deeprails 1.9.0__py3-none-any.whl → 1.10.0__py3-none-any.whl

deeprails/_client.py

@@ -21,7 +21,7 @@ from ._types import (
 )
 from ._utils import is_given, get_async_library
 from ._version import __version__
-from .resources import defend, monitor, evaluate
+from .resources import defend, monitor
 from ._streaming import Stream as Stream, AsyncStream as AsyncStream
 from ._exceptions import APIStatusError, DeeprailsError
 from ._base_client import (
@@ -45,7 +45,6 @@ __all__ = [
 class Deeprails(SyncAPIClient):
     defend: defend.DefendResource
     monitor: monitor.MonitorResource
-    evaluate: evaluate.EvaluateResource
     with_raw_response: DeeprailsWithRawResponse
     with_streaming_response: DeeprailsWithStreamedResponse
 
@@ -105,7 +104,6 @@ class Deeprails(SyncAPIClient):
 
         self.defend = defend.DefendResource(self)
         self.monitor = monitor.MonitorResource(self)
-        self.evaluate = evaluate.EvaluateResource(self)
         self.with_raw_response = DeeprailsWithRawResponse(self)
         self.with_streaming_response = DeeprailsWithStreamedResponse(self)
 
@@ -217,7 +215,6 @@ class Deeprails(SyncAPIClient):
 class AsyncDeeprails(AsyncAPIClient):
     defend: defend.AsyncDefendResource
     monitor: monitor.AsyncMonitorResource
-    evaluate: evaluate.AsyncEvaluateResource
     with_raw_response: AsyncDeeprailsWithRawResponse
     with_streaming_response: AsyncDeeprailsWithStreamedResponse
 
@@ -277,7 +274,6 @@ class AsyncDeeprails(AsyncAPIClient):
 
         self.defend = defend.AsyncDefendResource(self)
         self.monitor = monitor.AsyncMonitorResource(self)
-        self.evaluate = evaluate.AsyncEvaluateResource(self)
         self.with_raw_response = AsyncDeeprailsWithRawResponse(self)
         self.with_streaming_response = AsyncDeeprailsWithStreamedResponse(self)
 
@@ -390,28 +386,24 @@ class DeeprailsWithRawResponse:
     def __init__(self, client: Deeprails) -> None:
         self.defend = defend.DefendResourceWithRawResponse(client.defend)
         self.monitor = monitor.MonitorResourceWithRawResponse(client.monitor)
-        self.evaluate = evaluate.EvaluateResourceWithRawResponse(client.evaluate)
 
 
 class AsyncDeeprailsWithRawResponse:
     def __init__(self, client: AsyncDeeprails) -> None:
         self.defend = defend.AsyncDefendResourceWithRawResponse(client.defend)
         self.monitor = monitor.AsyncMonitorResourceWithRawResponse(client.monitor)
-        self.evaluate = evaluate.AsyncEvaluateResourceWithRawResponse(client.evaluate)
 
 
 class DeeprailsWithStreamedResponse:
     def __init__(self, client: Deeprails) -> None:
         self.defend = defend.DefendResourceWithStreamingResponse(client.defend)
         self.monitor = monitor.MonitorResourceWithStreamingResponse(client.monitor)
-        self.evaluate = evaluate.EvaluateResourceWithStreamingResponse(client.evaluate)
 
 
 class AsyncDeeprailsWithStreamedResponse:
     def __init__(self, client: AsyncDeeprails) -> None:
         self.defend = defend.AsyncDefendResourceWithStreamingResponse(client.defend)
         self.monitor = monitor.AsyncMonitorResourceWithStreamingResponse(client.monitor)
-        self.evaluate = evaluate.AsyncEvaluateResourceWithStreamingResponse(client.evaluate)
 
 
 Client = Deeprails

deeprails/_version.py

@@ -1,4 +1,4 @@
 # File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
 
 __title__ = "deeprails"
-__version__ = "1.9.0"  # x-release-please-version
+__version__ = "1.10.0"  # x-release-please-version

deeprails/resources/__init__.py

@@ -16,14 +16,6 @@ from .monitor import (
     MonitorResourceWithStreamingResponse,
     AsyncMonitorResourceWithStreamingResponse,
 )
-from .evaluate import (
-    EvaluateResource,
-    AsyncEvaluateResource,
-    EvaluateResourceWithRawResponse,
-    AsyncEvaluateResourceWithRawResponse,
-    EvaluateResourceWithStreamingResponse,
-    AsyncEvaluateResourceWithStreamingResponse,
-)
 
