oracle-ads 2.13.9rc1__py3-none-any.whl → 2.13.10rc0__py3-none-any.whl

ads/aqua/app.py

@@ -64,6 +64,7 @@ class AquaApp:
             set_auth("resource_principal")
         self._auth = default_signer({"service_endpoint": OCI_ODSC_SERVICE_ENDPOINT})
         self.ds_client = oc.OCIClientFactory(**self._auth).data_science
+        self.compute_client = oc.OCIClientFactory(**default_signer()).compute
         self.logging_client = oc.OCIClientFactory(**default_signer()).logging_management
         self.identity_client = oc.OCIClientFactory(**default_signer()).identity
         self.region = extract_region(self._auth)

ads/aqua/common/entities.py

@@ -157,6 +157,8 @@ class AquaMultiModelRef(Serializable):
         Optional environment variables to override during deployment.
     artifact_location : Optional[str]
         Artifact path of model in the multimodel group.
+    fine_tune_weights_location : Optional[str]
+        For fine tuned models, the artifact path of the modified model weights
     """
 
     model_id: str = Field(..., description="The model OCID to deploy.")
@@ -171,6 +173,9 @@ class AquaMultiModelRef(Serializable):
     artifact_location: Optional[str] = Field(
         None, description="Artifact path of model in the multimodel group."
     )
+    fine_tune_weights_location: Optional[str] = Field(
+        None, description="For fine tuned models, the artifact path of the modified model weights"
+    )
 
     class Config:
         extra = "ignore"

ads/aqua/common/enums.py

@@ -20,6 +20,12 @@ class Resource(ExtendedEnum):
     MODEL_VERSION_SET = "model-version-sets"
 
 
+class PredictEndpoints(ExtendedEnum):
+    CHAT_COMPLETIONS_ENDPOINT = "/v1/chat/completions"
+    TEXT_COMPLETIONS_ENDPOINT = "/v1/completions"
+    EMBEDDING_ENDPOINT = "/v1/embedding"
+
+
 class Tags(ExtendedEnum):
     TASK = "task"
     LICENSE = "license"
@@ -49,6 +55,7 @@ class InferenceContainerType(ExtendedEnum):
 class InferenceContainerTypeFamily(ExtendedEnum):
     AQUA_VLLM_CONTAINER_FAMILY = "odsc-vllm-serving"
     AQUA_VLLM_V1_CONTAINER_FAMILY = "odsc-vllm-serving-v1"
+    AQUA_VLLM_LLAMA4_CONTAINER_FAMILY = "odsc-vllm-serving-llama4"
     AQUA_TGI_CONTAINER_FAMILY = "odsc-tgi-serving"
     AQUA_LLAMA_CPP_CONTAINER_FAMILY = "odsc-llama-cpp-serving"
 
@@ -119,4 +126,9 @@ CONTAINER_FAMILY_COMPATIBILITY: Dict[str, List[str]] = {
         InferenceContainerTypeFamily.AQUA_VLLM_V1_CONTAINER_FAMILY,
         InferenceContainerTypeFamily.AQUA_VLLM_CONTAINER_FAMILY,
     ],
+    InferenceContainerTypeFamily.AQUA_VLLM_LLAMA4_CONTAINER_FAMILY: [
+        InferenceContainerTypeFamily.AQUA_VLLM_LLAMA4_CONTAINER_FAMILY,
+        InferenceContainerTypeFamily.AQUA_VLLM_V1_CONTAINER_FAMILY,
+        InferenceContainerTypeFamily.AQUA_VLLM_CONTAINER_FAMILY,
+    ],
 }

ads/aqua/common/utils.py

@@ -832,7 +832,9 @@ def get_params_dict(params: Union[str, List[str]]) -> dict:
     """
     params_list = get_params_list(params) if isinstance(params, str) else params
     return {
-        split_result[0]: split_result[1] if len(split_result) > 1 else UNKNOWN
+        split_result[0]: " ".join(split_result[1:])
+        if len(split_result) > 1
+        else UNKNOWN
         for split_result in (x.split() for x in params_list)
     }
 
@@ -881,7 +883,9 @@ def build_params_string(params: dict) -> str:
         A params string.
     """
     return (
-        " ".join(f"{name} {value}" for name, value in params.items()).strip()
+        " ".join(
+            f"{name} {value}" if value else f"{name}" for name, value in params.items()
+        ).strip()
         if params
         else UNKNOWN
     )
@@ -1158,9 +1162,11 @@ def validate_cmd_var(cmd_var: List[str], overrides: List[str]) -> List[str]:
 
 
 def build_pydantic_error_message(ex: ValidationError):
-    """Added to handle error messages from pydantic model validator.
+    """
+    Added to handle error messages from pydantic model validator.
     Combine both loc and msg for errors where loc (field) is present in error details, else only build error
-    message using msg field."""
+    message using msg field.
+    """
 
     return {
         ".".join(map(str, e["loc"])): e["msg"]
@@ -1185,67 +1191,71 @@ def is_pydantic_model(obj: object) -> bool:
 
 @cached(cache=TTLCache(maxsize=1, ttl=timedelta(minutes=5), timer=datetime.now))
 def load_gpu_shapes_index(
-    auth: Optional[Dict] = None,
+    auth: Optional[Dict[str, Any]] = None,
 ) -> GPUShapesIndex:
     """
-    Loads the GPU shapes index from Object Storage or a local resource folder.
+    Load the GPU shapes index, preferring the OS bucket copy over the local one.
 
-    The function first attempts to load the file from an Object Storage bucket using fsspec.
-    If the loading fails (due to connection issues, missing file, etc.), it falls back to
-    loading the index from a local file.
+    Attempts to read `gpu_shapes_index.json` from OCI Object Storage first;
+    if that succeeds, those entries will override the local defaults.
 
     Parameters
     ----------
-    auth: (Dict, optional). Defaults to None.
-        The default authentication is set using `ads.set_auth` API. If you need to override the
-        default, use the `ads.common.auth.api_keys` or `ads.common.auth.resource_principal` to create appropriate
-        authentication signer and kwargs required to instantiate IdentityClient object.
+    auth
+        Optional auth dict (as returned by `ads.common.auth.default_signer()`)
+        to pass through to `fsspec.open()`.
 
     Returns
     -------
-    GPUShapesIndex: The parsed GPU shapes index.
+    GPUShapesIndex
+        Merged index where any shape present remotely supersedes the local entry.
 
     Raises
     ------
-    FileNotFoundError: If the GPU shapes index cannot be found in either Object Storage or locally.
-    json.JSONDecodeError: If the JSON is malformed.
+    json.JSONDecodeError
+        If any of the JSON is malformed.
     """
     file_name = "gpu_shapes_index.json"
-    data: Dict[str, Any] = {}
 
-    # Check if the CONDA_BUCKET_NS environment variable is set.
+    # Try remote load
+    remote_data: Dict[str, Any] = {}
     if CONDA_BUCKET_NS:
         try:
             auth = auth or authutil.default_signer()
-            # Construct the object storage path. Adjust bucket name and path as needed.
             storage_path = (
                 f"oci://{CONDA_BUCKET_NAME}@{CONDA_BUCKET_NS}/service_pack/{file_name}"
             )
-            logger.debug("Loading GPU shapes index from Object Storage")
-            with fsspec.open(storage_path, mode="r", **auth) as file_obj:
-                data = json.load(file_obj)
-            logger.debug("Successfully loaded GPU shapes index.")
-        except Exception as ex:
             logger.debug(
-                f"Failed to load GPU shapes index from Object Storage. Details: {ex}"
-            )
-
-    # If loading from Object Storage failed, load from the local resource folder.
-    if not data:
-        try:
-            local_path = os.path.join(
-                os.path.dirname(__file__), "../resources", file_name
+                "Loading GPU shapes index from Object Storage: %s", storage_path
             )
-            logger.debug(f"Loading GPU shapes index from {local_path}.")
-            with open(local_path) as file_obj:
-                data = json.load(file_obj)
-            logger.debug("Successfully loaded GPU shapes index.")
-        except Exception as e:
+            with fsspec.open(storage_path, mode="r", **auth) as f:
+                remote_data = json.load(f)
             logger.debug(
-                f"Failed to load GPU shapes index from {local_path}. Details: {e}"
+                "Loaded %d shapes from Object Storage",
+                len(remote_data.get("shapes", {})),
             )
+        except Exception as ex:
+            logger.debug("Remote load failed (%s); falling back to local", ex)
+
+    # Load local copy
+    local_data: Dict[str, Any] = {}
+    local_path = os.path.join(os.path.dirname(__file__), "../resources", file_name)
+    try:
+        logger.debug("Loading GPU shapes index from local file: %s", local_path)
+        with open(local_path) as f:
+            local_data = json.load(f)
+        logger.debug(
+            "Loaded %d shapes from local file", len(local_data.get("shapes", {}))
+        )
+    except Exception as ex:
+        logger.debug("Local load GPU shapes index failed (%s)", ex)
+
+    # Merge: remote shapes override local
+    local_shapes = local_data.get("shapes", {})
+    remote_shapes = remote_data.get("shapes", {})
+    merged_shapes = {**local_shapes, **remote_shapes}
 
