Neptune + Keras integration

Experiment tracking for Keras-trained models.

What will you get with this integration?

• Log, organize, visualize, and compare ML experiments in a single place
• Monitor model training live
• Version and query production-ready models and associated metadata (e.g., datasets)
• Collaborate with the team and across the organization

What will be logged to Neptune?

• hyperparameters for every run,
• learning curves for losses and metrics during training,
• hardware consumption and stdout/stderr output during training,
• TensorFlow tensors as images to see model predictions live,
• training code and Git commit information,
• model weights,
• other metadata

Example charts in the Neptune UI with logged accuracy and loss

Resources

• Documentation
• Code example on GitHub
• Runs logged in the Neptune app
• Run example in Google Colab

Example

On the command line:

pip install neptune-tensorflow-keras

In Python:

import neptune
from neptune.integrations.tensorflow_keras import NeptuneCallback

# Start a run
run = neptune.init_run(
    project="common/tf-keras-integration",
    api_token=neptune.ANONYMOUS_API_TOKEN,
)

# Create a NeptuneCallback instance
neptune_cbk = NeptuneCallback(run=run)

# Pass the callback to model.fit()
model.fit(
    ...,
    callbacks=[neptune_cbk],
)

# Stop the run
run.stop()

Support

If you got stuck or simply want to talk to us, here are your options:

• Check our FAQ page
• You can submit bug reports, feature requests, or contributions directly to the repository.
• Chat! When in the Neptune application click on the blue message icon in the bottom-right corner and send a message. A real person will talk to you ASAP (typically very ASAP),
• You can just shoot us an email at support@neptune.ai