nemo-platform-plugin 0.1.0__py3-none-any.whl

nemo_platform/types/evaluation/tool_call_accuracy_metric.py

@@ -0,0 +1,70 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+from typing import Dict, List, Optional
+from datetime import datetime
+from typing_extensions import Literal
+
+from ..._models import BaseModel
+
+__all__ = ["ToolCallAccuracyMetric"]
+
+
+class ToolCallAccuracyMetric(BaseModel):
+    """RAGAS metric for measuring tool call accuracy."""
+
+    id: str
+
+    created_at: datetime
+
+    created_by: Optional[str] = None
+
+    entity_id: str
+    """Alias for id for backwards compatibility."""
+
+    parent: str
+    """Parent entity ID for nested entities."""
+
+    updated_at: datetime
+
+    updated_by: Optional[str] = None
+
+    workspace: str
+    """Workspace identifier"""
+
+    description: Optional[str] = None
+    """Human-readable description of the metric."""
+
+    input_template: Optional[Dict[str, object]] = None
+    """Optional Jinja template for rendering the input payload for RAGAS evaluation."""
+
+    labels: Optional[Dict[str, str]] = None
+    """Labels are key-value pairs that can be used for grouping and filtering."""
+
+    name: Optional[str] = None
+    """Entity name within the workspace"""
+
+    project: Optional[str] = None
+    """The name of the project associated with this entity."""
+
+    supported_job_types: Optional[List[Literal["online", "offline"]]] = None
+    """
+    A metric can evaluate model outputs for online evaluations or pre-generated
+    outputs for offline evaluations.
+    """
+
+    type: Optional[Literal["tool_call_accuracy"]] = None

nemo_platform/types/evaluation/tool_call_accuracy_metric_param.py

@@ -0,0 +1,44 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+from typing import Dict, List, Optional
+from typing_extensions import Literal
+
+from ..._models import BaseModel
+
+__all__ = ["ToolCallAccuracyMetricParam"]
+
+
+class ToolCallAccuracyMetricParam(BaseModel):
+    """Request type for ToolCallAccuracy metrics (no judge required)."""
+
+    description: Optional[str] = None
+    """Human-readable description of the metric."""
+
+    input_template: Optional[Dict[str, object]] = None
+    """Optional Jinja template for rendering the input payload for RAGAS evaluation."""
+
+    labels: Optional[Dict[str, str]] = None
+    """Labels are key-value pairs that can be used for grouping and filtering."""
+
+    supported_job_types: Optional[List[Literal["online", "offline"]]] = None
+    """
+    A metric can evaluate model outputs for online evaluations or pre-generated
+    outputs for offline evaluations.
+    """
+
+    type: Optional[Literal["tool_call_accuracy"]] = None

nemo_platform/types/evaluation/tool_call_accuracy_metric_param_param.py

@@ -0,0 +1,44 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+from __future__ import annotations
+
+from typing import Dict, List
+from typing_extensions import Literal, TypedDict
+
+__all__ = ["ToolCallAccuracyMetricParamParam"]
+
+
+class ToolCallAccuracyMetricParamParam(TypedDict, total=False):
+    """Request type for ToolCallAccuracy metrics (no judge required)."""
+
+    description: str
+    """Human-readable description of the metric."""
+
+    input_template: Dict[str, object]
+    """Optional Jinja template for rendering the input payload for RAGAS evaluation."""
+
+    labels: Dict[str, str]
+    """Labels are key-value pairs that can be used for grouping and filtering."""
+
+    supported_job_types: List[Literal["online", "offline"]]
+    """
+    A metric can evaluate model outputs for online evaluations or pre-generated
+    outputs for offline evaluations.
+    """
+
+    type: Literal["tool_call_accuracy"]