-    return GPUShapesIndex(**data)
+    return GPUShapesIndex(shapes=merged_shapes)
 
 
 def get_preferred_compatible_family(selected_families: set[str]) -> str:

ads/aqua/config/container_config.py

@@ -7,6 +7,7 @@ from typing import Dict, List, Optional
 from oci.data_science.models import ContainerSummary
 from pydantic import Field
 
+from ads.aqua import logger
 from ads.aqua.config.utils.serializer import Serializable
 from ads.aqua.constants import (
     SERVICE_MANAGED_CONTAINER_URI_SCHEME,
@@ -168,50 +169,47 @@ class AquaContainerConfig(Serializable):
             container_type = container.family_name
             usages = [x.upper() for x in container.usages]
             if "INFERENCE" in usages or "MULTI_MODEL" in usages:
+                # Extract additional configurations
+                additional_configurations = {}
+                try:
+                    additional_configurations = (
+                        container.workload_configuration_details_list[
+                            0
+                        ].additional_configurations
+                    )
+                except (AttributeError, IndexError) as ex:
+                    logger.debug(
+                        "Failed to extract `additional_configurations` for container '%s': %s",
+                        getattr(container, "container_name", "<unknown>"),
+                        ex,
+                    )
+
                 container_item.platforms.append(
-                    container.workload_configuration_details_list[
-                        0
-                    ].additional_configurations.get("platforms")
+                    additional_configurations.get("platforms")
                 )
                 container_item.model_formats.append(
-                    container.workload_configuration_details_list[
-                        0
-                    ].additional_configurations.get("modelFormats")
+                    additional_configurations.get("modelFormats")
                 )
+
+                # Parse environment variables from `additional_configurations`.
+                # Only keys present in the configuration will be added to the result.
+                config_keys = {
+                    "MODEL_DEPLOY_PREDICT_ENDPOINT": UNKNOWN,
+                    "MODEL_DEPLOY_HEALTH_ENDPOINT": UNKNOWN,
+                    "MODEL_DEPLOY_ENABLE_STREAMING": UNKNOWN,
+                    "PORT": UNKNOWN,
+                    "HEALTH_CHECK_PORT": UNKNOWN,
+                    "VLLM_USE_V1": UNKNOWN,
+                }
+
                 env_vars = [
-                    {
-                        "MODEL_DEPLOY_PREDICT_ENDPOINT": container.workload_configuration_details_list[
-                            0
-                        ].additional_configurations.get(
-                            "MODEL_DEPLOY_PREDICT_ENDPOINT", UNKNOWN
-                        )
-                    },
-                    {
-                        "MODEL_DEPLOY_HEALTH_ENDPOINT": container.workload_configuration_details_list[
-                            0
-                        ].additional_configurations.get(
-                            "MODEL_DEPLOY_HEALTH_ENDPOINT", UNKNOWN
-                        )
-                    },
-                    {
-                        "MODEL_DEPLOY_ENABLE_STREAMING": container.workload_configuration_details_list[
-                            0
-                        ].additional_configurations.get(
-                            "MODEL_DEPLOY_ENABLE_STREAMING", UNKNOWN
-                        )
-                    },
-                    {
-                        "PORT": container.workload_configuration_details_list[
-                            0
-                        ].additional_configurations.get("PORT", "")
-                    },
-                    {
-                        "HEALTH_CHECK_PORT": container.workload_configuration_details_list[
-                            0
-                        ].additional_configurations.get("HEALTH_CHECK_PORT", UNKNOWN),
-                    },
+                    {key: additional_configurations.get(key, default)}
+                    for key, default in config_keys.items()
+                    if key in additional_configurations
                 ]
-                container_spec = AquaContainerConfigSpec(
+
+                # Build container spec
+                container_item.spec = AquaContainerConfigSpec(
                     cli_param=container.workload_configuration_details_list[0].cmd,
                     server_port=str(
                         container.workload_configuration_details_list[0].server_port
@@ -236,13 +234,14 @@ class AquaContainerConfig(Serializable):
                         )
                     ),
                 )
-                container_item.spec = container_spec
+
             if "INFERENCE" in usages or "MULTI_MODEL" in usages:
                 inference_items[container_type] = container_item
             if "FINE_TUNE" in usages:
                 finetune_items[container_type] = container_item
             if "EVALUATION" in usages:
                 evaluate_items[container_type] = container_item
+
         return cls(
             inference=inference_items, finetune=finetune_items, evaluate=evaluate_items
         )

ads/aqua/extension/deployment_handler.py

@@ -2,16 +2,18 @@
 # 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/
 
-from typing import List, Union
+from typing import List, Optional, Union
 from urllib.parse import urlparse
 
 from tornado.web import HTTPError
 
+from ads.aqua.app import logger
+from ads.aqua.client.client import Client, ExtendedRequestError
 from ads.aqua.common.decorator import handle_exceptions
+from ads.aqua.common.enums import PredictEndpoints
 from ads.aqua.extension.base_handler import AquaAPIhandler
 from ads.aqua.extension.errors import Errors
-from ads.aqua.modeldeployment import AquaDeploymentApp, MDInferenceResponse
-from ads.aqua.modeldeployment.entities import ModelParams
+from ads.aqua.modeldeployment import AquaDeploymentApp
 from ads.config import COMPARTMENT_OCID
 
 
@@ -175,23 +177,107 @@ class AquaDeploymentHandler(AquaAPIhandler):
         )
 
