Merge remote-tracking branch 'origin/main' into feature/dtv2

Author: qiuosier

.github/workflows/run-unittests-default_setup.yml

@@ -1,4 +1,4 @@
-name: "[Py3.9-3.11] - Default Tests"
+name: "[Py3.9-3.12] - Default Tests"
 
 on:
   workflow_dispatch:

.github/workflows/run-unittests-py310-py311.yml

@@ -1,4 +1,4 @@
-name: "[Py3.10-3.11] - All Unit Tests"
+name: "[Py3.10-3.12] - All Unit Tests"
 
 on:
   workflow_dispatch:
@@ -23,6 +23,7 @@ permissions:
 # hack for https://github.com/actions/cache/issues/810#issuecomment-1222550359
 env:
   SEGMENT_DOWNLOAD_TIMEOUT_MINS: 5
+  ENV TF_USE_LEGACY_KERAS:  1
 
 jobs:
   test:
@@ -33,7 +34,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: ["3.10", "3.11"]
+        python-version: ["3.10", "3.11", "3.12"]
         name: ["unitary", "slow_tests"]
         include:
           - name: "unitary"

README-development.md

@@ -4,7 +4,7 @@
 The Oracle Accelerated Data Science (ADS) SDK used by data scientists and analysts for
 data exploration and experimental machine learning to democratize machine learning and
 analytics by providing easy-to-use, 
-performant, and user friendly tools that
+performant, and user-friendly tools that
 brings together the best of data science practices.
 
 The ADS SDK helps you connect to different data sources, perform exploratory data analysis,
@@ -176,7 +176,7 @@ pip install -r test-requirements.txt
 ```
 
 ### Step 2: Create local .env files 
-Running the local JuypterLab server requires setting OCI authentication, proxy, and OCI namespace parameters. Adapt this .env file with your specific OCI profile and OCIDs to set these variables.
+Running the local JupyterLab server requires setting OCI authentication, proxy, and OCI namespace parameters. Adapt this .env file with your specific OCI profile and OCIDs to set these variables.
 
