oracle-ads 2.11.16__py3-none-any.whl → 2.11.18__py3-none-any.whl

ads/aqua/modeldeployment/deployment.py

@@ -87,25 +87,26 @@ class AquaDeploymentApp(AquaApp):
 
     @telemetry(entry_point="plugin=deployment&action=create", name="aqua")
     def create(
-        self,
-        model_id: str,
-        instance_shape: str,
-        display_name: str,
-        instance_count: int = None,
-        log_group_id: str = None,
-        access_log_id: str = None,
-        predict_log_id: str = None,
-        compartment_id: str = None,
-        project_id: str = None,
-        description: str = None,
-        bandwidth_mbps: int = None,
-        web_concurrency: int = None,
-        server_port: int = None,
-        health_check_port: int = None,
-        env_var: Dict = None,
-        container_family: str = None,
-        memory_in_gbs: Optional[float] = None,
-        ocpus: Optional[float] = None,
+            self,
+            model_id: str,
+            instance_shape: str,
+            display_name: str,
+            instance_count: int = None,
+            log_group_id: str = None,
+            access_log_id: str = None,
+            predict_log_id: str = None,
+            compartment_id: str = None,
+            project_id: str = None,
+            description: str = None,
+            bandwidth_mbps: int = None,
+            web_concurrency: int = None,
+            server_port: int = None,
+            health_check_port: int = None,
+            env_var: Dict = None,
+            container_family: str = None,
+            memory_in_gbs: Optional[float] = None,
+            ocpus: Optional[float] = None,
+            model_file: Optional[str] = None,
     ) -> "AquaDeployment":
         """
         Creates a new Aqua deployment
@@ -150,6 +151,8 @@ class AquaDeploymentApp(AquaApp):
             The memory in gbs for the shape selected.
         ocpus: float
             The ocpu count for the shape selected.
+        model_file: str
+            The file used for model deployment.
         Returns
         -------
         AquaDeployment
@@ -212,21 +215,6 @@ class AquaDeploymentApp(AquaApp):
 
         env_var.update({"BASE_MODEL": f"{model_path_prefix}"})
 
-        model_format = aqua_model.freeform_tags.get(
-            Tags.MODEL_FORMAT, ModelFormat.SAFETENSORS.value
-        ).upper()
-        if model_format == ModelFormat.GGUF.value:
-            try:
-                model_file = aqua_model.custom_metadata_list.get(
-                    AQUA_MODEL_ARTIFACT_FILE
-                ).value
-            except ValueError as err:
-                raise AquaValueError(
-                    f"{AQUA_MODEL_ARTIFACT_FILE} key is not available in the custom metadata field "
-                    f"for model {aqua_model.id}."
-                ) from err
-            env_var.update({"BASE_MODEL_FILE": f"{model_file}"})
-
         if is_fine_tuned_model:
             _, fine_tune_output_path = get_model_by_reference_paths(
                 aqua_model.model_file_description
@@ -262,7 +250,7 @@ class AquaDeploymentApp(AquaApp):
         try:
             # Check if the container override flag is set. If set, then the user has chosen custom image
             if aqua_model.custom_metadata_list.get(
-                AQUA_DEPLOYMENT_CONTAINER_OVERRIDE_FLAG_METADATA_NAME
+                    AQUA_DEPLOYMENT_CONTAINER_OVERRIDE_FLAG_METADATA_NAME
             ).value:
                 is_custom_container = True
         except Exception:
@@ -281,6 +269,32 @@ class AquaDeploymentApp(AquaApp):
             f"Aqua Image used for deploying {aqua_model.id} : {container_image}"
         )
 