nemo_platform/types/evaluation/tool_call_accuracy_metric_response.py

@@ -0,0 +1,63 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+from typing import Dict, List, Optional
+from datetime import datetime
+from typing_extensions import Literal
+
+from ..._models import BaseModel
+
+__all__ = ["ToolCallAccuracyMetricResponse"]
+
+
+class ToolCallAccuracyMetricResponse(BaseModel):
+    """Response type for ToolCallAccuracy metrics (no judge required)."""
+
+    id: Optional[str] = None
+    """Entity name within the workspace"""
+
+    created_at: Optional[datetime] = None
+
+    description: Optional[str] = None
+    """Human-readable description of the metric."""
+
+    input_template: Optional[Dict[str, object]] = None
+    """Optional Jinja template for rendering the input payload for RAGAS evaluation."""
+
+    labels: Optional[Dict[str, str]] = None
+    """Labels are key-value pairs that can be used for grouping and filtering."""
+
+    name: Optional[str] = None
+    """Entity name within the workspace"""
+
+    parent: Optional[str] = None
+
+    project: Optional[str] = None
+    """The name of the project associated with this entity."""
+
+    supported_job_types: Optional[List[Literal["online", "offline"]]] = None
+    """
+    A metric can evaluate model outputs for online evaluations or pre-generated
+    outputs for offline evaluations.
+    """
+
+    type: Optional[Literal["tool_call_accuracy"]] = None
+
+    updated_at: Optional[datetime] = None
+
+    workspace: Optional[str] = None
+    """Workspace identifier"""

nemo_platform/types/evaluation/tool_calling_metric.py

@@ -0,0 +1,70 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+from typing import Dict, List, Optional
+from datetime import datetime
+from typing_extensions import Literal
+
+from ..._models import BaseModel
+
+__all__ = ["ToolCallingMetric"]
+
+
+class ToolCallingMetric(BaseModel):
+    """Persisted Tool Calling metric."""
+
+    id: str
+
+    created_at: datetime
+
+    created_by: Optional[str] = None
+
+    entity_id: str
+    """Alias for id for backwards compatibility."""
+
+    parent: str
+    """Parent entity ID for nested entities."""
+
+    reference: str
+    """The template for the ground truth reference to evaluate tool calling accuracy."""
+
+    updated_at: datetime
+
+    updated_by: Optional[str] = None
+
+    workspace: str
+    """Workspace identifier"""
+
+    description: Optional[str] = None
+    """Human-readable description of the metric."""
+
+    labels: Optional[Dict[str, str]] = None
+    """Labels are key-value pairs that can be used for grouping and filtering."""
+
+    name: Optional[str] = None
+    """Entity name within the workspace"""
+
+    project: Optional[str] = None
+    """The name of the project associated with this entity."""
+
+    supported_job_types: Optional[List[Literal["online", "offline"]]] = None
+    """
+    A metric can evaluate model outputs for online evaluations or pre-generated
+    outputs for offline evaluations.
+    """
+
+    type: Optional[Literal["tool-calling"]] = None

nemo_platform/types/evaluation/tool_calling_metric_param.py

@@ -0,0 +1,47 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+from typing import Dict, List, Optional
+from typing_extensions import Literal
+
+from ..._models import BaseModel
+
+__all__ = ["ToolCallingMetricParam"]
+
+
+class ToolCallingMetricParam(BaseModel):
+    """Request type for ToolCallingMetric.
+
+    Tool-calling accuracy metric for structured function calls.
+    """
+
+    reference: str
+    """The template for the ground truth reference to evaluate tool calling accuracy."""
+
+    description: Optional[str] = None
+    """Human-readable description of the metric."""
+
+    labels: Optional[Dict[str, str]] = None
+    """Labels are key-value pairs that can be used for grouping and filtering."""
+
+    supported_job_types: Optional[List[Literal["online", "offline"]]] = None
+    """
+    A metric can evaluate model outputs for online evaluations or pre-generated
+    outputs for offline evaluations.
+    """
+
+    type: Optional[Literal["tool-calling"]] = None