 ```
 CONDA_BUCKET_NS="your_conda_bucket"
@@ -248,7 +248,7 @@ All the unit tests can be found [here](https://github.com/oracle/accelerated-dat
 The following commands detail how the unit tests can be run.
 ```
 # Run all tests in AQUA project
-python -m pytest -q tests/unitary/with_extras/aqua/test_deployment.py
+python -m pytest -q tests/unitary/with_extras/aqua/*
 
 # Run all tests specific to a module within in AQUA project (ex. test_deployment.py, test_model.py, etc.)
 python -m pytest -q tests/unitary/with_extras/aqua/test_deployment.py

ads/aqua/__init__.py

@@ -14,7 +14,6 @@
     get_async_httpx_client,
     get_httpx_client,
 )
-from ads.aqua.common.utils import fetch_service_compartment
 from ads.config import OCI_RESOURCE_PRINCIPAL_VERSION
 
 ENV_VAR_LOG_LEVEL = "ADS_AQUA_LOG_LEVEL"
@@ -39,7 +38,3 @@ def set_log_level(log_level: str):
 
 if OCI_RESOURCE_PRINCIPAL_VERSION:
     set_auth("resource_principal")
-
-ODSC_MODEL_COMPARTMENT_OCID = (
-    os.environ.get("ODSC_MODEL_COMPARTMENT_OCID") or fetch_service_compartment()
-)

ads/aqua/app.py

@@ -6,10 +6,17 @@
 import os
 import traceback
 from dataclasses import fields
-from typing import Any, Dict, Optional, Union
+from datetime import datetime, timedelta
+from itertools import chain
+from typing import Any, Dict, List, Optional, Union
 
 import oci
-from oci.data_science.models import UpdateModelDetails, UpdateModelProvenanceDetails
+from cachetools import TTLCache, cached
+from oci.data_science.models import (
+    ContainerSummary,
+    UpdateModelDetails,
+    UpdateModelProvenanceDetails,
+)
 
 from ads import set_auth
 from ads.aqua import logger
@@ -22,6 +29,11 @@
     is_valid_ocid,
     load_config,
 )
+from ads.aqua.config.container_config import (
+    AquaContainerConfig,
+    AquaContainerConfigItem,
+)
+from ads.aqua.constants import SERVICE_MANAGED_CONTAINER_URI_SCHEME
 from ads.common import oci_client as oc
 from ads.common.auth import default_signer
 from ads.common.utils import UNKNOWN, extract_region, is_path_exists
@@ -238,7 +250,9 @@ def create_model_catalog(
             .with_custom_metadata_list(model_custom_metadata)
             .with_defined_metadata_list(model_taxonomy_metadata)
             .with_provenance_metadata(ModelProvenanceMetadata(training_id=UNKNOWN))
-            .with_defined_tags(**(defined_tags or {})) # Create defined tags when a model is created.
+            .with_defined_tags(
+                **(defined_tags or {})
+            )  # Create defined tags when a model is created.
             .create(
                 **kwargs,
             )
@@ -269,6 +283,44 @@ def if_artifact_exist(self, model_id: str, **kwargs) -> bool:
                 logger.info(f"Artifact not found in model {model_id}.")
                 return False
 
+    def get_config_from_metadata(
+        self, model_id: str, metadata_key: str
+    ) -> ModelConfigResult:
+        """Gets the config for the given Aqua model from model catalog metadata content.
+
+        Parameters
+        ----------
+        model_id: str
+            The OCID of the Aqua model.
+        metadata_key: str
+            The metadata key name where artifact content is stored
+        Returns
+        -------
+        ModelConfigResult
+            A Pydantic model containing the model_details (extracted from OCI) and the config dictionary.
+        """
+        config = {}
+        oci_model = self.ds_client.get_model(model_id).data
+        try:
+            config = self.ds_client.get_model_defined_metadatum_artifact_content(
+                model_id, metadata_key
+            ).data.content.decode("utf-8")
+            return ModelConfigResult(config=json.loads(config), model_details=oci_model)
+        except UnicodeDecodeError as ex:
+            logger.error(
+                f"Failed to decode content for '{metadata_key}' in defined metadata for model '{model_id}' : {ex}"
+            )
+        except json.JSONDecodeError as ex:
+            logger.error(
+                f"Invalid JSON format for '{metadata_key}' in defined metadata for model '{model_id}' : {ex}"
+            )
+        except Exception as ex:
+            logger.error(
+                f"Failed to retrieve defined metadata key '{metadata_key}' for model '{model_id}': {ex}"
+            )
+        return ModelConfigResult(config=config, model_details=oci_model)
+
+    @cached(cache=TTLCache(maxsize=1, ttl=timedelta(minutes=1), timer=datetime.now))
     def get_config(
         self,
         model_id: str,
@@ -307,22 +359,7 @@ def get_config(
             raise AquaRuntimeError(f"Target model {oci_model.id} is not an Aqua model.")
 
         config: Dict[str, Any] = {}
-
-        # if the current model has a service model tag, then
-        if Tags.AQUA_SERVICE_MODEL_TAG in oci_model.freeform_tags:
-            base_model_ocid = oci_model.freeform_tags[Tags.AQUA_SERVICE_MODEL_TAG]
-            logger.info(
-                f"Base model found for the model: {oci_model.id}. "
-                f"Loading {config_file_name} for base model {base_model_ocid}."
-            )
-            if config_folder == ConfigFolder.ARTIFACT:
-                artifact_path = get_artifact_path(oci_model.custom_metadata_list)
-            else:
-                base_model = self.ds_client.get_model(base_model_ocid).data
-                artifact_path = get_artifact_path(base_model.custom_metadata_list)
