oracle.oci.oci_ai_language_model – Manage a Model resource in Oracle Cloud Infrastructure

Note

This plugin is part of the oracle.oci collection (version 5.5.0).

You might already have this collection installed if you are using the ansible package. It is not included in ansible-core. To check whether it is installed, run ansible-galaxy collection list.

To install it, use: ansible-galaxy collection install oracle.oci.

To use it in a playbook, specify: oracle.oci.oci_ai_language_model.

New in version 2.9.0: of oracle.oci

Synopsis

  • This module allows the user to create, update and delete a Model resource in Oracle Cloud Infrastructure

  • For state=present, creates a new model for training and train the model with date provided.

  • This resource has the following action operations in the oracle.oci.oci_ai_language_model_actions module: change_compartment.

Requirements

The below requirements are needed on the host that executes this module.

Parameters

Parameter Choices/Defaults Comments
api_user
string
The OCID of the user, on whose behalf, OCI APIs are invoked. If not set, then the value of the OCI_USER_ID environment variable, if any, is used. This option is required if the user is not specified through a configuration file (See config_file_location). To get the user's OCID, please refer https://docs.us-phoenix-1.oraclecloud.com/Content/API/Concepts/apisigningkey.htm.
api_user_fingerprint
string
Fingerprint for the key pair being used. If not set, then the value of the OCI_USER_FINGERPRINT environment variable, if any, is used. This option is required if the key fingerprint is not specified through a configuration file (See config_file_location). To get the key pair's fingerprint value please refer https://docs.us-phoenix-1.oraclecloud.com/Content/API/Concepts/apisigningkey.htm.
api_user_key_file
string
Full path and filename of the private key (in PEM format). If not set, then the value of the OCI_USER_KEY_FILE variable, if any, is used. This option is required if the private key is not specified through a configuration file (See config_file_location). If the key is encrypted with a pass-phrase, the api_user_key_pass_phrase option must also be provided.
api_user_key_pass_phrase
string
Passphrase used by the key referenced in api_user_key_file, if it is encrypted. If not set, then the value of the OCI_USER_KEY_PASS_PHRASE variable, if any, is used. This option is required if the key passphrase is not specified through a configuration file (See config_file_location).
auth_purpose
string
    Choices:
  • service_principal
The auth purpose which can be used in conjunction with 'auth_type=instance_principal'. The default auth_purpose for instance_principal is None.
auth_type
string
    Choices:
  • api_key ←
  • instance_principal
  • instance_obo_user
  • resource_principal
  • security_token
The type of authentication to use for making API requests. By default auth_type="api_key" based authentication is performed and the API key (see api_user_key_file) in your config file will be used. If this 'auth_type' module option is not specified, the value of the OCI_ANSIBLE_AUTH_TYPE, if any, is used. Use auth_type="instance_principal" to use instance principal based authentication when running ansible playbooks within an OCI compute instance.
cert_bundle
string
The full path to a CA certificate bundle to be used for SSL verification. This will override the default CA certificate bundle. If not set, then the value of the OCI_ANSIBLE_CERT_BUNDLE variable, if any, is used.
compartment_id
string
The OCID for the models compartment.
Required for create using state=present.
Required for update when environment variable OCI_USE_NAME_AS_IDENTIFIER is set.
Required for delete when environment variable OCI_USE_NAME_AS_IDENTIFIER is set.
config_file_location
string
Path to configuration file. If not set then the value of the OCI_CONFIG_FILE environment variable, if any, is used. Otherwise, defaults to ~/.oci/config.
config_profile_name
string
The profile to load from the config file referenced by config_file_location. If not set, then the value of the OCI_CONFIG_PROFILE environment variable, if any, is used. Otherwise, defaults to the "DEFAULT" profile in config_file_location.
defined_tags
dictionary
Defined tags for this resource. Each key is predefined and scoped to a namespace. Example: `{"foo-namespace": {"bar-key": "value"}}`
This parameter is updatable.
description
string
A short description of the a model.
This parameter is updatable.
display_name
string
A user-friendly display name for the resource. It does not have to be unique and can be modified. Avoid entering confidential information.
Required for create, update, delete when environment variable OCI_USE_NAME_AS_IDENTIFIER is set.
This parameter is updatable when OCI_USE_NAME_AS_IDENTIFIER is not set.