+        model_formats_str = aqua_model.freeform_tags.get(
+            Tags.MODEL_FORMAT, ModelFormat.SAFETENSORS.value
+        ).upper()
+        model_format = model_formats_str.split(",")
+
+        # Figure out a better way to handle this in future release
+        if ModelFormat.GGUF.value in model_format and container_type_key.lower() == InferenceContainerTypeFamily.AQUA_LLAMA_CPP_CONTAINER_FAMILY:
+            if model_file is not None:
+                logger.info(
+                    f"Overriding {model_file} as model_file for model {aqua_model.id}."
+                )
+            else:
+                try:
+                    model_file = aqua_model.custom_metadata_list.get(
+                        AQUA_MODEL_ARTIFACT_FILE
+                    ).value
+                except ValueError as err:
+                    raise AquaValueError(
+                        f"{AQUA_MODEL_ARTIFACT_FILE} key is not available in the custom metadata field "
+                        f"for model {aqua_model.id}. Either register the model with a default model_file or pass "
+                        f"as a parameter when creating a deployment."
+                    ) from err
+
+            env_var.update({"BASE_MODEL_FILE": f"{model_file}"})
+            tags.update({Tags.MODEL_ARTIFACT_FILE: model_file})
+
         # todo: use AquaContainerConfig.from_container_index_json instead.
         # Fetch the startup cli command for the container
         # container_index.json will have "containerSpec" section which will provide the cli params for
@@ -312,8 +326,8 @@ class AquaDeploymentApp(AquaApp):
         if user_params:
             # todo: remove this check in the future version, logic to be moved to container_index
             if (
-                container_type_key.lower()
-                == InferenceContainerTypeFamily.AQUA_LLAMA_CPP_CONTAINER_FAMILY
+                    container_type_key.lower()
+                    == InferenceContainerTypeFamily.AQUA_LLAMA_CPP_CONTAINER_FAMILY
             ):
                 # AQUA_LLAMA_CPP_CONTAINER_FAMILY container uses uvicorn that required model/server params
                 # to be set as env vars
@@ -446,8 +460,8 @@ class AquaDeploymentApp(AquaApp):
         for model_deployment in model_deployments:
             oci_aqua = (
                 (
-                    Tags.AQUA_TAG in model_deployment.freeform_tags
-                    or Tags.AQUA_TAG.lower() in model_deployment.freeform_tags
+                        Tags.AQUA_TAG in model_deployment.freeform_tags
+                        or Tags.AQUA_TAG.lower() in model_deployment.freeform_tags
                 )
                 if model_deployment.freeform_tags
                 else False
@@ -501,8 +515,8 @@ class AquaDeploymentApp(AquaApp):
 
         oci_aqua = (
             (
-                Tags.AQUA_TAG in model_deployment.freeform_tags
-                or Tags.AQUA_TAG.lower() in model_deployment.freeform_tags
+                    Tags.AQUA_TAG in model_deployment.freeform_tags
+                    or Tags.AQUA_TAG.lower() in model_deployment.freeform_tags
             )
             if model_deployment.freeform_tags
             else False
@@ -519,8 +533,8 @@ class AquaDeploymentApp(AquaApp):
         log_group_name = ""
 
         logs = (
-            model_deployment.category_log_details.access
-            or model_deployment.category_log_details.predict
+                model_deployment.category_log_details.access
+                or model_deployment.category_log_details.predict
         )
         if logs:
             log_id = logs.log_id
@@ -575,9 +589,9 @@ class AquaDeploymentApp(AquaApp):
         return config
 
     def get_deployment_default_params(
-        self,
-        model_id: str,
-        instance_shape: str,
+            self,
+            model_id: str,
+            instance_shape: str,
     ) -> List[str]:
         """Gets the default params set in the deployment configs for the given model and instance shape.
 
@@ -609,8 +623,8 @@ class AquaDeploymentApp(AquaApp):
             )
 
         if (
-            container_type_key
-            and container_type_key in InferenceContainerTypeFamily.values()
+                container_type_key
+                and container_type_key in InferenceContainerTypeFamily.values()
         ):
             deployment_config = self.get_deployment_config(model_id)
             config_params = (
@@ -633,10 +647,10 @@ class AquaDeploymentApp(AquaApp):
         return default_params
 
     def validate_deployment_params(
-        self,
-        model_id: str,
-        params: List[str] = None,
-        container_family: str = None,
+            self,
+            model_id: str,
+            params: List[str] = None,
+            container_family: str = None,
     ) -> Dict:
         """Validate if the deployment parameters passed by the user can be overridden. Parameter values are not
         validated, only param keys are validated.
@@ -695,9 +709,9 @@ class AquaDeploymentApp(AquaApp):
 
