Working with checkpointsΒΆ

TODO: Explain the rationale behind checkpoints, and expand each section.

List the checkpoints available on the system:

$ lumi checkpoint list
================================================================================
|           id |                  name |       alias | source |         status |
================================================================================
| 48ed2350f5b2 |   Faster R-CNN w/COCO |    accurate | remote | NOT_DOWNLOADED |
| e3256ffb7e29 |      SSD w/Pascal VOC |        fast | remote | NOT_DOWNLOADED |
================================================================================

Inspect a checkpoint:

$ lumi checkpoint info accurate
Faster R-CNN w/COCO (48ed2350f5b2, accurate)
Base Faster R-CNN model trained with the full COCO dataset.

Model used: fasterrcnn
Dataset information
    Name: COCO
    Number of classes: 80

Creation date: 2018-03-21T20:04:59.785711
Luminoth version: v0.1.0

Source: remote (NOT_DOWNLOADED)
URL: https://github.com/tryolabs/luminoth/releases/download/v0.0.3/48ed2350f5b2.tar

Refresh the remote checkpoint index:

$ lumi checkpoint refresh
Retrieving remote index... done.
2 new remote checkpoints added.

Download a remote checkpoint:

$ lumi checkpoint download accurate
Downloading checkpoint...  [####################################]  100%
Importing checkpoint... done.
Checkpoint imported successfully.

Create a checkpoint:

$ lumi checkpoint create config.yml -e name='Faster R-CNN with cars' -e alias=cars
Creating checkpoint for given configuration...
Checkpoint b5c140450f48 created successfully.

Edit a checkpoint:

$ lumi checkpoint edit b5c140450f48 -e 'description=Model trained with COCO cars.'

Delete a checkpoint:

$ lumi checkpoint delete b5c140450f48
Checkpoint b5c140450f48 deleted successfully.

Export a checkpoint into a tar file, for easy sharing:

$ lumi checkpoint export 48ed2350f5b2
Checkpoint 48ed2350f5b2 exported successfully.

Import a previously-exported checkpoint:

$ lumi checkpoint import 48ed2350f5b2.tar