 __all__ = [
     "DefendResource",
@@ -38,10 +30,4 @@ __all__ = [
     "AsyncMonitorResourceWithRawResponse",
     "MonitorResourceWithStreamingResponse",
     "AsyncMonitorResourceWithStreamingResponse",
-    "EvaluateResource",
-    "AsyncEvaluateResource",
-    "EvaluateResourceWithRawResponse",
-    "AsyncEvaluateResourceWithRawResponse",
-    "EvaluateResourceWithStreamingResponse",
-    "AsyncEvaluateResourceWithStreamingResponse",
 ]

deeprails/types/__init__.py

@@ -2,12 +2,10 @@
 
 from __future__ import annotations
 
-from .evaluation import Evaluation as Evaluation
 from .defend_response import DefendResponse as DefendResponse
 from .monitor_response import MonitorResponse as MonitorResponse
 from .monitor_create_params import MonitorCreateParams as MonitorCreateParams
 from .monitor_update_params import MonitorUpdateParams as MonitorUpdateParams
-from .evaluate_create_params import EvaluateCreateParams as EvaluateCreateParams
 from .monitor_event_response import MonitorEventResponse as MonitorEventResponse
 from .monitor_detail_response import MonitorDetailResponse as MonitorDetailResponse
 from .monitor_retrieve_params import MonitorRetrieveParams as MonitorRetrieveParams

deeprails/types/monitor_detail_response.py

@@ -1,13 +1,108 @@
 # File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
 
-from typing import List, Optional
+from typing import Dict, List, Optional
 from datetime import datetime
 from typing_extensions import Literal
 
+from pydantic import Field as FieldInfo
+
 from .._models import BaseModel
-from .evaluation import Evaluation
 
-__all__ = ["MonitorDetailResponse", "Stats"]
+__all__ = ["MonitorDetailResponse", "Evaluation", "EvaluationModelInput", "Stats"]
+
+
+class EvaluationModelInput(BaseModel):
+    ground_truth: Optional[str] = None
+    """The ground truth for evaluating Ground Truth Adherence guardrail."""
+
+    system_prompt: Optional[str] = None
+    """The system prompt used to generate the output."""
+
+    user_prompt: Optional[str] = None
+    """The user prompt used to generate the output."""
+
+
+class Evaluation(BaseModel):
+    eval_id: str
+    """A unique evaluation ID."""
+
+    evaluation_status: Literal["in_progress", "completed", "canceled", "queued", "failed"]
+    """Status of the evaluation."""
+
+    api_model_input: EvaluationModelInput = FieldInfo(alias="model_input")
+    """A dictionary of inputs sent to the LLM to generate output.
+
+    The dictionary must contain at least a `user_prompt` field or a `system_prompt`
+    field. For ground_truth_adherence guardrail metric, `ground_truth` should be
+    provided.
+    """
+
+    api_model_output: str = FieldInfo(alias="model_output")
+    """Output generated by the LLM to be evaluated."""
+
+    run_mode: Literal["precision_plus", "precision", "smart", "economy"]
+    """Run mode for the evaluation.
+
+    The run mode allows the user to optimize for speed, accuracy, and cost by
+    determining which models are used to evaluate the event.
+    """
+
+    created_at: Optional[datetime] = None
+    """The time the evaluation was created in UTC."""
+
+    end_timestamp: Optional[datetime] = None
+    """The time the evaluation completed in UTC."""
+
+    error_message: Optional[str] = None
+    """Description of the error causing the evaluation to fail, if any."""
+
+    error_timestamp: Optional[datetime] = None
+    """The time the error causing the evaluation to fail was recorded."""
+
+    evaluation_result: Optional[Dict[str, object]] = None
+    """
+    Evaluation result consisting of average scores and rationales for each of the
+    evaluated guardrail metrics.
+    """
+
+    evaluation_total_cost: Optional[float] = None
+    """Total cost of the evaluation."""
+
+    guardrail_metrics: Optional[
+        List[
+            Literal[
+                "correctness",
+                "completeness",
+                "instruction_adherence",
+                "context_adherence",
+                "ground_truth_adherence",
+                "comprehensive_safety",
+            ]
+        ]
+    ] = None
+    """
+    An array of guardrail metrics that the model input and output pair will be
+    evaluated on.
+    """
+
+    api_model_used: Optional[str] = FieldInfo(alias="model_used", default=None)
+    """Model ID used to generate the output, like `gpt-4o` or `o3`."""
+
+    modified_at: Optional[datetime] = None
+    """The most recent time the evaluation was modified in UTC."""
+
+    nametag: Optional[str] = None
+    """An optional, user-defined tag for the evaluation."""
+
+    progress: Optional[int] = None
+    """Evaluation progress.
+
+    Values range between 0 and 100; 100 corresponds to a completed
+    `evaluation_status`.
+    """
+
+    start_timestamp: Optional[datetime] = None
+    """The time the evaluation started in UTC."""
 
