registry¶
Registry Commands
usage: ngc registry [--debug] [--format_type <fmt>] [-h]
{chart,collection,csp,image,label-set,model,resource,usage}
...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
registry¶
- {chart,collection,csp,image,label-set,model,resource,usage}
Possible choices: chart, collection, csp, image, label-set, model, resource, usage
Sub-commands¶
chart¶
Helm Chart Commands
ngc registry chart [--debug] [--format_type <fmt>] [-h]
{info,list,publish,pull,update-license-terms} ...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
chart¶
- {info,list,publish,pull,update-license-terms}
Possible choices: create, info, list, publish, pull, push, remove, update, update-license-terms
Sub-commands¶
list¶
List charts.
ngc registry chart list [--access-type <access_type>] [--column <column>]
[--debug] [--format_type <fmt>]
[--product-name <product_name>] [-h]
[<target>]
Positional Arguments¶
- <target>
Filter the search by allowing wildcards for charts. Format: org/[team/]chart_name[:version]. To target charts use 'org/[team/]chart_name'. Both name and version support the wildcards '*' and '?'. Version also supports character expressions ([a-z], [!ab], etc.). Examples: 'my_org/my_chart' - target my_chart in my_org repository. 'my_org/my_team/my_chart' - target my_chart in my_org/my_team repository. 'my_org/my_team/' - target all charts in my_org/my_team repository. 'my_org/my_chart' - target all chart versions for my_chart in my_org namespace. 'my_org/my_chart:[1-5]' - target versions 1-5 for my_chart in my_org namespace.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --column
Specify output column as column[=header], header is optional, default is repository[=Repository] or artifactVersion[=Version] for versions. Valid columns are created[=Created], createdBy[="Created By"], description[=Description], displayName[="Display Name"], guestAccess[="Guest Access"], labels[=Labels], name[=Name], org[=Org], public[=Public], size[=Size], team[=Team], updated[="Last Modified"], version[=Version]. Use quotes with spaces. Multiple column arguments are allowed.
- --access-type
Possible choices: EXCLUSIVE, LISTED, NOT_LISTED
Filter the list of resources to only resources that have specified access type.
- --product-name
Filter the list of resources to only resources that are under the product name. Multiple product-name arguments are allowed. Choose from: nvaie-vpaif, nv-cloud-functions, secrets-manager, base-command-platform, nv-ai-enterprise, nvaie-igx-safety-add-on, chipnemo-inference, ai-foundations, artifact-catalog, riva-virtual-assistant, nemo-llm-service, monai, tao, omniverse-cloud, picasso, data-services, bluefield-enterprise-sw, nv-quantum-cloud, nvaie-igx-yocto-add-on, fleet-command, forge, riva-enterprise, cuopt, met, bionemo-service, iam, nvaie-igx, e2, private-registry
info¶
Retrieve metadata for a chart or chart version.
ngc registry chart info [--debug] [--files] [--format_type <fmt>] [-h]
<target>
Positional Arguments¶
- <target>
Chart or chart version. Format: org/[team/]chart_name[:version].
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --files
List files in addition to details for a version.
update¶
Update a chart or chart version.
ngc registry chart update [--add-label <add-label>] [--built-by <builtBy>]
[--debug] [--display-name <dispName>]
[--format_type <fmt>] [--label <label>]
[--label-set <label-set>] [--logo <logo>]
[--overview-filename <path>]
[--publisher <publisher>]
[--remove-label <remove-label>]
[--short-desc <shortDesc>] [-h]
<target>
Positional Arguments¶
- <target>
Name of the chart to create. Format: org/[team/]chart_name.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --overview-filename
Overview. Provide the path to a file that contains the overview for the chart.
- --display-name
The name to display for the chart
- --add-label
Label for the chart to add. Can be used multiple times.Imperative label argument, not to be used with declarative label arguments --label or --label-set
- --remove-label
Label for the chart to remove. Can be used multiple times.Imperative label argument, not to be used with declarative label arguments --label or --label-set
- --label
Label for the chart to declare. Can be used multiple times.Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --label-set
Name of the label set for the chart to declare. Can be used multiple times. Format: org/[team/]name. Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --short-desc
A brief description of the chart
- --built-by
The entity responsible for building the chart
- --publisher
The entity responsible for creating the chart
- --logo
The URL of the image to set as the logo for the chart
remove¶
Remove a chart from the repository.
ngc registry chart remove [--debug] [--format_type <fmt>] [-h] [-y] <target>
Positional Arguments¶
- <target>
Chart or chart version. Format: org/[team/]chart_name[:version].
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- -y, --yes
Automatically say yes to all interactive questions.
pull¶
Download a chart version.
ngc registry chart pull [--debug] [--dest <path>] [--format_type <fmt>] [-h]
<target>
Positional Arguments¶
- <target>
Chart version. Format: org/[team/]chart[:version]. If no version specified, the latest version will be targeted.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --dest
Provide a destination to download the chart. Default: . (current directory)
create¶
Create a chart's metadata.
ngc registry chart create [--built-by <builtBy>] [--debug]
[--display-name <dispName>] [--format_type <fmt>]
[--label <label>] [--label-set <label-set>]
[--logo <logo>] [--overview-filename <path>]
[--publisher <publisher>] --short-desc <shortDesc>
[-h]
<target>
Positional Arguments¶
- <target>
Name of the chart to create. Format: org/[team/]chart_name.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --overview-filename
Overview. Provide the path to a file that contains the overview for the chart.
- --display-name
The name to display for the chart
- --label
Label for the chart to declare. Can be used multiple times.Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --label-set
Name of the label set for the chart to declare. Can be used multiple times. Format: org/[team/]name. Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --built-by
The entity responsible for building the chart
- --publisher
The entity responsible for creating the chart
- --logo
The URL of the image to set as the logo for the chart
Required named arguments¶
- --short-desc
A brief description of the chart
push¶
Push (upload) a chart.
ngc registry chart push [--debug] [--dry-run] [--format_type <fmt>]
[--source <path>] [-h]
<target>
Positional Arguments¶
- <target>
Chart version. Format: org/[team/]chart_name:version. Alternative format using file name: org/[team/]chart_name-1.2.3.tgz
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --source
The path to the directory containing the packaged chart. If not specified, the chart will be uploaded from the current directory.
- --dry-run
List file paths, total upload size and file count without performing the upload.
update-license-terms¶
Update an artifact's license terms.
ngc registry chart update-license-terms [--clear] [--debug]
[--format_type <fmt>]
[--license-terms-file <filename>] [-h]
org/[team/]chart[:version]
Positional Arguments¶
- org/[team/]chart[:version]
The the target chart and version you want to publish to. Optional when getting publishing status using the --status flag. Format: org/[team/]chart[:version]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --license-terms-file
Publish the object with a specific license term defined in JSON file. File format: [{'licenseId': <id>, 'licenseVersion': <version>,'needsAcceptance': true/false,'governingTerms': <text>}]
- --clear
Whether to clear an artifact's license terms.
publish¶
Publish a chart from the NGC chart registry to catalog. If no version is provided, the latest is assumed.
ngc registry chart publish [--access-type <access_type>] [--allow-guest]
[--debug] [--discoverable] [--format_type <fmt>]
[--license-terms-file <filename>] [--metadata-only]
[--nspect-id NSPECT_ID]
[--product-name <product_name>] [--public]
[--source org/[team/]chart[:version]]
[--status <workflow_id>] [--version-only]
[--visibility-only] [-h]
[org/[team/]chart[:version]]
Positional Arguments¶
- org/[team/]chart[:version]
The the target chart and version you want to publish to. Optional when getting publishing status using the --status flag. Format: org/[team/]chart[:version]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --source
The source chart and version you want to publish. Format: org/[team/]chart[:version]
- --metadata-only
Only perform a shallow copy of the metadata instead of a deep copy of the objects referenced
- --version-only
Only copy the specified version of the object without copying any metadata
- --visibility-only
Only change the visibility qualities of the target. Metadata and version files are not affected.
- --allow-guest
Allow anonymous users to download the published object
- --discoverable
Allow the published object to be discoverable in searches
- --public
Allow access to the published object by everyone instead of just those with specific roles
- --product-name
Publish the object under a Product. Choose from: nvaie-vpaif, nv-cloud-functions, secrets-manager, base-command-platform, nv-ai-enterprise, nvaie-igx-safety-add-on, chipnemo-inference, ai-foundations, artifact-catalog, riva-virtual-assistant, nemo-llm-service, monai, tao, omniverse-cloud, picasso, data-services, bluefield-enterprise-sw, nv-quantum-cloud, nvaie-igx-yocto-add-on, fleet-command, forge, riva-enterprise, cuopt, met, bionemo-service, iam, nvaie-igx, e2, private-registry
- --status
Get the status of publishing based on provide workflow id.
- --access-type
Possible choices: EXCLUSIVE, LISTED, NOT_LISTED
Publish the object with a specific access type. Choose from: LISTED, EXCLUSIVE, NOT_LISTED
- --license-terms-file
Publish the object with a specific license term defined in JSON file. File format: [{'licenseId': <id>, 'licenseVersion': <version>,'needsAcceptance': true/false,'governingTerms': <text>}]
- --nspect-id
nSpect ID of artifact
collection¶
Collection Commands
ngc registry collection [--debug] [--format_type <fmt>] [-h]
{find,info,list,publish} ...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
collection¶
- {find,info,list,publish}
Possible choices: create, find, info, list, publish, remove, update
Sub-commands¶
create¶
Create a collection.
ngc registry collection create [--add-chart <chart>] [--add-image <image>]
[--add-model <model>]
[--add-resource <resource>] [--built-by <name>]
--category <category> [--debug]
[--display-name <name>] [--format_type <fmt>]
[--label <label>] [--label-set <label-set>]
[--logo <logo>] [--overview-filename <path>]
[--publisher <publisher>] --short-desc <desc>
[-h]
<target>
Positional Arguments¶
- <target>
Collection to create. Format: Org/[team/]name.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --display-name
Human-readable name for the collection.
- --label-set
Name of the label set for the collection to declare. Can be used multiple times. Format: org/[team/]name. Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --label
Label for the collection to declare. Can be used multiple times.Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --logo
A link to the logo for the collection.
- --overview-filename
A markdown file with an overview of the collection.
- --built-by
Name of the owner of this collection.
- --publisher
The publishing organization.
- --add-image
Name of an image to include in the collection. Can be used multiple times. Format: org/[team/]name.
- --add-model
Name of a model to include in the collection. Can be used multiple times. Format: org/[team/]name.
- --add-resource
Name of a resource to include in the collection. Can be used multiple times. Format: org/[team/]name.
- --add-chart
Name of a chart to include in the collection. Can be used multiple times. Format: org/[team/]name.
Required named arguments¶
- --short-desc
A brief description of the collection.
- --category
Possible choices: ADVANCED, ANNOTATION, AUTOMATIC_SPEECH_RECOGNITION, AUTOMOTIVE, BEGINNER, DEEP_LEARNING, FINANCE, HEALTHCARE, HIGH_PERFORMANCE_COMPUTING, IMAGE_CLASSIFICATION, IMAGE_SYNTHESIS, INFRASTRUCTURE, INTELLIGENT_VIDEO_ANALYTICS, MACHINE_LEARNING, MANUFACTURING, NATURAL_LANGUAGE_PROCESSING, OBJECT_DETECTION, OIL_AND_GAS, RETAIL, ROBOTICS, SMART_CITIES, SUPERCOMPUTING, TELECOMMUNICATIONS, WINDOWS_RTX_ACCELERATED_MODELS
Field for describing collection's use case. Choices are: HEALTHCARE, SMART_CITIES, RETAIL, MANUFACTURING, SUPERCOMPUTING, ROBOTICS, AUTOMOTIVE, TELECOMMUNICATIONS, FINANCE, OIL_AND_GAS, NATURAL_LANGUAGE_PROCESSING, AUTOMATIC_SPEECH_RECOGNITION, INTELLIGENT_VIDEO_ANALYTICS, OBJECT_DETECTION, IMAGE_CLASSIFICATION, ANNOTATION, IMAGE_SYNTHESIS, MACHINE_LEARNING, DEEP_LEARNING, INFRASTRUCTURE, HIGH_PERFORMANCE_COMPUTING, WINDOWS_RTX_ACCELERATED_MODELS, BEGINNER, ADVANCED
info¶
Display information about a collection in the registry.
ngc registry collection info [--debug] [--format_type <fmt>] [-h] <target>
Positional Arguments¶
- <target>
Collection. Format: org/[team/]collection_name.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
list¶
Display a list of available collections in the registry.
ngc registry collection list [--access-type <access_type>] [--column <column>]
[--debug] [--format_type <fmt>]
[--product-name <product_name>] [-h]
[<target>]
Positional Arguments¶
- <target>
Filter the search by allowing wildcards for Collection(s). Format: org/[team/]collection_name. To target Collection(s), use "org/[team/]collection_name". Org, team, and name support the wildcards "*" and "?". Examples: "my_org/my_collection" - target my_collection in my_org namespace. "my_org/my_team/my_collection" - target my_collection in my_org/my_team namespace. "my_org/my_team/" - target all collections in my_org/my_team namespace. "my_org/my_collection" - target collections starting with my_collection in my_org namespace.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --column
Specify output column as column[=header], header is optional, default is name[=Name]. Valid columns are display_name[="Display Name"], name[=Name], created_date[="Created Date"], updated_date[="Updated Date"], category[=Category], org[=Org], team[=Team], labels[=Labels], accessType[="Access Type"], productNames[="Associated Products"]. Use quotes with spaces. Multiple column arguments are allowed.
- --access-type
Possible choices: EXCLUSIVE, LISTED, NOT_LISTED
Filter the list of collections to only collections that have specified access type.
