backmerged

Author: KshitizLohia

ads/common/auth.py

@@ -416,7 +416,7 @@ def create_signer(
     >>> auth = ads.auth.create_signer(auth_type="security_token", config=config) # security token authentication created based on provided config
     """
     if signer or signer_callable:
-        configuration = ads.telemetry.update_oci_client_config()
+        configuration = ads.telemetry.update_oci_client_config(config)
         if signer_callable:
             signer = signer_callable(**signer_kwargs)
         signer_dict = {
@@ -479,7 +479,7 @@ def default_signer(client_kwargs: Optional[Dict] = None) -> Dict:
     """
     auth_state = AuthState()
     if auth_state.oci_signer or auth_state.oci_signer_callable:
-        configuration = ads.telemetry.update_oci_client_config()
+        configuration = ads.telemetry.update_oci_client_config(auth_state.oci_config)
         signer = auth_state.oci_signer
         if auth_state.oci_signer_callable:
             signer_kwargs = auth_state.oci_signer_kwargs or {}

ads/model/artifact_uploader.py

@@ -128,8 +128,11 @@ class LargeArtifactUploader(ArtifactUploader):
         The OCI Object Storage URI where model artifacts will be copied to.
         The `bucket_uri` is only necessary for uploading large artifacts which
         size is greater than 2GB. Example: `oci://<bucket_name>@<namespace>/prefix/`.
+
         .. versionadded:: 2.8.10
-        If artifact_path is object storage path to a zip archive, bucket_uri will be ignored.
+
+            If artifact_path is object storage path to a zip archive, bucket_uri will be ignored.
+
     dsc_model: OCIDataScienceModel
         The data scince model instance.
     overwrite_existing_artifact: bool
@@ -173,8 +176,11 @@ def __init__(
             The OCI Object Storage URI where model artifacts will be copied to.
             The `bucket_uri` is only necessary for uploading large artifacts from local which
             size is greater than 2GB. Example: `oci://<bucket_name>@<namespace>/prefix/`.
+
             .. versionadded:: 2.8.10
-            If `artifact_path` is object storage path to a zip archive, `bucket_uri` will be ignored.
+
+                If `artifact_path` is object storage path to a zip archive, `bucket_uri` will be ignored.
+
         auth: (Dict, optional). Defaults to `None`.
             The default authetication is set using `ads.set_auth` API.
             If you need to override the default, use the `ads.common.auth.api_keys` or

ads/model/datascience_model.py

@@ -549,8 +549,11 @@ def create(self, **kwargs) -> "DataScienceModel":
                 The OCI Object Storage URI where model artifacts will be copied to.
                 The `bucket_uri` is only necessary for uploading large artifacts which
                 size is greater than 2GB. Example: `oci://<bucket_name>@<namespace>/prefix/`.
+
                 .. versionadded:: 2.8.10
-                If `artifact` is provided as an object storage path to a zip archive, `bucket_uri` will be ignored.
+
+                    If `artifact` is provided as an object storage path to a zip archive, `bucket_uri` will be ignored.
+
             overwrite_existing_artifact: (bool, optional). Defaults to `True`.
                 Overwrite target bucket artifact if exists.
             remove_existing_artifact: (bool, optional). Defaults to `True`.
@@ -639,8 +642,11 @@ def upload_artifact(
             The OCI Object Storage URI where model artifacts will be copied to.
             The `bucket_uri` is only necessary for uploading large artifacts which
             size is greater than 2GB. Example: `oci://<bucket_name>@<namespace>/prefix/`.
+
             .. versionadded:: 2.8.10
-            If `artifact` is provided as an object storage path to a zip archive, `bucket_uri` will be ignored.
+
+                If `artifact` is provided as an object storage path to a zip archive, `bucket_uri` will be ignored.
+
         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/opctl/distributed/cli.py

