Modeling view

The Modeling view lets you build a deep learning model yourself or work with a colleague on the same model. There are no limits to how you can build the model - it’s as easy to make a simple one as to make a complex, multi-layered one.

The Modeling view consists of two pages, the Modeling canvas, and the Experiments overview.

No coding
There is no coding needed, just drag and drop to build your model. No coding means it’s simpler and faster but also fewer bugs.

Modeling canvas

In the Modeling canvas you’ll find everything you need to build a deep learning model in one place. You can drill down into an experiment, but there is no dense information that is hard to digest. Only the relevant information is seen.

You’ll see your model on the canvas and all settings on the side, with Run settings, Dataset settings, and all building blocks that you need to build an AI model as well as pre-built deep learning architectures called pretrained blocks.

You can switch between experiments in the drop-down or go back to the Experiments overview by clicking the Experiments overview button.

Experiments overview

The Experiments overview consists of a grid (or table) view where you see all created experiments in the current project.

The experiments are by default grouped by status; created, running, failed or completed.

You can collapse the groups to make the overview less cluttered, which is good if you’ve got a lot of experiments.

Experiment cards

On all completed and running experiments, you can see a loss graph. Look at it to check if the curve converges properly. The graph also makes it easy to compare eperiments quickly.

The card also includes info on Creator, Dataset, Dataset version, Loss function, and tags (if you have tagged the experiments).

Filter out experiments

You can filter out experiments you want to see based on:

Sort your experiments

You can sort your experiments by their status or time created.

Was this page helpful?
Yes No