- --product-name
Filter the list of collections to only resources that are under the product name. Multiple product-name arguments are allowed. Choose from: nvaie-vpaif, nv-cloud-functions, secrets-manager, base-command-platform, nv-ai-enterprise, nvaie-igx-safety-add-on, chipnemo-inference, ai-foundations, artifact-catalog, riva-virtual-assistant, nemo-llm-service, monai, tao, omniverse-cloud, picasso, data-services, bluefield-enterprise-sw, nv-quantum-cloud, nvaie-igx-yocto-add-on, fleet-command, forge, riva-enterprise, cuopt, met, bionemo-service, iam, nvaie-igx, e2, private-registry
update¶
Update a collection.
ngc registry collection update [--add-chart <chart>] [--add-image <image>]
[--add-label <add-label>] [--add-model <model>]
[--add-resource <resource>] [--built-by <name>]
[--category <category>] [--debug]
[--display-name <name>] [--format_type <fmt>]
[--label <label>] [--label-set <label-set>]
[--logo <logo<] [--overview-filename <path>]
[--publisher <publisher>]
[--remove-chart <chart>]
[--remove-image <image>]
[--remove-label <remove-label>]
[--remove-model <model>]
[--remove-resource <resource>]
[--short-desc <desc>] [-h]
<target>
Positional Arguments¶
- <target>
Collection to create. Format: Org/[team/]name.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --display-name
Human-readable name for the collection.
- --add-label
Label for the collection to add. Can be used multiple times.Imperative label argument, not to be used with declarative label arguments --label or --label-set
- --remove-label
Label for the collection to remove. Can be used multiple times.Imperative label argument, not to be used with declarative label arguments --label or --label-set
- --label-set
Name of the label set for the collection to declare. Can be used multiple times. Format: org/[team/]name. Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --label
Label for the collection to declare. Can be used multiple times.Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --logo
A link to the logo for the collection.
- --overview-filename
A markdown file with an overview of the collection.
- --built-by
Name of the owner of this collection.
- --publisher
The publishing organization.
- --short-desc
A brief description of the collection.
- --add-image
Name of an image to include in the collection. Can be used multiple times. Format: org/[team/]name.
- --add-model
Name of a model to include in the collection. Can be used multiple times. Format: org/[team/]name.
- --add-resource
Name of a resource to include in the collection. Can be used multiple times. Format: org/[team/]name.
- --add-chart
Name of a chart to include in the collection. Can be used multiple times. Format: org/[team/]name.
- --category
Possible choices: ADVANCED, ANNOTATION, AUTOMATIC_SPEECH_RECOGNITION, AUTOMOTIVE, BEGINNER, DEEP_LEARNING, FINANCE, HEALTHCARE, HIGH_PERFORMANCE_COMPUTING, IMAGE_CLASSIFICATION, IMAGE_SYNTHESIS, INFRASTRUCTURE, INTELLIGENT_VIDEO_ANALYTICS, MACHINE_LEARNING, MANUFACTURING, NATURAL_LANGUAGE_PROCESSING, OBJECT_DETECTION, OIL_AND_GAS, RETAIL, ROBOTICS, SMART_CITIES, SUPERCOMPUTING, TELECOMMUNICATIONS, WINDOWS_RTX_ACCELERATED_MODELS
Field for describing collection's use case. Choices are: HEALTHCARE, SMART_CITIES, RETAIL, MANUFACTURING, SUPERCOMPUTING, ROBOTICS, AUTOMOTIVE, TELECOMMUNICATIONS, FINANCE, OIL_AND_GAS, NATURAL_LANGUAGE_PROCESSING, AUTOMATIC_SPEECH_RECOGNITION, INTELLIGENT_VIDEO_ANALYTICS, OBJECT_DETECTION, IMAGE_CLASSIFICATION, ANNOTATION, IMAGE_SYNTHESIS, MACHINE_LEARNING, DEEP_LEARNING, INFRASTRUCTURE, HIGH_PERFORMANCE_COMPUTING, WINDOWS_RTX_ACCELERATED_MODELS, BEGINNER, ADVANCED
- --remove-image
An image to be removed from the collection.
- --remove-model
A model to be removed from the collection.
- --remove-resource
A resource to be removed from the collection.
- --remove-chart
A chart to be removed from the collection.
remove¶
Remove a collection.
ngc registry collection remove [--debug] [--format_type <fmt>] [-h] [-y]
<target>
Positional Arguments¶
- <target>
Collection to remove. Format: org/[team/]collection_name
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- -y, --yes
Automatically confirm removal to interactive prompts.
find¶
Get a list of collections containing the specified artifact.
ngc registry collection find [--debug] [--format_type <fmt>] [-h]
<artifact_type> <artifact_target>
Positional Arguments¶
- <artifact_type>
Possible choices: CHART, IMAGE, MODEL, RESOURCE
Type of artifact to look for. Choices: MODEL, CHART, RESOURCE, IMAGE
- <artifact_target>
Target artifact to look for. Format: org/[team/]artifact_name.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
publish¶
Publish a collection from the NGC private registry to catalog.
ngc registry collection publish [--access-type <access_type>] [--allow-guest]
[--debug] [--discoverable]
[--format_type <fmt>] [--metadata-only]
[--product-name <product_name>] [--public]
[--source org/[team/]collection_name]
[--visibility-only] [-h]
org/[team/]collection_name
Positional Arguments¶
- org/[team/]collection_name
Collection. Format: org/[team/]collection_name.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --source
The source collection you want to publish. Format: org/[team/]collection_name
- --allow-guest
Allow anonymous users to download the published object
- --discoverable
Allow the published object to be discoverable in searches
- --public
Allow access to the published object by everyone instead of just those with specific roles
- --metadata-only
Only perform a shallow copy of the metadata instead of a deep copy of the objects referenced
- --visibility-only
Only change the visibility qualities of the target. Metadata and version files are not affected.
- --access-type
Possible choices: EXCLUSIVE, LISTED, NOT_LISTED
Publish the object with a specific access type. Choose from: LISTED, EXCLUSIVE, NOT_LISTED
- --product-name
Publish the object under a Product. Choose from: nvaie-vpaif, nv-cloud-functions, secrets-manager, base-command-platform, nv-ai-enterprise, nvaie-igx-safety-add-on, chipnemo-inference, ai-foundations, artifact-catalog, riva-virtual-assistant, nemo-llm-service, monai, tao, omniverse-cloud, picasso, data-services, bluefield-enterprise-sw, nv-quantum-cloud, nvaie-igx-yocto-add-on, fleet-command, forge, riva-enterprise, cuopt, met, bionemo-service, iam, nvaie-igx, e2, private-registry
csp¶
Cloud Service Provider (CSP) Commands
ngc registry csp [--debug] [--format_type <fmt>] [-h]
{create,create-settings,info,info-settings,list,remove,remove-settings,update,update-settings}
...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
csp¶
- {create,create-settings,info,info-settings,list,remove,remove-settings,update,update-settings}
Possible choices: create, create-settings, info, info-settings, list, remove, remove-settings, update, update-settings
Sub-commands¶
info¶
Get CSP info.
ngc registry csp info [--debug] [--format_type <fmt>] [-h] <name>
Positional Arguments¶
- <name>
CSP name to fetch.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
list¶
List CSPs.
ngc registry csp list [--debug] [--enabled-only] [--format_type <fmt>] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --enabled-only
Show only enabled CSPs
create¶
Create a new CSP entry.
ngc registry csp create [--debug] [--description <description>]
[--display-name <display_name>] [--enable]
[--format_type <fmt>] [--label <label>]
[--logo <logo>] [-h]
<name>
Positional Arguments¶
- <name>
New entry key to add to list of available CSPs.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --description
Description for the CSP.
- --display-name
Display name for the CSP.
- --logo
Link to the CSP logo.
- --enable
Make the CSP entry available for deployment.
- --label
Label for the CSP. To specify more than one label, use multiple --label arguments.
update¶
Update an existing CSP entry
ngc registry csp update [--debug] [--description <description>] [--disable]
[--display-name <display_name>] [--enable]
[--format_type <fmt>] [--label <label>]
[--logo <logo>] [-h]
<name>
Positional Arguments¶
- <name>
CSP entry to update.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --description
Description for the CSP.
- --display-name
Display name for the CSP.
- --logo
Link to the CSP logo.
- --label
Label for the CSP. To specify more than one label, use multiple --label arguments.
- --enable
Make the CSP entry available for deployment.
- --disable
Make the CSP entry unavailable for deployment.
remove¶
Remove an existing CSP.
ngc registry csp remove [--debug] [--format_type <fmt>] [-h] [-y] <name>
Positional Arguments¶
- <name>
Name of CSP to remove.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- -y, --yes
Automatically confirm removal to interactive prompts.
info-settings¶
Get CSP deployment constraints info
ngc registry csp info-settings [--debug] [--format_type <fmt>] [-h] <name>
Positional Arguments¶
- <name>
Name of the CSP to fetch deployment constraint settings for
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
create-settings¶
Create CSP deployment constraints
ngc registry csp create-settings [--debug] [--disk-default <disk_default>]
--disk-max <disk_max> --disk-min <disk_min>
[--format_type <fmt>]
[--gpu-default <gpu_default>]
[--gpu-default-type <gpu_default_type>]
--gpu-max <gpu_max> --gpu-min <gpu_min>
--gpu-type <gpu_type> [-h]
<name>
Positional Arguments¶
- <name>
CSP to create deployment constraints for
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --gpu-default
Default number of GPUs to use when not otherwise specified
- --gpu-default-type
Default GPU type to use when not otherwise specified
- --disk-default
Default disk size in GBs to use when not otherwise specified
Required named arguments¶
- --gpu-min
Minimum allowed number of GPUs
- --gpu-max
Maximum allowed number of GPUs
- --gpu-type
GPU card type allowed. To specify more than one type, use multiple --gpu-type arguments.
- --disk-min
Minimum allowed disk allocation in GBs
- --disk-max
Maximum allowed disk allocation in GBs
update-settings¶
Update CSP deployment constraints
ngc registry csp update-settings [--debug] [--disk-default <disk_default>]
[--disk-max <disk_max>]
[--disk-min <disk_min>] [--format_type <fmt>]
[--gpu-default <gpu_default>]
[--gpu-default-type <gpu_default_type>]
[--gpu-max <gpu_max>] [--gpu-min <gpu_min>]
[--gpu-type <gpu_type>] [-h]
<name>
Positional Arguments¶
- <name>
CSP to update deployment constraints for
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --gpu-min
Minimum allowed number of GPUs
- --gpu-max
Maximum allowed number of GPUs
- --gpu-default
Default number of GPUs to use when not otherwise specified
- --gpu-type
GPU card type allowed. To specify more than one type, use multiple --gpu-type arguments.
- --gpu-default-type
Default GPU type to use when not otherwise specified
- --disk-min
Minimum allowed disk allocation in GBs
- --disk-max
Maximum allowed disk allocation in GBs
- --disk-default
Default disk size in GBs to use when not otherwise specified
remove-settings¶
Remove the deployment constraints for a CSP
ngc registry csp remove-settings [--debug] [--format_type <fmt>] [-h] [-y]
<name>
Positional Arguments¶
- <name>
CSP to remove deployment constraints for
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- -y, --yes
Automatically confirm removal to interactive prompts.
image¶
Container Image Registry Commands
ngc registry image [--debug] [--format_type <fmt>] [-h]
{deploy,info,list,publickey,publish,pull,scan,update-license-terms}
...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
image¶
- {deploy,info,list,publickey,publish,pull,scan,update-license-terms}
Possible choices: create, deploy, info, list, publickey, publish, pull, push, remove, scan, sign, update, update-license-terms
Sub-commands¶
deploy¶
Manage interactive container deployments for images.
ngc registry image deploy [--debug] [--format_type <fmt>] [-h]
{create,info,list,remove,start,update} ...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
deploy¶
- {create,info,list,remove,start,update}
Possible choices: create, info, list, remove, start, update
Sub-commands¶
start¶
Create interactive deployment of an image from the NGC catalog to a cloud service provider (CSP). Default parameters can be overwritten with the provided flag options.
ngc registry image deploy start [--debug] [--disk <disk>] [--dry-run]
[--format_type <fmt>] [--gpu <gpu>]
[--gpu-type <gpu-type>] [-h]
org/[team/]image_name:tag <csp>
Positional Arguments¶
- org/[team/]image_name:tag
Image and tag to use for deployment.
- <csp>
Cloud service provider to deploy to.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --gpu
Number of GPUs to use.
- --gpu-type
Type of GPUs to use.
- --disk
Amount of disk space to allocate in GBs.
- --dry-run
If enabled, don't automatically open the link in a browser.
create¶
Create the default deployment parameter set for an image deployment.
ngc registry image deploy create [--debug] [--disk <disk>]
[--format_type <fmt>] [--gpu <gpu>]
[--gpu-type <gpu-type>] [-h]
<org>/[<team>/]<image> <csp>
Positional Arguments¶
- <org>/[<team>/]<image>
Image and tag to use for deployment.
- <csp>
Cloud service provider to deploy to.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --gpu
Number of GPUs to use.
- --gpu-type
Type of GPUs to use.
- --disk
Amount of disk space to allocate in GBs.
remove¶
Delete deployment parameters for an image and unmark as being deployable.
ngc registry image deploy remove [--debug] [--format_type <fmt>] [-h] [-y]
<org>/[<team>/]<image> <csp>
Positional Arguments¶
- <org>/[<team>/]<image>
Image whose deployment parameters should be removed.
- <csp>
Cloud service provider to deploy to.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- -y, --yes
Automatically say yes to all interactive questions.
info¶
Get the default deployments parameters for a CSP deployment of an image.
ngc registry image deploy info [--debug] [--exclude-csp-defaults]
[--format_type <fmt>] [-h]
<org>/[<team>/]<image> <csp>
Positional Arguments¶
- <org>/[<team>/]<image>
Image to lookup.
- <csp>
CSP default deployment to get.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --exclude-csp-defaults
Exclude CSP-level defaults.
update¶
Update an existing deployment parameter set for an image and CSP.
ngc registry image deploy update [--debug] [--disk <disk>]
[--format_type <fmt>] [--gpu <gpu>]
[--gpu-type <gpu-type>] [-h]
<org>/[<team>/]<image> <csp>
Positional Arguments¶
- <org>/[<team>/]<image>
Image and tag to use for deployment.
- <csp>
Cloud service provider to deploy to.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --gpu
Number of GPUs to use.
- --gpu-type
Type of GPUs to use.