@@ -30,7 +30,14 @@ def commands():
     "-f",
     help="Distributed training framework type",
     type=click.Choice(
-        ["dask", "horovod-tensorflow", "horovod-pytorch", "pytorch", "tensorflow"]
+        [
+            "dask",
+            "horovod-tensorflow",
+            "horovod-pytorch",
+            "pytorch",
+            "tensorflow",
+            "ray",
+        ]
     ),
     default=None,
     required=True,

docs/source/release_notes.rst

@@ -2,6 +2,15 @@
 Release Notes
 =============
 
+2.8.11
+------
+Release date: October 18, 2023
+
+* Added support to mount file systems in Data Science notebook sessions and jobs.
+* Added support to cancel all job runs in the ADS ``api`` and ``opctl`` commands.
+* Updated ``ads.set_auth()`` to use both ``config`` and ``signer`` when provided.
+* Fixed a bug when initializing distributed training artifacts with "Ray" framework.
+
 2.8.10
 ------
 Release date: September 27, 2023

docs/source/user_guide/jobs/infra_and_runtime.rst

@@ -152,6 +152,61 @@ see also `ADS Logging <../logging/logging.html>`_.
 
 With logging configured, you can call :py:meth:`~ads.jobs.DataScienceJobRun.watch` method to stream the logs.
 
+
+Mounting File Systems
+---------------------
+
+Data Science Job supports mounting multiple types of file systems,
+see `Data Science Job Mounting File Systems <https://docs.oracle.com/en-us/iaas/data-science/using/jobs-create.htm>`_. A maximum number of 5 file systems are
+allowed to be mounted for each Data Science Job. You can specify a list of file systems to be mounted
+by calling :py:meth:`~ads.jobs.DataScienceJob.with_storage_mount()`. For each file system to be mounted,
+you need to pass a dictionary with ``src`` and ``dest`` as keys. For example, you can pass
+``<mount_target_ip_address>@<export_path>`` as the value for ``src`` to mount OCI File Storage and you can also
+pass ``oci://<bucket_name>@<namespace>/<prefix>`` to mount OCI Object Storage. The value of
+``dest`` indicates the path and directory to which you want to mount the file system and must be in the
+format as ``<destination_path>/<destination_directory_name>``. The ``<destination_directory_name>`` is required
+while the ``<destination_path>`` is optional. The ``<destination_path>`` must start with character ``/`` if provided.
+If not, the file systems will be mounted to ``/mnt/<destination_directory_name>`` by default.
+
+
+.. tabs::
+
+  .. code-tab:: python
+    :caption: Python
+
+    from ads.jobs import DataScienceJob
+
+    infrastructure = (
+        DataScienceJob()
+        .with_log_group_id("<log_group_ocid>")
+        .with_log_id("<log_ocid>")
+        .with_storage_mount(
+          {
+            "src" : "<mount_target_ip_address>@<export_path>",
+            "dest" : "<destination_path>/<destination_directory_name>"
+          }, # mount oci file storage to path "<destination_path>/<destination_directory_name>"
+          {
+            "src" : "oci://<bucket_name>@<namespace>/<prefix>",
+            "dest" : "<destination_directory_name>"
+          } # mount oci object storage to path "/mnt/<destination_directory_name>"
+        )
+    )
+
+  .. code-tab:: yaml
+    :caption: YAML
+
+    kind: infrastructure
+    type: dataScienceJob
+    spec:
+      logGroupId: <log_group_ocid>
+      logId: <log_ocid>
+      storageMount:
+      - src: <mount_target_ip_address>@<export_path>
+        dest: <destination_path>/<destination_directory_name>
+      - src: oci://<bucket_name>@<namespace>/<prefix>
+        dest: <destination_directory_name>
+
+
 Runtime
 =======
 

