can just look on the dashboard. We obviously need to play with comt a bit more before we can have real feedback on it but so far it works amazing and seems like a product that team like ours can't leave without.

Pros:

- Easy integration on multiple environments

- Great dashboard

- Responsive on Slack for any questions/problems

Cons:

- No mobile app

- Documentation could be a bit more polished

Aviran Mulain

What an awesome product! This is a remarkable development in machine learning engineering and so far it does a fantastic job for my team and myself. Totally recommended!

Pros:

Got a model training in 1m!

Cons:

Can't find any

Eden Dolev

Shared this with the data science and MLE teams in my company and they love it. Hard to imagine how we didn't use anything like it until now.

Pros:

straightforward installation and integration, makes tracking and comparing experiments very easy

Cons:

Not open source

Ðietrich 🇪🇺🥨
How is this different to the already established (and open source) OpenML ( https://github.com/openml/OpenML ) ?
Yuecheng Zhu
@trtm From looking at Github, it looks like the only relationship is the letters "ML".
Gideon Mendels
Maker
@trtm Overall OpenML seems to run your models for you. We allow you to train your models anywhere you want (keeping your sensitive data private).