Merge branch 'main' into v2.11.8

VipulMascarenhas · VipulMascarenhas · commit 4ce8a8eab0ec · 2024-04-24T15:36:56.000-07:00
diff --git a/ads/aqua/__init__.py b/ads/aqua/__init__.py
@@ -14,7 +14,7 @@
 
 
 def get_logger_level():
-    """Retrieves logging level from environment variable `LOG_LEVEL`."""
+    """Retrieves logging level from environment variable `ADS_AQUA_LOG_LEVEL`."""
     level = os.environ.get(ENV_VAR_LOG_LEVEL, "INFO").upper()
     return level
 
diff --git a/ads/aqua/base.py b/ads/aqua/base.py
@@ -163,7 +163,6 @@ def create_model_version_set(
         tag = Tags.AQUA_FINE_TUNING.value
 
         if not model_version_set_id:
-            tag = Tags.AQUA_FINE_TUNING.value  # TODO: Fix this
             try:
                 model_version_set = ModelVersionSet.from_name(
                     name=model_version_set_name,
diff --git a/ads/aqua/cli.py b/ads/aqua/cli.py
@@ -17,6 +17,7 @@
 from ads.aqua.finetune import AquaFineTuningApp
 from ads.aqua.model import AquaModelApp
 from ads.config import NB_SESSION_OCID
+from ads.common.utils import LOG_LEVELS
 
 
 class AquaCommand:
@@ -47,14 +48,20 @@ def __init__(
             or 'ERROR' if not set. Example values include 'DEBUG', 'INFO',
             'WARNING', 'ERROR', and 'CRITICAL'.
         """
+        if log_level.upper() not in LOG_LEVELS:
+            logger.error(
+                f"Log level should be one of {LOG_LEVELS}. Setting default to ERROR."
+            )
+            log_level = "ERROR"
         set_log_level(log_level)
         # gracefully exit if env var is not set
         if not ODSC_MODEL_COMPARTMENT_OCID:
-            logger.error(
+            logger.debug(
                 "ODSC_MODEL_COMPARTMENT_OCID environment variable is not set for Aqua."
             )
             if NB_SESSION_OCID:
                 logger.error(
-                    f"Aqua is not available for the notebook session {NB_SESSION_OCID}."
+                    f"Aqua is not available for the notebook session {NB_SESSION_OCID}. For more information, "
+                    f"please refer to the documentation."
                 )
             sys.exit(1)
diff --git a/ads/aqua/deployment.py b/ads/aqua/deployment.py
@@ -22,6 +22,9 @@
     UNKNOWN_DICT,
     get_resource_name,
     get_model_by_reference_paths,
+    get_ocid_substring,
+    AQUA_MODEL_TYPE_SERVICE,
+    AQUA_MODEL_TYPE_CUSTOM,
 )
 from ads.aqua.finetune import FineTuneCustomMetadata
 from ads.aqua.data import AquaResourceIdentifier
@@ -391,11 +394,13 @@ def create(
             .with_runtime(container_runtime)
         ).deploy(wait_for_completion=False)
 
-        model_type = "custom" if is_fine_tuned_model else "service"
-        deployment_id = deployment.dsc_model_deployment.id
-        telemetry_kwargs = (
-            {"ocid": deployment_id[-8:]} if len(deployment_id) > 8 else {}
+        model_type = (
+            AQUA_MODEL_TYPE_CUSTOM if is_fine_tuned_model else AQUA_MODEL_TYPE_SERVICE
         )
+        deployment_id = deployment.dsc_model_deployment.id
+        # we arbitrarily choose last 8 characters of OCID to identify MD in telemetry
+        telemetry_kwargs = {"ocid": get_ocid_substring(deployment_id, key_len=8)}
+
         # tracks unique deployments that were created in the user compartment
         self.telemetry.record_event_async(
             category=f"aqua/{model_type}/deployment",
@@ -461,10 +466,11 @@ def list(self, **kwargs) -> List["AquaDeployment"]:
                 state = model_deployment.lifecycle_state.upper()
                 if state in ["ACTIVE", "FAILED"]:
                     # tracks unique deployments that were listed in the user compartment
+                    # we arbitrarily choose last 8 characters of OCID to identify MD in telemetry
                     self.telemetry.record_event_async(
                         category=f"aqua/deployment",
                         action="list",
-                        detail=deployment_id[-8:] if len(deployment_id) > 8 else "",
+                        detail=get_ocid_substring(deployment_id, key_len=8),
                         value=state,
                     )
 