 
-class AquaDeploymentInferenceHandler(AquaAPIhandler):
-    @staticmethod
-    def validate_predict_url(endpoint):
-        try:
-            url = urlparse(endpoint)
-            if url.scheme != "https":
-                return False
-            if not url.netloc:
-                return False
-            return url.path.endswith("/predict")
-        except Exception:
-            return False
+class AquaDeploymentStreamingInferenceHandler(AquaAPIhandler):
+    def _get_model_deployment_response(
+        self,
+        model_deployment_id: str,
+        payload: dict,
+        route_override_header: Optional[str],
+    ):
+        """
+        Returns the model deployment inference response in a streaming fashion.
+
+        This method connects to the specified model deployment endpoint and
+        streams the inference output back to the caller, handling both text
+        and chat completion endpoints depending on the route override.
+
+        Parameters
+        ----------
+        model_deployment_id : str
+            The OCID of the model deployment to invoke.
+            Example: 'ocid1.datasciencemodeldeployment.iad.oc1.xxxyz'
+
+        payload : dict
+            Dictionary containing the model inference parameters.
+            Same example for text completions:
+                {
+                    "max_tokens": 1024,
+                    "temperature": 0.5,
+                    "prompt": "what are some good skills deep learning expert. Give us some tips on how to structure interview with some coding example?",
+                    "top_p": 0.4,
+                    "top_k": 100,
+                    "model": "odsc-llm",
+                    "frequency_penalty": 1,
+                    "presence_penalty": 1,
+                    "stream": true
+                }
+
+        route_override_header : Optional[str]
+            Optional override for the inference route, used for routing between
+            different endpoint types (e.g., chat vs. text completions).
+            Example: '/v1/chat/completions'
+
+        Returns
+        -------
+        Generator[str]
+            A generator that yields strings of the model's output as they are received.
+
+        Raises
+        ------
+        HTTPError
+            If the request to the model deployment fails or if streaming cannot be established.
+        """
+
+        model_deployment = AquaDeploymentApp().get(model_deployment_id)
+        endpoint = model_deployment.endpoint + "/predictWithResponseStream"
+        endpoint_type = model_deployment.environment_variables.get(
+            "MODEL_DEPLOY_PREDICT_ENDPOINT", PredictEndpoints.TEXT_COMPLETIONS_ENDPOINT
+        )
+        aqua_client = Client(endpoint=endpoint)
+
+        if PredictEndpoints.CHAT_COMPLETIONS_ENDPOINT in (
+            endpoint_type,
+            route_override_header,
+        ):
+            try:
+                for chunk in aqua_client.chat(
+                    messages=payload.pop("messages"),
+                    payload=payload,
+                    stream=True,
+                ):
+                    try:
+                        yield chunk["choices"][0]["delta"]["content"]
+                    except Exception as e:
+                        logger.debug(
+                            f"Exception occurred while parsing streaming response: {e}"
+                        )
+            except ExtendedRequestError as ex:
+                raise HTTPError(400, str(ex))
+            except Exception as ex:
+                raise HTTPError(500, str(ex))
+
+        elif endpoint_type == PredictEndpoints.TEXT_COMPLETIONS_ENDPOINT:
+            try:
+                for chunk in aqua_client.generate(
+                    prompt=payload.pop("prompt"),
+                    payload=payload,
+                    stream=True,
+                ):
+                    try:
+                        yield chunk["choices"][0]["text"]
+                    except Exception as e:
+                        logger.debug(
+                            f"Exception occurred while parsing streaming response: {e}"
+                        )
+            except ExtendedRequestError as ex:
+                raise HTTPError(400, str(ex))
+            except Exception as ex:
+                raise HTTPError(500, str(ex))
 
     @handle_exceptions
-    def post(self, *args, **kwargs):  # noqa: ARG002
+    def post(self, model_deployment_id):
         """
-        Handles inference request for the Active Model Deployments
+        Handles streaming inference request for the Active Model Deployments
         Raises
         ------
         HTTPError
@@ -205,32 +291,29 @@ class AquaDeploymentInferenceHandler(AquaAPIhandler):
         if not input_data:
             raise HTTPError(400, Errors.NO_INPUT_DATA)
 
-        endpoint = input_data.get("endpoint")
-        if not endpoint:
-            raise HTTPError(400, Errors.MISSING_REQUIRED_PARAMETER.format("endpoint"))
-
-        if not self.validate_predict_url(endpoint):
-            raise HTTPError(400, Errors.INVALID_INPUT_DATA_FORMAT.format("endpoint"))
-
         prompt = input_data.get("prompt")
-        if not prompt:
-            raise HTTPError(400, Errors.MISSING_REQUIRED_PARAMETER.format("prompt"))
+        messages = input_data.get("messages")
 
-        model_params = (
-            input_data.get("model_params") if input_data.get("model_params") else {}
-        )
-        try:
-            model_params_obj = ModelParams(**model_params)
-        except Exception as ex:
+        if not prompt and not messages:
             raise HTTPError(
-                400, Errors.INVALID_INPUT_DATA_FORMAT.format("model_params")
-            ) from ex
-
-        return self.finish(
-            MDInferenceResponse(prompt, model_params_obj).get_model_deployment_response(
-                endpoint
+                400, Errors.MISSING_REQUIRED_PARAMETER.format("prompt/messages")
             )
+        if not input_data.get("model"):
+            raise HTTPError(400, Errors.MISSING_REQUIRED_PARAMETER.format("model"))
+        route_override_header = self.request.headers.get("route", None)
+        self.set_header("Content-Type", "text/event-stream")
+        response_gen = self._get_model_deployment_response(
+            model_deployment_id, input_data, route_override_header
         )
+        try:
+            for chunk in response_gen:
+                self.write(chunk)
+                self.flush()
+            self.finish()
+        except Exception as ex:
+            self.set_status(ex.status_code)
+            self.write({"message": "Error occurred", "reason": str(ex)})
+            self.finish()
 
 
 class AquaDeploymentParamsHandler(AquaAPIhandler):
@@ -294,5 +377,5 @@ __handlers__ = [
     ("deployments/?([^/]*)", AquaDeploymentHandler),
     ("deployments/?([^/]*)/activate", AquaDeploymentHandler),
     ("deployments/?([^/]*)/deactivate", AquaDeploymentHandler),
-    ("inference", AquaDeploymentInferenceHandler),
+    ("inference/stream/?([^/]*)", AquaDeploymentStreamingInferenceHandler),
 ]

ads/aqua/extension/ui_handler.py

@@ -15,7 +15,7 @@ from ads.aqua.extension.errors import Errors
 from ads.aqua.extension.utils import validate_function_parameters
 from ads.aqua.model.entities import ImportModelDetails
 from ads.aqua.ui import AquaUIApp
-from ads.config import COMPARTMENT_OCID
+from ads.config import COMPARTMENT_OCID, IS_BYOR_ENABLED
 
 
 @dataclass
@@ -82,6 +82,10 @@ class AquaUIHandler(AquaAPIhandler):
             return self.is_bucket_versioned()
         elif paths.startswith("aqua/containers"):
             return self.list_containers()
+        elif paths.startswith("aqua/capacityreservations/enabled"):
+            return self.is_capacity_reservations_enabled()
+        elif paths.startswith("aqua/capacityreservations"):
+            return self.list_capacity_reservations()
         else:
             raise HTTPError(400, f"The request {self.request.path} is invalid.")
 
@@ -103,6 +107,19 @@ class AquaUIHandler(AquaAPIhandler):
             AquaUIApp().list_log_groups(compartment_id=compartment_id, **kwargs)
         )
 
+    def is_capacity_reservations_enabled(self):
+        """Checks if the tenant is whitelisted for BYOR (Bring your own reservation) feature."""
+        return self.finish({"status": str(IS_BYOR_ENABLED).strip().lower() == "true"})
+
+    def list_capacity_reservations(self, **kwargs):
+        """Lists users compute reservations in a specified compartment."""
+        compartment_id = self.get_argument("compartment_id", default=COMPARTMENT_OCID)
+        return self.finish(
+            AquaUIApp().list_capacity_reservations(
+                compartment_id=compartment_id, **kwargs
+            )
+        )
+
     def list_logs(self, log_group_id: str, **kwargs):
         """Lists the specified log group's log objects."""
         return self.finish(AquaUIApp().list_logs(log_group_id=log_group_id, **kwargs))
@@ -279,4 +296,5 @@ __handlers__ = [
     ("bucket/versioning/?([^/]*)", AquaUIHandler),
     ("containers/?([^/]*)", AquaUIHandler),
     ("cli/?([^/]*)", AquaCLIHandler),
+    ("capacityreservations/?([^/]*)", AquaUIHandler),
 ]

ads/aqua/model/model.py

@@ -83,6 +83,10 @@ from ads.aqua.model.entities import (
     ModelValidationResult,
 )
 from ads.aqua.model.enums import MultiModelSupportedTaskType
+from ads.aqua.model.utils import (
+    extract_base_model_from_ft,
+    extract_fine_tune_artifacts_path,
+)
 from ads.common.auth import default_signer
 from ads.common.oci_resource import SEARCH_TYPE, OCIResource
 from ads.common.utils import (
@@ -311,12 +315,27 @@ class AquaModelApp(AquaApp):
             #         "Currently only service models are supported for multi model deployment."
             #     )
 