- --disk
Amount of disk space to allocate in GBs.
list¶
Get list of available CSP deployments for an artifact.
ngc registry image deploy list [--debug] [--format_type <fmt>] [-h]
<org>/[<team>/]<image>
Positional Arguments¶
- <org>/[<team>/]<image>
Target artifact to get list of CSP deployments for.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
list¶
List container images accessible by the user.
ngc registry image list [--access-type <access_type>] [--column <column>]
[--debug] [--format_type <fmt>]
[--product-name <product_name>] [--signed] [-h]
[<org>/[<team>/]<image>[:<tags>]]
Positional Arguments¶
- <org>/[<team>/]<image>[:<tags>]
Filter the search by allowing the wildcards '*' and '?'. In order to list all the tags for <image>, pattern should be <org>/[<team>/]<image>:*. Examples: "my_org/my_image" - target my_image in my_org namespace. "my_org/my_team/my_image" - target my_image in my_org/my_team namespace. "my_org/my_team/" - target all images in my_org/my_team namespace. "my_org/my_image" - target images starting with my_image in my_org namespace. "my_org/my_image:[1-5]" - target versions 1-5 for my_image in my_org namespace.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --column
Specify output column as column[=header], header is optional, default is repository[=Repository] or tag[=Tag] for versions. Valid columns are created[="Created Date"], description[=Description], name[=Name], org[=Org], permission[=Permission], shared[=Shared], size[="Image Size"], tag[="Latest Tag"], team[=Team], updated[="Updated Date"], labels[=Labels], signed[="Signed Tag?"]. Use quotes with spaces. Multiple column arguments are allowed.
- --signed
Show only container images that are signed
- --access-type
Possible choices: EXCLUSIVE, LISTED, NOT_LISTED
Filter the list of resources to only resources that have specified access type.
- --product-name
Filter the list of resources to only resources that are under the product name. Multiple flags supported. Choose from: nvaie-vpaif, nv-cloud-functions, secrets-manager, base-command-platform, nv-ai-enterprise, nvaie-igx-safety-add-on, chipnemo-inference, ai-foundations, artifact-catalog, riva-virtual-assistant, nemo-llm-service, monai, tao, omniverse-cloud, picasso, data-services, bluefield-enterprise-sw, nv-quantum-cloud, nvaie-igx-yocto-add-on, fleet-command, forge, riva-enterprise, cuopt, met, bionemo-service, iam, nvaie-igx, e2, private-registry
remove¶
Remove an image repository or specific image with an image tag.
ngc registry image remove [--debug] [--format_type <fmt>] [-h] [-y]
<org>/[<team>/]<image>[:<tags>]
Positional Arguments¶
- <org>/[<team>/]<image>[:<tags>]
Filter the search by allowing the wildcards '*' and '?'.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- -y, --yes
Automatically say yes to all interactive questions.
info¶
Display information about an image repository or tagged image.
ngc registry image info [--debug] [--details] [--format_type <fmt>]
[--history] [--layers] [--scan] [-h]
<org>/[<team>/]<image>[:<tag>]
Positional Arguments¶
- <org>/[<team>/]<image>[:<tag>]
Name of the image repository or tagged image.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --layers
Show the layers of a tagged image.
- --history
Show the history of a tagged image.
- --details
Show the details of an image repository.
- --scan
Show the scan details of a tagged image. If no tag is provided, the most recent tag is used.
pull¶
Pull a container image from the NGC image registry. If no tag is provided, 'latest' is assumed.
ngc registry image pull [--debug] [--format_type <fmt>] [--scan <file>] [-h]
<org>/[<team>/]<image>[:<tag>]
Positional Arguments¶
- <org>/[<team>/]<image>[:<tag>]
Name of the image repository or tagged image.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --scan
Download the image scan report as a CSV file instead of pulling the image. If no tag is provided, 'latest' is assumed.
update¶
Update image repository metadata
ngc registry image update [--add-label <add-label>] [--built-by <name>]
[--debug] [--desc <desc>] [--display-name <name>]
[--format_type <fmt>] [--label <label>]
[--label-set <label-set>] [--logo <url>]
[--multinode] [--no-multinode]
[--overview <file.md>] [--publisher <publisher>]
[--remove-label <remove-label>] [-h]
<org>/[<team>/]<image>[:<tag>]
Positional Arguments¶
- <org>/[<team>/]<image>[:<tag>]
Name of the image repository or tagged image.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --desc
Description for the target image.
- --overview
Documentation (text or markdown file) for the image.
- --add-label
Label for the repository to add. Can be used multiple times.Imperative label argument, not to be used with declarative label arguments --label or --label-set
- --remove-label
Label for the repository to remove. Can be used multiple times.Imperative label argument, not to be used with declarative label arguments --label or --label-set
- --label
Label for the repository to declare. Can be used multiple times.Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --label-set
Name of the label set for the repository to declare. Can be used multiple times. Format: org/[team/]name. Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --logo
A URL pointing to the logo for the repository.
- --publisher
The person or entity publishing the image.
- --built-by
The person who built the container image.
- --display-name
Different name to display for the image.
- --multinode
Marks an image as supporting multinode.
- --no-multinode
Marks an image as not supporting multinode.
push¶
Push a container image to the NGC image registry. If no tag is provided, 'latest' is assumed.
ngc registry image push [--built-by <name>] [--debug] [--desc <desc>]
[--display-name <name>] [--format_type <fmt>]
[--label <label>] [--label-set <label-set>]
[--logo <url>] [--multinode] [--overview <file.md>]
[--publisher <publisher>] [-h] [-y]
<org>/[<team>/]<image>[:<tag>]
Positional Arguments¶
- <org>/[<team>/]<image>[:<tag>]
Name of the image repository or tagged image.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- -y, --yes
Automatically say yes to tagging the image if not already tagged in correct format.
- --desc
Description for the target image.
- --overview
Documentation (text or markdown file) for the image.
- --label
Label for the repository to declare. Can be used multiple times.Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --label-set
Name of the label set for the repository to declare. Can be used multiple times. Format: org/[team/]name. Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --logo
A URL pointing to the logo for the repository.
- --publisher
The person or entity publishing the image.
- --built-by
The person who built the container image.
- --multinode
Marks an image as supporting multinode.
- --display-name
Different name to display for the image.
create¶
Create a top level metadata repository in the NGC image registry.
ngc registry image create [--built-by <name>] [--debug] [--desc <desc>]
[--display-name <name>] [--format_type <fmt>]
[--label <label>] [--label-set <label-set>]
[--logo <url>] [--multinode] [--overview <file.md>]
[--publisher <publisher>] [-h]
<org>/[<team>/]<image>
Positional Arguments¶
- <org>/[<team>/]<image>
Name of the image repository.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --desc
Description for the target image.
- --overview
Documentation (text or markdown file) for the image.
- --label
Label for the repository to declare. Can be used multiple times.Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --label-set
Name of the label set for the repository to declare. Can be used multiple times. Format: org/[team/]name. Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --logo
A URL pointing to the logo for the repository.
- --publisher
The person or entity publishing the image.
- --built-by
The person who built the container image.
- --multinode
Marks an image as supporting multinode.
- --display-name
Different name to display for the image.
scan¶
Scan a container image from the NGC image registry.
ngc registry image scan [--debug] [--format_type <fmt>] [-h]
[<org>/[<team>/]<image>:<tag>]
Positional Arguments¶
- <org>/[<team>/]<image>:<tag>
Filter the search by allowing the wildcards '*' and '?'. In order to list all the tags for <image>, pattern should be <org>/[<team>/]<image>:*. Examples: "my_org/my_image" - target my_image in my_org namespace. "my_org/my_team/my_image" - target my_image in my_org/my_team namespace. "my_org/my_team/" - target all images in my_org/my_team namespace. "my_org/my_image" - target images starting with my_image in my_org namespace. "my_org/my_image:[1-5]" - target versions 1-5 for my_image in my_org namespace.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
publickey¶
Return the public key used to sign images for local validation.
ngc registry image publickey [--debug] [--format_type <fmt>] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
publish¶
Publish an image from the NGC image registry to catalog. If no version is provided, 'latest' is assumed.
ngc registry image publish [--access-type <access_type>] [--allow-guest]
[--debug] [--discoverable] [--format_type <fmt>]
[--license-terms-file <filename>] [--metadata-only]
[--nspect-id NSPECT_ID]
[--product-name <product_name>] [--public] [--sign]
[--source org/[team/]image_name[:version]]
[--status <workflow_id>] [--version-only]
[--visibility-only] [-h]
[org/[team/]image_name[:version]]
Positional Arguments¶
- org/[team/]image_name[:version]
The target image and version you want to publish to. Optional when getting publishing status using the --status flag. Format: org/[team/]image_name[:version]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --source
The source image and version you want to publish. Format: org/[team/]image_name[:version]
- --metadata-only
Only perform a shallow copy of the metadata instead of a deep copy of the objects referenced
- --version-only
Only copy the specified version of the object without copying any metadata
- --visibility-only
Only change the visibility qualities of the target. Metadata and version files are not affected.
- --allow-guest
Allow anonymous users to download the published object
- --discoverable
Allow the published object to be discoverable in searches
- --public
Allow access to the published object by everyone instead of just those with specific roles
- --sign
Have the published image cryptographically signed by NVIDIA.
- --product-name
Publish the object under a Product. Choose from: nvaie-vpaif, nv-cloud-functions, secrets-manager, base-command-platform, nv-ai-enterprise, nvaie-igx-safety-add-on, chipnemo-inference, ai-foundations, artifact-catalog, riva-virtual-assistant, nemo-llm-service, monai, tao, omniverse-cloud, picasso, data-services, bluefield-enterprise-sw, nv-quantum-cloud, nvaie-igx-yocto-add-on, fleet-command, forge, riva-enterprise, cuopt, met, bionemo-service, iam, nvaie-igx, e2, private-registry
- --access-type
Possible choices: EXCLUSIVE, LISTED, NOT_LISTED
Publish the object with a specific access type. Choose from: LISTED, EXCLUSIVE, NOT_LISTED
- --status
Get the status of publishing based on provide workflow id.
- --license-terms-file
Publish the object with a specific license term defined in JSON file. File format: [{'licenseId': <id>, 'licenseVersion': <version>,'needsAcceptance': true/false,'governingTerms': <text>}]
- --nspect-id
nSpect ID of artifact
sign¶
Have the image cryptographically signed by NVIDIA.
ngc registry image sign [--debug] [--format_type <fmt>] [-h]
<org>/[<team>/]<image>[:<tags>]
Positional Arguments¶
- <org>/[<team>/]<image>[:<tags>]
Name of the image repository or tagged image.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
update-license-terms¶
Update an artifact's license terms.
ngc registry image update-license-terms [--clear] [--debug]
[--format_type <fmt>]
[--license-terms-file <filename>] [-h]
org/[team/]image_name[:version]
Positional Arguments¶
- org/[team/]image_name[:version]
The target image and version you want to publish to. Optional when getting publishing status using the --status flag. Format: org/[team/]image_name[:version]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --license-terms-file
Publish the object with a specific license term defined in JSON file. File format: [{'licenseId': <id>, 'licenseVersion': <version>,'needsAcceptance': true/false,'governingTerms': <text>}]
- --clear
Whether to clear an artifact's license terms.
label-set¶
Label-Set Commands
ngc registry label-set [--debug] [--format_type <fmt>] [-h] {info,list} ...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
label-set¶
- {info,list}
Possible choices: create, info, list, remove, update
Sub-commands¶
list¶
List label-set(s) in a namespace.
ngc registry label-set list [--column <column>] [--debug]
[--format_type <fmt>]
[--resource-type <resourceType>] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --column
Specify output column as column[=header], header is optional, default is name[=Name]. Valid columns are displayName[="Display Name"], org[=Org], team[=Team], resourceType[="Resource Type"], readOnly[="Read Only"], isGlobal[=Global], labels[=Labels]. Use quotes with spaces. Multiple column arguments are allowed.
- --resource-type
Possible choices: COLLECTION, CONTAINER, DATASET, HELM_CHART, JOB, JOB_TEMPLATE, MODEL, ORGANIZATION, PRODUCT, RESOURCE, RESULTSET, TEAM, USER, WORKSPACE
Filter global label sets by resource type. Allowed values: COLLECTION, CONTAINER, DATASET, HELM_CHART, JOB, JOB_TEMPLATE, MODEL, ORGANIZATION, PRODUCT, RESOURCE, RESULTSET, TEAM, USER, WORKSPACE
info¶
Get information about a label-set.
ngc registry label-set info [--debug] [--format_type <fmt>]
[--resource-type <resourceType>] [-h]
<target>
Positional Arguments¶
- <target>
Label set name. Format: org/[team/]name.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --resource-type
Possible choices: COLLECTION, CONTAINER, DATASET, HELM_CHART, JOB, JOB_TEMPLATE, MODEL, ORGANIZATION, PRODUCT, RESOURCE, RESULTSET, TEAM, USER, WORKSPACE
Filter global label sets by resource type. Allowed values: COLLECTION, CONTAINER, DATASET, HELM_CHART, JOB, JOB_TEMPLATE, MODEL, ORGANIZATION, PRODUCT, RESOURCE, RESULTSET, TEAM, USER, WORKSPACE
update¶
Update a label-set.
ngc registry label-set update [--add-label <label>] [--debug]
[--display-name <dispName>]
[--format_type <fmt>] [--label <label>]
[--remove-label <label>] [-h]
<target>
Positional Arguments¶
- <target>
Name of the label-set to update. Format: org/[team/]name.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --display-name
The name to display for the label-set
- --label
DEPRECATED - please use '--add-label' instead. Label to be added to the label-set. Use quotes with spaces. Multiple add-label arguments are allowed.
- --add-label
Label to be added to the label-set. Use quotes with spaces. Multiple add-label arguments are allowed.
- --remove-label
Label to be removed from the label-set. Use quotes with spaces. Multiple remove-label arguments are allowed.
remove¶
Remove a label-set in a namespace.
ngc registry label-set remove [--debug] [--format_type <fmt>] [-h] [-y]
<target>
Positional Arguments¶
- <target>
Name of the label-set to remove. Format: org/[team/]name.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- -y, --yes
Automatically say yes to all interactive questions.
create¶
Create a label-set in a namespace.
ngc registry label-set create [--debug] --display-name <dispName>
[--format_type <fmt>] [--label <label>] [-h]
<target>
Positional Arguments¶
- <target>
Name of the label-set to create. Format: org/[team/]name.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --label
Label in the label-set. Use quotes with spaces. Multiple label arguments are allowed.
