We updated the Experiment Tab with tons of cool features. Filter & search experiments with granularity, drag-n-drop to reorder columns, and search for columns in the columns panel. We also added custom experiment names so you can name your favorite experiment shirley.
The home for data science collaboration
A community oriented data science platform for collaboration.
Based on Open Source tools and open formats.
Made with 🐕 for data scientists.
Get started. Fast.
Build your own ML project:
Start the tutorial
Frequently asked questions:
Learn more
See your new superpowers:
Example repo
Latest updates
New
📒 Notebook Diffing
This feature was requested by many of you, and I'm happy to say you can now compare your notebooks in a meaningful way, making it easy to understand what inputs and outputs changed. No need to manually parse the JSON file to understand what's going on in your notebooks.
New
🚀 Data Science Pull Requests
Data Science Pull Requests (DS PRs) are the natural extension of Pull Requests (PRs) to the realm of data science. You can now discuss, compare, and review changes to experiments, data, and models in your pull requests. After you finish your review, you can merge your data, models, and artifacts automatically.
"No one can whistle a symphony. It takes a whole orchestra to play it."
When people work smarter, and work together, they can build awesome things.
Data scientists are struggling with this.
The basis for frictionless collaboration is the ability to understand what others have done, and the ability to pick up where they left off.
DAGsHub is a web platform based on open source tools, optimized for data science and oriented towards the open source community.
It is a central location where projects can be hosted, discovered, and collaborated on.
Superpowers you get with DAGsHub
Collaborate effectively
Leverage the same best practices used in software engineering to get more done. DAGsHub helps
automate your workflow,...
so you can focus on work instead of coordination.
Share intermediate results from your pipeline with any collaborator, instead of re-running all the pre-processing code, and re-run only parts of the pipeline that you changed.
Ease of use also means it takes less time to get new team members and collaborators caught up and contributing.
Share intermediate results from your pipeline with any collaborator, instead of re-running all the pre-processing code, and re-run only parts of the pipeline that you changed.
Ease of use also means it takes less time to get new team members and collaborators caught up and contributing.
Reproduce results easily
Reproduce any experiment from previous versions of your project. It’s as simple as switching a branch and running a single command....
Experiment with different hyper parameters in parallel. Each run will automagically save its results in an organized and easy to access location.
Data & pipeline versioning
Maintaining giant spreadsheets of hyper parameters and inconsistent naming conventions are a thing of the past....
DVC tracks data & code versions. Every time you run a stage in your pipeline, your data is saved
automatically, so you don’t need to worry about accidentally ruining your data.
You can visualize your pipeline (as DAGs) and see how it changes over time, and get any version of your model with a click.
You can visualize your pipeline (as DAGs) and see how it changes over time, and get any version of your model with a click.
Simplicity with Open Source & open formats