The repository for all your experiments.

Cubyc automates experiment tracking, so you can focus on the real data science.

Run Experiments Track Functions Query Runs

from cubyc import Run

run = Run("github.com/team/project.git", params={"lr": 0.1}):

# Your experiment code here
...
run.log({"accuracy": 0.95})

from cubyc import Run

@Run.track("github.com/team/project.git")
def func(learning_rate):
    # Your experiment code here
    ...
    yield {"accuracy": 0.95}

SELECT 
    config.lr, 
    logs.accuracy
FROM 
    config
INNER JOIN 
    logs ON config.id = logs.id

Why Cubyc?

It's just Python
With a bit of SQL
Git for versioning and collaboration
And Automatic experiment tracking
Privacy for your data
While open-source and free

Bring Your Own Repository

Avoid data lock-ins and privacy risks. Store, track, and collaborate on your experiments with Git like you do with code.

Set Up Cubyc

Git comparison

Git graph

Automatic Versioning

Stop writing boilerplate code and config files. Automatically version and reproduce your experiment runs.

Start Tracking

SQL-Based Analysis

Break free from clunky UIs and dashboards. Query your runs with SQL and build custom reports using your favorite tools.

SQL Docs

Git comparison