Skip to content

LiveDeployment.md

Latest commit

 

History

History
242 lines (168 loc) · 8.06 KB

LiveDeployment.md

File metadata and controls

242 lines (168 loc) · 8.06 KB

LiveDeployment

Method HTTP request Release Stage
create POST /v2/models/liveDeployments Private Beta
get GET /v2/models/liveDeployments/{liveDeploymentRid} Private Beta
replace PUT /v2/models/liveDeployments/{liveDeploymentRid} Private Beta
transform_json POST /v2/models/liveDeployments/{liveDeploymentRid}/transformJson Public Beta

create

Creates a new live deployment for a model version with the specified runtime configuration. The deployment will begin provisioning compute resources and deploying the target model version.

Parameters

Name Type Description Notes
deployment_type CreateLiveDeploymentTarget The target model source for the live deployment. Determines which model and version selection strategy to use when creating the deployment.
runtime_configuration LiveDeploymentRuntimeConfiguration The compute resource configuration for the deployment.
preview Optional[PreviewMode] Enables the use of preview functionality. [optional]

Return type

LiveDeployment

Example

from foundry_sdk import FoundryClient
import foundry_sdk
from pprint import pprint

client = FoundryClient(auth=foundry_sdk.UserTokenAuth(...), hostname="example.palantirfoundry.com")

# CreateLiveDeploymentTarget | The target model source for the live deployment. Determines which model and version selection strategy to use when creating the deployment.
deployment_type = None
# LiveDeploymentRuntimeConfiguration | The compute resource configuration for the deployment.
runtime_configuration = {
    "minReplicas": 1,
    "maxReplicas": 3,
    "cpu": 1.0,
    "memory": "256MiB",
    "threadCount": 32,
}
# Optional[PreviewMode] | Enables the use of preview functionality.
preview = None


try:
    api_response = client.models.LiveDeployment.create(
        deployment_type=deployment_type,
        runtime_configuration=runtime_configuration,
        preview=preview,
    )
    print("The create response:\n")
    pprint(api_response)
except foundry_sdk.PalantirRPCException as e:
    print("HTTP error when calling LiveDeployment.create: %s\n" % e)

Authorization

See README

HTTP response details

Status Code Type Description Content Type
200 LiveDeployment The created LiveDeployment application/json

[Back to top] [Back to API list] [Back to Model list] [Back to README]

get

Retrieves a live deployment by its Resource Identifier (RID), including its deployed model version and runtime configuration.

Parameters

Name Type Description Notes
live_deployment_rid LiveDeploymentRid
preview Optional[PreviewMode] Enables the use of preview functionality. [optional]

Return type

LiveDeployment

Example

from foundry_sdk import FoundryClient
import foundry_sdk
from pprint import pprint

client = FoundryClient(auth=foundry_sdk.UserTokenAuth(...), hostname="example.palantirfoundry.com")

# LiveDeploymentRid
live_deployment_rid = None
# Optional[PreviewMode] | Enables the use of preview functionality.
preview = None


try:
    api_response = client.models.LiveDeployment.get(live_deployment_rid, preview=preview)
    print("The get response:\n")
    pprint(api_response)
except foundry_sdk.PalantirRPCException as e:
    print("HTTP error when calling LiveDeployment.get: %s\n" % e)

Authorization

See README

HTTP response details

Status Code Type Description Content Type
200 LiveDeployment application/json

[Back to top] [Back to API list] [Back to Model list] [Back to README]

replace

Updates the runtime configuration of the live deployment. The deployment will apply the new configuration to the running replicas.

Parameters

Name Type Description Notes
live_deployment_rid LiveDeploymentRid
runtime_configuration LiveDeploymentRuntimeConfiguration The compute resource configuration for the deployment.
preview Optional[PreviewMode] Enables the use of preview functionality. [optional]

Return type

LiveDeployment

Example

from foundry_sdk import FoundryClient
import foundry_sdk
from pprint import pprint

client = FoundryClient(auth=foundry_sdk.UserTokenAuth(...), hostname="example.palantirfoundry.com")

# LiveDeploymentRid
live_deployment_rid = None
# LiveDeploymentRuntimeConfiguration | The compute resource configuration for the deployment.
runtime_configuration = {
    "minReplicas": 1,
    "maxReplicas": 3,
    "cpu": 1.0,
    "memory": "256MiB",
    "threadCount": 32,
}
# Optional[PreviewMode] | Enables the use of preview functionality.
preview = None


try:
    api_response = client.models.LiveDeployment.replace(
        live_deployment_rid, runtime_configuration=runtime_configuration, preview=preview
    )
    print("The replace response:\n")
    pprint(api_response)
except foundry_sdk.PalantirRPCException as e:
    print("HTTP error when calling LiveDeployment.replace: %s\n" % e)

Authorization

See README

HTTP response details

Status Code Type Description Content Type
200 LiveDeployment The replaced LiveDeployment application/json

[Back to top] [Back to API list] [Back to Model list] [Back to README]

transform_json

Performs inference on the live deployment.

Parameters

Name Type Description Notes
live_deployment_rid LiveDeploymentRid
input Dict[str, Any] The input data for the model inference. The structure should match the model's transform API specification, where each key is an input name and the value is the corresponding input data.
preview Optional[PreviewMode] Enables the use of preview functionality. [optional]

Return type

TransformLiveDeploymentResponse

Example

from foundry_sdk import FoundryClient
import foundry_sdk
from pprint import pprint

client = FoundryClient(auth=foundry_sdk.UserTokenAuth(...), hostname="example.palantirfoundry.com")

# LiveDeploymentRid
live_deployment_rid = None
# Dict[str, Any] | The input data for the model inference. The structure should match the model's transform API specification, where each key is an input name and the value is the corresponding input data.
input = {"input_df": [{"feature_1": 1.0, "feature_2": 2}]}
# Optional[PreviewMode] | Enables the use of preview functionality.
preview = None


try:
    api_response = client.models.LiveDeployment.transform_json(
        live_deployment_rid, input=input, preview=preview
    )
    print("The transform_json response:\n")
    pprint(api_response)
except foundry_sdk.PalantirRPCException as e:
    print("HTTP error when calling LiveDeployment.transform_json: %s\n" % e)

Authorization

See README

HTTP response details

Status Code Type Description Content Type
200 TransformLiveDeploymentResponse application/json

[Back to top] [Back to API list] [Back to Model list] [Back to README]