diff --git a/ads/aqua/utils.py b/ads/aqua/utils.py
@@ -73,6 +73,8 @@
 READY_TO_DEPLOY_STATUS = "ACTIVE"
 READY_TO_FINE_TUNE_STATUS = "TRUE"
 AQUA_GA_LIST = ["id19sfcrra6z"]
+AQUA_MODEL_TYPE_SERVICE = "service"
+AQUA_MODEL_TYPE_CUSTOM = "custom"
 
 
 class LifecycleStatus(Enum):
@@ -741,3 +743,9 @@ def known_realm():
 
     """
     return os.environ.get("CONDA_BUCKET_NS") in AQUA_GA_LIST
+
+
+def get_ocid_substring(ocid: str, key_len: int) -> str:
+    """This helper function returns the last n characters of the ocid specified by key_len parameter.
+    If ocid is None or length is less than key_len, it returns an empty string."""
+    return ocid[-key_len:] if ocid and len(ocid) > key_len else ""
diff --git a/ads/common/utils.py b/ads/common/utils.py
@@ -102,6 +102,8 @@
 # The number of worker processes to use in parallel for uploading individual parts of a multipart upload.
 DEFAULT_PARALLEL_PROCESS_COUNT = 9
 
+LOG_LEVELS = ["NOTSET", "DEBUG", "INFO", "WARNING", "ERROR", "CRITICAL"]
+
 
 class FileOverwriteError(Exception):  # pragma: no cover
     pass
@@ -1751,7 +1753,7 @@ def get_log_links(
 ) -> str:
     """
     This method returns the web console link for the given log ids.
-    
+
     Parameters
     ----------
     log_group_id: str, required
@@ -1776,7 +1778,9 @@ def get_log_links(
         query_range = f'''search "{compartment_id}/{log_group_id}/{log_id}"'''
         query_source = f"source='{source_id}'"
         sort_condition = f"sort by datetime desc&regions={region}"
-        search_query = f"search?searchQuery={query_range} | {query_source} | {sort_condition}"
+        search_query = (
+            f"search?searchQuery={query_range} | {query_source} | {sort_condition}"
+        )
         console_link_url = f"https://cloud.oracle.com/logging/{search_query}"
     elif log_group_id:
         console_link_url = f"https://cloud.oracle.com/logging/log-groups/{log_group_id}?region={region}"
diff --git a/pyproject.toml b/pyproject.toml
@@ -63,7 +63,7 @@ dependencies = [
   "jinja2>=2.11.2",
   "matplotlib>=3.1.3",
   "numpy>=1.19.2",
-  "oci>=2.113.0",
+  "oci>=2.125.3",
   "ocifs>=1.1.3",
   "pandas>1.2.1; python_version<'3.9'", # starting pandas v2.1.0 requires-python = '>=3.9'
   "pandas>=2.2.0; python_version>='3.9'",
diff --git a/tests/unitary/with_extras/aqua/test_cli.py b/tests/unitary/with_extras/aqua/test_cli.py
@@ -77,15 +77,17 @@ def test_aqua_command_without_compartment_env_var(self, name, session_ocid):
                 reload(ads.aqua.cli)
                 with patch("ads.aqua.cli.set_log_level") as mock_setting_log:
                     with patch("ads.aqua.logger.error") as mock_logger_error:
-                        AquaCommand()
-                        mock_setting_log.assert_called_with(
-                            TestAquaCLI.DEFAULT_AQUA_CLI_LOGGING_LEVEL
-                        )
-                        mock_logger_error.assert_any_call(
-                            "ODSC_MODEL_COMPARTMENT_OCID environment variable is not set for Aqua."
-                        )
-                        if session_ocid:
-                            mock_logger_error.assert_any_call(
-                                f"Aqua is not available for the notebook session {session_ocid}."
+                        with patch("ads.aqua.logger.debug") as mock_logger_debug:
+                            AquaCommand()
+                            mock_setting_log.assert_called_with(
+                                TestAquaCLI.DEFAULT_AQUA_CLI_LOGGING_LEVEL
                             )
-                        mock_exit.assert_called_with(1)
+                            mock_logger_debug.assert_any_call(
+                                "ODSC_MODEL_COMPARTMENT_OCID environment variable is not set for Aqua."
+                            )
+                            if session_ocid:
+                                mock_logger_error.assert_any_call(
+                                    f"Aqua is not available for the notebook session {session_ocid}. For more information, "
+                                    f"please refer to the documentation."
+                                )
+                            mock_exit.assert_called_with(1)
diff --git a/tests/unitary/with_extras/aqua/test_finetuning.py b/tests/unitary/with_extras/aqua/test_finetuning.py
@@ -21,8 +21,8 @@
 from ads.model.datascience_model import DataScienceModel
 from ads.model.model_metadata import ModelCustomMetadata
 
