What is Layer?
Layer is a collaborative machine learning platform where you can build, train, track and share your ML models. It helps you create production-grade ML pipelines with a seamless local<>cloud transition while enabling collaboration with semantic versioning, extensive artifact logging and dynamic reporting.
How to use Layer?
Add a @model decorator to your training function to train and register your model to Layer. This will help you manage the lifecycle of your model with model versioning and extensive experiment tracking.
from layer.decorators import model
model = ...
Add a @dataset decorator to your dataset creation function. Layer will semantically version your training data so that you can have reproducible ML pipelines.
from layer.decorators import dataset
df = ...
Layer Projects are purpose-built containers of your ML entities; datasets and models. You can think of them like git repos except they are specifically designed for machine learning.
Meta Store for Datasets and Models
You can register your datasets and models to share with your team mates. Anyone in your team can fetch and reuse the entity for their own project.
Extensive Artifact Logging
You can log your parameters, charts, metrics, plots to Layer Projects. Layer will help you compare them between different versions of datasets and models enabling experiment tracking and governence.
Share with your team or anyone!
You can share your project with your team or anyone on the internet. It enables you collaborate with anyone. See all of the public projects here.
Who should use Layer?
We built Layer to empower data teams. It can be used by:
- Data scientists to build, train machine learning models and track experiments.
- ML/Data engineers to streamline machine learning operations.