aliases: name
force_create
boolean
    Choices:
  • no ←
  • yes
Whether to attempt non-idempotent creation of a resource. By default, create resource is an idempotent operation, and doesn't create the resource if it already exists. Setting this option to true, forcefully creates a copy of the resource, even if it already exists.This option is mutually exclusive with key_by.
freeform_tags
dictionary
Simple key-value pair that is applied without any predefined name, type or scope. Exists for cross-compatibility only. Example: `{"bar-key": "value"}`
This parameter is updatable.
key_by
list / elements=string
The list of attributes of this resource which should be used to uniquely identify an instance of the resource. By default, all the attributes of a resource are used to uniquely identify a resource.
model_details
dictionary
Required for create using state=present.
classification_mode
dictionary
Applicable when model_type is 'TEXT_CLASSIFICATION'
classification_mode
string / required
    Choices:
  • MULTI_CLASS
  • MULTI_LABEL
classification Modes
version
string
Optional if nothing specified latest base model will be used for training. Supported versions can be found at /modelTypes/{modelType}
language_code
string
supported language default value is en
model_type
string / required
    Choices:
  • PRE_TRAINED_KEYPHRASE_EXTRACTION
  • PRE_TRAINED_HEALTH_NLU
  • PRE_TRAINED_UNIVERSAL
  • NAMED_ENTITY_RECOGNITION
  • PRE_TRAINED_LANGUAGE_DETECTION
  • PRE_TRAINED_NAMED_ENTITY_RECOGNITION
  • PRE_TRAINED_SENTIMENT_ANALYSIS
  • PRE_TRAINED_PHI
  • PRE_TRAINED_TEXT_CLASSIFICATION
  • TEXT_CLASSIFICATION
  • PRE_TRAINED_SUMMARIZATION
  • PRE_TRAINED_PII
Model type
version
string
Optional pre trained model version. if nothing specified latest pre trained model will be used. Supported versions can be found at /modelTypes/{modelType}
Applicable when model_type is one of ['NAMED_ENTITY_RECOGNITION', 'PRE_TRAINED_PII', 'PRE_TRAINED_PHI', 'PRE_TRAINED_TEXT_CLASSIFICATION', 'PRE_TRAINED_NAMED_ENTITY_RECOGNITION', 'PRE_TRAINED_HEALTH_NLU', 'PRE_TRAINED_LANGUAGE_DETECTION', 'PRE_TRAINED_KEYPHRASE_EXTRACTION', 'PRE_TRAINED_SENTIMENT_ANALYSIS', 'PRE_TRAINED_SUMMARIZATION', 'PRE_TRAINED_UNIVERSAL']
model_id
string
unique model OCID.
Required for update using state=present when environment variable OCI_USE_NAME_AS_IDENTIFIER is not set.
Required for delete using state=absent when environment variable OCI_USE_NAME_AS_IDENTIFIER is not set.

