Guild AI automates TensorFlow and Keras deep learning workflow, letting you focus on optimizing your models and getting them into production as quickly as possible. The command line toolset, supports your deep learning work end-to-end — from model acquisition, through training and testing, to deployment. You can even publish your models for your collaborators to use!
Guild AI automatically records each training experiment, including saved models, logs, generated artifacts, and run metadata. Use this information to compare runs and inform your next steps. Is your model ready to release, or do you need to train with more data or different hyperparameters? The information you need to make smart decisions is always on hand!
Guild AI lets you effortlessly compare training results. Along with model accuracy and loss, Guild AI provides information essential to making informed comparisons including hyperparameters, model architecture, training steps, and data. Guild even lets you compare generated artifacts like images and audio to help you assess each outcome.
When your model is ready to release, Guild AI has a command that makes it easy. If you're model is destined for production, you can seamlessly deploy to Cloud Machine Learning Engine. If you want to share your results with colleagues, you can package and publish your model so that others can easily reproduce and learn from your work.
Guild AI automatically saves each training run so you don't have to worry about losing valuable results. Everything associated with a run is available for study and use — saved models, TensorFlow event logs, generated content, and run metadata.
Compare model performance at-a-glance
Guild AI makes it easy to compare model accuracy and loss alongside essential metadata like model architecture, training steps, and hyperparameters. With this information you can make informed decisions about how to improve your model and whether it's ready for release.
Visualize runs with TensorBoard
Guild AI seamlessly integrates TensorBoard for visualization of TensorFlow event logs. Use TensorBoard to compare loss and accuracy at recorded steps, visualize model statistics, images, and audio, and explore graph definitions.
Enhance your Keras training
If you develop your models using Keras, Guild AI provides enhanced features that let you train models with automatic TensorFlow event logging (used to visualize your training in TensorBoard) and automatic experiment tracking without modifying your Keras source code.
Explore generated content
Guild AI lets you quickly explore content generated from your models, including images and audio. Because Guild saves each experiment, including those that generate files, you can easily compare model performance across runs.
Train and predict on Cloud ML
Guild AI makes it easy to train and run your models on Google's Cloud Machine Learning Engine (Cloud ML). Train your model locally for proof-of-concept and then, to leverage the performance of distributed training on Google's advanced hardware, move your training to the cloud with a single command.
Package and distribute your models
Guild AI lets you package and share your models with colleagues and other deep learners. Making it easy to discover, install, and use your models, encouraging feedback, which can lead to unexpected insight and improvements.
If you’d like to try Guild AI, it takes just five commands to install Guild AI, train a model, and visualize your results. For more information on installing and using Guild AI, see our installation guide.
1. Install Guild AI
pip install guildai
2. Initialize Guild AI
3. Install a model package
guild install mnist
4. Train a model
guild train mnist-softmax
5. View the results
Go deeper with Guild AI
Go deeper into the Quick Start material above with a step-by-step tutorial on training MNIST models and comparing results.Go deeper with Guild AI
Discover the models
Guild AI supports an ever-growing ecosystem of TensorFlow and Keras models that you can install and train with a few simple commands.Discover the models
Browse the docs
If you're interested in a complete picture of Guild AI, start by browsing its comprehensives documentation.Browse the docs