add datasetio to distribution

llama_stack/providers/remote/datasetio/nvidia/config.py

@@ -24,11 +24,6 @@ class NvidiaDatasetIOConfig(BaseModel):
         description="The NVIDIA dataset namespace.",
     )
 
-    access_policies: Optional[dict] = Field(
-        default_factory=lambda: os.getenv("NVIDIA_ACCESS_POLICIES", {}),
-        description="The NVIDIA access policies.",
-    )
-
     project_id: Optional[str] = Field(
         default_factory=lambda: os.getenv("NVIDIA_PROJECT_ID", "test-project"),
         description="The NVIDIA project ID.",
@@ -46,8 +41,6 @@ class NvidiaDatasetIOConfig(BaseModel):
             default_values.append("project_id='test-project'")
         if os.getenv("NVIDIA_DATASET_NAMESPACE") is None:
             default_values.append("dataset_namespace='default'")
-        if os.getenv("NVIDIA_ACCESS_POLICIES") is None:
-            default_values.append("access_policies='{}'")
         if os.getenv("NVIDIA_DATASETS_URL") is None:
             default_values.append("datasets_url='http://nemo.test'")
 
@@ -64,8 +57,5 @@ class NvidiaDatasetIOConfig(BaseModel):
             "api_key": "${env.NVIDIA_API_KEY:}",
             "user_id": "${env.NVIDIA_USER_ID:llama-stack-user}",
             "dataset_namespace": "${env.NVIDIA_DATASET_NAMESPACE:default}",
-            "access_policies": "${env.NVIDIA_ACCESS_POLICIES:}",
             "project_id": "${env.NVIDIA_PROJECT_ID:test-project}",
-            "customizer_url": "${env.NVIDIA_CUSTOMIZER_URL:}",
-            "output_model_dir": "${env.NVIDIA_OUTPUT_MODEL_DIR:test-example-model@v1}",
         }

llama_stack/providers/remote/datasetio/nvidia/datasetio.py

@@ -4,14 +4,13 @@
 # This source code is licensed under the terms described in the LICENSE file in
 # the root directory of this source tree.
 
-from typing import Any, Dict, Literal, Optional
+from typing import Any, Dict, List, Optional
 
 import aiohttp
 
 from llama_stack.apis.common.content_types import URL
 from llama_stack.apis.common.type_system import ParamType
-from llama_stack.apis.datasets.datasets import Dataset, ListDatasetsResponse
-from llama_stack.apis.resource import ResourceType
+from llama_stack.apis.datasetio import IterrowsResponse
 from llama_stack.schema_utils import webmethod
 
 from .config import NvidiaDatasetIOConfig
@@ -20,8 +19,6 @@ from .config import NvidiaDatasetIOConfig
 class NvidiaDatasetIOAdapter:
     """Nvidia NeMo DatasetIO API."""
 
-    type: Literal[ResourceType.dataset.value] = ResourceType.dataset.value
-
     def __init__(self, config: NvidiaDatasetIOConfig):
         self.config = config
         self.headers = {}
@@ -80,19 +77,6 @@ class NvidiaDatasetIOAdapter:
         """
         ...
 
-    @webmethod(route="/datasets/{dataset_id:namespace}", method="GET")
-    async def get_dataset(
-        self,
-        dataset_id: str,
-    ) -> Optional[Dataset]:
-        raise NotImplementedError("Not implemented")
-
-    @webmethod(route="/datasets", method="GET")
-    async def list_datasets(
-        self,
-    ) -> ListDatasetsResponse:
-        raise NotImplementedError("Not implemented")
-
     @webmethod(route="/datasets/{dataset_id:path}", method="POST")
     async def update_dataset(
         self,
@@ -112,3 +96,14 @@ class NvidiaDatasetIOAdapter:
         namespace: Optional[str] = "default",
     ) -> None:
         raise NotImplementedError("Not implemented")
+
+    async def iterrows(
+        self,
+        dataset_id: str,
+        start_index: Optional[int] = None,
+        limit: Optional[int] = None,
+    ) -> IterrowsResponse:
+        raise NotImplementedError("Not implemented")
+
+    async def append_rows(self, dataset_id: str, rows: List[Dict[str, Any]]) -> None:
+        raise NotImplementedError("Not implemented")