     @staticmethod
     def _find_restricted_params(
-        default_params: Union[str, List[str]],
-        user_params: Union[str, List[str]],
-        container_family: str,
+            default_params: Union[str, List[str]],
+            user_params: Union[str, List[str]],
+            container_family: str,
     ) -> List[str]:
         """Returns a list of restricted params that user chooses to override when creating an Aqua deployment.
         The default parameters coming from the container index json file cannot be overridden.

ads/common/auth.py

@@ -73,7 +73,7 @@ class AuthState(metaclass=SingletonMeta):
         self.oci_key_profile = self.oci_key_profile or os.environ.get(
             "OCI_CONFIG_PROFILE", DEFAULT_PROFILE
         )
-        self.oci_config = self.oci_config or {}
+        self.oci_config = self.oci_config or {"region": os.environ["OCI_RESOURCE_REGION"]} if os.environ.get("OCI_RESOURCE_REGION") else {}
         self.oci_signer_kwargs = self.oci_signer_kwargs or {}
         self.oci_client_kwargs = self.oci_client_kwargs or {}
 
@@ -82,7 +82,7 @@ def set_auth(
     auth: Optional[str] = AuthType.API_KEY,
     oci_config_location: Optional[str] = DEFAULT_LOCATION,
     profile: Optional[str] = DEFAULT_PROFILE,
-    config: Optional[Dict] = {},
+    config: Optional[Dict] = {"region": os.environ["OCI_RESOURCE_REGION"]} if os.environ.get("OCI_RESOURCE_REGION") else {},
     signer: Optional[Any] = None,
     signer_callable: Optional[Callable] = None,
     signer_kwargs: Optional[Dict] = {},
@@ -678,7 +678,7 @@ class ResourcePrincipal(AuthSignerGenerator):
         >>> signer_generator = AuthFactory().signerGenerator(AuthType.RESOURCE_PRINCIPAL)
         >>> signer_generator(signer_args).create_signer()
         """
-        configuration = ads.telemetry.update_oci_client_config()
+        configuration = ads.telemetry.update_oci_client_config(AuthState().oci_config)
         signer_dict = {
             "config": configuration,
             "signer": oci.auth.signers.get_resource_principals_signer(),
@@ -739,7 +739,7 @@ class InstancePrincipal(AuthSignerGenerator):
         >>> signer_generator = AuthFactory().signerGenerator(AuthType.INSTANCE_PRINCIPAL)
         >>> signer_generator(signer_args).create_signer()
         """
-        configuration = ads.telemetry.update_oci_client_config()
+        configuration = ads.telemetry.update_oci_client_config(AuthState().oci_config)
         signer_dict = {
             "config": configuration,
             "signer": oci.auth.signers.InstancePrincipalsSecurityTokenSigner(

ads/jobs/builders/infrastructure/dsc_job.py

@@ -312,7 +312,7 @@ class DSCJob(OCIDataScienceMixin, oci.data_science.models.Job):
         logger.debug(oci_model)
         res = self.client.create_job(oci_model)
         self.update_from_oci_model(res.data)
-        if self.lifecycle_state == "ACTIVE":
+        if not self.artifact:
             return
         try:
             if issubclass(self.artifact.__class__, Artifact):
@@ -377,13 +377,12 @@ class DSCJob(OCIDataScienceMixin, oci.data_science.models.Job):
         """
         runs = self.run_list()
         for run in runs:
-            if force_delete:
-                if run.lifecycle_state in [
-                    DataScienceJobRun.LIFECYCLE_STATE_ACCEPTED,
-                    DataScienceJobRun.LIFECYCLE_STATE_IN_PROGRESS,
-                    DataScienceJobRun.LIFECYCLE_STATE_NEEDS_ATTENTION,
-                ]:
-                    run.cancel(wait_for_completion=True)
+            if force_delete and run.lifecycle_state in [
+                DataScienceJobRun.LIFECYCLE_STATE_ACCEPTED,
+                DataScienceJobRun.LIFECYCLE_STATE_IN_PROGRESS,
+                DataScienceJobRun.LIFECYCLE_STATE_NEEDS_ATTENTION,
+            ]:
+                run.cancel(wait_for_completion=True)
             run.delete()
         self.client.delete_job(self.id)
         return self
@@ -488,7 +487,9 @@ class DSCJob(OCIDataScienceMixin, oci.data_science.models.Job):
             oci.data_science.models.DefaultJobConfigurationDetails().swagger_types.keys()
         )
         env_config_swagger_types = {}
-        if hasattr(oci.data_science.models, "OcirContainerJobEnvironmentConfigurationDetails"):
+        if hasattr(
+            oci.data_science.models, "OcirContainerJobEnvironmentConfigurationDetails"
+        ):
             env_config_swagger_types = (
                 oci.data_science.models.OcirContainerJobEnvironmentConfigurationDetails().swagger_types.keys()
             )
@@ -502,7 +503,7 @@ class DSCJob(OCIDataScienceMixin, oci.data_science.models.Job):
                 value = kwargs.pop(key)
                 if key in [
                     ContainerRuntime.CONST_CMD,
-                    ContainerRuntime.CONST_ENTRYPOINT
+                    ContainerRuntime.CONST_ENTRYPOINT,
                 ] and isinstance(value, str):
                     value = ContainerRuntimeHandler.split_args(value)
                 env_config_kwargs[key] = value
@@ -536,9 +537,13 @@ class DSCJob(OCIDataScienceMixin, oci.data_science.models.Job):
 