aliases: id
project_id
string
The OCID of the project to associate with the model.
Required for create using state=present.
realm_specific_endpoint_template_enabled
boolean
    Choices:
  • no
  • yes
Enable/Disable realm specific endpoint template for service client. By Default, realm specific endpoint template is disabled. If not set, then the value of the OCI_REALM_SPECIFIC_SERVICE_ENDPOINT_TEMPLATE_ENABLED variable, if any, is used.
region
string
The Oracle Cloud Infrastructure region to use for all OCI API requests. If not set, then the value of the OCI_REGION variable, if any, is used. This option is required if the region is not specified through a configuration file (See config_file_location). Please refer to https://docs.us-phoenix-1.oraclecloud.com/Content/General/Concepts/regions.htm for more information on OCI regions.
state
string
    Choices:
  • present ←
  • absent
The state of the Model.
Use state=present to create or update a Model.
Use state=absent to delete a Model.
tenancy
string
OCID of your tenancy. If not set, then the value of the OCI_TENANCY variable, if any, is used. This option is required if the tenancy OCID is not specified through a configuration file (See config_file_location). To get the tenancy OCID, please refer https://docs.us-phoenix-1.oraclecloud.com/Content/API/Concepts/apisigningkey.htm
test_strategy
dictionary
strategy_type
string / required
    Choices:
  • TEST_AND_VALIDATION_DATASET
This information will define the test strategy different datasets for test and validation(optional) dataset.
testing_dataset
dictionary / required
dataset_id
string
Data Science Labelling Service OCID
Required when dataset_type is 'DATA_SCIENCE_LABELING'
dataset_type
string / required
    Choices:
  • DATA_SCIENCE_LABELING
  • OBJECT_STORAGE
Possible data sets
location_details
dictionary
Required when dataset_type is 'OBJECT_STORAGE'
bucket_name
string / required
Object storage bucket name
location_type
string / required
    Choices:
  • OBJECT_LIST
Possible object storage location types
namespace_name
string / required
Object storage namespace
object_names
list / elements=string / required
Array of files which need to be processed in the bucket
validation_dataset
dictionary
dataset_id
string
Data Science Labelling Service OCID
Required when dataset_type is 'DATA_SCIENCE_LABELING'
dataset_type
string / required
    Choices:
  • DATA_SCIENCE_LABELING
  • OBJECT_STORAGE
Possible data sets
location_details
dictionary
Required when dataset_type is 'OBJECT_STORAGE'
bucket_name
string / required
Object storage bucket name
location_type
string / required
    Choices:
  • OBJECT_LIST
Possible object storage location types
namespace_name
string / required
Object storage namespace
object_names
list / elements=string / required
Array of files which need to be processed in the bucket
training_dataset
dictionary
dataset_id
string
Data Science Labelling Service OCID
Required when dataset_type is 'DATA_SCIENCE_LABELING'
dataset_type
string / required
    Choices:
  • DATA_SCIENCE_LABELING
  • OBJECT_STORAGE
Possible data sets
location_details
dictionary
Required when dataset_type is 'OBJECT_STORAGE'
bucket_name
string / required
Object storage bucket name
location_type
string / required
    Choices:
  • OBJECT_LIST
Possible object storage location types
namespace_name
string / required
Object storage namespace
object_names
list / elements=string / required
Array of files which need to be processed in the bucket
wait
boolean
    Choices:
  • no
  • yes ←
Whether to wait for create or delete operation to complete.
wait_timeout
integer
Time, in seconds, to wait when wait=yes. Defaults to 1200 for most of the services but some services might have a longer wait timeout.

Examples

- name: Create model
  oci_ai_language_model:
    # required
    compartment_id: "ocid1.compartment.oc1..xxxxxxEXAMPLExxxxxx"
    project_id: "ocid1.project.oc1..xxxxxxEXAMPLExxxxxx"
    model_details:
      # required
      model_type: PRE_TRAINED_KEYPHRASE_EXTRACTION

      # optional
      language_code: language_code_example
      version: version_example

    # optional
    training_dataset:
      # required
      dataset_id: "ocid1.dataset.oc1..xxxxxxEXAMPLExxxxxx"
      dataset_type: DATA_SCIENCE_LABELING
    test_strategy:
      # required
      strategy_type: TEST_AND_VALIDATION_DATASET
      testing_dataset:
        # required
        dataset_id: "ocid1.dataset.oc1..xxxxxxEXAMPLExxxxxx"
        dataset_type: DATA_SCIENCE_LABELING

        # optional
      validation_dataset:
        # required
        dataset_id: "ocid1.dataset.oc1..xxxxxxEXAMPLExxxxxx"
        dataset_type: DATA_SCIENCE_LABELING
    display_name: display_name_example
    description: description_example
    freeform_tags: {'Department': 'Finance'}
    defined_tags: {'Operations': {'CostCenter': 'US'}}

- name: Update model
  oci_ai_language_model:
    # required
    model_id: "ocid1.model.oc1..xxxxxxEXAMPLExxxxxx"

    # optional
    display_name: display_name_example
    description: description_example
    freeform_tags: {'Department': 'Finance'}
    defined_tags: {'Operations': {'CostCenter': 'US'}}

