{"id":4576,"date":"2022-11-10T17:50:11","date_gmt":"2022-11-11T01:50:11","guid":{"rendered":"https:\/\/live-cometml.pantheonsite.io\/?p=4576"},"modified":"2025-04-24T17:16:30","modified_gmt":"2025-04-24T17:16:30","slug":"new-integration-comet-and-ray","status":"publish","type":"post","link":"https:\/\/www.comet.com\/site\/blog\/new-integration-comet-and-ray\/","title":{"rendered":"New Integration: Comet and Ray"},"content":{"rendered":"\n
\"\"\/<\/figure>\n\n\n\n

<\/p>\n\n\n\n

\n

We\u2019re excited to announce another excellent integration with Comet \u2014 Ray<\/a>! This integration allows data scientists to leverage Comet\u2019s experiment tracking and visualization tools with Ray\u2019s incredible library for scaling compute intensive Machine Learning workloads.<\/p>\n

About Ray<\/h1>\n

Ray<\/a> is an open source project that makes it simple to scale any compute-intensive Python workload \u2014 from deep learning to production model serving.<\/p>\n

With a rich set of libraries for Hyperparameter Tuning<\/a>, Reinforcement Learning<\/a>, Model Serving<\/a> and integrations built on a flexible distributed execution framework, Ray makes distributed computing easy and accessible to every engineer.<\/p>\n

About the Integration<\/h1>\n

Ray now ships with a CometLoggerCallback<\/code>. You can now log metrics, hyperparameters, source code, and much more from your Ray Trials to the Comet UI.<\/p>\n

Once you\u2019ve set up your account and configured your Comet API Key<\/a> within your project, you install Ray Tune and Comet using pip<\/code><\/p>\n

pip install comet_ml\npip install \"ray[tune]\"<\/span><\/pre>\n
Configure your project with your Comet credentials.<\/p>\n
export COMET_API_KEY=\"Your API Key\"\nexport COMET_WORKSPACE=\"Your Workspace Name\"\nexport COMET_PROJECT_NAME =\"Your Project Name\"<\/span><\/pre>\n
Once you have set up your credentials, simply import the CometLoggerCallback<\/code> from Ray!<\/p>\n
The following code snippet shows how you can easily log Ray Trials to the Comet UI.<\/p>\n
\"\"\"Examples logging Tune runs to comet.ml\"\"\"\nimport comet_ml\nimport argparse\nimport numpy as np\n\nfrom ray import tune\nfrom ray.tune.integration.comet import CometLoggerCallback\n\n\ndef train_function(config, checkpoint_dir=None):\n    for i in range(30):\n        loss = config[\"mean\"] + config[\"sd\"] * np.random.randn()\n        tune.report(loss=loss)\n\n\ndef tune_function():\n    analysis = tune.run(\n        train_function,\n        name=\"comet-ray-demo\",\n        metric=\"loss\",\n        mode=\"min\",\n        callbacks=[\n            CometLoggerCallback(\n                tags=[\"my-trial\"]\n            )\n        ],\n        config={\"mean\": tune.grid_search([1, 2, 3]), \"sd\": tune.uniform(0.2, 0.8)},\n    )\n    return analysis.best_config\n\nbest_config = tune_function()<\/pre>\n
Your runs will now be reported by Comet. You\u2019ll then be able to take advantage of Comet\u2019s rich visualization and analysis capabilities<\/a> to gain insights from your Ray experiments.<\/p>\n
\n
\n
\"\"<\/figure>
Ray Tune Runs Logged to Comet<\/picture><\/div>\n<\/div>\n<\/figure>\n
Getting Started<\/h1>\n
Getting started with this integration is really easy and the following resources should help you start logging Ray Trials to Comet in no time!<\/p>\n