-class FineTuningTestCase(TestCase):
 
+class FineTuningTestCase(TestCase):
     SERVICE_COMPARTMENT_ID = "ocid1.compartment.oc1..<OCID>"
 
     def setUp(self):
@@ -61,20 +61,20 @@ def test_create_fine_tuning(
         mock_job_create,
         mock_job_id,
         mock_job_name,
-        mock_job_run
+        mock_job_run,
     ):
         custom_metadata_list = ModelCustomMetadata()
         custom_metadata_list.add(
             key=FineTuneCustomMetadata.SERVICE_MODEL_ARTIFACT_LOCATION.value,
-            value="test_service_model_artifact_location"
+            value="test_service_model_artifact_location",
         )
         custom_metadata_list.add(
             key=FineTuneCustomMetadata.SERVICE_MODEL_DEPLOYMENT_CONTAINER.value,
-            value="test_service_model_deployment_container"
+            value="test_service_model_deployment_container",
         )
         custom_metadata_list.add(
             key=FineTuneCustomMetadata.SERVICE_MODEL_FINE_TUNE_CONTAINER.value,
-            value="test_service_model_fine_tune_container"
+            value="test_service_model_fine_tune_container",
         )
 
         ft_source = MagicMock()
@@ -84,10 +84,7 @@ def test_create_fine_tuning(
         ft_source.custom_metadata_list = custom_metadata_list
         mock_get_source.return_value = ft_source
 
-        mock_mvs_create.return_value = (
-            "test_experiment_id",
-            "test_experiment_name"
-        )
+        mock_mvs_create.return_value = ("test_experiment_id", "test_experiment_name")
 
         ft_model = MagicMock()
         ft_model.id = "test_ft_model_id"
@@ -97,10 +94,7 @@ def test_create_fine_tuning(
 
         mock_get_finetuning_config.return_value = {
             "shape": {
-                "VM.GPU.A10.1": {
-                    "batch_size": 1,
-                    "replica": 1
-                },
+                "VM.GPU.A10.1": {"batch_size": 1, "replica": 1},
             }
         }
         mock_get_container_image.return_value = "test_container_image"
@@ -122,53 +116,49 @@ def test_create_fine_tuning(
             ft_name="test_ft_name",
             dataset_path="oci://ds_bucket@namespace/prefix/dataset.jsonl",
             report_path="oci://report_bucket@namespace/prefix/",
-            ft_parameters={
-                "epochs":1,
-                "learning_rate":0.02
-            },
+            ft_parameters={"epochs": 1, "learning_rate": 0.02},
             shape_name="VM.GPU.A10.1",
             replica=1,
             validation_set_size=0.2,
             block_storage_size=1,
             experiment_name="test_experiment_name",
         )
 