+            # check if model is a fine-tuned model and if so, add the fine tuned weights path to the fine_tune_weights_location pydantic field
+            is_fine_tuned_model = (
+                Tags.AQUA_FINE_TUNED_MODEL_TAG in source_model.freeform_tags
+            )
+
+            if is_fine_tuned_model:
+                model.model_id, model.model_name = extract_base_model_from_ft(
+                    source_model
+                )
+                model_artifact_path, model.fine_tune_weights_location = (
+                    extract_fine_tune_artifacts_path(source_model)
+                )
+
+            else:
+                # Retrieve model artifact for base models
+                model_artifact_path = source_model.artifact
+
             display_name_list.append(display_name)
 
             self._extract_model_task(model, source_model)
 
-            # Retrieve model artifact
-            model_artifact_path = source_model.artifact
             if not model_artifact_path:
                 raise AquaValueError(
                     f"Model '{display_name}' (ID: {model.model_id}) has no artifacts. "
@@ -367,7 +386,8 @@ class AquaModelApp(AquaApp):
                 raise AquaValueError(
                     "The selected models are associated with different container families: "
                     f"{list(selected_models_deployment_containers)}."
-                    "For multi-model deployment, all models in the group must share the same container family."
+                    "For multi-model deployment, all models in the group must belong to the same container "
+                    "family or to compatible container families."
                 )
         else:
             deployment_container = selected_models_deployment_containers.pop()

ads/aqua/model/utils.py

@@ -0,0 +1,52 @@
+#!/usr/bin/env python
+# Copyright (c) 2025 Oracle and/or its affiliates.
+# Licensed under the Universal Permissive License v 1.0 as shown at https://oss.oracle.com/licenses/upl/
+"""AQUA model utils"""
+
+from typing import Dict, Optional, Tuple
+
+from ads.aqua.common.entities import AquaMultiModelRef
+from ads.aqua.common.errors import AquaValueError
+from ads.aqua.common.utils import get_model_by_reference_paths
+from ads.aqua.finetuning.constants import FineTuneCustomMetadata
+from ads.common.object_storage_details import ObjectStorageDetails
+from ads.model.datascience_model import DataScienceModel
+
+
+def extract_base_model_from_ft(aqua_model: DataScienceModel) -> Tuple[str, str]:
+    """Extracts the model_name and base model OCID (config_source_id) OCID for a fine-tuned model"""
+
+    config_source_id = aqua_model.custom_metadata_list.get(
+        FineTuneCustomMetadata.FINE_TUNE_SOURCE
+    ).value
+    model_name = aqua_model.custom_metadata_list.get(
+        FineTuneCustomMetadata.FINE_TUNE_SOURCE_NAME
+    ).value
+
+    if not config_source_id or not model_name:
+        raise AquaValueError(
+            f"Either {FineTuneCustomMetadata.FINE_TUNE_SOURCE} or {FineTuneCustomMetadata.FINE_TUNE_SOURCE_NAME} is missing "
+            f"from custom metadata for the model {config_source_id}"
+        )
+
+    return config_source_id, model_name
+
+
+def extract_fine_tune_artifacts_path(aqua_model: DataScienceModel) -> Tuple[str, str]:
+    """Extracts the fine tuning source (fine_tune_output_path) and base model path from the DataScienceModel Object"""
+
+    base_model_path, fine_tune_output_path = get_model_by_reference_paths(
+        aqua_model.model_file_description
+    )
+
+    if not fine_tune_output_path or not ObjectStorageDetails.is_oci_path(
+        fine_tune_output_path
+    ):
+        raise AquaValueError(
+            "Fine tuned output path is not available in the model artifact."
+        )
+
+    os_path = ObjectStorageDetails.from_path(fine_tune_output_path)
+    fine_tune_output_path = os_path.filepath.rstrip("/")
+
+    return base_model_path, fine_tune_output_path

ads/aqua/modeldeployment/__init__.py

@@ -1,8 +1,6 @@
 #!/usr/bin/env python
-# -*- coding: utf-8 -*-
-# Copyright (c) 2024 Oracle and/or its affiliates.
+# Copyright (c) 2025 Oracle and/or its affiliates.
 # Licensed under the Universal Permissive License v 1.0 as shown at https://oss.oracle.com/licenses/upl/
 from ads.aqua.modeldeployment.deployment import AquaDeploymentApp
-from ads.aqua.modeldeployment.inference import MDInferenceResponse
 
-__all__ = ["AquaDeploymentApp", "MDInferenceResponse"]
+__all__ = ["AquaDeploymentApp"]

ads/aqua/modeldeployment/deployment.py

@@ -17,7 +17,11 @@ from ads.aqua.common.entities import (
     ComputeShapeSummary,
     ContainerPath,
 )
-from ads.aqua.common.enums import InferenceContainerTypeFamily, ModelFormat, Tags
+from ads.aqua.common.enums import (
+    InferenceContainerTypeFamily,
+    ModelFormat,
+    Tags,
+)
 from ads.aqua.common.errors import AquaRuntimeError, AquaValueError
 from ads.aqua.common.utils import (
     DEFINED_METADATA_TO_FILE_MAP,
@@ -25,7 +29,6 @@ from ads.aqua.common.utils import (
     build_pydantic_error_message,
     get_combined_params,
     get_container_params_type,
-    get_model_by_reference_paths,
     get_ocid_substring,
     get_params_dict,
     get_params_list,
@@ -46,9 +49,12 @@ from ads.aqua.constants import (
     UNKNOWN_DICT,
 )
 from ads.aqua.data import AquaResourceIdentifier
-from ads.aqua.finetuning.finetuning import FineTuneCustomMetadata
 from ads.aqua.model import AquaModelApp
 from ads.aqua.model.constants import AquaModelMetadataKeys, ModelCustomMetadataFields
+from ads.aqua.model.utils import (
+    extract_base_model_from_ft,
+    extract_fine_tune_artifacts_path,
+)
 from ads.aqua.modeldeployment.entities import (
     AquaDeployment,
     AquaDeploymentConfig,
@@ -211,6 +217,7 @@ class AquaDeploymentApp(AquaApp):
             )
         else:
             model_ids = [model.model_id for model in create_deployment_details.models]
+
             try:
                 model_config_summary = self.get_multimodel_deployment_config(
                     model_ids=model_ids, compartment_id=compartment_id
@@ -343,22 +350,6 @@ class AquaDeploymentApp(AquaApp):
         config_source_id = create_deployment_details.model_id
         model_name = aqua_model.display_name
 
-        is_fine_tuned_model = Tags.AQUA_FINE_TUNED_MODEL_TAG in aqua_model.freeform_tags
-
-        if is_fine_tuned_model:
-            try:
-                config_source_id = aqua_model.custom_metadata_list.get(
-                    FineTuneCustomMetadata.FINE_TUNE_SOURCE
-                ).value
-                model_name = aqua_model.custom_metadata_list.get(
-                    FineTuneCustomMetadata.FINE_TUNE_SOURCE_NAME
-                ).value
-            except ValueError as err:
-                raise AquaValueError(
-                    f"Either {FineTuneCustomMetadata.FINE_TUNE_SOURCE} or {FineTuneCustomMetadata.FINE_TUNE_SOURCE_NAME} is missing "
-                    f"from custom metadata for the model {config_source_id}"
-                ) from err
-
         # set up env and cmd var
         env_var = create_deployment_details.env_var or {}
         cmd_var = create_deployment_details.cmd_var or []
@@ -378,19 +369,11 @@ class AquaDeploymentApp(AquaApp):
 
         env_var.update({"BASE_MODEL": f"{model_path_prefix}"})
 
-        if is_fine_tuned_model:
-            _, fine_tune_output_path = get_model_by_reference_paths(
-                aqua_model.model_file_description
-            )
-
-            if not fine_tune_output_path:
-                raise AquaValueError(
-                    "Fine tuned output path is not available in the model artifact."
-                )
-
-            os_path = ObjectStorageDetails.from_path(fine_tune_output_path)
-            fine_tune_output_path = os_path.filepath.rstrip("/")
+        is_fine_tuned_model = Tags.AQUA_FINE_TUNED_MODEL_TAG in aqua_model.freeform_tags
 