nemo_platform/types/evaluation/tool_calling_metric_param_param.py

@@ -0,0 +1,47 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+from __future__ import annotations
+
+from typing import Dict, List
+from typing_extensions import Literal, Required, TypedDict
+
+__all__ = ["ToolCallingMetricParamParam"]
+
+
+class ToolCallingMetricParamParam(TypedDict, total=False):
+    """Request type for ToolCallingMetric.
+
+    Tool-calling accuracy metric for structured function calls.
+    """
+
+    reference: Required[str]
+    """The template for the ground truth reference to evaluate tool calling accuracy."""
+
+    description: str
+    """Human-readable description of the metric."""
+
+    labels: Dict[str, str]
+    """Labels are key-value pairs that can be used for grouping and filtering."""
+
+    supported_job_types: List[Literal["online", "offline"]]
+    """
+    A metric can evaluate model outputs for online evaluations or pre-generated
+    outputs for offline evaluations.
+    """
+
+    type: Literal["tool-calling"]

nemo_platform/types/evaluation/tool_calling_metric_response.py

@@ -0,0 +1,63 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+from typing import Dict, List, Optional
+from datetime import datetime
+from typing_extensions import Literal
+
+from ..._models import BaseModel
+
+__all__ = ["ToolCallingMetricResponse"]
+
+
+class ToolCallingMetricResponse(BaseModel):
+    """Response type for ToolCallingMetric."""
+
+    reference: str
+    """The template for the ground truth reference to evaluate tool calling accuracy."""
+
+    id: Optional[str] = None
+    """Entity name within the workspace"""
+
+    created_at: Optional[datetime] = None
+
+    description: Optional[str] = None
+    """Human-readable description of the metric."""
+
+    labels: Optional[Dict[str, str]] = None
+    """Labels are key-value pairs that can be used for grouping and filtering."""
+
+    name: Optional[str] = None
+    """Entity name within the workspace"""
+
+    parent: Optional[str] = None
+
+    project: Optional[str] = None
+    """The name of the project associated with this entity."""
+
+    supported_job_types: Optional[List[Literal["online", "offline"]]] = None
+    """
+    A metric can evaluate model outputs for online evaluations or pre-generated
+    outputs for offline evaluations.
+    """
+
+    type: Optional[Literal["tool-calling"]] = None
+
+    updated_at: Optional[datetime] = None
+
+    workspace: Optional[str] = None
+    """Workspace identifier"""

nemo_platform/types/evaluation/topic_adherence_metric.py

@@ -0,0 +1,92 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+from typing import Dict, List, Optional
+from datetime import datetime
+from typing_extensions import Literal
+
+from .model import Model
+from ..._models import BaseModel
+from .inference_params import InferenceParams
+
+__all__ = ["TopicAdherenceMetric"]
+
+
+class TopicAdherenceMetric(BaseModel):
+    """RAGAS metric for measuring topic adherence."""
+
+    id: str
+
+    created_at: datetime
+
+    created_by: Optional[str] = None
+
+    entity_id: str
+    """Alias for id for backwards compatibility."""
+
+    judge_model: Model
+    """Model definition for use without persisting to the Models API."""
+
+    parent: str
+    """Parent entity ID for nested entities."""
+
+    updated_at: datetime
+
+    updated_by: Optional[str] = None
+
+    workspace: str
+    """Workspace identifier"""
+
+    description: Optional[str] = None
+    """Human-readable description of the metric."""
+
+    ignore_request_failure: Optional[bool] = None
+    """
+    If True, request failures to the judge model are ignored and the metric result
+    is marked as NaN. Parse/output formatting failures are always converted to NaN.
+    """
+
+    inference: Optional[InferenceParams] = None
+    """Parameters for model inference.
+
+    Extra fields can be supplied for additional options applied to the inference
+    request directly. Fields not supported by the model may cause inference errors
+    during evaluation.
+    """
+
+    input_template: Optional[Dict[str, object]] = None
+    """Optional Jinja template for rendering the input payload for RAGAS evaluation."""
+
+    labels: Optional[Dict[str, str]] = None
+    """Labels are key-value pairs that can be used for grouping and filtering."""
+
+    metric_mode: Optional[Literal["f1", "precision", "recall"]] = None
+    """The mode for computing topic adherence score."""
+
+    name: Optional[str] = None
+    """Entity name within the workspace"""
+
+    project: Optional[str] = None
+    """The name of the project associated with this entity."""
+
+    supported_job_types: Optional[List[Literal["online", "offline"]]] = None
+    """
+    A metric can evaluate model outputs for online evaluations or pre-generated
+    outputs for offline evaluations.
+    """
+
+    type: Optional[Literal["topic_adherence"]] = None