- name: Update model using name (when environment variable OCI_USE_NAME_AS_IDENTIFIER is set)
  oci_ai_language_model:
    # required
    compartment_id: "ocid1.compartment.oc1..xxxxxxEXAMPLExxxxxx"
    display_name: display_name_example

    # optional
    description: description_example
    freeform_tags: {'Department': 'Finance'}
    defined_tags: {'Operations': {'CostCenter': 'US'}}

- name: Delete model
  oci_ai_language_model:
    # required
    model_id: "ocid1.model.oc1..xxxxxxEXAMPLExxxxxx"
    state: absent

- name: Delete model using name (when environment variable OCI_USE_NAME_AS_IDENTIFIER is set)
  oci_ai_language_model:
    # required
    compartment_id: "ocid1.compartment.oc1..xxxxxxEXAMPLExxxxxx"
    display_name: display_name_example
    state: absent

Return Values

Common return values are documented here, the following are the fields unique to this module:

Key Returned Description
model
complex
on success
Details of the Model resource acted upon by the current operation

Sample:
{'compartment_id': 'ocid1.compartment.oc1..xxxxxxEXAMPLExxxxxx', 'defined_tags': {'Operations': {'CostCenter': 'US'}}, 'description': 'description_example', 'display_name': 'display_name_example', 'evaluation_results': {'class_metrics': [{'f1': 3.4, 'label': 'label_example', 'precision': 3.4, 'recall': 3.4, 'support': 3.4}], 'confusion_matrix': {'matrix': {}}, 'entity_metrics': [{'f1': 3.4, 'label': 'label_example', 'precision': 3.4, 'recall': 3.4}], 'labels': [], 'metrics': {'accuracy': 3.4, 'macro_f1': 3.4, 'macro_precision': 3.4, 'macro_recall': 3.4, 'micro_f1': 3.4, 'micro_precision': 3.4, 'micro_recall': 3.4, 'weighted_f1': 3.4, 'weighted_precision': 3.4, 'weighted_recall': 3.4}, 'model_type': 'NAMED_ENTITY_RECOGNITION'}, 'freeform_tags': {'Department': 'Finance'}, 'id': 'ocid1.resource.oc1..xxxxxxEXAMPLExxxxxx', 'lifecycle_details': 'lifecycle_details_example', 'lifecycle_state': 'DELETING', 'model_details': {'classification_mode': {'classification_mode': 'MULTI_CLASS', 'version': 'version_example'}, 'language_code': 'language_code_example', 'model_type': 'NAMED_ENTITY_RECOGNITION', 'version': 'version_example'}, 'project_id': 'ocid1.project.oc1..xxxxxxEXAMPLExxxxxx', 'system_tags': {}, 'test_strategy': {'strategy_type': 'TEST_AND_VALIDATION_DATASET', 'testing_dataset': {'dataset_id': 'ocid1.dataset.oc1..xxxxxxEXAMPLExxxxxx', 'dataset_type': 'OBJECT_STORAGE', 'location_details': {'bucket_name': 'bucket_name_example', 'location_type': 'OBJECT_LIST', 'namespace_name': 'namespace_name_example', 'object_names': []}}, 'validation_dataset': {'dataset_id': 'ocid1.dataset.oc1..xxxxxxEXAMPLExxxxxx', 'dataset_type': 'OBJECT_STORAGE', 'location_details': {'bucket_name': 'bucket_name_example', 'location_type': 'OBJECT_LIST', 'namespace_name': 'namespace_name_example', 'object_names': []}}}, 'time_created': '2013-10-20T19:20:30+01:00', 'time_updated': '2013-10-20T19:20:30+01:00', 'training_dataset': {'dataset_id': 'ocid1.dataset.oc1..xxxxxxEXAMPLExxxxxx', 'dataset_type': 'OBJECT_STORAGE', 'location_details': {'bucket_name': 'bucket_name_example', 'location_type': 'OBJECT_LIST', 'namespace_name': 'namespace_name_example', 'object_names': []}}, 'version': 'version_example'}
 
compartment_id
string
on success
The OCID for the model's compartment.

