Model output

MLReef pipelines

In order to understand how and where MLReef places your model output, it makes sense to first understand the general pipeline flow within MLReef.

When you start an experiment pipeline (as a matter of fact, any pipeline in MLReef behaves very similarly), a machine will be spawned

Model output path & files

When you develop your custom models, you can decide what the output can or should look like. There are no restrictions or limitations set by MLReef. In most cases, we propose the following outputs:

  1. Log files
  2. Model binary file

You have to set up the output-path in the parameters of your code repository to save the files after the pipeline finishes.

The output files are saved as artifacts you can download them. Go to the tab experiments, you can find your experiment in the table, click it. Then you can see the details of that particular experiments, you should see in the left menu, Details, Training,Files, click in Files and then click the button Download, wait for a minute and you will receive a zip file named artifacts.zip.