Comet.ML is the GitHub for Machine Learning Models
Overview
- Comet.ml lets you track your machine learning models, code, experiments and even hyperparameters
- You only need to add the tracking code to your preferred tool
- It supports all the popular tools and libraries like R, python, TensorFlow, Keras, among others
Introduction
GitHub has gained unparalleled popularity over the years for it’s amazing flexibility in allowing teams to collaborate and contribute to projects.
Along the same lines comes Comet.ML, a tool that enables data scientists and machine learning practitioners to automatically track their machine learning code, experiments, hyperparameters, and model results. It additionally creates a graph of all the results which makes it easier to visualize and compare all your experiments.
It has been released today from beta and is available to use for everyone.
Source: Medium
You can integrate Comet into your ML pipeline fairly easily. To get started, you only need to add the Comet.ml tracking code to whatever tool you’re using for building the model. One of the best parts about this service is that it doesn’t matter where you train your model (it could be on your machine, on the cloud, etc.).
It allows the user to choose from any of the below tools and libraries:
Comet works with GitHub and other git service providers. Once you have run your experiments and finalized your best model, you can generate a pull request straight to your GitHub repository.
To install it on your machine via pip, follow the below command, depending on your version of python:
pip install comet_ml pip3 install comet_ml
Get a high level overview of how Comet.ml captures your model’s results and visualizes it below:
There are different services available depending on your requirements. They range from a free version (the most popular option) to a business version. Check out and sign up for Comet.ml here!
Our take on this
We are loving this release from Comet.ml! As a data scientist, you don’t need to change the tool you’re working on or the process you usually follow. Instead, you have a supplementary service that let’s you see deeper into how well your experiments are going by comparing model to model.
Tuning the hyperparameters of your model to get better results, or to compare it to previous models, is a tedious process. Comet.ml helps immensely in that aspect as well.
It’s a bit like Tensorboard with code and hyperparameter tracking. Go ahead and use it today and let us know your experience in the comments below!
Subscribe to AVBytes here to get regular data science, machine learning and AI updates in your inbox!
Senior Editor at Analytics Vidhya. Data visualization practitioner who loves reading and delving deeper into the data science and machine learning arts. Always looking for new ways to improve processes using ML and AI.