Required named arguments¶
- --display-name
The name to display for the label-set.
model¶
Model Commands
ngc registry model [--debug] [--format_type <fmt>] [-h]
{deploy,download-version,download-version-signature,info,list,playground,public-key,publish,sign,update-license-terms}
...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
model¶
- {deploy,download-version,download-version-signature,info,list,playground,public-key,publish,sign,update-license-terms}
Possible choices: commit-version, create, deploy, download-version, download-version-signature, info, list, playground, public-key, publish, remove, sign, update, update-license-terms, upload-version
Sub-commands¶
deploy¶
Manage interactive container deployments for models.
ngc registry model deploy [--debug] [--format_type <fmt>] [-h]
{create,info,list,remove,start,update} ...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
deploy¶
- {create,info,list,remove,start,update}
Possible choices: create, info, list, remove, start, update
Sub-commands¶
start¶
Start interactive deployment of a model from the NGC catalog to a cloud service provider (CSP). Default parameters can be overwritten with the provided flag options.
ngc registry model deploy start [--debug] [--disk <disk>] [--dry-run]
[--format_type <fmt>] [--gpu <gpu>]
[--gpu-type <gpu-type>]
[--image <org>/[<team>/]<image>:<tag>] [-h]
<org>/[<team>/]<model_name>:<version> <csp>
Positional Arguments¶
- <org>/[<team>/]<model_name>:<version>
Model and version to use for deployment.
- <csp>
Cloud service provider to deploy to.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --image
Image and tag to use as base for deployment if different from default.
- --gpu
Number of GPUs to use.
- --gpu-type
Type of GPU to use.
- --disk
Amount of disk space to allocate in GBs.
- --dry-run
Just show the deployment URL and the list of files that would be deployed.
create¶
Create the default deployment parameter set for a model deployment. For each flag parameter not provided, the CSP default will be used.
ngc registry model deploy create [--debug] [--disk <disk>]
[--format_type <fmt>] [--gpu <gpu>]
[--gpu-type <gpu-type>] [-h]
<org>/[<team>/]<model_name>
<org>/[<team>/]<image>:<tag> <csp>
Positional Arguments¶
- <org>/[<team>/]<model_name>
Model to use for deployment.
- <org>/[<team>/]<image>:<tag>
Base image to use for deployments.
- <csp>
Cloud service provider to deploy to.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --gpu
Number of GPUs to use.
- --gpu-type
Type of GPU to use.
- --disk
Amount of disk space to allocate in GBs.
remove¶
Delete deployment parameters for a model and unmark as being deployable.
ngc registry model deploy remove [--debug] [--format_type <fmt>] [-h] [-y]
<org>/[<team>/]<model_name> <csp>
Positional Arguments¶
- <org>/[<team>/]<model_name>
Model whose deployment parameters should be removed.
- <csp>
Cloud service provider to deploy to.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- -y, --yes
Automatically say yes to all interactive questions.
info¶
Get the default deployments parameters for a CSP deployment of an model.
ngc registry model deploy info [--debug] [--exclude-csp-defaults]
[--format_type <fmt>] [-h]
<org>/[<team>/]<model_name> <csp>
Positional Arguments¶
- <org>/[<team>/]<model_name>
Model to lookup.
- <csp>
CSP deployment parameters to get.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --exclude-csp-defaults
Exclude CSP-level defaults.
update¶
Update the default deployment parameter set for a model and CSP.
ngc registry model deploy update [--debug] [--disk <disk>]
[--format_type <fmt>] [--gpu <gpu>]
[--gpu-type <gpu-type>]
[--image <org>/[<team>/]<image>:<tag>] [-h]
<org>/[<team>/]<model_name> <csp>
Positional Arguments¶
- <org>/[<team>/]<model_name>
Model to use for deployment.
- <csp>
Cloud service provider to deploy to.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --image
Base image to use for deployments.
- --gpu
Number of GPUs to use.
- --gpu-type
Type of GPU to use.
- --disk
Amount of disk space to allocate in GBs.
list¶
Get list of available CSP deployments for an artifact.
ngc registry model deploy list [--debug] [--format_type <fmt>] [-h]
<org>/[<team>/]<model_name>
Positional Arguments¶
- <org>/[<team>/]<model_name>
Model to use for deployment.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
playground¶
Manage playground for model.
ngc registry model playground [--debug] [--format_type <fmt>] [-h] {info} ...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
playground¶
- {info}
Possible choices: info
Sub-commands¶
info¶
Get the info of the playground for a model.
ngc registry model playground info [--debug] [--format_type <fmt>] [-h]
<org>/[<team>/]<model_name>
Positional Arguments¶
- <org>/[<team>/]<model_name>
Modelname of the playground to lookup
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
list¶
List model(s) or model version(s).
ngc registry model list [--access-type <access_type>] [--column <column>]
[--debug] [--format_type <fmt>]
[--product-name <product_name>] [--signed]
[--sort <order>] [-h]
[<target>]
Positional Arguments¶
- <target>
Filter the search by allowing wildcards for Model(s) or model version(s). Format: org/[team/]name[:version]. To target model version(s), use "org/[team/]name:version". To target model(s), use "org/[team/]name". Name and version also supports the wildcards "*" and "?". Examples: "my_org/my_model" - target my_model in my_org namespace. "my_org/my_team/my_model" - target my_model in my_org/my_team namespace. "my_org/my_team/" - target all models in my_org/my_team namespace. "my_org/my_model" - target models starting with my_model in my_org namespace.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --column
Specify output column as column[=header], header is optional, default is repository[=Repository] or version[=Version] for versions. Valid columns are name[=Name], org[=Org], team[=Team], description[=Description], updated[="Last Modified"], created[="Created Date"], shared[=Shared], size[="File Size"], repository[=Repository], version[="Latest Version"], application[=Application], framework[=Framework], precision[=Precision], permission[=Permission], accuracy[=Accuracy], epochs[=Epochs], batch[="Batch Size"], gpu[="GPU Model"], memory[="Memory Footprint"], status[=Status], labels[=Labels]. Use quotes with spaces. Multiple column arguments are allowed.
- --sort
Possible choices: ALPHABETIC_ASC, ALPHABETIC_DESC, BUILT_BY_ASC, BUILT_BY_DESC, CREATED_DATE_ASC, CREATED_DATE_DESC, DISPLAY_NAME_ASC, DISPLAY_NAME_DESC, PUBLISHER_ASC, PUBLISHER_DESC, RESOURCE_ID_ASC, RESOURCE_ID_DESC, SEMVER_ASC, SEMVER_DESC, UPDATED_DATE_ASC, UPDATED_DATE_DESC
Sort model versions. Allowed values: ALPHABETIC_ASC, ALPHABETIC_DESC, CREATED_DATE_ASC, CREATED_DATE_DESC, SEMVER_ASC, SEMVER_DESC, DISPLAY_NAME_ASC, DISPLAY_NAME_DESC, UPDATED_DATE_ASC, UPDATED_DATE_DESC, PUBLISHER_ASC, PUBLISHER_DESC, BUILT_BY_ASC, BUILT_BY_DESC, RESOURCE_ID_ASC, RESOURCE_ID_DESC. Default: SEMVER_DESC.
- --access-type
Possible choices: EXCLUSIVE, LISTED, NOT_LISTED
Filter the list of resources to only resources that have specified access type.
- --product-name
Filter the list of resources to only resources that are under the product name. Multiple product-name arguments are allowed. Choose from: nvaie-vpaif, nv-cloud-functions, secrets-manager, base-command-platform, nv-ai-enterprise, nvaie-igx-safety-add-on, chipnemo-inference, ai-foundations, artifact-catalog, riva-virtual-assistant, nemo-llm-service, monai, tao, omniverse-cloud, picasso, data-services, bluefield-enterprise-sw, nv-quantum-cloud, nvaie-igx-yocto-add-on, fleet-command, forge, riva-enterprise, cuopt, met, bionemo-service, iam, nvaie-igx, e2, private-registry
- --signed
Show only models that are signed
info¶
Retrieve metadata for a model or model version.
ngc registry model info [--credentials] [--debug] [--files]
[--format_type <fmt>] [--metrics] [-h]
<target>
Positional Arguments¶
- <target>
Model or model version. Format: org/[team/]model_name[:version]. To target a model version, use "org/[team/]model_name:version". To target a model, use "org/[team/]model_name".
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --files
List files in addition to details for a version.
- --credentials
List model credentials in addition to details for a version.
- --metrics
List model credentials in addition to details for a version. DEPRECATED; will be removed after May 2021. Please use '--credentials' instead.
create¶
Create a model.
ngc registry model create --application <app> [--bias-filename <path>]
[--built-by <name>] [--debug]
[--display-name <name>]
[--explainability-filename <path>] --format <fmt>
[--format_type <fmt>] --framework <fwk>
[--label <label>] [--label-set <label-set>]
[--logo <url>] [--overview-filename <path>]
--precision <prec> [--privacy-filename <path>]
[--public-dataset-license <lcs>]
[--public-dataset-link <link>]
[--public-dataset-name <name>] [--publisher <name>]
[--safety-security-filename <path>] --short-desc
<desc> [-h]
<target>
Positional Arguments¶
- <target>
Model. Format: org/[team/]model_name.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --overview-filename
Overview. Provide the path to a file that contains the overview for the model.
- --bias-filename
Bias. Provide the path to a file that contains the bias in the model.
- --explainability-filename
Explainability. Provide the path to a file that contains the explainability for this model.
- --privacy-filename
Privacy. Provide the path to a file that contains the privacy for this model.
- --safety-security-filename
Safety and Security. Provide the path to a file that contains the safety and security in the model.
- --display-name
Display name for the model.
- --label
Label for the model to declare. Can be used multiple times.Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --label-set
Name of the label set for the model to declare. Can be used multiple times. Format: org/[team/]name. Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --logo
URL for the model logo image.
- --public-dataset-name
Name of public dataset used in the model.
- --public-dataset-link
Link to public dataset used in the model.
- --public-dataset-license
License for public dataset used in the model.
- --built-by
Builder of the model.
- --publisher
Publisher of the model.
Required named arguments¶
- --application
Model application.
- --framework
Framework used to train the model.
- --format
Format of the model.
- --precision
Precision the model was trained with.
- --short-desc
Short description of the model.
update¶
Update a model or model version.
ngc registry model update [--accuracy-reached <accuracy>]
[--add-label <add-label>] [--application <app>]
[--batch-size <size>] [--bias-filename <path>]
[--built-by <name>] [--debug] [--desc <desc>]
[--display-name <name>]
[--explainability-filename <path>] [--format <fmt>]
[--format_type <fmt>] [--framework <fwk>]
[--gpu-model <model>] [--label <label>]
[--label-set <label-set>] [--logo <url>]
[--memory-footprint <footprint>]
[--num-epochs <num>] [--overview-filename <path>]
[--precision <prec>] [--privacy-filename <path>]
[--public-dataset-license <lcs>]
[--public-dataset-link <url>]
[--public-dataset-name <name>] [--publisher <name>]
[--remove-label <remove-label>]
[--safety-security-filename <path>] [--set-latest]
[--short-desc <desc>] [-h]
<target>
Positional Arguments¶
- <target>
Model or model version. Format: org/[team/]model_name[:version]. To target a model version, use "org/[team/]model_name:version". To target a model, use "org/[team/]model_name".
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --application
Model application.
- --framework
Framework used to train the model.
- --format
Format of the model.
- --precision
Precision the model was trained with.
- --short-desc
Short description of the model.
- --desc
Description for the model version.
- --overview-filename
Overview. Provide the path to a file that contains the overview for the model.
- --bias-filename
Bias. Provide the path to a file that contains the bias in the model.
- --explainability-filename
Explainability. Provide the path to a file that contains the explainability for this model.
- --privacy-filename
Privacy. Provide the path to a file that contains the privacy for this model.
- --safety-security-filename
Safety and Security. Provide the path to a file that contains the safety and security in the model.
- --display-name
Display name for the model.
- --add-label
Label for the model to add. Can be used multiple times.Imperative label argument, not to be used with declarative label arguments --label or --label-set
- --remove-label
Label for the model to remove. Can be used multiple times.Imperative label argument, not to be used with declarative label arguments --label or --label-set
- --label
Label for the model to declare. Can be used multiple times.Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --label-set
Name of the label set for the model to declare. Can be used multiple times. Format: org/[team/]name. Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --logo
URL for the model logo image.
- --public-dataset-name
Name of public dataset used in the model.
- --public-dataset-link
Link to public dataset used in the model.
- --public-dataset-license
License for public dataset used in the model.
- --built-by
Builder of the model.
- --publisher
Publisher of the model.
- --gpu-model
The GPU used to train the model version.
- --memory-footprint
The memory footprint of the model version.
- --num-epochs
The number of epochs for the model version.
- --batch-size
The batch size of the model version.
- --accuracy-reached
Accuracy reached with model version.
- --set-latest
Set this version to be the latest version.
remove¶
Remove a model or model version.
ngc registry model remove [--debug] [--format_type <fmt>] [-h] [-y] <target>
Positional Arguments¶
- <target>
Model or model version. Format: org/[team/]model_name[:version]. To target a model version, use "org/[team/]model_name:version". To target a model, use "org/[team/]model_name".
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- -y, --yes
Automatically say yes to all interactive questions.
download-version¶
Download a model version.
ngc registry model download-version [--debug] [--dest <path>]
[--exclude <wildcard>] [--file <wildcard>]
[--format_type <fmt>] [-h]
<target>
Positional Arguments¶
- <target>
Model version. Format: org/[team/]model_name[:version]. If no version specified, the latest version will be targeted.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --dest
Destination to download the current model. Default: .
- --file
Specify individual files to download from the model. Supports standard Unix shell-style wildcards like (?, [abc], [!a-z], etc..) May be used multiple times in the same command.