nemo_platform/types/evaluation/topic_adherence_metric_param.py

@@ -0,0 +1,69 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+from typing import Dict, List, Union, Optional
+from typing_extensions import Literal, TypeAlias
+
+from .model import Model
+from ..._models import BaseModel
+from .model_ref import ModelRef
+from .inference_params import InferenceParams
+
+__all__ = ["TopicAdherenceMetricParam", "JudgeModel"]
+
+JudgeModel: TypeAlias = Union[Model, ModelRef]
+
+
+class TopicAdherenceMetricParam(BaseModel):
+    """Request type for TopicAdherence metrics."""
+
+    judge_model: JudgeModel
+    """The judge model configuration."""
+
+    description: Optional[str] = None
+    """Human-readable description of the metric."""
+
+    ignore_request_failure: Optional[bool] = None
+    """
+    If True, request failures to the judge model are ignored and the metric result
+    is marked as NaN. Parse/output formatting failures are always converted to NaN.
+    """
+
+    inference: Optional[InferenceParams] = None
+    """Parameters for model inference.
+
+    Extra fields can be supplied for additional options applied to the inference
+    request directly. Fields not supported by the model may cause inference errors
+    during evaluation.
+    """
+
+    input_template: Optional[Dict[str, object]] = None
+    """Optional Jinja template for rendering the input payload for RAGAS evaluation."""
+
+    labels: Optional[Dict[str, str]] = None
+    """Labels are key-value pairs that can be used for grouping and filtering."""
+
+    metric_mode: Optional[Literal["f1", "precision", "recall"]] = None
+    """The mode for computing topic adherence score."""
+
+    supported_job_types: Optional[List[Literal["online", "offline"]]] = None
+    """
+    A metric can evaluate model outputs for online evaluations or pre-generated
+    outputs for offline evaluations.
+    """
+
+    type: Optional[Literal["topic_adherence"]] = None

nemo_platform/types/evaluation/topic_adherence_metric_param_param.py

@@ -0,0 +1,70 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
+
+from __future__ import annotations
+
+from typing import Dict, List, Union
+from typing_extensions import Literal, Required, TypeAlias, TypedDict
+
+from .model_ref import ModelRef
+from .model_param import ModelParam
+from .inference_params_param import InferenceParamsParam
+
+__all__ = ["TopicAdherenceMetricParamParam", "JudgeModel"]
+
+JudgeModel: TypeAlias = Union[ModelParam, ModelRef]
+
+
+class TopicAdherenceMetricParamParam(TypedDict, total=False):
+    """Request type for TopicAdherence metrics."""
+
+    judge_model: Required[JudgeModel]
+    """The judge model configuration."""
+
+    description: str
+    """Human-readable description of the metric."""
+
+    ignore_request_failure: bool
+    """
+    If True, request failures to the judge model are ignored and the metric result
+    is marked as NaN. Parse/output formatting failures are always converted to NaN.
+    """
+
+    inference: InferenceParamsParam
+    """Parameters for model inference.
+
+    Extra fields can be supplied for additional options applied to the inference
+    request directly. Fields not supported by the model may cause inference errors
+    during evaluation.
+    """
+
+    input_template: Dict[str, object]
+    """Optional Jinja template for rendering the input payload for RAGAS evaluation."""
+
+    labels: Dict[str, str]
+    """Labels are key-value pairs that can be used for grouping and filtering."""
+
+    metric_mode: Literal["f1", "precision", "recall"]
+    """The mode for computing topic adherence score."""
+
+    supported_job_types: List[Literal["online", "offline"]]
+    """
+    A metric can evaluate model outputs for online evaluations or pre-generated
+    outputs for offline evaluations.
+    """
+
+    type: Literal["topic_adherence"]