 
 class Stats(BaseModel):

{deeprails-1.9.0.dist-info → deeprails-1.10.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: deeprails
-Version: 1.9.0
+Version: 1.10.0
 Summary: The official Python library for the deeprails API
 Project-URL: Homepage, https://docs.deeprails.com/
 Project-URL: Repository, https://github.com/deeprails/deeprails-sdk-python

{deeprails-1.9.0.dist-info → deeprails-1.10.0.dist-info}/RECORD

@@ -1,6 +1,6 @@
 deeprails/__init__.py,sha256=Rz38DnPkRJzvTJjgsLPjYyEId-pJndorgdE4Yhcq0J4,2692
 deeprails/_base_client.py,sha256=3bTH_cn_59gEjfZdcRS6g6Er4ZobAgTCRsnQmD_viNQ,67050
-deeprails/_client.py,sha256=oinOakaCufsSKAjWC2JOwT4bCeEmRB-xQCT9FNAG1rM,16125
+deeprails/_client.py,sha256=BzJOTEiC2RszdX3GCBEk_R5k6e48rzP-XlbCN1Cplr4,15563
 deeprails/_compat.py,sha256=DQBVORjFb33zch24jzkhM14msvnzY7mmSmgDLaVFUM8,6562
 deeprails/_constants.py,sha256=S14PFzyN9-I31wiV7SmIlL5Ga0MLHxdvegInGdXH7tM,462
 deeprails/_exceptions.py,sha256=N99bKrZNjDzgEDrvdw99WO-zpuPeZySaUUKNLEUz8mU,3226
@@ -11,7 +11,7 @@ deeprails/_resource.py,sha256=7RXX5KZr4j0TIE66vnduHp7p9Yf9X0FyDDECuvRHARg,1118
 deeprails/_response.py,sha256=yj0HJDU91WPpiczwi6CBOLAl_bqf4I_I96vWMAwx6Fg,28806
 deeprails/_streaming.py,sha256=hCp5bK9dyw2TyrVL69m-6qGC-QtGYwhXmFzITCWPgAs,10112
 deeprails/_types.py,sha256=XR3mad9NsGqZsjrd1VVJ657-4O4kwyw9Qzg4M3i6Vh0,7239
-deeprails/_version.py,sha256=wxi1GAD191IfMIMELfuuMOJr3v9EY7XoLbXOaNNw-dY,161
+deeprails/_version.py,sha256=dE6ROKmhSXt4B73GxgSswrtaxSRJBdQihEVLUUaQ-M0,162
 deeprails/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 deeprails/_utils/__init__.py,sha256=7fch0GT9zpNnErbciSpUNa-SjTxxjY6kxHxKMOM4AGs,2305
 deeprails/_utils/_compat.py,sha256=D8gtAvjJQrDWt9upS0XaG9Rr5l1QhiAx_I_1utT_tt0,1195
@@ -26,26 +26,23 @@ deeprails/_utils/_transform.py,sha256=NjCzmnfqYrsAikUHQig6N9QfuTVbKipuP3ur9mcNF-
 deeprails/_utils/_typing.py,sha256=N_5PPuFNsaygbtA_npZd98SVN1LQQvFTKL6bkWPBZGU,4786
 deeprails/_utils/_utils.py,sha256=0dDqauUbVZEXV0NVl7Bwu904Wwo5eyFCZpQThhFNhyA,12253
 deeprails/lib/.keep,sha256=wuNrz-5SXo3jJaJOJgz4vFHM41YH_g20F5cRQo0vLes,224
-deeprails/resources/__init__.py,sha256=ha0jL9Et8fHzPdkTa7ecihYapOC4I6O-PHf2X9igprE,1491
+deeprails/resources/__init__.py,sha256=JBzEBVMLg2n1w2QJGYCsyjPHY6zHZ0G3xaWuuFoa5Pg,1015
 deeprails/resources/defend.py,sha256=e6NzUJziYx__FkGSVNeaqzA8sZU0L5dY5Nob1IPCxc4,28186
-deeprails/resources/evaluate.py,sha256=6SJ3iOH3uYPz3k3MS_G_P75zgSu4bBIurazHQ8byEl0,13025
