Home

neptune.new

This documentation describes the new version of Neptune with an improved Python API and UI. Learn more about it from the release blogpost.

If you are looking for our legacy Neptune documentation, go here: https://docs-legacy.neptune.ai/

Neptune in 3 minutes

Try Neptune on Colab with zero setup and see results in the UI

Get a quick feel of how monitoring and keeping track of runs can look like.

What does Neptune do?

Neptune is a lightweight run management tool that helps you keep track of your machine learning runs.

Some of the most common Neptune use cases:

How does Neptune work?

Following snippets are just one example.

If you want more quick-start examples then go to quick starts.

Step 1: Installation

Install neptune-client. Check the docs for more help.

pip install neptune-client

Step 2: Prepare your training script

Create a new file main.py, then copy and paste the following script into it:

import neptune.new as neptune
run = neptune.init(project="your_workspace/your_project")
# Track metadata and hyperparameters of your run
run["JIRA"] = "NPT-952"
run["algorithm"] = "ConvNet"
params = {
"batch_size": 64,
"dropout": 0.2,
"learning_rate": 0.001,
"optimizer": "Adam"
}
run["parameters"] = params
# Track the training process by logging your training metrics
for epoch in range(100):
run["train/accuracy"].log(epoch * 0.6)
run["train/loss"].log(epoch * 0.4)
# Log the final results
run["f1_score"] = 0.67

Step 3: Run this example in your terminal

python main.py

You will get a link to Neptune Web UI, open it and take a look at how your runs are visualized.

Remember to set your API token before you run the script: export NEPTUNE_API_TOKEN="YOUR_API_TOKEN"

Instead of specifying the project as a parameter in neptune.init() you can also set it as an environment variable: export NEPTUNE_PROJECT="your_workspace/your_project"

Step 4: See everything in the UI

Check other quick-starts as well: quick starts.

Discover Neptune

  • Example project: See how example project looks in Neptune

  • YouTube channel: Provides hands-on videos that showcase key Neptune features.

  • Neptune blog: Provides in-depth articles about best practices in machine learning experimentation (among other things)

  • neptune-client: Neptune client is an open source Python library that lets you integrate your Python scripts with Neptune.

  • Questions? Send an email to [email protected] by email or click the chat icon in the bottom right corner.