- --exclude
Exclude files or directories from the downloaded model. Supports standard Unix shell-style wildcards like (?, [abc], [!a-z], etc..). May be used multiple times in the same command.
upload-version¶
Upload a model version.
ngc registry model upload-version [--accuracy-reached <accuracy>]
[--base-version <version>]
[--batch-size <size>]
[--credentials-file <file>] [--debug]
[--desc <desc>] [--dry-run]
[--format_type <fmt>] [--gpu-model <model>]
[--link <url>] [--link-type <type>]
[--memory-footprint <footprint>]
[--metrics-file <file>] [--no-commit]
[--num-epochs <num>] [--source <path>] [-h]
<target>
Positional Arguments¶
- <target>
Model version. Format: org/[team/]model_name:version.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --source
Provide source directory of the model or path of single file to be uploaded. Default: .
- --gpu-model
The GPU used to train the model version.
- --memory-footprint
The memory footprint of the model version.
- --num-epochs
The number of epochs for the model version.
- --batch-size
The batch size of the model version.
- --accuracy-reached
Accuracy reached with model version.
- --desc
Description for the model version.
- --dry-run
List file paths, total upload size and file count without performing the upload.
- --link-type
Type of link to a resource or other toolsets for the model. Choices: NGC, Github, Other.
- --link
Link to resource or other toolsets for the model
- --credentials-file
A JSON file containing a single object with 'name' and 'attributes' fields. Attributes are a list of key-value pair each takes the form of {'key': KEY, 'value': VALUE}. A maximum of twelve attributes may be used per file, and up to three files may be specified.
- --metrics-file
A JSON file containing a single object with 'name' and 'attributes' fields. Attributes are a list of key-value pair each takes the form of {'key': KEY, 'value': VALUE}. A maximum of twelve attributes may be used per file, and up to three files may be specified. DEPRECATED; will be removed after May 2021. Please use --credentials-file instead.
- --base-version
Use an existing model version as the base: include all its files, and have any overlapping files overwritten by those in the source.
- --no-commit
Upload model but do not complete the version, allowing subsequent uploads.
commit-version¶
Commit a model version.
ngc registry model commit-version [--debug] [--format_type <fmt>] [-h]
<target>
Positional Arguments¶
- <target>
Model version. Format: org/[team/]model_name:version.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
publish¶
Publish a model from the NGC model registry to catalog. If no version is provided, 'latest' is assumed.
ngc registry model publish [--access-type <access_type>] [--allow-guest]
[--debug] [--discoverable] [--format_type <fmt>]
[--license-terms-file <filename>] [--metadata-only]
[--product-name <product_name>] [--public] [--sign]
[--source org/[team/]model_name[:version]]
[--status <workflow_id>] [--upload-pending]
[--version-only] [--visibility-only] [-h]
[org/[team/]model_name[:version]]
Positional Arguments¶
- org/[team/]model_name[:version]
The the target model and version you want to publish to. Optional when getting publishing status using the --status flag. Format: org/[team/]model_name[:version]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --source
The source model and version you want to publish. Format: org/[team/]model_name[:version]
- --metadata-only
Only perform a shallow copy of the metadata instead of a deep copy of the objects referenced
- --version-only
Only copy the specified version of the object without copying any metadata
- --visibility-only
Only change the visibility qualities of the target. Metadata and version files are not affected.
- --allow-guest
Allow anonymous users to download the published object
- --discoverable
Allow the published object to be discoverable in searches
- --public
Allow access to the published object by everyone instead of just those with specific roles
- --product-name
Publish the object under a Product. Choose from: nvaie-vpaif, nv-cloud-functions, secrets-manager, base-command-platform, nv-ai-enterprise, nvaie-igx-safety-add-on, chipnemo-inference, ai-foundations, artifact-catalog, riva-virtual-assistant, nemo-llm-service, monai, tao, omniverse-cloud, picasso, data-services, bluefield-enterprise-sw, nv-quantum-cloud, nvaie-igx-yocto-add-on, fleet-command, forge, riva-enterprise, cuopt, met, bionemo-service, iam, nvaie-igx, e2, private-registry
- --status
Get the status of publishing based on provide workflow id.
- --access-type
Possible choices: EXCLUSIVE, LISTED, NOT_LISTED
Publish the object with a specific access type. Choose from: LISTED, EXCLUSIVE, NOT_LISTED
- --upload-pending
Allows published model version to continue to be updated with files.Commit version is requied to complete the upload and allow download.
- --license-terms-file
Publish the object with a specific license term defined in JSON file. File format: [{'licenseId': <id>, 'licenseVersion': <version>,'needsAcceptance': true/false,'governingTerms': <text>}]
- --sign
Publish the object and sign the version.
update-license-terms¶
Update an artifact's license terms.
ngc registry model update-license-terms [--clear] [--debug]
[--format_type <fmt>]
[--license-terms-file <filename>] [-h]
org/[team/]model_name[:version]
Positional Arguments¶
- org/[team/]model_name[:version]
Model. Format: org/[team/]model_name.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --license-terms-file
Publish the object with a specific license term defined in JSON file. File format: [{'licenseId': <id>, 'licenseVersion': <version>,'needsAcceptance': true/false,'governingTerms': <text>}]
- --clear
Whether to clear an artifact's license terms.
sign¶
Request a model version to get signed.
ngc registry model sign [--debug] [--format_type <fmt>] [-h] <target>
Positional Arguments¶
- <target>
Model version. Format: org/[team/]model_name:version.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
download-version-signature¶
Download the signature file of a model.
ngc registry model download-version-signature [--debug] [--dest <path>]
[--format_type <fmt>] [-h]
<target>
Positional Arguments¶
- <target>
Model version. Format: org/[team/]model_name[:version]. If no version specified, the latest version will be targeted.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --dest
Destination to download the current model version signature. Default: .
public-key¶
Download the public key used to sign models.
ngc registry model public-key [--debug] [--dest <path>] [--format_type <fmt>]
[-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --dest
Destination to download public key. Default: .
resource¶
Resource Commands
ngc registry resource [--debug] [--format_type <fmt>] [-h]
{deploy,download-version,download-version-signature,info,list,public-key,publish,sign,update-license-terms}
...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
resource¶
- {deploy,download-version,download-version-signature,info,list,public-key,publish,sign,update-license-terms}
Possible choices: create, deploy, download-version, download-version-signature, info, list, public-key, publish, remove, sign, update, update-license-terms, upload-version
Sub-commands¶
deploy¶
Manage interactive container deployments for resources.
ngc registry resource deploy [--debug] [--format_type <fmt>] [-h]
{create,info,list,remove,start,update} ...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
deploy¶
- {create,info,list,remove,start,update}
Possible choices: create, info, list, remove, start, update
Sub-commands¶
start¶
Start interactive deployment of a resource from the NGC catalog to a cloud service provider (CSP). Default parameters can be overwritten with the provided flag options.
ngc registry resource deploy start [--debug] [--disk <disk>] [--dry-run]
[--format_type <fmt>] [--gpu <gpu>]
[--gpu-type <gpu-type>]
[--image <org>/[<team>/]<image>:<tag>] [-h]
<org>/[<team>/]<resource_name>:<version>
<csp>
Positional Arguments¶
- <org>/[<team>/]<resource_name>:<version>
Resource and version to use for deployment.
- <csp>
Cloud service provider to deploy to.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --image
Image and tag to use as base for deployment if different from default.
- --gpu
Number of GPUs to use.
- --gpu-type
Type of GPU to use.
- --disk
Amount of disk space to allocate in GBs.
- --dry-run
Just show the deployment URL and the list of files that would be deployed.
create¶
Create the default deployment parameter set for a resource deployment. For each flag parameter not provided, the CSP default will be used.
ngc registry resource deploy create [--debug] [--disk <disk>]
[--format_type <fmt>] [--gpu <gpu>]
[--gpu-type <gpu-type>] [-h]
<org>/[<team>/]<resource_name>
<org>/[<team>/]<image>:<tag> <csp>
Positional Arguments¶
- <org>/[<team>/]<resource_name>
Resource to use for deployment.
- <org>/[<team>/]<image>:<tag>
Base image to use for deployments.
- <csp>
Cloud service provider to deploy to.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --gpu
Number of GPUs to use.
- --gpu-type
Type of GPU to use.
- --disk
Amount of disk space to allocate in GBs.
remove¶
Delete deployment parameters for a resource and unmark as being deployable.
ngc registry resource deploy remove [--debug] [--format_type <fmt>] [-h] [-y]
<org>/[<team>/]<resource_name> <csp>
Positional Arguments¶
- <org>/[<team>/]<resource_name>
Resource whose deployment parameters should be removed.
- <csp>
Cloud service provider to deploy to.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- -y, --yes
Automatically say yes to all interactive questions.
info¶
Get the default deployments parameters for a CSP deployment of a resource.
ngc registry resource deploy info [--debug] [--exclude-csp-defaults]
[--format_type <fmt>] [-h]
<org>/[<team>/]<resource_name> <csp>
Positional Arguments¶
- <org>/[<team>/]<resource_name>
Resource to lookup.
- <csp>
CSP deployment parameters to get.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --exclude-csp-defaults
Exclude CSP-level defaults.
update¶
Update the default deployment parameter set for a resource and CSP.
ngc registry resource deploy update [--debug] [--disk <disk>]
[--format_type <fmt>] [--gpu <gpu>]
[--gpu-type <gpu-type>]
[--image <org>/[<team>/]<image>:<tag>]
[-h]
<org>/[<team>/]<resource_name> <csp>
Positional Arguments¶
- <org>/[<team>/]<resource_name>
Resource to use for deployment.
- <csp>
Cloud service provider to deploy to.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --image
Base image to use for deployments.
- --gpu
Number of GPUs to use.
- --gpu-type
Type of GPU to use.
- --disk
Amount of disk space to allocate in GBs.
list¶
Get list of available CSP deployments for an artifact.
ngc registry resource deploy list [--debug] [--format_type <fmt>] [-h]
<org>/[<team>/]<resource_name>
Positional Arguments¶
- <org>/[<team>/]<resource_name>
Resource to use for deployment.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
list¶
List resources or resource version(s).
ngc registry resource list [--access-type <access_type>] [--column <column>]
[--debug] [--format_type <fmt>]
[--product-name <product_name>] [--signed] [-h]
[<target>]
Positional Arguments¶
- <target>
Filter the search by allowing wildcards for resources or resource version(s). Format: [org/[team/]]name[:version]. To target resource version(s), use "[org/[team/]]name:version". To target resources use "[org/[team/]]name". Both name and version support the wildcards "*" and "?". Version also supports character expressions ([a-z], [!ab], etc.). Examples: "my_org/my_resource" - target my_resource in my_org namespace. "my_org/my_team/my_resource" - target my_resource in my_org/my_team namespace. "my_org/my_team/" - target all resources in my_org/my_team namespace. "my_org/my_resource" - target all resource versions for my_resource in my_org namespace. "my_org/my_resource:[1-5]" - target versions 1-5 for my_resource in my_org namespace.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --column
Specify output column as column[=header], header is optional, default is repository[=Repository] or latest_version[=Latest Version] for versions. Valid columns are name[=Name], org[=Org], team[=Team], description[=Description], updated[="Last Modified"], created[="Created Date"], shared[=Shared], size[="File Size"], latest_version[="Latest Version"], version[=Version], application[=Application], framework[=Framework], precision[=Precision], permission[=Permission], accuracy[=Accuracy], epochs[=Epochs], batch[="Batch Size"], gpu[="GPU Model"], memory[="Memory Footprint"], status[=Status], labels[=Labels]. Use quotes with spaces. Multiple column arguments are allowed.
- --access-type
Possible choices: EXCLUSIVE, LISTED, NOT_LISTED
Filter the list of resources to only resources that have specified access type.
- --product-name
Filter the list of resources to only resources that are under the product name. Multiple product-name arguments are allowed. Choose from: nvaie-vpaif, nv-cloud-functions, secrets-manager, base-command-platform, nv-ai-enterprise, nvaie-igx-safety-add-on, chipnemo-inference, ai-foundations, artifact-catalog, riva-virtual-assistant, nemo-llm-service, monai, tao, omniverse-cloud, picasso, data-services, bluefield-enterprise-sw, nv-quantum-cloud, nvaie-igx-yocto-add-on, fleet-command, forge, riva-enterprise, cuopt, met, bionemo-service, iam, nvaie-igx, e2, private-registry
- --signed
Show only models that are signed
info¶
Retrieve metadata for a resource or resource version.
ngc registry resource info [--debug] [--files] [--format_type <fmt>] [-h]
<target>
Positional Arguments¶
- <target>
Resource or resource version. Format: org/[team/]resource_name[:version]. To target a resource version, use "org/[team/]name:version". To target a resource, use "org/[team/]name".
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --files
List files in addition to details for a version.
create¶
Create a resource.
ngc registry resource create [--advanced-filename <path>] --application <app>
[--built-by <name>] [--debug]
[--display-name <name>] --format <fmt>
[--format_type <fmt>] --framework <fwk>
[--label <label>] [--label-set <label-set>]
[--logo <url>] [--overview-filename <path>]
[--performance-filename <path>] --precision
<prec> [--public-dataset-license <lcs>]
[--public-dataset-link <url>]
[--public-dataset-name <name>]
[--publisher <name>]
[--quick-start-guide-filename <path>]
[--setup-filename <path>] --short-desc <desc>
[-h]
<target>
Positional Arguments¶
- <target>
Resource. Format: org/[team/]resource_name.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --overview-filename
Overview. Provide the path to a file that contains the overview for the resource.
- --display-name
Display name.
- --label
Label for the resource to declare. Can be used multiple times.Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --label-set
Name of the label set for the resource to declare. Can be used multiple times. Format: org/[team/]name. Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --logo
URL for the resource logo image.
- --public-dataset-name
Name of public dataset used.
- --public-dataset-link
Link to public dataset used.
- --public-dataset-license
License for public dataset used.
- --built-by
Builder of the resource.
- --publisher
Publisher of the resource.
- --advanced-filename
Advanced guide. Provide the path to a file that contains the "Advanced Guide" for the resource.
- --performance-filename
Performance data. Provide the path to a file that contains the performance data for the resource.
- --quick-start-guide-filename
Quick start information. Provide the path to a file that contains the "Quick Start Guide" information for the resource.
- --setup-filename
Setup instructions. Provide the path to a file that contains the setup instructions for the resource.