+        if is_fine_tuned_model:
+            config_source_id, model_name = extract_base_model_from_ft(aqua_model)
+            _, fine_tune_output_path = extract_fine_tune_artifacts_path(aqua_model)
             env_var.update({"FT_MODEL": f"{fine_tune_output_path}"})
 
         container_type_key = self._get_container_type_key(
@@ -647,6 +630,12 @@ class AquaDeploymentApp(AquaApp):
             config_data = {"params": params, "model_path": artifact_path_prefix}
             if model.model_task:
                 config_data["model_task"] = model.model_task
+
+            if model.fine_tune_weights_location:
+                config_data["fine_tune_weights_location"] = (
+                    model.fine_tune_weights_location
+                )
+
             model_config.append(config_data)
             model_name_list.append(model.model_name)
 
@@ -805,6 +794,9 @@ class AquaDeploymentApp(AquaApp):
         # we arbitrarily choose last 8 characters of OCID to identify MD in telemetry
         telemetry_kwargs = {"ocid": get_ocid_substring(deployment_id, key_len=8)}
 
+        if Tags.BASE_MODEL_CUSTOM in tags:
+            telemetry_kwargs["custom_base_model"] = True
+
         # tracks unique deployments that were created in the user compartment
         self.telemetry.record_event_async(
             category=f"aqua/{model_type}/deployment",
@@ -948,7 +940,6 @@ class AquaDeploymentApp(AquaApp):
         model_deployment = self.ds_client.get_model_deployment(
             model_deployment_id=model_deployment_id, **kwargs
         ).data
-
         oci_aqua = (
             (
                 Tags.AQUA_TAG in model_deployment.freeform_tags
@@ -993,7 +984,6 @@ class AquaDeploymentApp(AquaApp):
         aqua_deployment = AquaDeployment.from_oci_model_deployment(
             model_deployment, self.region
         )
-
         if Tags.MULTIMODEL_TYPE_TAG in model_deployment.freeform_tags:
             aqua_model_id = model_deployment.freeform_tags.get(
                 Tags.AQUA_MODEL_ID_TAG, UNKNOWN
@@ -1024,7 +1014,6 @@ class AquaDeploymentApp(AquaApp):
             aqua_deployment.models = [
                 AquaMultiModelRef(**metadata) for metadata in multi_model_metadata
             ]
-
         return AquaDeploymentDetail(
             **vars(aqua_deployment),
             log_group=AquaResourceIdentifier(

ads/aqua/ui.py

@@ -90,6 +90,26 @@ class AquaUIApp(AquaApp):
         res = self.logging_client.list_logs(log_group_id=log_group_id, **kwargs).data
         return sanitize_response(oci_client=self.logging_client, response=res)
 
+    @telemetry(entry_point="plugin=ui&action=list_capacity_reservations", name="aqua")
+    def list_capacity_reservations(self, **kwargs) -> list:
+        """
+        Lists users compute reservations in a specified compartment
+
+        Returns
+        -------
+            json representation of `oci.core.models.ComputeCapacityReservationSummary`.
+
+        """
+        compartment_id = kwargs.pop("compartment_id", COMPARTMENT_OCID)
+        logger.info(f"Loading Capacity reservations from compartment: {compartment_id}")
+
+        reservations = self.compute_client.list_compute_capacity_reservations(
+            compartment_id=compartment_id, **kwargs
+        )
+        return sanitize_response(
+            oci_client=self.compute_client, response=reservations.data
+        )
+
     @telemetry(entry_point="plugin=ui&action=list_compartments", name="aqua")
     def list_compartments(self) -> str:
         """Lists the compartments in a tenancy specified by TENANCY_OCID env variable. This is a pass through the OCI list_compartments

ads/common/oci_client.py

@@ -1,7 +1,6 @@
 #!/usr/bin/env python
-# -*- coding: utf-8; -*-
 
-# Copyright (c) 2021, 2024 Oracle and/or its affiliates.
+# Copyright (c) 2021, 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 logging
@@ -9,20 +8,20 @@ import logging
 import oci.artifacts
 from oci.ai_language import AIServiceLanguageClient
 from oci.artifacts import ArtifactsClient
+from oci.core import ComputeClient, VirtualNetworkClient
 from oci.data_catalog import DataCatalogClient
 from oci.data_flow import DataFlowClient
 from oci.data_labeling_service import DataLabelingManagementClient
 from oci.data_labeling_service_dataplane import DataLabelingClient
 from oci.data_science import DataScienceClient
 from oci.identity import IdentityClient
+from oci.limits import LimitsClient
+from oci.logging import LoggingManagementClient
 from oci.marketplace import MarketplaceClient
 from oci.object_storage import ObjectStorageClient
 from oci.resource_search import ResourceSearchClient
 from oci.secrets import SecretsClient
 from oci.vault import VaultsClient
-from oci.logging import LoggingManagementClient
-from oci.core import VirtualNetworkClient
-from oci.limits import LimitsClient
 
 logger = logging.getLogger(__name__)
 
@@ -69,6 +68,7 @@ class OCIClientFactory:
             "secret": SecretsClient,
             "vault": VaultsClient,
             "identity": IdentityClient,
+            "compute": ComputeClient,
             "ai_language": AIServiceLanguageClient,
             "data_labeling_dp": DataLabelingClient,
             "data_labeling_cp": DataLabelingManagementClient,
@@ -114,6 +114,10 @@ class OCIClientFactory:
     def object_storage(self):
         return self.create_client("object_storage")
 
+    @property
+    def compute(self):
+        return self.create_client("compute")
+
     @property
     def identity(self):
         return self.create_client("identity")

ads/config.py

@@ -14,6 +14,7 @@ OCI_ODSC_SERVICE_ENDPOINT = os.environ.get("OCI_ODSC_SERVICE_ENDPOINT")
 OCI_IDENTITY_SERVICE_ENDPOINT = os.environ.get("OCI_IDENTITY_SERVICE_ENDPOINT")
 NB_SESSION_COMPARTMENT_OCID = os.environ.get("NB_SESSION_COMPARTMENT_OCID")
 PROJECT_OCID = os.environ.get("PROJECT_OCID") or os.environ.get("PIPELINE_PROJECT_OCID")
+IS_BYOR_ENABLED = os.environ.get("ALLOWLISTED_FOR_BYOR", False)
 NB_SESSION_OCID = os.environ.get("NB_SESSION_OCID")
 USER_OCID = os.environ.get("USER_OCID")
 OCI_RESOURCE_PRINCIPAL_VERSION = os.environ.get("OCI_RESOURCE_PRINCIPAL_VERSION")

ads/opctl/operator/lowcode/forecast/model/prophet.py

@@ -43,22 +43,27 @@ def _fit_model(data, params, additional_regressors):
     from prophet import Prophet
 
     monthly_seasonality = params.pop("monthly_seasonality", False)
-    data_floor = params.pop("min", None)
-    data_cap = params.pop("max", None)
-    if data_cap or data_floor:
+
+    has_min = "min" in params
+    has_max = "max" in params
+    if has_min or has_max:
         params["growth"] = "logistic"
+        data_floor = params.pop("min", None)
+        data_cap = params.pop("max", None)
+    
     model = Prophet(**params)
     if monthly_seasonality:
         model.add_seasonality(name="monthly", period=30.5, fourier_order=5)
         params["monthly_seasonality"] = monthly_seasonality
     for add_reg in additional_regressors:
         model.add_regressor(add_reg)
-    if data_floor:
+        
+    if has_min:
         data["floor"] = float(data_floor)
-        params["floor"] = data_floor
-    if data_cap:
+        params["min"] = data_floor
+    if has_max:
         data["cap"] = float(data_cap)
-        params["cap"] = data_cap
+        params["max"] = data_cap
 