         if env_config_kwargs:
             env_config_kwargs["jobEnvironmentType"] = "OCIR_CONTAINER"
-            env_config_override = kwargs.get("job_environment_configuration_override_details", {})
+            env_config_override = kwargs.get(
+                "job_environment_configuration_override_details", {}
+            )
             env_config_override.update(env_config_kwargs)
-            kwargs["job_environment_configuration_override_details"] = env_config_override
+            kwargs["job_environment_configuration_override_details"] = (
+                env_config_override
+            )
 
         wait = kwargs.pop("wait", False)
         run = DataScienceJobRun(**kwargs, **self.auth).create()
@@ -894,10 +899,14 @@ class DataScienceJobRun(
         return self
 
     def delete(self, force_delete: bool = False):
-        if force_delete:
+        if force_delete and self.status in [
+            DataScienceJobRun.LIFECYCLE_STATE_ACCEPTED,
+            DataScienceJobRun.LIFECYCLE_STATE_IN_PROGRESS,
+            DataScienceJobRun.LIFECYCLE_STATE_NEEDS_ATTENTION,
+        ]:
             self.cancel(wait_for_completion=True)
         super().delete()
-        return
+        return self
 
 
 # This is for backward compatibility

ads/jobs/builders/infrastructure/dsc_job_runtime.py

@@ -181,9 +181,9 @@ class RuntimeHandler:
             "jobType": self.data_science_job.job_type,
         }
         if runtime.maximum_runtime_in_minutes:
-            job_configuration_details[
-                "maximum_runtime_in_minutes"
-            ] = runtime.maximum_runtime_in_minutes
+            job_configuration_details["maximum_runtime_in_minutes"] = (
+                runtime.maximum_runtime_in_minutes
+            )
         job_configuration_details["environment_variables"] = self._translate_env(
             runtime
         )
@@ -310,7 +310,7 @@ class RuntimeHandler:
         for extraction in extractions:
             runtime_spec.update(extraction(dsc_job))
         return self.RUNTIME_CLASS(self._format_env_var(runtime_spec))
-    
+
     def _extract_properties(self, dsc_job) -> dict:
         """Extract the job runtime properties from data science job.
 
@@ -968,23 +968,10 @@ class ContainerRuntimeHandler(RuntimeHandler):
         payload["job_environment_configuration_details"] = job_env_config
         return payload
 
-    def _translate_artifact(self, runtime: Runtime):
-        """Specifies a dummy script as the job artifact.
-        runtime is not used in this method.
-
-        Parameters
-        ----------
-        runtime : Runtime
-            This is not used.
-
-        Returns
-        -------
-        str
-            Path to the dummy script.
-        """
-        return os.path.join(
-            os.path.dirname(__file__), "../../templates", "container.py"
-        )
+    def _translate_artifact(self, runtime: ContainerRuntime):
+        """Additional artifact for the container"""
+        if runtime.artifact_uri:
+            return ScriptArtifact(runtime.artifact_uri, runtime)
 