Sample:
ocid1.compartment.oc1..xxxxxxEXAMPLExxxxxx
 
defined_tags
dictionary
on success
Defined tags for this resource. Each key is predefined and scoped to a namespace. Example: `{"foo-namespace": {"bar-key": "value"}}`

Sample:
{'Operations': {'CostCenter': 'US'}}
 
description
string
on success
A short description of the Model.

Sample:
description_example
 
display_name
string
on success
A user-friendly display name for the resource. It does not have to be unique and can be modified. Avoid entering confidential information.

Sample:
display_name_example
 
evaluation_results
complex
on success

   
class_metrics
complex
on success
List of text classification metrics
     
f1
float
on success
F1-score, is a measure of a model's accuracy on a dataset

Sample:
3.4
     
label
string
on success
Text classification label

Sample:
label_example
     
precision
float
on success
Precision refers to the number of true positives divided by the total number of positive predictions (i.e., the number of true positives plus the number of false positives)

Sample:
3.4
     
recall
float
on success
Measures the model's ability to predict actual positive classes. It is the ratio between the predicted true positives and what was actually tagged. The recall metric reveals how many of the predicted classes are correct.

Sample:
3.4
     
support
float
on success
number of samples in the test set

Sample:
3.4
   
confusion_matrix
complex
on success
class level confusion matrix
     
matrix
dictionary
on success
confusion matrix data
   
entity_metrics
complex
on success
List of entity metrics
     
f1
float
on success
F1-score, is a measure of a model's accuracy on a dataset

Sample:
3.4
     
label
string
on success
Entity label

Sample:
label_example
     
precision
float
on success
Precision refers to the number of true positives divided by the total number of positive predictions (i.e., the number of true positives plus the number of false positives)

Sample:
3.4
     
recall
float
on success
Measures the model's ability to predict actual positive classes. It is the ratio between the predicted true positives and what was actually tagged. The recall metric reveals how many of the predicted classes are correct.

Sample:
3.4
   
labels
list / elements=string
on success
labels
   
metrics
complex
on success

     
accuracy
float
on success
The fraction of the labels that were correctly recognised .

Sample:
3.4
     
macro_f1
float
on success
F1-score, is a measure of a model's accuracy on a dataset

Sample:
3.4
     
macro_precision
float
on success
Precision refers to the number of true positives divided by the total number of positive predictions (i.e., the number of true positives plus the number of false positives)

Sample:
3.4
     
macro_recall
float
on success
Measures the model's ability to predict actual positive classes. It is the ratio between the predicted true positives and what was actually tagged. The recall metric reveals how many of the predicted classes are correct.

Sample:
3.4
     
micro_f1
float
on success
F1-score, is a measure of a model's accuracy on a dataset

Sample:
3.4
     
micro_precision
float
on success
Precision refers to the number of true positives divided by the total number of positive predictions (i.e., the number of true positives plus the number of false positives)

Sample:
3.4
     
micro_recall
float
on success
Measures the model's ability to predict actual positive classes. It is the ratio between the predicted true positives and what was actually tagged. The recall metric reveals how many of the predicted classes are correct.

Sample:
3.4
     
weighted_f1
float
on success
F1-score, is a measure of a model's accuracy on a dataset

Sample:
3.4
     
weighted_precision
float
on success
Precision refers to the number of true positives divided by the total number of positive predictions (i.e., the number of true positives plus the number of false positives)

Sample:
3.4
     
weighted_recall
float
on success
Measures the model's ability to predict actual positive classes. It is the ratio between the predicted true positives and what was actually tagged. The recall metric reveals how many of the predicted classes are correct.

Sample:
3.4
   
model_type
string
on success
Model type

Sample:
NAMED_ENTITY_RECOGNITION
 
freeform_tags
dictionary
on success
Simple key-value pair that is applied without any predefined name, type or scope. Exists for cross-compatibility only. Example: `{"bar-key": "value"}`

Sample:
{'Department': 'Finance'}
 
id
string
on success
Unique identifier model OCID of a model that is immutable on creation

Sample:
ocid1.resource.oc1..xxxxxxEXAMPLExxxxxx
 
lifecycle_details
string
on success
A message describing the current state in more detail. For example, can be used to provide actionable information for a resource in failed state.