Required named arguments¶
- --application
Application of the resource. Allowed values: CLASSIFICATION, OBJECT_DETECTION, SEGMENTATION, TRANSLATION, TEXT_TO_SPEECH, RECOMMENDER, SENTIMENT, NLP, KUBEFLOW_PIPELINE, OTHER, SPEAKER_VERIFICATION, PUNCTUATION_AND_CAPITALIZATION, DIALOG_STATE_TRACKING, SPEAKER_DIARIZATION, SPEAKER_RECOGNITION, SPEECH_SYNTHESIS, VOICE_ACTIVITY_DETECTION, SPEECH_RECOGNITION, LANGUAGE_MODELING, NAMED_ENTITY_RECOGNITION, TEXT_CLASSIFICATION, TEXT_PROCESSING, QUESTION_ANSWERING, CONTENT_GENERATION.
- --framework
Framework used in the resource. Allowed values: TensorFlow, Caffe2, CNTK, Torch, PyTorch, MXNet, Keras, Other, TransferLearningToolkit, TensorFlow2.
- --format
Format of the model (checkpoint) generated by the resource.
- --precision
Precision supported by the resource. Allowed values: FP16, FP32, INT8, FPBOTH, OTHER, ALL.
- --short-desc
Short description.
update¶
Update a resource or resource version.
ngc registry resource update [--accuracy-reached <accuracy>]
[--add-label <add-label>]
[--advanced-filename <path>]
[--application <app>] [--batch-size <size>]
[--built-by <name>] [--debug] [--desc <desc>]
[--display-name <name>] [--format <fmt>]
[--format_type <fmt>] [--framework <fwk>]
[--gpu-model <model>] [--label <label>]
[--label-set <label-set>] [--logo <url>]
[--memory-footprint <footprint>]
[--num-epochs <num>] [--overview-filename <path>]
[--performance-filename <path>]
[--precision <prec>]
[--public-dataset-license <lcs>]
[--public-dataset-link <url>]
[--public-dataset-name <name>]
[--publisher <name>]
[--quick-start-guide-filename <path>]
[--release-notes-filename <path>]
[--remove-label <remove-label>]
[--setup-filename <path>] [--short-desc <desc>]
[-h]
<target>
Positional Arguments¶
- <target>
Resource or resource version. Format: org/[team/]resource_name[:version]. To target a resource version, use "org/[team/]name:version". To target a resource, use "org/[team/]name".
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --application
Application of the resource. Allowed values: CLASSIFICATION, OBJECT_DETECTION, SEGMENTATION, TRANSLATION, TEXT_TO_SPEECH, RECOMMENDER, SENTIMENT, NLP, KUBEFLOW_PIPELINE, OTHER, SPEAKER_VERIFICATION, PUNCTUATION_AND_CAPITALIZATION, DIALOG_STATE_TRACKING, SPEAKER_DIARIZATION, SPEAKER_RECOGNITION, SPEECH_SYNTHESIS, VOICE_ACTIVITY_DETECTION, SPEECH_RECOGNITION, LANGUAGE_MODELING, NAMED_ENTITY_RECOGNITION, TEXT_CLASSIFICATION, TEXT_PROCESSING, QUESTION_ANSWERING, CONTENT_GENERATION.
- --framework
Framework used in the resource. Allowed values: TensorFlow, Caffe2, CNTK, Torch, PyTorch, MXNet, Keras, Other, TransferLearningToolkit, TensorFlow2.
- --format
Format of the model (checkpoint) generated by the resource.
- --precision
Precision supported by the resource. Allowed values: FP16, FP32, INT8, FPBOTH, OTHER, ALL.
- --short-desc
Short description.
- --overview-filename
Overview. Provide the path to a file that contains the overview for the resource.
- --display-name
Display name.
- --add-label
Label for the resource to add. Can be used multiple times.Imperative label argument, not to be used with declarative label arguments --label or --label-set
- --remove-label
Label for the resource to remove. Can be used multiple times.Imperative label argument, not to be used with declarative label arguments --label or --label-set
- --label
Label for the resource to declare. Can be used multiple times.Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --label-set
Name of the label set for the resource to declare. Can be used multiple times. Format: org/[team/]name. Declarative label argument, not to be used with declarative label arguments —add-label or —remove-label
- --logo
URL for the resource logo image.
- --public-dataset-name
Name of public dataset used.
- --public-dataset-link
Link to public dataset used.
- --public-dataset-license
License for public dataset used.
- --built-by
Builder of the resource.
- --publisher
Publisher of the resource.
- --desc
Full description of resource version.
- --accuracy-reached
Accuracy reached with resource version.
- --batch-size
The batch size of the resource version.
- --gpu-model
The GPU used to train the resource version.
- --memory-footprint
The memory footprint of the resource version.
- --num-epochs
The number of epochs for the resource version.
- --release-notes-filename
Release notes. Provide the path to a file that contains the release notes for the resource.
- --advanced-filename
Advanced guide. Provide the path to a file that contains the "Advanced Guide" for the resource.
- --performance-filename
Performance data. Provide the path to a file that contains the performance data for the resource.
- --quick-start-guide-filename
Quick start information. Provide the path to a file that contains the "Quick Start Guide" information for the resource.
- --setup-filename
Setup instructions. Provide the path to a file that contains the setup instructions for the resource.
remove¶
Remove a resource or resource version.
ngc registry resource remove [--debug] [--format_type <fmt>] [-h] [-y]
<target>
Positional Arguments¶
- <target>
Resource or resource version. Format: org/[team/]resource_name[:version]. To target a resource version, use "org/[team/]name:version". To target a resource, use "org/[team/]name".
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- -y, --yes
Automatically say yes to all interactive questions.
download-version¶
Download a resource version.
ngc registry resource download-version [--debug] [--dest <path>]
[--exclude <wildcard>]
[--file <wildcard>]
[--format_type <fmt>] [-h]
<target>
Positional Arguments¶
- <target>
Resource version. Format: org/[team/]resource_name[:version]. If no version specified, the latest version will be targeted.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --dest
Provide a destination to download the resource. Default: .
- --file
Specify individual files to download from the resource. Supports standard Unix shell-style wildcards like (?, [abc], [!a-z], etc..) May be used multiple times in the same command.
- --exclude
Exclude files or directories from the downloaded resource. Supports standard Unix shell-style wildcards like (?, [abc], [!a-z], etc..). May be used multiple times in the same command.
upload-version¶
Upload a resource version.
ngc registry resource upload-version [--accuracy-reached <accuracy>]
[--base-version <version>]
[--batch-size <size>] [--debug]
[--desc <desc>] [--dry-run]
[--format_type <fmt>]
[--gpu-model <model>]
[--memory-footprint <footprint>]
[--num-epochs <num>]
[--performance-filename <path>]
[--quick-start-guide-filename <path>]
[--release-notes-filename <path>]
[--setup-filename <path>]
[--source <path>] [-h]
<target>
Positional Arguments¶
- <target>
Resource version. Format: org/[team/]resource_name:version.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --accuracy-reached
Accuracy reached with resource version.
- --batch-size
The batch size of the resource version.
- --gpu-model
The GPU used to train the resource version.
- --memory-footprint
The memory footprint of the resource version.
- --num-epochs
The number of epochs for the resource version.
- --desc
Full description of resource version.
- --performance-filename
Performance data. Provide the path to a file that contains the performance data for the resource.
- --quick-start-guide-filename
Quick start information. Provide the path to a file that contains the "Quick Start Guide" information for the resource.
- --setup-filename
Setup instructions. Provide the path to a file that contains the setup instructions for the resource.
- --release-notes-filename
Release notes. Provide the path to a file that contains the release notes for the resource.
- --source
Provide source directory of the resource or path of single file to be uploaded; if omitted, current directory will be used.
- --dry-run
List file paths, total upload size and file count without performing the upload.
- --base-version
Use an existing model version as the base: include all its files, and have any overlapping files overwritten by those in the source.
publish¶
Publish a resource from the NGC resource registry to catalog. If no version is provided, the latest is assumed.
ngc registry resource publish [--access-type <access_type>] [--allow-guest]
[--debug] [--discoverable] [--format_type <fmt>]
[--license-terms-file <filename>]
[--metadata-only] [--nspect-id NSPECT_ID]
[--product-name <product_name>] [--public]
[--sign]
[--source org/[team/]resource_name[:version]]
[--status <workflow_id>] [--version-only]
[--visibility-only] [-h]
[org/[team/]resource_name[:version]]
Positional Arguments¶
- org/[team/]resource_name[:version]
The the target resource and version you want to publish to. Optional when getting publishing status using the --status flag. Format: org/[team/]resource_name[:version]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --source
The source resource and version you want to publish. Format: org/[team/]resource_name[:version]
- --metadata-only
Only perform a shallow copy of the metadata instead of a deep copy of the objects referenced
- --version-only
Only copy the specified version of the object without copying any metadata
- --visibility-only
Only change the visibility qualities of the target. Metadata and version files are not affected.
- --allow-guest
Allow anonymous users to download the published object
- --discoverable
Allow the published object to be discoverable in searches
- --public
Allow access to the published object by everyone instead of just those with specific roles
- --product-name
Publish the object under a Product. Choose from: nvaie-vpaif, nv-cloud-functions, secrets-manager, base-command-platform, nv-ai-enterprise, nvaie-igx-safety-add-on, chipnemo-inference, ai-foundations, artifact-catalog, riva-virtual-assistant, nemo-llm-service, monai, tao, omniverse-cloud, picasso, data-services, bluefield-enterprise-sw, nv-quantum-cloud, nvaie-igx-yocto-add-on, fleet-command, forge, riva-enterprise, cuopt, met, bionemo-service, iam, nvaie-igx, e2, private-registry
- --status
Get the status of publishing based on provide workflow id.
- --access-type
Possible choices: EXCLUSIVE, LISTED, NOT_LISTED
Publish the object with a specific access type. Choose from: LISTED, EXCLUSIVE, NOT_LISTED
- --license-terms-file
Publish the object with a specific license term defined in JSON file. File format: [{'licenseId': <id>, 'licenseVersion': <version>,'needsAcceptance': true/false,'governingTerms': <text>}]
- --sign
Publish the object and sign the version.
- --nspect-id
nSpect ID of artifact
update-license-terms¶
Update an artifact's license terms.
ngc registry resource update-license-terms [--clear] [--debug]
[--format_type <fmt>]
[--license-terms-file <filename>]
[-h]
org/[team/]resource_name[:version]
Positional Arguments¶
- org/[team/]resource_name[:version]
The the target resource and version you want to publish to. Optional when getting publishing status using the --status flag. Format: org/[team/]resource_name[:version]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --license-terms-file
Publish the object with a specific license term defined in JSON file. File format: [{'licenseId': <id>, 'licenseVersion': <version>,'needsAcceptance': true/false,'governingTerms': <text>}]
- --clear
Whether to clear an artifact's license terms.
sign¶
Request a resource version to get signed.
ngc registry resource sign [--debug] [--format_type <fmt>] [-h] <target>
Positional Arguments¶
- <target>
Resource version. Format: org/[team/]resource_name[:version]. If no version specified, the latest version will be targeted.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
download-version-signature¶
Download the signature file of a resource.
ngc registry resource download-version-signature [--debug] [--dest <path>]
[--format_type <fmt>] [-h]
<target>
Positional Arguments¶
- <target>
Resource version. Format: org/[team/]resource_name[:version]. If no version specified, the latest version will be targeted.
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --dest
Destination to download the current resource version signature. Default: .
public-key¶
Download the public key used to sign resources.
ngc registry resource public-key [--debug] [--dest <path>]
[--format_type <fmt>] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
- --dest
Destination to download public key. Default: .
usage¶
Registry Usage Commands
ngc registry usage [--debug] [--format_type <fmt>] [-h] {info} ...
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
usage¶
- {info}
Possible choices: info
Sub-commands¶
info¶
Show Registry Storage usage.
ngc registry usage info [--debug] [--format_type <fmt>] [-h]
Named Arguments¶
- --debug
Enable debug mode.
- --format_type
Possible choices: ascii, csv, json
Specify the output format type. Supported formats are: ['ascii', 'csv', 'json']. Only commands that produce tabular data support csv format. Default: ascii
Examples¶
How to list repository images¶
$ ngc registry image list
+------------+------------+------------+------------+---------+------------+
| Name | Repository | Latest Tag | Image Size | Updated | Permission |
| | | | | Date | |
+------------+------------+------------+------------+---------+------------+
| busyboxss1 | nvexpl/bus | signed2 | 746.73 KB | Mar 08, | unlocked |
| 111 | ybox | | | 2021 | |
| digits | nvexpl/dig | 21.04-tens | 5.91 GB | Apr 13, | unlocked |
| | its | orflow-py3 | | 2021 | |
| One Long | nvexpl/mk/ | 3.4 | 745.89 KB | Aug 05, | unlocked |
| Display | alpine | | | 2020 | |
| Name | | | | | |
| pusherTest | nvexpl/mk/ | nginx-21.0 | 51.12 MB | Jan 07, | unlocked |
| | pusher | 1 | | 2021 | |
| This is | nvexpl/mk/ | 1 | 0 B | Apr 08, | unlocked |
| One Long | alpine_int | | | 2019 | |
| Container | egration | | | | |
...
How to list repository images with column arguments¶
$ ngc registry image list --column name --column tag
+--------------------------+-------------------------+-------------------------+
| Repository | Name | Latest Tag |
+--------------------------+-------------------------+-------------------------+
| nvexmpl/busybox | busyboxss1111 | signed2 |
| nvexmpl/digits | digits | 21.04-tensorflow-py3 |
| nvexmpl/mk/alpine | One Long Display Name | 3.4 |
| nvexmpl/mk/pusher | pusherTest | nginx-21.01 |
| nvexmpl/mk/alpine_integr | This is One Long | 1 |
| ation | Container Name | |
...