docs/source/user_guide/jobs/tabs/infra_config.rst

@@ -22,6 +22,17 @@
         .with_shape_config_details(memory_in_gbs=16, ocpus=1)
         # Minimum/Default block storage size is 50 (GB).
         .with_block_storage_size(50)
+        # A maximum number of 5 file systems are allowed to be mounted for a job.
+        .with_storage_mount(
+          {
+            "src" : "<mount_target_ip_address>@<export_path>",
+            "dest" : "<destination_path>/<destination_directory_name>"
+          }, # mount oci file storage to path "<destination_path>/<destination_directory_name>"
+          {
+            "src" : "oci://<bucket_name>@<namespace>/<prefix>",
+            "dest" : "<destination_directory_name>"
+          } # mount oci object storage to path "/mnt/<destination_directory_name>"
+        )
     )
 
   .. code-tab:: yaml
@@ -40,3 +51,8 @@
         ocpus: 1
       shapeName: VM.Standard.E3.Flex
       subnetId: <subnet_ocid>
+      storageMount:
+      - src: <mount_target_ip_address>@<export_path>
+        dest: <destination_path>/<destination_directory_name>
+      - src: oci://<bucket_name>@<namespace>/<prefix>
+        dest: <destination_directory_name>

docs/source/user_guide/jobs/tabs/llama2_full.rst

@@ -14,42 +14,32 @@
             .with_compartment_id("<compartment_ocid>")
             .with_project_id("<project_ocid>")
             .with_subnet_id("<subnet_ocid>")
-            .with_shape_name("VM.GPU.A10.1")
+            .with_shape_name("VM.GPU.A10.2")
             .with_block_storage_size(256)
         )
         .with_runtime(
             PyTorchDistributedRuntime()
             # Specify the service conda environment by slug name.
-            .with_service_conda("pytorch20_p39_gpu_v1")
+            .with_service_conda("pytorch20_p39_gpu_v2")
             .with_git(
               url="https://github.com/facebookresearch/llama-recipes.git",
-              commit="03faba661f079ee1ecaeb66deaa6bdec920a7bab"
+              commit="1aecd00924738239f8d86f342b36bacad180d2b3"
             )
             .with_dependency(
               pip_pkg=" ".join([
-                "'accelerate>=0.21.0'",
-                "appdirs",
-                "loralib",
-                "bitsandbytes==0.39.1",
-                "black",
-                "'black[jupyter]'",
-                "datasets",
-                "fire",
-                "'git+https://github.com/huggingface/peft.git'",
-                "'transformers>=4.31.0'",
-                "sentencepiece",
-                "py7zr",
-                "scipy",
-                "optimum"
+                "--extra-index-url https://download.pytorch.org/whl/cu118 torch==2.1.0",
+                "git+https://github.com/huggingface/peft.git@15a013af5ff5660b9377af24d3eee358213d72d4"
+                "appdirs==1.4.4",
+                "llama-recipes==0.0.1",
+                "py7zr==0.20.6",
               ])
             )
             .with_output("/home/datascience/outputs", "oci://bucket@namespace/outputs/$JOB_RUN_OCID")
             .with_command(" ".join([
-              "torchrun llama_finetuning.py",
+              "torchrun examples/finetuning.py",
               "--enable_fsdp",
               "--pure_bf16",
               "--batch_size_training 1",
-              "--micro_batch_size 1",
               "--model_name $MODEL_NAME",
               "--dist_checkpoint_root_folder /home/datascience/outputs",
               "--dist_checkpoint_folder fine-tuned"
@@ -87,36 +77,26 @@
         spec:
           git:
             url: https://github.com/facebookresearch/llama-recipes.git
-            commit: 03faba661f079ee1ecaeb66deaa6bdec920a7bab
+            commit: 1aecd00924738239f8d86f342b36bacad180d2b3
           command: >-
             torchrun llama_finetuning.py
             --enable_fsdp
             --pure_bf16
             --batch_size_training 1
-            --micro_batch_size 1
             --model_name $MODEL_NAME
             --dist_checkpoint_root_folder /home/datascience/outputs
             --dist_checkpoint_folder fine-tuned
           replicas: 2
           conda:
             type: service
-            slug: pytorch20_p39_gpu_v1
+            slug: pytorch20_p39_gpu_v2
           dependencies:
             pipPackages: >-
-              'accelerate>=0.21.0'
-              appdirs
-              loralib
-              bitsandbytes==0.39.1
-              black
-              'black[jupyter]'
-              datasets
-              fire
-              'git+https://github.com/huggingface/peft.git'
-              'transformers>=4.31.0'
-              sentencepiece
-              py7zr
-              scipy
-              optimum
+              --extra-index-url https://download.pytorch.org/whl/cu118 torch==2.1.0
+              git+https://github.com/huggingface/peft.git@15a013af5ff5660b9377af24d3eee358213d72d4
+              llama-recipes==0.0.1
+              appdirs==1.4.4
+              py7zr==0.20.6
           outputDir: /home/datascience/outputs
           outputUri: oci://bucket@namespace/outputs/$JOB_RUN_OCID
           env:

docs/source/user_guide/jobs/tabs/quick_start_job.rst

@@ -24,6 +24,17 @@
             .with_shape_config_details(memory_in_gbs=16, ocpus=1)
             # Minimum/Default block storage size is 50 (GB).
             .with_block_storage_size(50)
+            # A maximum number of 5 file systems are allowed to be mounted for a job.
+            .with_storage_mount(
+              {
+                "src" : "<mount_target_ip_address>@<export_path>",
+                "dest" : "<destination_path>/<destination_directory_name>"
+              }, # mount oci file storage to path "<destination_path>/<destination_directory_name>"
+              {
+                "src" : "oci://<bucket_name>@<namespace>/<prefix>",
+                "dest" : "<destination_directory_name>"
+              } # mount oci object storage to path "/mnt/<destination_directory_name>"
+            )
         )
         .with_runtime(
             PythonRuntime()
@@ -59,6 +70,11 @@
             ocpus: 1
           shapeName: VM.Standard.E3.Flex
           subnetId: <subnet_ocid>
+          storageMount:
+          - src: <mount_target_ip_address>@<export_path>
+            dest: <destination_path>/<destination_directory_name>
+          - src: oci://<bucket_name>@<namespace>/<prefix>
+            dest: <destination_directory_name>
       runtime:
         kind: runtime
         type: python

docs/source/user_guide/model_registration/_template/deploy.rst

@@ -7,12 +7,11 @@ You can use the ``.deploy()`` method to deploy a model. You must first save the
 
 The ``.deploy()`` method returns a ``ModelDeployment`` object.  Specify deployment attributes such as display name, instance type, number of instances,  maximum router bandwidth, and logging groups.  The API takes the following parameters:
 
-See `API documentation <../../ads.model.html#id1>`__ for more details about the parameters.
-
+See :py:meth:`~ads.model.GenericModel.deploy`  for more details about the parameters.
 
 .. admonition:: Tips
    :class: note
 
    * Providing ``deployment_access_log_id`` and ``deployment_predict_log_id`` helps in debugging your model inference setup.
-   * Default Load Balancer configuration has bandwidth of 10 Mbps. `Refer service document to help you choose the right setup. <https://docs.oracle.com/en-us/iaas/data-science/using/model_dep_create.htm>`_ 
-   * Check for supported instance shapes `here <https://docs.oracle.com/en-us/iaas/data-science/using/overview.htm#supported-shapes>`_ .
+   * Default Load Balancer configuration has bandwidth of 10 Mbps. `Refer service document to help you choose the right setup. <https://docs.oracle.com/en-us/iaas/data-science/using/model_dep_create.htm>`_
+   * Check for supported instance shapes `here <https://docs.oracle.com/en-us/iaas/data-science/using/supported-shapes.htm>`_ .