-        else:
-            logger.info(f"Loading {config_file_name} for model {oci_model.id}...")
-            artifact_path = get_artifact_path(oci_model.custom_metadata_list)
+        artifact_path = get_artifact_path(oci_model.custom_metadata_list)
         if not artifact_path:
             logger.debug(
                 f"Failed to get artifact path from custom metadata for the model: {model_id}"
@@ -337,6 +374,9 @@ def get_config(
         config_file_path = os.path.join(config_path, config_file_name)
         if is_path_exists(config_file_path):
             try:
+                logger.info(
+                    f"Loading config: `{config_file_name}` from `{config_path}`"
+                )
                 config = load_config(
                     config_path,
                     config_file_name=config_file_name,
@@ -355,6 +395,85 @@ def get_config(
 
         return ModelConfigResult(config=config, model_details=oci_model)
 
+    def get_container_image(self, container_type: str = None) -> str:
+        """
+        Gets the latest smc container complete image name from the given container type.
+
+        Parameters
+        ----------
+        container_type: str
+            type of container, can be either odsc-vllm-serving, odsc-llm-fine-tuning, odsc-llm-evaluate
+
+        Returns
+        -------
+        str:
+            A complete container name along with version. ex: dsmc://odsc-vllm-serving:0.7.4.1
+        """
+
+        containers = self.list_service_containers()
+        container = next(
+            (c for c in containers if c.is_latest and c.family_name == container_type),
+            None,
+        )
+        if not container:
+            raise AquaValueError(f"Invalid container type : {container_type}")
+        container_image = (
+            SERVICE_MANAGED_CONTAINER_URI_SCHEME
+            + container.container_name
+            + ":"
+            + container.tag
+        )
+        return container_image
+
+    @cached(cache=TTLCache(maxsize=20, ttl=timedelta(minutes=30), timer=datetime.now))
+    def list_service_containers(self) -> List[ContainerSummary]:
+        """
+        List containers from containers.conf in OCI Datascience control plane
+        """
+        containers = self.ds_client.list_containers().data
+        return containers
+
+    def get_container_config(self) -> AquaContainerConfig:
+        """
+        Fetches latest containers from containers.conf in OCI Datascience control plane
+
+        Returns
+        -------
+        AquaContainerConfig
+            An Object that contains latest container info for the given container family
+
+        """
+        return AquaContainerConfig.from_service_config(
+            service_containers=self.list_service_containers()
+        )
+
+    def get_container_config_item(
+        self, container_family: str
+    ) -> AquaContainerConfigItem:
+        """
+        Fetches latest container for given container_family_name from containers.conf in OCI Datascience control plane
+
+        Returns
+        -------
+        AquaContainerConfigItem
+            An Object that contains latest container info for the given container family
+
+        """
+
+        aqua_container_config = self.get_container_config()
+        inference_config = aqua_container_config.inference.values()
+        ft_config = aqua_container_config.finetune.values()
+        eval_config = aqua_container_config.evaluate.values()
+        container = next(
+            (
+                container
+                for container in chain(inference_config, ft_config, eval_config)
+                if container.family.lower() == container_family.lower()
+            ),
+            None,
+        )
+        return container
+
     @property
     def telemetry(self):
         if not self._telemetry:

ads/aqua/cli.py

@@ -1,23 +1,20 @@
 #!/usr/bin/env python
-# -*- coding: utf-8 -*--
 
-# Copyright (c) 2024 Oracle and/or its affiliates.
+# Copyright (c) 2024, 2025 Oracle and/or its affiliates.
 # Licensed under the Universal Permissive License v 1.0 as shown at https://oss.oracle.com/licenses/upl/
 import os
 
 from ads.aqua import (
     ENV_VAR_LOG_LEVEL,
-    ODSC_MODEL_COMPARTMENT_OCID,
     logger,
     set_log_level,
 )
-from ads.aqua.common.errors import AquaCLIError, AquaConfigError
+from ads.aqua.common.errors import AquaCLIError
 from ads.aqua.evaluation import AquaEvaluationApp
 from ads.aqua.finetuning import AquaFineTuningApp
 from ads.aqua.model import AquaModelApp
 from ads.aqua.modeldeployment import AquaDeploymentApp
 from ads.common.utils import LOG_LEVELS
-from ads.config import NB_SESSION_OCID
 
 
 class AquaCommand:
@@ -82,16 +79,6 @@ def __init__(
 
         set_log_level(aqua_log_level)
 
-        if not ODSC_MODEL_COMPARTMENT_OCID:
-            if NB_SESSION_OCID:
-                raise AquaConfigError(
-                    f"Aqua is not available for the notebook session {NB_SESSION_OCID}. For more information, "
-                    f"please refer to the documentation."
-                )
-            raise AquaConfigError(
-                "ODSC_MODEL_COMPARTMENT_OCID environment variable is not set for Aqua."
-            )
-
     @staticmethod
     def _validate_value(flag, value):
         """Check if the given value for bool flag is valid.