-        aqua_ft_summary = self.app.create(
-            **create_aqua_ft_details
-        )
+        aqua_ft_summary = self.app.create(**create_aqua_ft_details)
 
         assert asdict(aqua_ft_summary) == {
-            'console_url': f'https://cloud.oracle.com/data-science/models/{ft_model.id}?region={self.app.region}',
-            'experiment': {
-                'id': f'{mock_mvs_create.return_value[0]}',
-                'name': f'{mock_mvs_create.return_value[1]}',
-                'url': f'https://cloud.oracle.com/data-science/model-version-sets/{mock_mvs_create.return_value[0]}?region={self.app.region}'
+            "console_url": f"https://cloud.oracle.com/data-science/models/{ft_model.id}?region={self.app.region}",
+            "experiment": {
+                "id": f"{mock_mvs_create.return_value[0]}",
+                "name": f"{mock_mvs_create.return_value[1]}",
+                "url": f"https://cloud.oracle.com/data-science/model-version-sets/{mock_mvs_create.return_value[0]}?region={self.app.region}",
             },
-            'id': f'{ft_model.id}',
-            'job': {
-                'id': f'{mock_job_id.return_value}',
-                'name': f'{mock_job_name.return_value}',
-                'url': f'https://cloud.oracle.com/data-science/jobs/{mock_job_id.return_value}?region={self.app.region}'
+            "id": f"{ft_model.id}",
+            "job": {
+                "id": f"{mock_job_id.return_value}",
+                "name": f"{mock_job_name.return_value}",
+                "url": f"https://cloud.oracle.com/data-science/jobs/{mock_job_id.return_value}?region={self.app.region}",
             },
-            'lifecycle_details': f'{ft_job_run.lifecycle_details}',
-            'lifecycle_state': f'{ft_job_run.lifecycle_state}',
-            'name': f'{ft_model.display_name}',
-            'parameters': {
-                'epochs': 1,
-                'learning_rate': 0.02
+            "lifecycle_details": f"{ft_job_run.lifecycle_details}",
+            "lifecycle_state": f"{ft_job_run.lifecycle_state}",
+            "name": f"{ft_model.display_name}",
+            "parameters": {
+                "epochs": 1,
+                "learning_rate": 0.02,
+                "sample_packing": "True",
             },
-            'source': {
-                'id': f'{ft_source.id}',
-                'name': f'{ft_source.display_name}',
-                'url': f'https://cloud.oracle.com/data-science/models/{ft_source.id}?region={self.app.region}'
+            "source": {
+                "id": f"{ft_source.id}",
+                "name": f"{ft_source.display_name}",
+                "url": f"https://cloud.oracle.com/data-science/models/{ft_source.id}?region={self.app.region}",
             },
-            'tags': {
-                'aqua_finetuning': 'aqua_finetuning',
-                'finetuning_experiment_id': f'{mock_mvs_create.return_value[0]}',
-                'finetuning_job_id': f'{mock_job_id.return_value}',
-                'finetuning_source': f'{ft_source.id}'
+            "tags": {
+                "aqua_finetuning": "aqua_finetuning",
+                "finetuning_experiment_id": f"{mock_mvs_create.return_value[0]}",
+                "finetuning_job_id": f"{mock_job_id.return_value}",
+                "finetuning_source": f"{ft_source.id}",
             },
-            'time_created': f'{ft_model.time_created}'
+            "time_created": f"{ft_model.time_created}",
         }
 
     def test_exit_code_message(self):
