Skip to content

checkpoint inspect

Returns detailed information about a Checkpoint: its inception time, metadata, status, file content list, and more. It outputs the information in YAML format by default.

Terminal window
flexai checkpoint inspect <checkpoint_name> [--json]
Required

Name of the Checkpoint to inspect.

Examples
  • mistral-500-checkpoint
--json
<boolean>
Optional
Flag

Output the information in JSON format.

Examples
  • --json
Terminal window
flexai checkpoint inspect mistral-500-checkpoint
FieldDescriptionData Type
kindThe type of resourceString
metadataMetadata about the checkpointObject
metadata.nameThe name of the checkpointString
metadata.idThe unique identifier of the checkpointString (UUID)
metadata.creatorUserIDThe user ID of the checkpoint creatorString (UUID)
metadata.ownerOrgIDThe organization ID that owns the checkpointString (UUID)
specCheckpoint contents detailsObject
spec.fromLocalFilesA list with the paths of the files used to create this checkpointString List
spec.storageProviderThe name of the Remote Storage Provider Connection, if anyString
spec.sourcePathThe path to the bucket and file or directory on the Remote Storage Provider Connection, if anyString
statusStatus information of the checkpointObject
status.statusThe current status of the checkpointString
status.storageProviderIDThe ID of the Remote Storage Provider Connection used to upload the checkpoint, if anyString (UUID)
status.sizeThe total size of the checkpointString (File Size)
status.filesA list of files with their paths and sizesObject List
status.files.pathThe path of the file within the checkpointString (File Path)
status.files.sizeThe size of the fileString (File Size)
status.metadata.nameThe human-readable name of the checkpointString
status.metadata.nodeThe Node where the checkpoint was createdString
status.metadata.stepThe training step at which the checkpoint was createdNumber
status.metadata.trainLossThe training loss at the time the checkpoint was createdNumber
status.metadata.evalLossThe evaluation loss at the time the checkpoint was createdNumber
status.metadata.modelThe name of the base model used in the checkpointString
status.metadata.versionThe version of the model used in the checkpoint, such as 4.53.2String
status.metadata.inferenceReadyIndicates whether the checkpoint is ready for inference or not. Meaning it includes all necessary files and metadata required for inference to tasksBoolean
status.createdAtThe timestamp when the checkpoint was createdString (ISO 8601)
status.updatedAtThe timestamp when the checkpoint was last updatedString (ISO 8601)
kind: Checkpoint
metadata:
name: test-nanogpt-run-1
id: 431a0ecb-cd8f-4508-8ae3-0990833a4f16
creatorUserID: bd67af19-2599-4a57-832e-a1ac042f48be
ownerOrgID: 270a5476-b91a-442f-8a13-852ef7bb5b94
spec:
fromLocalFiles:
- output/checkpoint_20250203_130043/ckpt.pt
storageProvider: ""
sourcePath: ""
status:
status: available
files:
- path: ckpt.pt
size: 343.79 MB
metadata:
name: ""
node: ""
step: 0
trainLoss: 0
evalLoss: 0
model: ""
version: ""
inferenceReady: false
storageProviderID: 00000000-0000-0000-0000-000000000000
size: 343.79 MB
createdAt: "2025-02-03T13:14:37.730471Z"
updatedAt: "2025-02-03T13:14:54.591148Z"