 deeprails/resources/monitor.py,sha256=bW7HCn8lDpWqaPd11J_tVnGSduWTX3sfXlDYYycvk8k,22524
-deeprails/types/__init__.py,sha256=-Dsqdk7bVrDKj3ryCVbwIfIqSLFfF8pOu3JU7O9-8TA,1254
+deeprails/types/__init__.py,sha256=PlF7idbYiAj3xyJYstHDaBer-rLUHe6eUEfRfJKisvk,1124
 deeprails/types/defend_create_workflow_params.py,sha256=AmyoU1z3a0U39pqVtVnFsM86ysK1nYIx15Tolwb4PKA,2102
 deeprails/types/defend_response.py,sha256=VoePIT9RKN795y-3ZvoFmzSTCetqkZZh6iQLYjJEFoY,1603
 deeprails/types/defend_submit_event_params.py,sha256=yL_rLUGKlZjXHGbdi8h9ZItb4sICQ2POO_o0VbUtans,1539
 deeprails/types/defend_update_workflow_params.py,sha256=QH2k7EDMLub3mW1lPV5SUoKDHW_T2arSo-RGHLterwo,373
-deeprails/types/evaluate_create_params.py,sha256=GGdFXjYjyLD5O1fn4bwDwDKI0EwvciJNF-uYmvCUBwA,2001
-deeprails/types/evaluation.py,sha256=3ajWcd7oYRtYZDOe0fVQHeVRN6-ymjA7F47VduwTfP4,3328
 deeprails/types/monitor_create_params.py,sha256=kTSj-PhuzpT-HPDTQJemRWfd8w32foUMH9FQZj8symk,384
-deeprails/types/monitor_detail_response.py,sha256=pts1JThENDUQhF0xKsPvOY3C5fUU-z8HW6ZVCKwmq8g,1939
+deeprails/types/monitor_detail_response.py,sha256=s8dzFFaU9uI-GHCfUxCN9yLhpmq4LTu7CSww9z0SLvk,5028
 deeprails/types/monitor_event_response.py,sha256=-cnugHD_3QeeZRMbo6aQBirqSPgKIKpaD2qNkgxCeCA,565
 deeprails/types/monitor_response.py,sha256=LjnJVYniee1hgvZu8RT-9jX4xd0Ob_yvq4NBOxVn59c,950
 deeprails/types/monitor_retrieve_params.py,sha256=PEsRmbd-81z4pJvhfi4JbrQWNzmeiLkoNsTUoPZ6kFY,352
 deeprails/types/monitor_submit_event_params.py,sha256=YetTV8HzmDGNnSYoKZp8tv3u9L6oYn4tnvWOppTRfOQ,2164
 deeprails/types/monitor_update_params.py,sha256=gJyFFxT_u_iWABknuKnLpPl9r-VPfCcGtOAmh6sPwUw,550
 deeprails/types/workflow_event_response.py,sha256=mIzOCnYJg4TDSq_tG_0WfA0_Gmc9-0q-befyookfUFM,867
-deeprails-1.9.0.dist-info/METADATA,sha256=XP4pijd9tcmRhIrw4j10axapTSyQ48VzYK_WCqiGkCc,12080
-deeprails-1.9.0.dist-info/WHEEL,sha256=C2FUgwZgiLbznR-k0b_5k3Ai_1aASOXDss3lzCUsUug,87
-deeprails-1.9.0.dist-info/licenses/LICENSE,sha256=rFTxPcYE516UQLju2SCY1r2pSDDfodL0-ZvxF_fgueg,11339
-deeprails-1.9.0.dist-info/RECORD,,
+deeprails-1.10.0.dist-info/METADATA,sha256=oMlAwTn3V1FCFTw1ErsGTS_clu3q1dtpn3vQAssiJv0,12081
+deeprails-1.10.0.dist-info/WHEEL,sha256=C2FUgwZgiLbznR-k0b_5k3Ai_1aASOXDss3lzCUsUug,87
+deeprails-1.10.0.dist-info/licenses/LICENSE,sha256=rFTxPcYE516UQLju2SCY1r2pSDDfodL0-ZvxF_fgueg,11339
+deeprails-1.10.0.dist-info/RECORD,,