     model.fit(data)
     return model
@@ -304,7 +309,7 @@ class ProphetOperatorModel(ForecastOperatorBaseModel):
             # Global Expl
             g_expl = self.drop_horizon(expl_df).mean()
             g_expl.name = s_id
-            global_expl.append(g_expl)
+            global_expl.append(np.abs(g_expl))
         self.global_explanation = pd.concat(global_expl, axis=1)
         self.formatted_global_explanation = (
             self.global_explanation / self.global_explanation.sum(axis=0) * 100

ads/telemetry/client.py

@@ -8,6 +8,7 @@ import threading
 import traceback
 import urllib.parse
 from typing import Optional
+import concurrent.futures
 
 import oci
 
@@ -16,7 +17,8 @@ from ads.config import DEBUG_TELEMETRY
 from .base import TelemetryBase
 
 logger = logging.getLogger(__name__)
-
+THREAD_POOL_SIZE = 16
+thread_pool = concurrent.futures.ThreadPoolExecutor(max_workers=THREAD_POOL_SIZE)
 
 class TelemetryClient(TelemetryBase):
     """Represents a telemetry python client providing functions to record an event.
@@ -102,7 +104,7 @@ class TelemetryClient(TelemetryBase):
 
     def record_event_async(
         self, category: str = None, action: str = None, detail: str = None, **kwargs
-    ):
+    )-> None:
         """Send a head request to generate an event record.
 
         Parameters
@@ -117,9 +119,4 @@ class TelemetryClient(TelemetryBase):
         Thread
             A started thread to send a head request to generate an event record.
         """
-        thread = threading.Thread(
-            target=self.record_event, args=(category, action, detail), kwargs=kwargs
-        )
-        thread.daemon = True
-        thread.start()
-        return thread
+        thread_pool.submit(self.record_event, args=(category, action, detail), kwargs=kwargs)

{oracle_ads-2.13.9rc1.dist-info → oracle_ads-2.13.10rc0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: oracle_ads
-Version: 2.13.9rc1
+Version: 2.13.10rc0
 Summary: Oracle Accelerated Data Science SDK
 Keywords: Oracle Cloud Infrastructure,OCI,Machine Learning,ML,Artificial Intelligence,AI,Data Science,Cloud,Oracle,GenAI,Generative AI,Forecast,Anomaly,Document Understanding
 Author: Oracle Data Science
@@ -15,7 +15,7 @@ Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
 License-File: LICENSE.txt
-Requires-Dist: PyYAML>=6
+Requires-Dist: PyYAML>=6.0.1
 Requires-Dist: asteval>=0.9.25
 Requires-Dist: cerberus>=1.3.4
 Requires-Dist: cloudpickle>=1.6.0
@@ -99,13 +99,14 @@ Requires-Dist: ipython>=7.23.1, <8.0 ; extra == "notebook"
 Requires-Dist: ipywidgets~=7.6.3 ; extra == "notebook"
 Requires-Dist: lightgbm ; extra == "onnx"
 Requires-Dist: onnx>=1.12.0,<=1.15.0 ; extra == "onnx" and ( python_version < '3.12')
-Requires-Dist: onnx>=1.12.0 ; extra == "onnx" and ( python_version >= '3.12')
-Requires-Dist: onnxmltools>=1.10.0 ; extra == "onnx"
+Requires-Dist: onnx~=1.17.0 ; extra == "onnx" and ( python_version >= '3.12')
+Requires-Dist: onnxmltools~=1.13.0 ; extra == "onnx"
 Requires-Dist: onnxruntime~=1.17.0,!=1.16.0 ; extra == "onnx" and ( python_version < '3.12')
-Requires-Dist: onnxruntime ; extra == "onnx" and ( python_version >= '3.12')
+Requires-Dist: onnxruntime~=1.22.0 ; extra == "onnx" and ( python_version >= '3.12')
 Requires-Dist: oracle_ads[viz] ; extra == "onnx"
 Requires-Dist: protobuf ; extra == "onnx"
-Requires-Dist: skl2onnx>=1.10.4 ; extra == "onnx"
+Requires-Dist: skl2onnx>=1.10.4 ; extra == "onnx" and ( python_version < '3.12')
+Requires-Dist: skl2onnx~=1.18.0 ; extra == "onnx" and ( python_version >= '3.12')
 Requires-Dist: tf2onnx ; extra == "onnx"
 Requires-Dist: xgboost<=1.7 ; extra == "onnx"
 Requires-Dist: conda-pack ; extra == "opctl"

