Overview¶

The Comet R SDK, called cometr, is an open source library. The source code can be found at github.com/comet-ml/cometr. Any fixes or enhancements are welcome.

To install the stable CRAN version:

install.packages("cometr")

To install the latest development version from GitHub:

install.packages("remotes")
remotes::install_github("comet-ml/cometr")

See the README for more information on how to set up and use.

Quick setup¶

After you have aquired your Comet API key and done the one-time setup, then you are ready to dive into Machine Learning in R.

Specifically, you need to have defined your ~/.comet.yml file, which might look something like this:

COMET_WORKSPACE: my-comet-user-name
COMET_PROJECT_NAME: keras
COMET_API_KEY: my-comet-api-key

In R, you'll need to install the cometr package:

install.packages("cometr")

Now you are ready to begin running Machine Learning R scripts with Comet.

Comet machine learning examples in R¶

First, you might explore these Jupyter Notebooks that demonstrate Comet with R:

Comet-R-nnet.ipynb

Here are some basic R script examples to get you started:

Note that each example might require additional R packages. See the script's source for details.

After installing necessary packages, you can run a particular one using the following command (for example):

$ Rscript keras-example.R

Each Comet R Experiment has a unique URL, and all Experiments are grouped under your workspace or project. There, you can interactively create visualizations, analyze the performance of many models, share your results, and work with collaborators.

If you have any questions, consider these resources:

The rest of this page dives into the Comet R Experiment, the object at the heart of your machine learning.

Comet R Experiment¶

The Experiment object is where the logging of the machine learning code occurs.

Experiment

Comet R API¶

Functions for creating, deleting, and querying Comet.

Advanced functions¶

Apr. 11, 2024