     def _translate_env_config(self, runtime: Runtime) -> dict:
         """Converts runtime properties to ``OcirContainerJobEnvironmentConfigurationDetails`` payload required by OCI Data Science job.
@@ -1007,7 +994,7 @@ class ContainerRuntimeHandler(RuntimeHandler):
             property = runtime.get_spec(key, None)
             if key in [
                 ContainerRuntime.CONST_CMD,
-                ContainerRuntime.CONST_ENTRYPOINT
+                ContainerRuntime.CONST_ENTRYPOINT,
             ] and isinstance(property, str):
                 property = self.split_args(property)
             if property is not None:
@@ -1063,7 +1050,7 @@ class ContainerRuntimeHandler(RuntimeHandler):
             spec[ContainerRuntime.CONST_ENV_VAR] = envs
 
         return spec
-    
+
     def _extract_properties(self, dsc_job) -> dict:
         """Extract the runtime properties from data science job.
 
@@ -1078,10 +1065,10 @@ class ContainerRuntimeHandler(RuntimeHandler):
             A runtime specification dictionary for initializing a runtime.
         """
         spec = super()._extract_envs(dsc_job)
-        
+
         job_env_config = getattr(dsc_job, "job_environment_configuration_details", None)
         job_env_type = getattr(job_env_config, "job_environment_type", None)
-        
+
         if not (job_env_config and job_env_type == "OCIR_CONTAINER"):
             raise IncompatibleRuntime()
 

ads/jobs/builders/runtimes/artifact.py

@@ -183,11 +183,6 @@ class ScriptArtifact(Artifact):
         if os.path.isdir(source):
             basename = os.path.basename(str(source).rstrip("/"))
             source = str(source).rstrip("/")
-            # Runtime must have entrypoint if the source is a directory
-            if self.runtime and not self.runtime.entrypoint:
-                raise ValueError(
-                    "Please specify entrypoint when script source is a directory."
-                )
             output = os.path.join(self.temp_dir.name, basename)
             shutil.make_archive(
                 output, "zip", os.path.dirname(source), base_dir=basename

ads/jobs/builders/runtimes/container_runtime.py

@@ -56,6 +56,7 @@ class ContainerRuntime(MultiNodeRuntime):
     CONST_CMD = "cmd"
     CONST_IMAGE_DIGEST = "imageDigest"
     CONST_IMAGE_SIGNATURE_ID = "imageSignatureId"
+    CONST_SCRIPT_PATH = "scriptPathURI"
     attribute_map = {
         CONST_IMAGE: CONST_IMAGE,
         CONST_ENTRYPOINT: CONST_ENTRYPOINT,
@@ -121,7 +122,7 @@ class ContainerRuntime(MultiNodeRuntime):
     def image_digest(self) -> str:
         """The container image digest."""
         return self.get_spec(self.CONST_IMAGE_DIGEST)
-    
+
     def with_image_digest(self, image_digest: str) -> "ContainerRuntime":
         """Sets the digest of custom image.
 
@@ -136,12 +137,12 @@ class ContainerRuntime(MultiNodeRuntime):
             The runtime instance.
         """
         return self.set_spec(self.CONST_IMAGE_DIGEST, image_digest)
-    
+
     @property
     def image_signature_id(self) -> str:
         """The container image signature id."""
         return self.get_spec(self.CONST_IMAGE_SIGNATURE_ID)
-    
+
     def with_image_signature_id(self, image_signature_id: str) -> "ContainerRuntime":
         """Sets the signature id of custom image.
 
@@ -217,3 +218,25 @@ class ContainerRuntime(MultiNodeRuntime):
             entrypoint=["bash", "--login", "-c"],
             cmd="{Container CMD. For MLflow and Operator will be auto generated}",
         )
+
+    @property
+    def artifact_uri(self) -> str:
+        """The URI of the source code"""
+        return self.get_spec(self.CONST_SCRIPT_PATH)
+
+    def with_artifact(self, uri: str):
+        """Specifies the artifact to be added to the container.
+
+        Parameters
+        ----------
+        uri : str
+            URI to the source code script, which can be any URI supported by fsspec,
+            including http://, https:// and OCI object storage.
+            For example: oci://your_bucket@your_namespace/path/to/script.py
+
+        Returns
+        -------
+        self
+            The runtime instance.
+        """
+        return self.set_spec(self.CONST_SCRIPT_PATH, uri)