{oracle_ads-2.13.9rc1.dist-info → oracle_ads-2.13.10rc0.dist-info}/RECORD

@@ -1,23 +1,23 @@
 ads/__init__.py,sha256=OxHySbHbMqPgZ8sUj33Bxy-smSiNgRjtcSUV77oBL08,3787
 ads/cli.py,sha256=WkOpZv8jWgFYN9BNkt2LJBs9KzJHgFqq3pIymsqc8Q4,4292
-ads/config.py,sha256=yrCvWEEYcMwWkk9_6IJJZnxbvrOVzsQNMBrCJVafYU8,8106
+ads/config.py,sha256=ZFZpp0SgKgB1-7yQ6MEdVdyJ2nwzixbH91sJ4_XnG8Y,8170
 ads/aqua/__init__.py,sha256=7DjwtmZaX-_atIkmZu6XQKHqJUEeemJGR2TlxzMHSXs,973
-ads/aqua/app.py,sha256=KesfIyVm3T8mj3ugsdVSp05b9RwQAEVw7QN1UB4o4qU,18397
+ads/aqua/app.py,sha256=PmG8-XnOfMRJ0GQgJDl6TSGF95EBTrMXGRO8p1FG43Y,18475
 ads/aqua/cli.py,sha256=8S0JnhWY9IBZjMyB-5r4I-2nl-WK6yw1iirPsAXICF0,3358
 ads/aqua/constants.py,sha256=dUl02j5XTAG6sL7XJ9HS5fT0Z869WceRLFIbBwCzmtw,5081
 ads/aqua/data.py,sha256=HfxLfKiNiPJecMQy0JAztUsT3IdZilHHHOrCJnjZMc4,408
-ads/aqua/ui.py,sha256=AyX1vFW9f6hoyKN55M6s4iKBLHsOHC41hwRjDfD4NlI,20191
+ads/aqua/ui.py,sha256=PYyr46ewx9Qygcsv6BryUF6rLHU0t5YjUgKSb1uZK2Y,20971
 ads/aqua/client/__init__.py,sha256=-46EcKQjnWEXxTt85bQzXjA5xsfoBXIGm_syKFlVL1c,178
 ads/aqua/client/client.py,sha256=zlscNhFZVgGnkJ-aj5iZ5v5FedOzpQc4RJDxGPl9VvQ,31388
 ads/aqua/client/openai_client.py,sha256=Gi8nSrtPAUOjxRNu-6UUAqtxWyQIQ5CAvatnm7XfnaM,12501
 ads/aqua/common/__init__.py,sha256=rZrmh1nho40OCeabXCNWtze-mXi-PGKetcZdxZSn3_0,204
 ads/aqua/common/decorator.py,sha256=JEN6Cy4DYgQbmIR3ShCjTuBMCnilDxq7jkYMJse1rcM,4112
-ads/aqua/common/entities.py,sha256=2_Sv07SvekPmU77DfQIycdv0UJOaT6TZM5WyNJyq7GM,9188
-ads/aqua/common/enums.py,sha256=rTZDOQzTfcgwEl7gjVY3_JotHXkz7wB_edEIB0i6AeQ,3739
+ads/aqua/common/entities.py,sha256=ZHlnW2gf6gQ-TmRCjRcobR5qWoM3Vj7RaGYpAIzFKro,9474
+ads/aqua/common/enums.py,sha256=fBJczGd4Trv9Lf0O4_TAEDWYkp5ou9oGeKhxdSW2-84,4274
 ads/aqua/common/errors.py,sha256=QONm-2jKBg8AjgOKXm6x-arAV1KIW9pdhfNN1Ys21Wo,3044
-ads/aqua/common/utils.py,sha256=z93NqufjGzmEpsd7VmSvIpFUawcaoLjBFPSiBCjq2Wk,42001
+ads/aqua/common/utils.py,sha256=OBFfvtAyI-CkM0ezF8unfYH596KzNkul-o8XMFgwep0,41833
 ads/aqua/config/__init__.py,sha256=2a_1LI4jWtJpbic5_v4EoOUTXCAH7cmsy9BW5prDHjU,179
-ads/aqua/config/container_config.py,sha256=WkxaBZ-6TlKXbhrLD5q-BAmXXZp_crLoZGp8TNtbGHg,9844
+ads/aqua/config/container_config.py,sha256=W6V5VZOLyEAbiY3rEbHyv4KY8yRhj8Qp6k1nWX_FwWg,9590
 ads/aqua/config/evaluation/__init__.py,sha256=2a_1LI4jWtJpbic5_v4EoOUTXCAH7cmsy9BW5prDHjU,179
 ads/aqua/config/evaluation/evaluation_service_config.py,sha256=NuaQoLVYPHJiWjGfq1-F6-DK0DyOAGjVS87K1SXFVvw,4497
 ads/aqua/config/utils/__init__.py,sha256=2a_1LI4jWtJpbic5_v4EoOUTXCAH7cmsy9BW5prDHjU,179
@@ -36,7 +36,7 @@ ads/aqua/extension/aqua_ws_msg_handler.py,sha256=VDa9vQOsYKX6flsUkDEx6nl-5MFCH5R
 ads/aqua/extension/base_handler.py,sha256=W-eBXn9XYypCZuY84e9cSKRuY0CDyuou_znV6Yn9YzU,3047
 ads/aqua/extension/common_handler.py,sha256=okjFJlJv0FLXsMM1td6upqqA6tJEJIj1IIfTiughC5Q,3809
 ads/aqua/extension/common_ws_msg_handler.py,sha256=PAy98ZsM8VAXcy11ahsuam3QUDdmE-Hz4F5pISVkNHY,1242
-ads/aqua/extension/deployment_handler.py,sha256=Q5EHfAWcEqiE9rH0lQeFXPn0WQdwiRlrl4lZI1OXPqo,10394
+ads/aqua/extension/deployment_handler.py,sha256=9ncV1H1pWnyoZuF4iva7n4xpSpGZauZJs2455Vxg2YE,13929
 ads/aqua/extension/deployment_ws_msg_handler.py,sha256=JX3ZHRtscrflSxT7ZTEEI_p_owtk3m5FZq3QXE96AGY,2013
 ads/aqua/extension/errors.py,sha256=4LbzZdCoDEtOcrVI-1dgiza4oAYGof6w5LbN6HqroYk,1396
 ads/aqua/extension/evaluation_handler.py,sha256=fJH73fa0xmkEiP8SxKL4A4dJgj-NoL3z_G-w_WW2zJs,4353
@@ -44,7 +44,7 @@ ads/aqua/extension/evaluation_ws_msg_handler.py,sha256=dv0iwOSTxYj1kQ1rPEoDmGgFB
 ads/aqua/extension/finetune_handler.py,sha256=97obbhITswTrBvl88g7gk4GvF2SUHBGUAq4rOylFbtQ,3079
 ads/aqua/extension/model_handler.py,sha256=LlfBqGI4YVXio0gUnqi7Tpe3yfkc7-ToZCcQ3cds6rY,14094
 ads/aqua/extension/models_ws_msg_handler.py,sha256=VyPbtBZrbRMIYJqOy5DR7j4M4qJK1RBqkxX6RbIvPGE,1851
-ads/aqua/extension/ui_handler.py,sha256=Q0LkrV6VtVUI4GpNgqJQt8SGzxHzp4X5hdHF6KgPp9M,11217
+ads/aqua/extension/ui_handler.py,sha256=OAyRgEQ26rpNA5CwBGG6YXM9qCuPrZAUXW490a-fkxs,12136
 ads/aqua/extension/ui_websocket_handler.py,sha256=oLFjaDrqkSERbhExdvxjLJX0oRcP-DVJ_aWn0qy0uvo,5084
 ads/aqua/extension/utils.py,sha256=-uppIKADKl8TFzZB2QWEIei_wtVwWN63qffhuh4Q_KA,5159
 ads/aqua/extension/models/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -57,12 +57,12 @@ ads/aqua/model/__init__.py,sha256=j2iylvERdANxgrEDp7b_mLcKMz1CF5Go0qgYCiMwdos,27
 ads/aqua/model/constants.py,sha256=oOAb4ulsdWBtokCE5SPX7wg8X8SaisLPayua58zhWfY,1856
 ads/aqua/model/entities.py,sha256=JiKB8SnaUxerRMlwrgpyfQLRuTOB8I14J-Rg5RFPwqw,12660
 ads/aqua/model/enums.py,sha256=bN8GKmgRl40PQrTmd1r-Pqr9VXTIV8gic5-3SAGNnwg,1152
-ads/aqua/model/model.py,sha256=AjsM7o5Dcas4G5imdCQ1VX2Y5bCooMZvDQBKQX8KTUA,88217
-ads/aqua/modeldeployment/__init__.py,sha256=RJCfU1yazv3hVWi5rS08QVLTpTwZLnlC8wU8diwFjnM,391
+ads/aqua/model/model.py,sha256=nq9PehumfFyh_UZDhltVl7Vi_L4TMM9KCx6yqZL3oDw,89027
+ads/aqua/model/utils.py,sha256=8lYaD51JpAPb7U_xQzRSzFflRefU5Yz6vA80cuZOL0w,2087
+ads/aqua/modeldeployment/__init__.py,sha256=k5fGG2b6mae-uiQyuTAHqPjlzcUyJ4NFaF-uoMnK5Uc,277
 ads/aqua/modeldeployment/constants.py,sha256=lJF77zwxmlECljDYjwFAMprAUR_zctZHmawiP-4alLg,296
-ads/aqua/modeldeployment/deployment.py,sha256=hWpjomp9UyRRsUDAa9VN5ezpqNOpLB7lOZJX8mYwDjI,56265
+ads/aqua/modeldeployment/deployment.py,sha256=V0ZJY63NXE2uqv_Lm5zdT6wSiqXR7QydZh-IL0ObGEo,55603
 ads/aqua/modeldeployment/entities.py,sha256=qwNH-8eHv-C2QPMITGQkb6haaJRvZ5c0i1H0Aoxeiu4,27100
-ads/aqua/modeldeployment/inference.py,sha256=rjTF-AM_rHLzL5HCxdLRTrsaSMdB-SzFYUp9dIy5ejw,2109
 ads/aqua/modeldeployment/utils.py,sha256=Aky4WZ5E564JVZ96X9RYJz_KlB_cAHGzV6mihtd3HV8,22009
 ads/aqua/resources/gpu_shapes_index.json,sha256=-6rSkyQ04T1z_Yfr3cxGPI7NAtgTwG7beIEjLYuMMIc,1948
 ads/aqua/server/__init__.py,sha256=fswoO0kX0hrp2b1owF4f-bv_OodntvvUY3FvhL6FCMk,179
@@ -98,7 +98,7 @@ ads/common/model_artifact_schema.json,sha256=aNwC9bW5HHMXJK-XAWV56RosqOqiCkzKHBi
 ads/common/model_export_util.py,sha256=rQjL_bb0ecGV2fj0jQXS2-aRbXl4ONDwuNVqdr3DiAQ,26574
 ads/common/model_metadata.py,sha256=Md1soURHww8GHMG3q_HV0RHVb6dPtg9FZ_7Wmd9L-Yc,641
 ads/common/object_storage_details.py,sha256=bvqIyB-zLpr5NMnZW8YtSupVH3RpWLBgbR3wPYlQhPU,9531
-ads/common/oci_client.py,sha256=SmME1qdCFtOdMtC7-703C732e4lEK71kNfTSqBRFkSM,6053
+ads/common/oci_client.py,sha256=HcGUWdc0GQLksJmn6GIrcBJvuhcanwhMlJhdt85_JIA,6164
 ads/common/oci_datascience.py,sha256=biBgm-udtSYRL46XYfBFJjpkPFcw2ew-xvp3rbbpwmI,1698
 ads/common/oci_logging.py,sha256=U0HRAUkpnycGpo2kWMrT3wjQVFZaWqLL6pZ2B6_epsM,41925
 ads/common/oci_mixin.py,sha256=mhja5UomrhXH43uB0jT-u2KaT37op9tM-snxvtGfc40,34548
@@ -728,7 +728,7 @@ ads/opctl/operator/lowcode/forecast/model/factory.py,sha256=5a9A3ql-bU412BiTB20o
 ads/opctl/operator/lowcode/forecast/model/forecast_datasets.py,sha256=2NsWE2WtD_O7uAXw42_3tmG3vb5lk3mdnzCZTph4hao,18903
 ads/opctl/operator/lowcode/forecast/model/ml_forecast.py,sha256=t5x6EBxOd7XwfT3FGdt-n9gscxaHMm3R2A4Evvxbj38,9646
 ads/opctl/operator/lowcode/forecast/model/neuralprophet.py,sha256=-AS3PPd8Fqn1uaMybJwTnFbmIfUxNPDlgYjGtjy9-E8,19944
-ads/opctl/operator/lowcode/forecast/model/prophet.py,sha256=jb8bshJf5lDdGJkNH-2SrwN4tdHImP7iD9I8KS4EmZU,17321
+ads/opctl/operator/lowcode/forecast/model/prophet.py,sha256=U0M3M7yzy7ea-mEG1c-dELL6t1Pf_LnnnbrZB08CG_Y,17402
 ads/opctl/operator/lowcode/forecast/whatifserve/__init__.py,sha256=JNDDjLrNorKXMHUuXMifqXea3eheST-lnrcwCl2bWrk,242
 ads/opctl/operator/lowcode/forecast/whatifserve/deployment_manager.py,sha256=w42anuqAScEQ0vBG3vW4LVLNq1bPdpAWGQEmNhMwZ08,12052
 ads/opctl/operator/lowcode/forecast/whatifserve/score.py,sha256=JjEDtrqUfL4x9t-vvafXMLNwY9-vgc6QPX_Ee-wmI58,8709
@@ -807,7 +807,7 @@ ads/secrets/oracledb.py,sha256=VAbsY206gc_Ru8FBOCKNGnIlX4VprIIQ9PmBoujRy_k,6146
 ads/secrets/secrets.py,sha256=k_f3hwh5RI7Hv7zEKDDuxh7kGB-F29bK15jxgu4sxVw,15197
 ads/telemetry/__init__.py,sha256=_pKx4hDK7DUXHjvAq4Zbeg6wl9_UkgGE7O77KWezu8g,230
 ads/telemetry/base.py,sha256=6kKVSfTxx2Dr0XL7R8OfjTy3hJoJpKMvP6D7RdVddl4,2199
-ads/telemetry/client.py,sha256=33xG082IGoWCbhD5xmYkDgI2znqL9LgqLM2kfvbg6fw,4975
+ads/telemetry/client.py,sha256=71hUqn-fF24zlb84ew5A0gugdkhBmzn30JEuNHDnMyM,5000
 ads/telemetry/telemetry.py,sha256=xCu64tUEzYzFaXEMxnMdtQJPme5-9sW8l3uxibN4E04,7930
 ads/templates/dataflow_pyspark.jinja2,sha256=JiTuaGt5LC7z1mrDoDIbZ5UskBETLpm0lMpn5VZYNGs,254
 ads/templates/dataflow_sparksql.jinja2,sha256=8MvGzj7PWgUNo-9dg6zil8WTpBL3eNcR815fz64v1yM,466
@@ -851,8 +851,8 @@ ads/type_discovery/unknown_detector.py,sha256=yZuYQReO7PUyoWZE7onhhtYaOg6088wf1y
 ads/type_discovery/zipcode_detector.py,sha256=3AlETg_ZF4FT0u914WXvTT3F3Z6Vf51WiIt34yQMRbw,1421
 ads/vault/__init__.py,sha256=x9tMdDAOdF5iDHk9u2di_K-ze5Nq068x25EWOBoWwqY,245
 ads/vault/vault.py,sha256=hFBkpYE-Hfmzu1L0sQwUfYcGxpWmgG18JPndRl0NOXI,8624
-oracle_ads-2.13.9rc1.dist-info/entry_points.txt,sha256=9VFnjpQCsMORA4rVkvN8eH6D3uHjtegb9T911t8cqV0,35
-oracle_ads-2.13.9rc1.dist-info/licenses/LICENSE.txt,sha256=zoGmbfD1IdRKx834U0IzfFFFo5KoFK71TND3K9xqYqo,1845
-oracle_ads-2.13.9rc1.dist-info/WHEEL,sha256=G2gURzTEtmeR8nrdXUJfNiB3VYVxigPQ-bEQujpNiNs,82
-oracle_ads-2.13.9rc1.dist-info/METADATA,sha256=322tybBGPwpm9eAuQEcS9dQ_7qwMye0YfNFu75RxQhs,16656
-oracle_ads-2.13.9rc1.dist-info/RECORD,,
+oracle_ads-2.13.10rc0.dist-info/entry_points.txt,sha256=9VFnjpQCsMORA4rVkvN8eH6D3uHjtegb9T911t8cqV0,35
+oracle_ads-2.13.10rc0.dist-info/licenses/LICENSE.txt,sha256=zoGmbfD1IdRKx834U0IzfFFFo5KoFK71TND3K9xqYqo,1845
+oracle_ads-2.13.10rc0.dist-info/WHEEL,sha256=G2gURzTEtmeR8nrdXUJfNiB3VYVxigPQ-bEQujpNiNs,82
+oracle_ads-2.13.10rc0.dist-info/METADATA,sha256=MZ4kGFRBnDsst32Mrvla6SR2ufMpOvR9p48bBBuQkKs,16782
+oracle_ads-2.13.10rc0.dist-info/RECORD,,

