Experiment tracking hub bringing organization and collaboration to data science projects.
No credit card required, free for individuals and teams, takes 5 minutes to get started.
#1. PRODUCT MANIFESTO
We work with the tools data scientists use and don’t change the way they do things. Data Science's inherent messiness is fundamental to creativity.
We put a lot of attention into designing the UI and UX of what we build.
Data scientists use Neptune to store their ideas and results. We make sure they are backed up and secured.
#2. HOW IT WORKS
You can record the metrics, data versions, data exploration notebooks, hyper parameters, images or even trained machine learning models and prediction distribution charts. Your code is saved automatically, whether you remembered to version it or not.
You can experiment at any pace without ever losing the control over the process. You do your magic, Neptune will take care of the rest.
# Track your model training from Python with Neptuneimport neptune
neptune.send_metric(’auc’, score) neptune.send_image(’roc_auc_curve’, PIL_image) neptune.send_artifact(’my_model.h5’)
# Host runs and notebooks at Neptune
neptune@ubuntu:~$ neptune sync –with tensorboard –dir logdir
neptune@ubuntu:~$ neptune sync –with mlflow –dir mlruns
neptune@ubuntu:~$ neptune sync –with jupyter –filepath data_exploration.ipynb
We will not tell you how to do things, you know how you like to work. With minimal overhead on your side Neptune will help you organize all the artefacts related to your projects in a way designed specifically for the data science domain.
Give different roles to different people in the organisation and assign them to selected projects.
Since everything related to your work is in one place, every piece of work can be shared and discussed. Any change is visible to everyone instantly once created. You can collaborate on new ideas, give feedback, spot bugs and gain insights quickly, discuss and comment learning curves.
Invite other people to collaborate in your projects or decide to work individually. Make your work public and show results of your work to the world or keep it private. It up to you to choose.
Neptune is loved by teams and individuals
Symmetrical is speeding up the machine learning model training process thank to the process organization.
At Symmetrical we develop a marketplace for financial products by connecting consumers to the right products in real-time. We discovered neptune.ml to be an invaluable tool for running machine learning experiments in an extremely fast manner. Even more importantly, we learned that by using it we can significantly shorten the amount of time we spend on setting up and running experiments. We are constantly working on improving the way in which we conduct data science experiments and neptune.ml has a big chance of becoming the must-have in our pipeline for developing, testing and fine-tuning algorithms.
Collaboration is enabling deepsense.ai to build top quality machine learning models for their clients.
At deepsense.ai we provide machine learning and deep learning solutions and consultancy for market leaders such as BCG, IBM, Juniper, EY, Nielsen, NVIDIA, L'Oréal. Using Neptune enables us to cooperate more closely with our customers and eliminate most of the problems related to project communication. We limited the number of meetings to the ones that were really necessary to make strategic decisions since clients have real-time access to what we do and we collaborate on a regular basis. When everything is tracked and organized in the one knowledge center we don’t need to create much additional documentation for our clients. We also realized that with Neptune we were able to deliver our projects faster and we optimized the time spent onboarding, which we found very beneficial.
up to 3 collaborators per private project
Would you like to deploy Neptune on your cloud or own infrastructure?
#5. DATA SECURITY
User Data at Rest
User Data in Transit
All traffic over open network, including any data transfers, is encrypted.