How to remove a registry image¶
$ ngc registry image remove example_org/digits:2.0
Successfully removed image 'example_org/digits:2.0'
How to remove a registry image repository¶
$ ngc registry image remove example_org/digits
Successfully removed image repository 'example_org/digits'
How to get information about a tagged image¶
$ ngc registry image info example_org/digits:3.0
--------------------------------------------------
Image Information
Name: example_org/digits:3.0
Architecture: amd64
Schema Version: 1
--------------------------------------------------
How to get information about an image repository¶
$ ngc registry image info example_org/digits
--------------------------------------------------
Image Repository Information
Name: digits
Short Description: A short description of the repository
Labels: computation, machine_learning
Public: No
Shared with organizations: example_org
Shared with teams:
Last Updated: 2019-05-06T20:10:42.650Z
Latest Image Size: 14.18 MB
Latest Tag: 3.0
Tags:
3.0
2.0
1.0
--------------------------------------------------
How to pull an image from the NGC image registry¶
NGC automatically logs you in to the image registry when pulling or pushing images.
Docker is required to be installed on the client system.
Pulling an image from a registry requires you to specify the path:
$ ngc registry image list
+---------------------+------------------+------------+------------+--------------+------------+
| Name | Repository | Latest Tag | Image Size | Updated | Permission |
| | | | | Date | |
+---------------------+------------------+------------+------------+--------------+------------+
| ... | ... | ... | ... | ... | ... |
| | | | | | |
| tensorrt | example_org/tensorrt | 19.05 | 1.91 GB | May 09, 2019 | unlocked |
| | | | | | |
| ... | ... | ... | ... | ... | ... |
+---------------------+------------------+------------+------------+--------------+------------+
$ ngc registry image pull example_org/tensorrt:19.05
Logging in to https://nvcr.io... login successful.
19.05-py3: Pulling from example_org/tensorrt
7e6591854262: Downloading [======================> ] 19.32MB/43.75MB
089d60cb4e0a: Download complete
9c461696bc09: Download complete
45085432511a: Download complete
7e6591854262: Downloading [======================> ] 19.32MB/43.75MB
089d60cb4e0a: Download complete
9c461696bc09: Download complete
45085432511a: Download complete
7e6591854262: Downloading [========================> ] 21.53MB/43.75MB
089d60cb4e0a: Download complete
...
Digest: sha256:0422d59d3aece1ea72e2a5699d234c9c266636a6c0cdd3f1adc2b7773302c88e
Status: Downloaded newer image for nvcr.io/example_org/tensorrt:19.05-py3
NOTE that if you do not provide a tag when pulling or pushing an image, 'latest' will be assumed:
$ ngc registry image pull example_org/tensorrt
Logging in to https://nvcr.io... login successful.
Error: manifest for nvcr.io/example_org/tensorrt:latest not found.
How to update an image from the NGC image registry¶
Updating an image from a registry requires you to specify the path:
$ ngc registry image list
+-----------------+----------------------+------------+------------+--------------+------------+
| Name | Repository | Latest Tag | Image Size | Updated | Permission |
| | | | | Date | |
+-----------------+----------------------+------------+------------+--------------+------------+
| ... | ... | ... | ... | ... | ... |
| | | | | | |
| tensorrt | example_org/tensorrt | 19.05 | 1.91 GB | May 09, 2019 | unlocked |
| | | | | | |
| ... | ... | ... | ... | ... | ... |
+-----------------+----------------------+------------+------------+--------------+------------+
$ ngc registry image update example_org/tensorrt --desc new_description --label new_label --logo new_logo --overview new_file --publisher new_pub --built-by new_built
Updating repository metadata
Repository metadata updated.
How to push an image to the NGC image registry¶
Pushing an image is similar to pulling, except that your image name (tag) must include the registry URL as well as the path to your image in the registry. These paths often include your organization name and (if applicable) team name. The registry URL is optional when telling the NGC CLI about the image, and if the registry URL is not present in your tag, the NGC CLI can help you rename it before pushing.
Docker is required to be installed on the client system.
$ docker images
REPOSITORY TAG IMAGE ID CREATED SIZE
example_org/my-image latest d07226900e72 13 days ago 38.6MB
$ ngc registry image push example_org/my-image:latest
Logging in to https://nvcr.io... login successful.
The push refers to repository [nvcr.io/example_org/my-image]
Error: tag does not exist: nvcr.io/example_org/my-image:latest
Image must be tagged in format: <registry>/<org>/[<team>/]<image-name>[:<tag>]
Would you like to tag 'example_org/my-image:latest' as 'nvcr.io/example_org/my-image:latest' and push? [y/N]y
The push refers to repository [nvcr.io/example_org/my-image]
01f1c15e33e7: Mounted from example_org/checkpoint_counter_example
da3ae26fc4ba: Layer already exists
a464c54f93a9: Layer already exists
latest: digest: sha256:7eb9230b4666a11c1d241c15cae5110a31c56e5befa1f7956e9a2639a4baabe0 size: 947
Successfully pushed nvcr.io/example_org/my-image:latest to the registry.
$ docker images
REPOSITORY TAG IMAGE ID CREATED SIZE
example_org/my-image latest d07226900e72 13 days ago 38.6MB
nvcr.io/example_org/test-sm-img latest d07226900e72 13 days ago 38.6MB
How to list models¶
$ ngc registry model list
+---------+---------+---------+---------+---------+---------+---------+---------+
| Name | Reposit | Latest | Applica | Framewo | Precisi | Last Mo | Permiss |
| | ory | Version | tion | rk | on | dified | ion |
+---------+---------+---------+---------+---------+---------+---------+---------+
| denoise | nvidia/ | | | | | May 14, | unlocke |
| r | ngx/den | | | | | 2019 | d |
| | oiser | | | | | | |
| testmod | nvidia/ | | | | | May 14, | unlocke |
| el | cim/tes | | | | | 2019 | d |
| | tmodel | | | | | | |
| testmod | nvidia/ | 54323 | none | cudaaaa | | Feb 25, | unlocke |
| el123 | cim/tes | | | | | 2020 | d |
How to list models using column arguments¶
$ ngc registry model list --column version --column permission
+-------------------------+-----------------------+------------+
| Repository | Latest Version | Permission |
+-------------------------+-----------------------+------------+
| nvexpl/ngx/denoiser | | unlocked |
| nvexpl/cim/testmodel | | unlocked |
| nvexpl/cim/testmodel1 | 54323 | unlocked |
| nvexpl/cim/heart_ultras | | unlocked |
| ound | | |
| nvexpl/ngx/dldenoiser | | unlocked |
| nvexpl/ngx/dlaa | | unlocked |
| nvexpl/ngx/latest | | unlocked |
| nvexpl/ngx/all | | unlocked |
| nvexpl/ngx/config | 2 | unlocked |
| nvexpl/ngx/dlupscale | | unlocked |
How to get information about a model¶
$ ngc registry model info example_model_name
--------------------------------------------------
Model Information
Name: example_model_name
Domain(TODO Application): Recommender
Framework: Tensorflow
Model Format: ProtoText
Precision: FP16
Description:
This is a longer description. blah blah blah blah blah
Short Description: short desc blah blah
Display Name:
Logo: link/to/logo
Org: example_org
Team:
Built By: TODO
Owner Name: TODO
Publisher: TODO
Created Date: 2018-11-29T20:14:15.046Z
Updated Date: 2018-11-29T20:14:15.046Z
Labels:
label1
label2
Latest Version ID: 3
Latest Version Size (bytes): 116246
Public Dataset Used:
Name: ds name
Link: link to ds
License: ds license
--------------------------------------------------
How to create a model¶
$ ngc registry model create example_org/example_model_name --short-desc 'A short description here ' \
--application 'Recommender' --framework 'Tensorflow' --format 'ProtoText' \
--precision 'FP16' --public-dataset-link 'www.example.com' --public-dataset-name 'my dataset' \
--public-dataset-license 'ds license' --label 'label1' --label 'label2' \
--overview-filename /path/to/overview.md
Successfully created model 'example_org/example_model_name'.
--------------------------------------------------
Model Information
Name: example_model_name
Application: Recommender
Framework: Tensorflow
Model Format: ProtoText
Precision: FP16
Short Description: A short description here
Display Name:
Org: example_org
Team:
Built By:
Owner Name:
Publisher:
Created Date: 2018-11-29T20:14:15.046Z
Updated Date: 2018-11-29T20:14:15.046Z
Labels:
label1
label2
Latest Version ID:
Latest Version Size (bytes):
Public Dataset Used:
Name: my dataset
Link: www.example.com
License: ds license
Overview:
# overview file contents here
--------------------------------------------------
How to upload a model version¶
$ ngc registry model upload-version example_org/example_model_name:v1 --gpu-model 'GV-100' \
--accuracy-reached 76.77 --batch-size 0 --memory-footprint '1024KB' \
--num-epochs 91 --source tox.out
Uploaded 113.66 KB, 1/1 files in 1s, Avg Upload speed: 113.66 KB/s, Curr Upload Speed: 113.66 KB/s
----------------------------------------------------
Model id: example_model_name[version=v1]
Upload status: Completed
Uploaded local path: /home/jmaddux/projects/ngc_cli/test_new_cmds/ngc-cli/tox.out
Total files uploaded: 1
Total uploaded size: 113.66 KB
Started at: 2018-11-29 14:17:42.302443
Completed at: 2018-11-29 14:17:52.310996
Duration taken: 1s
----------------------------------------------------
How to list model versions¶
$ ngc registry model list "example_org/example_model_name:*"
+---------+----------+--------+------------+-----------+------------------+-----------+-------+-----------------+------------------------+
| Version | Accuracy | Epochs | Batch Size | GPU Model | Memory Footprint | File Size | Owner | Status | Created Date |
+---------+----------+--------+------------+-----------+------------------+-----------+-------+-----------------+------------------------+
| v1 | 76.77 | 91 | 0.0 | GV-100 | 1024KB | 113.52 KB | | UPLOAD_COMPLETE |2018-11-29 20:17:08 UTC |
+---------+----------+--------+------------+-----------+------------------+-----------+-------+-----------------+------------------------+
How to list model versions using column arguments¶
$ ngc registry model list "example_org/example_model_name:*" --column size --column status
+---------+-----------+-----------------+
| Version | File Size | Status |
+---------+-----------+-----------------+
| v1 | 113.52 KB | UPLOAD_COMPLETE |
+---------+-----------+-----------------+
How to get information about a model version¶
$ ngc registry model info example_org/example_model_name:v1
--------------------------------------------------
Model Version Information
Id: v1
Batch Size: 0
Memory Footprint: 1024KB
Number Of Epochs: 91
GPU Model: GV-100
Owner Name:
Created Date: 2018-11-29T20:17:08.218Z
Description:
Status: UPLOAD_COMPLETE
Total File Count: 1
Total Size: 2.03 KB
--------------------------------------------------
How to list files for a model¶
$ ngc registry model example_org/example_model_name:v1 --files
--------------------------------------------------
Model Version Information
Id: v1
Batch Size: 0
Memory Footprint: 1024KB
Number Of Epochs: 91
GPU Model: GV-100
Owner Name:
Created Date: 2018-11-29T20:17:08.218Z
Description:
Status: UPLOAD_COMPLETE
Total File Count: 1
Total Size: 2.03 KB
File List:
tox.out - 2.03 KB
--------------------------------------------------
How to update a model's version¶
$ ngc registry model update example_org/example_model_name:v1 --desc \
'This is an updated description. Blah blah blah'
Successfully updated model version'example_model_name.v1'.
--------------------------------------------------
Model Version Information
Id: v1
Batch Size: 0.0
Memory Footprint (TODO units): 1024KB
Number Of Epochs: 91
GPU Model: GV-100
Owner Name:
Created Date: 2018-11-29T20:17:08.218Z
Description:
This is an updated description. blah blah blah
Status: UPLOAD_COMPLETE
Total File Count: 1
Total Size: 2.03 KB
--------------------------------------------------
How to remove a model¶
$ ngc registry model remove example_org/example_model_name
Re-enter model name to confirm removal from org-example_org: example_model_name
Successfully removed model 'example_model_name'.
How to remove a version from a model¶
$ ngc registry model remove example_org/example_model_name:v1
Are you sure you would like to remove example_org/example_model_name:v1? [y/n]y
Successfully removed model version 'example_org/example_model_name:v1'.
How to update a model¶
Each of the attributes of a model can be updated so long as they are
specified. See the full list with ngc registry model update -h.
$ ngc registry model update example_org/example_model_name --display-name "My Deep Model"
--------------------------------------------------
Model Information
Name: example_model_name
Domain: Recommender
Framework: TensorFlow
Model Format: ProtoText
Precision: FP16
Description: A model for deep learning and recommendation
Short Description: A description of the model
Display Name: My Deep Model
Logo: /path/to/url/rec.jpg
Org: example_org
Team:
Built By:
Owner Name:
Publisher:
Created Date: 2018-11-08T16:58:31.285Z
Updated Date: 2018-11-08T16:58:31.285Z
Labels:
foo
Latest Version ID: v1
Latest Version Size: 2.01 KB
Public Dataset Used:
Name: my dataset
Link: www.example.com
License: ds license
--------------------------------------------------
How to download a version of a model¶
$ ngc registry model download-version example_org/my_ml_model:v1
----------------------------------------------------
Transfer id: my_ml_model
Download status: Completed
Downloaded local path: /a/path/to/the/model
Total files downloaded: 5
Total downloaded size: 200 MB
Started at: 2018-11-28 15:13:32.119121
Completed at: 2018-11-28 15:13:33.129846
Duration taken: 10s
----------------------------------------------------
See ngc registry -h for a full list of options.
How to create a resource¶
See help strings for resource commands with ngc registry resource -h.
$ ngc registry resource create example_org/example_resource \
--application OBJECT_DETECTION \
--format FMT \
--framework TensorFlow \
--precision FP16 \
--short-desc "A short description of my example_resource"
Successfully created resource 'example_org/example_resource'.