ads/aqua/modeldeployment/inference.py

@@ -1,74 +0,0 @@
-#!/usr/bin/env python
-
-# 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 json
-from dataclasses import dataclass, field
-
-import requests
-
-from ads.aqua.app import AquaApp
-from ads.aqua.modeldeployment.entities import ModelParams
-from ads.common.auth import default_signer
-from ads.telemetry import telemetry
-
-
-@dataclass
-class MDInferenceResponse(AquaApp):
-    """Contains APIs for Aqua Model deployments Inference.
-
-    Attributes
-    ----------
-
-    model_params: Dict
-    prompt: string
-
-    Methods
-    -------
-    get_model_deployment_response(self, **kwargs) -> "String"
-        Creates an instance of model deployment via Aqua
-    """
-
-    prompt: str = None
-    model_params: field(default_factory=ModelParams) = None
-
-    @telemetry(entry_point="plugin=inference&action=get_response", name="aqua")
-    def get_model_deployment_response(self, endpoint):
-        """
-        Returns MD inference response
-
-        Parameters
-        ----------
-        endpoint: str
-            MD predict url
-        prompt: str
-            User prompt.
-
-        model_params: (Dict, optional)
-            Model parameters to be associated with the message.
-            Currently supported VLLM+OpenAI parameters.
-
-            --model-params '{
-                "max_tokens":500,
-                "temperature": 0.5,
-                "top_k": 10,
-                "top_p": 0.5,
-                "model": "/opt/ds/model/deployed_model",
-                ...}'
-
-        Returns
-        -------
-        model_response_content
-        """
-
-        params_dict = self.model_params.to_dict()
-        params_dict = {
-            key: value for key, value in params_dict.items() if value is not None
-        }
-        body = {"prompt": self.prompt, **params_dict}
-        request_kwargs = {"json": body, "headers": {"Content-Type": "application/json"}}
-        response = requests.post(
-            endpoint, auth=default_signer()["signer"], **request_kwargs
-        )
-        return json.loads(response.content)