16.3. Getting Started
Thie document describes clara commands.
Create a model from a directory. The Model Directory must conform to TRITON server specifications.
16.3.1.1. Usage
$ clara create model -p [MODEL_DIRECTORY] -t [MODEL_TYPE] [--raw]
16.3.1.2. Flags
Name |
Shorthand |
Default |
Usage |
|---|---|---|---|
path |
p |
Model path |
|
type |
t |
Model Type {“tensorflow”, “tensorrt”, “pytorch”, “unknown”} |
|
raw |
r |
false |
Show output in raw format |
Lists information about models
16.3.2.1. Usage
$ clara list models [--quiet]
16.3.2.2. Flags
Name |
Shorthand |
Default |
Usage |
|---|---|---|---|
quiet |
q |
false |
Only show model ids |
Delete models.
16.3.3.1. Usage
$ clara delete models [MODEL_ID]...
16.3.3.2. Arguments
Name |
Default |
Usage |
|---|---|---|
MODEL_ID |
The full or partial unique identifier of the model. Partial unique identifiers can be used so long as it matches from left-to-right and uniquely identifies a model. |
All the model ids passed as argument are deleted.
Creates a job from a name, pipeline id, job priority and file or directory.
16.3.4.1. Usage
$ clara create job -n NAME -p PIPELINE_ID [-f [FILE | DIRECTORY]] [-i JOB_PRIORITY] [--start]
16.3.4.2. Flags
Name |
Shorthand |
Default |
Usage |
|---|---|---|---|
name |
n |
Name of the job |
|
pipeline |
p |
Pipeline id |
|
priority |
i |
normal |
Job Priority {“low”, “normal”, “high”, “immediate}” |
files |
f |
Filename or directory to files used to create the job |
|
start |
s |
false |
Start the job (without any arguments) after creating it |
raw |
r |
false |
Show output in raw format |
Starts a created job from job id.
16.3.5.1. Usage
$ clara start job -j JOB_ID [-a KEY=VALUE]...
16.3.5.2. Flags
Name |
Shorthand |
Default |
Usage |
|---|---|---|---|
args |
a |
Argument to the job in KEY=VALUE format |
|
job |
j |
Job id |
Creates a clara pipeline.
16.3.6.1. Usage
$ clara create pipeline -p PIPELINE_DEFINITION_FILE
16.3.6.2. Flags
Name |
Shorthand |
Default |
Usage |
|---|---|---|---|
path |
p |
Pipeline definition file path |
|
raw |
r |
false |
Show output in raw format |
Lists pipelines.
16.3.7.1. Usage
$ clara list pipelines
16.3.7.2. Flags
Name |
Shorthand |
Default |
Usage |
|---|---|---|---|
quiet |
q |
false |
Only show pipeline ids |
Prints definition of the pipeline.
16.3.8.1. Usage
$ clara describe pipeline -p PIPELINE_ID
16.3.8.2. Flags
Name |
Shorthand |
Default |
Usage |
|---|---|---|---|
pipeline |
p |
The full or partial unique identifier of the Clara pipeline. Partial unique identifiers can be used so long as it matches from left-to-right and uniquely identifies a pipeline. |
Lists jobs.
16.3.9.1. Usage
$ clara list jobs
16.3.9.2. Flags
Name |
Shorthand |
Default |
Usage |
|---|---|---|---|
quiet |
q |
false |
Only show job ids |
all |
false |
List all jobs including canceled ones |
|
canceled |
false |
List only canceled jobs |
|
faulted |
false |
List only faulted jobs |
Submit reuqest to cancel job(s).
16.3.10.1. Usage
$ clara cancel jobs [-j JOB_ID1,JOB_ID2...] [--all]
$ clara cancel jobs [-j JOB_ID1,JOB_ID2...] [--all]
16.3.10.2. Flags
Name |
Shorthand |
Default |
Usage |
|---|---|---|---|
all |
false |
Cancel all jobs |
16.3.10.3. Arguments
Name |
Default |
Usage |
|---|---|---|
JOB_ID |
The full or partial unique identifier of the Clara job. Partial unique identifiers can be used so long as it matches from left-to-right and uniquely identifies a job. |
Uploads payloads
16.3.11.1. Usage
$ clara upload payload -p PAYLOAD_ID -f {FILE | DIRECTORY}
16.3.11.2. Flags
Name |
Shorthand |
Default |
Usage |
|---|---|---|---|
files |
f |
Payload file or directory to read from |
|
payload |
p |
Payload identifier |
Delete payloads.
16.3.12.1. Usage
$ clara delete payloads -p PAYLOAD_ID1[,PAYLOAD_ID2...]
$ clara delete payload -p PAYLOAD_ID1[,PAYLOAD_ID2...]
16.3.12.2. Arguments
Name |
Default |
Usage |
|---|---|---|
PAYLOAD_ID |
Full unique identifier of the clara payload. |
Download files associated with the selected job.
16.3.13.1. Usage
$ clara download JOB_ID:SRC_PATH [DEST_PATH] [flags]
16.3.13.2. Arguments
Name |
Default |
Usage |
|---|---|---|
JOB_ID |
The full or partial unique identifier of the Clara job from which to download content from. Partial unique identifiers can be used so long as it matches from left-to-right and uniquely identifies a job. |
|
SRC_PATH |
Source path of the file(s) to download. File names are relative to the root of the payload. Supports simple glob style pattern matching. |
|
DEST_PATH |
|
Optional destination path where the files will be downloaded. By default file(s) will be downloaded to the current directory. |
Prints the information and status of the job.
16.3.14.1. Usage
$ clara describe job -j JOB_ID
16.3.14.2. Flags
Name |
Shorthand |
Default |
Usage |
|---|---|---|---|
job |
j |
The full or partial unique identifier of the Clara job from which to download content from. Partial unique identifiers can be used so long as it matches from left-to-right and uniquely identifies a job. |
Configures NGC credentials for use by CLI.
16.3.15.1. Usage
If –orgteam is not specified, the orgteam defaults to ‘nvidia/clara’ which points to the publicly available Clara NGC artifacts
$ clara config [flags]
16.3.15.2. Flags
Name |
Shorthand |
Default |
Usage |
|---|---|---|---|
key |
NGC API Key |
||
orgteam |
nvidia/clara |
Org and team (input format: “{org}[/{team}]” team is optional) |
|
yes |
y |
false |
Bypass the prompt |
Read clara job operator logs.
16.3.16.1. Usage
$ clara logs [flags]
16.3.16.2. Flags
Name |
Shorthand |
Default |
Usage |
|---|---|---|---|
job |
j |
Clara Job Id |
|
operator |
o |
Clara Job Operator |