--------------------------------------------------
Resource Information
Name: example_resource
Application: OBJECT_DETECTION
Training Framework: TensorFlow
Model Format: FMT
Precision: FP16
Short Description: A short description of my example_resource
Public Dataset Used:
Name:
Link:
License:
Display Name:
Logo:
Org: example_org
Team:
Built By:
Publisher:
Created Date: 2020-04-01 19:03:33 UTC
Updated Date: 2020-04-01 19:03:33 UTC
Can Guest Download: False
Read Only: False
Labels:
Latest Version ID:
Latest Version Size (bytes):
Overview:
Setup:
Quick Start Guide:
Advanced Guide:
Performance:
--------------------------------------------------
How to upload a new resource version¶
$ ngc registry resource upload-version example_org/example_resource:v1 \
--accuracy-reached 99 \
--batch-size 25 \
--gpu-model V100 \
--memory-footprint 8GB \
--num-epochs 2 \
--source path/to/resource/dir
Uploaded 100 KB, 7/10 files in 3s, Avg Upload speed: 33.33 KB/s, Curr Upload Speed: 0 B/s
----------------------------------------------------
Resource ID: example_resource[version=v1]
Upload status: Completed
Uploaded local path (resource): /path/to/resource/dir
Total files uploaded: 10
Total transferred: 100 KB
Started at: 2020-04-01 14:12:01.155986
Completed at: 2020-04-01T14:12:05.162686Z
Duration taken: 4s
----------------------------------------------------
How to get information about a resource or resource version¶
$ ngc registry resource info example_org/example_resource
--------------------------------------------------
Resource Information
Name: example_resource
Application: OBJECT_DETECTION
Training Framework: TensorFlow
Model Format: FMT
Precision: FP16
Short Description: A short description of my example_resource
Public Dataset Used:
Name:
Link:
License:
Display Name:
Logo:
Org: example_org
Team:
Built By:
Publisher:
Created Date: 2020-04-01 19:03:33 UTC
Updated Date: 2020-04-01 19:12:05 UTC
Can Guest Download: False
Read Only: False
Labels:
Latest Version ID: v1
Latest Version Size (bytes): 10240
Overview:
Setup:
Quick Start Guide:
Advanced Guide:
Performance:
--------------------------------------------------
$ ngc registry resource info example_org/example_resource:v1
--------------------------------------------------
Resource Version Information
Id: v1
Batch Size: 25
Memory Footprint: 8GB
Number Of Epochs: 2
Accuracy Reached: 99.0
GPU Model: V100
Created By: stg-bed4agevt4237955bukje3t22e
Created Date: 2020-04-01 19:12:00 UTC
Status: UPLOAD_COMPLETE
Total File Count: 64
Total Size: 24.27 MB
Description:
Performance:
Quick Start Guide:
Release Notes:
Setup:
--------------------------------------------------
How to list resources¶
$ ngc registry resource list
+------------------+------------------+----------------+------------------+------------+------------+---------------+------------+
| Name | Repository | Latest Version | Application | Framework | Precision | Last Modified | Permission |
+------------------+------------------+----------------+------------------+------------+------------+---------------+------------+
| example_resource | example_org/exam | v1 | OBJECT_DETECTION | TensorFlow | FP16 | Feb 14, 2020 | unlocked |
| | ple_resource | | | | | | |
| | v1_5 | | | | | | |
+------------------+------------------+----------------+------------------+------------+------------+---------------+------------+
How to list resources with column arguments¶
$ ngc registry resource list --column name --column version
+------------------+------------------+----------------+
| Repository | Name | Latest Version |
+------------------+------------------+----------------+
| example_org/exam | example_resource | v1 |
| ple_resource | | |
| v1_5 | | |
+------------------+------------------+----------------+
How to download a resource version¶
$ ngc registry resource download-version example_org/example_resorurce:v1 --dest ./tmp
Downloaded 23.74 MB in 4s, Download speed: 5.92 MB/s
----------------------------------------------------
Transfer id: example_resource_vv1 Download status: Completed.
Downloaded local path: /home/a_user/tmp/example_resource_vv2
Total files downloaded: 10
Total downloaded size: 101.24 KB
Started at: 2020-04-01 14:19:39.517996
Completed at: 2020-04-01 14:19:43.525729
Duration taken: 2s
----------------------------------------------------
How to remove a resource or resource version¶
$ ngc registry resource remove example_org/example_resource
Are you sure you would like to remove example_org/example_resource? [y/n]y
Successfully removed resource 'example_org/example_resource'.
$ ngc registry resource remove example_org/example_resource:v1
Are you sure you would like to remove example_org/example_resource:v1? [y/n]y
Successfully removed model version 'example_org/example_resource:v1'.
How to update a resource or resource version¶
$ ngc registry resource update example_org/example_resource \
--framework PyTorch \
--precision FP32
Successfully updated resource 'example_org/example_resource'.
--------------------------------------------------
Resource Information
Name: example_resource
Application: OBJECT_DETECTION
Training Framework: PyTorch
Model Format: FMT
Precision: FP32
Short Description: A short description of my example_resource
Public Dataset Used:
Name:
Link:
License:
Display Name:
Logo:
Org: example_org
Team:
Built By:
Publisher:
Created Date: 2020-04-01 19:03:33 UTC
Updated Date: 2020-04-01 19:28:15 UTC
Can Guest Download: False
Read Only: False
Labels:
Latest Version ID: v2
Latest Version Size (bytes): 10240
Overview:
Setup:
Quick Start Guide:
Advanced Guide:
Performance:
--------------------------------------------------
$ ngc registry resource update example_org/example_resource:v2 --desc "a new description here" --batch-size 100
Successfully updated resource version 'example_org/example_resource:v2'.
--------------------------------------------------
Resource Version Information
Id: v2
Batch Size: 100
Memory Footprint: 8GB
Number Of Epochs: 2
Accuracy Reached: 99.0
GPU Model: V100
Created By: stg-bed4agevt4237955bukje3t22e
Created Date: 2020-04-01 19:23:09 UTC
Status: UPLOAD_COMPLETE
Total File Count: 10
Total Size: 100 KB
Description:
a new description here
Performance:
Quick Start Guide:
Release Notes:
Setup:
--------------------------------------------------
How to list helm charts¶
$ ngc registry chart list '*nginx*'
+---------+---------+---------+--------+---------+---------+---------+---------+
| Name | Reposit | Version | Size | Created | Descrip | Created | Last Mo |
| | ory | | | By | tion | Date | dified |
+---------+---------+---------+--------+---------+---------+---------+---------+
| nginx- | nvexmpl | 1.26.2 | 109.19 | ukp3qoa | | Apr 27, | Apr 27, |
| ingress | /nginx- | | KB | svi9h8v | | 2020 | 2020 |
| | ingress | | | vo3fkg9 | | | |
| | | | | afo9f | | | |
+---------+---------+---------+--------+---------+---------+---------+---------+
How to list helm charts with column arguments¶
$ ngc registry chart list '*nginx*' --column name --column labels
+-----------------------+---------------+--------+
| Repository | Name | Labels |
+-----------------------+---------------+--------+
| nvexmpl/nginx-ingress | nginx-ingress | |
+-----------------------+---------------+--------+
How to remove a registry chart¶
$ ngc registry chart remove example_org/digits:2.0
Successfully removed chart 'example_org/digits:2.0'
How to remove a registry chart repository¶
$ ngc registry chart remove example_org/digits
Successfully removed chart 'example_org/digits:2.0'
Successfully removed chart 'example_org/digits:1.0'
Successfully removed chart repository 'example_org/digits'
How to get information about a chart version¶
$ ngc registry chart info nvexmpl/nginx-ingress:1.26.2
--------------------------------------------------
Chart Version Information
Created Date: 2020-04-27 21:03:09 UTC
Updated Date: 2020-04-27 21:03:09 UTC
Version ID: 1.26.2
Total File Count: 68
Total Size: 109.19 KB
Status: UPLOAD_COMPLETE
--------------------------------------------------
How to get information about a chart repository¶
$ ngc registry chart info nvexmpl/nginx-ingress
--------------------------------------------------
Chart Information
Name: nginx-ingress
Short Description:
Display Name:
Team:
Publisher:
Built By:
Labels:
Logo:
Created Date: 2020-04-27 21:03:09 UTC
Updated Date: 2020-04-27 21:03:09 UTC
Read Only: False
Latest Version ID: 1.26.2
Latest Version Size (bytes): 111814
Overview:
An nginx Ingress controller that uses ConfigMap to store the nginx configuration.
--------------------------------------------------
How to pull a chart from the NGC chart registry¶
Pulling a chart from a registry requires you to specify the path, which includes org name, team name (if any), chart name, and version. Note that if you do not provide a version when pulling a chart, 'latest' will be assumed:
$ ngc registry chart list
+--------------------------------+--------------------------------+----------+----------+--------------------------------+------------------------------+--------------+---------------+
| Name | Repository | Version | Size | Created By | Description | Created Date | Last Modified |
+--------------------------------+--------------------------------+----------+----------+--------------------------------+------------------------------+--------------+---------------+
| ... | ... | ... | ... | ... | ... | ... | ... |
| digits | example_org/digits | 3.0 | 9 KB | | A Helm Chart for digits | Oct 23, 2020 | Oct 28, 2020 |
| ... | ... | ... | ... | ... | ... | ... | ... |
+--------------------------------+--------------------------------+----------+----------+--------------------------------+------------------------------+--------------+---------------+
$ ngc registry chart pull example_org/digits:3.0 --dest output/
Successfully pulled chart version '/Users/me/output/digits-3.0.tgz'.
$ ngc registry chart pull example_org/digits --dest output/
No version specified, downloading latest version: 3.0.
Successfully pulled chart version '/Users/me/output/digits-3.0.tgz'.
How to update a chart from the NGC chart registry¶
Updating a chart from a registry requires you to specify the path:
$ ngc registry chart update example_org/digits --short-desc new_description --label new_label --logo new_logo --publisher new_pub --built-by new_built
Successfully updated chart 'example_org/digits'.
--------------------------------------------------
Chart Information
Name: digits
Short Description: new_description
Display Name:
Team:
Publisher: new_pub
Built By: new_built
Labels:
new_label
Logo: new_logo
Created Date: 2020-10-23 16:00:37 UTC
Updated Date: 2020-11-05 20:13:35 UTC
Read Only: False
Latest Version ID: 0.4.0
Latest Version Size (bytes): 9268
Overview:
This is a general overview of the chart. It can contain lots of information about its usage.
--------------------------------------------------
How to push a chart to the NGC chart registry¶
Before you can push a chart, you have to create the overall chart record:
$ ngc registry chart create example_org/digits --short-desc "A Helm Chart for digits" --overview-filename overview.txt
Successfully created chart 'example_org/digits'.
--------------------------------------------------
Chart Information
Name: digits
Short Description: A Helm Chart for digits
Display Name:
Team:
Publisher:
Built By:
Labels:
Logo:
Created Date: 2020-11-05 21:38:27 UTC
Updated Date: 2020-11-05 21:38:27 UTC
Read Only: False
Latest Version ID:
Latest Version Size (bytes):
Overview:
This is a general overview of the chart. It can contain lots of information about its usage.
--------------------------------------------------
Once the chart is created, you can push versions of it to the helm repository:
$ ngc registry chart push example_org/digits:1.0
Successfully pushed chart version 'digits:1.0'.
--------------------------------------------------
Chart Version Information
Created Date: 2020-10-23 16:00:37 UTC
Updated Date: 2020-10-23 16:00:37 UTC
Version ID: 1.0
Total File Count: 11
Total Size: 10.25 KB
Status: UPLOAD_COMPLETE
--------------------------------------------------
How to list collections¶
$ ngc registry collection list
+-------------------------------------+-------------------------------------+--------------------------+--------------------------+
| Display Name | Name | Created | Updated |
+-------------------------------------+-------------------------------------+--------------------------+--------------------------+
| None | nvidia/ngc-web/ngcfegzg2xtmq | 2020-08-12T17:40:54.681Z | 2021-03-29T18:09:58.760Z |
| None | nvidia/ngcfe9hg8gl7u | 2020-08-12T17:40:54.689Z | 2021-03-29T18:09:58.055Z |
| None | nvidia/ngc-web/ngcfe2qc1y2p4 | 2020-08-12T17:52:22.250Z | 2021-03-29T18:10:01.759Z |
| None | nvidia/ngcfeykzaahcm | 2020-08-12T17:52:22.298Z | 2021-03-29T18:10:00.681Z |
| None | nvidia/ngc-web/ngcfejyeogzwm | 2020-08-12T17:54:35.579Z | 2021-03-29T18:09:58.939Z |
| None | nvidia/ngcfellsbe7xx | 2020-08-12T17:54:35.581Z | 2021-03-29T18:09:58.941Z |
| None | nvidia/ngc-web/ngcfec6weygws | 2020-08-12T18:02:01.309Z | 2021-03-29T18:09:58.423Z |
...
How to remove a collection¶
$ ngc registry collection remove example_org/collectionname
Are you sure you would like to remove example_org/collectionname? [y/n]y
Successfully removed collection 'example_org/collectionname'.
How to get collection information¶
$ ngc registry collection info example_org/collectionname
--------------------------------------------------
Collection Information
Name: collectionname
Org: example_org
Team:
Display Name:
Logo:
Built By:
Publisher:
Created Date: 2021-05-03T19:47:20.044Z
Updated Date: 2021-05-03T19:47:20.044Z
Short Description: Dummy description
Category: BEGINNER
Labels:
Images:
Charts:
Models:
Resources:
Overview:
--------------------------------------------------
How to create a collection¶
When you create a collection, all of the entities you wish to link to the collection (images, helm charts, etc.) must already exist.
$ ngc registry collection create example_org/collectionname --short-desc "Dummy description" --category beginner --add-image orgname/imagename
--------------------------------------------------
Collection Information
Name: collectionname
Org: example_org
Team:
Display Name:
Logo:
Built By:
Publisher:
Created Date: 2021-05-03T19:47:20.044Z
Updated Date: 2021-05-03T19:47:20.044Z
Short Description: Dummy description
Category: BEGINNER
Labels:
Images:
orgname/imagename
Charts:
Models:
Resources:
Overview:
--------------------------------------------------
How to update a collection¶
Any of the fields of a collection can be updated after it's been created. You can use --help for a full list of argument options.
$ ngc registry collection update example_org/collectionname --short-desc "New dummy description" --add-chart orgname/chartname --remove-image orgname/imagename
--------------------------------------------------
Collection Information
Name: collectionname
Org: example_org
Team:
Display Name:
Logo:
Built By:
Publisher:
Created Date: 2021-05-03T19:47:20.044Z
Updated Date: 2021-05-03T19:47:20.044Z
Short Description: New dummy description
Category: BEGINNER
Labels:
Images:
Charts:
orgname/chartname
Models:
Resources:
Overview:
--------------------------------------------------