Sample:
lifecycle_details_example
 
lifecycle_state
string
on success
The state of the model.

Sample:
DELETING
 
model_details
complex
on success

   
classification_mode
complex
on success

     
classification_mode
string
on success
classification Modes

Sample:
MULTI_CLASS
     
version
string
on success
Optional if nothing specified latest base model will be used for training. Supported versions can be found at /modelTypes/{modelType}

Sample:
version_example
   
language_code
string
on success
supported language default value is en

Sample:
language_code_example
   
model_type
string
on success
Model type

Sample:
NAMED_ENTITY_RECOGNITION
   
version
string
on success
Optional if nothing specified latest base model will be used for training. Supported versions can be found at /modelTypes/{modelType}

Sample:
version_example
 
project_id
string
on success
The OCID of the project to associate with the model.

Sample:
ocid1.project.oc1..xxxxxxEXAMPLExxxxxx
 
system_tags
dictionary
on success
Usage of system tag keys. These predefined keys are scoped to namespaces. Example: `{ "orcl-cloud": { "free-tier-retained": "true" } }`
 
test_strategy
complex
on success

   
strategy_type
string
on success
This information will define the test strategy different datasets for test and validation(optional) dataset.

Sample:
TEST_AND_VALIDATION_DATASET
   
testing_dataset
complex
on success

     
dataset_id
string
on success
Data Science Labelling Service OCID

Sample:
ocid1.dataset.oc1..xxxxxxEXAMPLExxxxxx
     
dataset_type
string
on success
Possible data sets

Sample:
OBJECT_STORAGE
     
location_details
complex
on success

       
bucket_name
string
on success
Object storage bucket name

Sample:
bucket_name_example
       
location_type
string
on success
Possible object storage location types

Sample:
OBJECT_LIST
       
namespace_name
string
on success
Object storage namespace

Sample:
namespace_name_example
       
object_names
list / elements=string
on success
Array of files which need to be processed in the bucket
   
validation_dataset
complex
on success

     
dataset_id
string
on success
Data Science Labelling Service OCID

Sample:
ocid1.dataset.oc1..xxxxxxEXAMPLExxxxxx
     
dataset_type
string
on success
Possible data sets

Sample:
OBJECT_STORAGE
     
location_details
complex
on success

       
bucket_name
string
on success
Object storage bucket name

Sample:
bucket_name_example
       
location_type
string
on success
Possible object storage location types

Sample:
OBJECT_LIST
       
namespace_name
string
on success
Object storage namespace

Sample:
namespace_name_example
       
object_names
list / elements=string
on success
Array of files which need to be processed in the bucket
 
time_created
string
on success
The time the the model was created. An RFC3339 formatted datetime string.

Sample:
2013-10-20T19:20:30+01:00
 
time_updated
string
on success
The time the model was updated. An RFC3339 formatted datetime string.

Sample:
2013-10-20T19:20:30+01:00
 
training_dataset
complex
on success

   
dataset_id
string
on success
Data Science Labelling Service OCID

Sample:
ocid1.dataset.oc1..xxxxxxEXAMPLExxxxxx
   
dataset_type
string
on success
Possible data sets

Sample:
OBJECT_STORAGE
   
location_details
complex
on success

     
bucket_name
string
on success
Object storage bucket name

Sample:
bucket_name_example
     
location_type
string
on success
Possible object storage location types

Sample:
OBJECT_LIST
     
namespace_name
string
on success
Object storage namespace

Sample:
namespace_name_example
     
object_names
list / elements=string
on success
Array of files which need to be processed in the bucket
 
version
string
on success
For pre trained models this will identify model type version used for model creation For custom identifying the model by model id is difficult. This param provides ease of use for end customer. <<service>>::<<service-name>>_<<model-type-version>>::<<custom model on which this training has to be done>> ex: ai-lang::NER_V1::CUSTOM-V0

Sample:
version_example


Authors

  • Oracle (@oracle)