deeprails/resources/evaluate.py

@@ -1,334 +0,0 @@
-# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
-
-from __future__ import annotations
-
-from typing import List
-from typing_extensions import Literal
-
-import httpx
-
-from ..types import evaluate_create_params
-from .._types import Body, Omit, Query, Headers, NotGiven, omit, not_given
-from .._utils import maybe_transform, async_maybe_transform
-from .._compat import cached_property
-from .._resource import SyncAPIResource, AsyncAPIResource
-from .._response import (
-    to_raw_response_wrapper,
-    to_streamed_response_wrapper,
-    async_to_raw_response_wrapper,
-    async_to_streamed_response_wrapper,
-)
-from .._base_client import make_request_options
-from ..types.evaluation import Evaluation
-
-__all__ = ["EvaluateResource", "AsyncEvaluateResource"]
-
-
-class EvaluateResource(SyncAPIResource):
-    @cached_property
-    def with_raw_response(self) -> EvaluateResourceWithRawResponse:
-        """
-        This property can be used as a prefix for any HTTP method call to return
-        the raw response object instead of the parsed content.
-
-        For more information, see https://www.github.com/deeprails/deeprails-sdk-python#accessing-raw-response-data-eg-headers
-        """
-        return EvaluateResourceWithRawResponse(self)
-
-    @cached_property
-    def with_streaming_response(self) -> EvaluateResourceWithStreamingResponse:
-        """
-        An alternative to `.with_raw_response` that doesn't eagerly read the response body.
-
-        For more information, see https://www.github.com/deeprails/deeprails-sdk-python#with_streaming_response
-        """
-        return EvaluateResourceWithStreamingResponse(self)
-
-    def create(
-        self,
-        *,
-        model_input: evaluate_create_params.ModelInput,
-        model_output: str,
-        run_mode: Literal["precision_plus", "precision", "smart", "economy"],
-        guardrail_metrics: List[
-            Literal[
-                "correctness",
-                "completeness",
-                "instruction_adherence",
-                "context_adherence",
-                "ground_truth_adherence",
-                "comprehensive_safety",
-            ]
-        ]
-        | Omit = omit,
-        model_used: str | Omit = omit,
-        nametag: str | Omit = omit,
-        # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.
-        # The extra values given here take precedence over values defined on the client or passed to this method.
-        extra_headers: Headers | None = None,
-        extra_query: Query | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = not_given,
-    ) -> Evaluation:
-        """
-        Use this endpoint to evaluate a model's input and output pair against selected
-        guardrail metrics
-
-        Args:
-          model_input: A dictionary of inputs sent to the LLM to generate output. The dictionary must
-              contain at least a `user_prompt` field or a `system_prompt` field. For
-              ground_truth_adherence guardrail metric, `ground_truth` should be provided.
-
-          model_output: Output generated by the LLM to be evaluated.
-
-          run_mode: Run mode for the evaluation. The run mode allows the user to optimize for speed,
-              accuracy, and cost by determining which models are used to evaluate the event.
-              Available run modes include `precision_plus`, `precision`, `smart`, and
-              `economy`. Defaults to `smart`.
-
-          guardrail_metrics: An array of guardrail metrics that the model input and output pair will be
-              evaluated on. For non-enterprise users, these will be limited to the allowed
-              guardrail metrics.
-
-          model_used: Model ID used to generate the output, like `gpt-4o` or `o3`.
-
-          nametag: An optional, user-defined tag for the evaluation.
-
-          extra_headers: Send extra headers
-
-          extra_query: Add additional query parameters to the request
-
-          extra_body: Add additional JSON properties to the request
-
-          timeout: Override the client-level default timeout for this request, in seconds
-        """
-        return self._post(
-            "/evaluate",
-            body=maybe_transform(
-                {
-                    "model_input": model_input,
-                    "model_output": model_output,
-                    "run_mode": run_mode,
-                    "guardrail_metrics": guardrail_metrics,
-                    "model_used": model_used,
-                    "nametag": nametag,
-                },
-                evaluate_create_params.EvaluateCreateParams,
-            ),
-            options=make_request_options(
-                extra_headers=extra_headers, extra_query=extra_query, extra_body=extra_body, timeout=timeout
-            ),
-            cast_to=Evaluation,
-        )
-
-    def retrieve(
-        self,
-        eval_id: str,
-        *,
-        # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.
-        # The extra values given here take precedence over values defined on the client or passed to this method.
-        extra_headers: Headers | None = None,
-        extra_query: Query | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = not_given,
-    ) -> Evaluation:
-        """
-        Use this endpoint to retrieve the evaluation record for a given evaluation ID
-
-        Args:
-          extra_headers: Send extra headers
-
-          extra_query: Add additional query parameters to the request
-
-          extra_body: Add additional JSON properties to the request
-
-          timeout: Override the client-level default timeout for this request, in seconds
-        """
-        if not eval_id:
-            raise ValueError(f"Expected a non-empty value for `eval_id` but received {eval_id!r}")
-        return self._get(
-            f"/evaluate/{eval_id}",
-            options=make_request_options(
-                extra_headers=extra_headers, extra_query=extra_query, extra_body=extra_body, timeout=timeout
-            ),
-            cast_to=Evaluation,
-        )
-
-
-class AsyncEvaluateResource(AsyncAPIResource):
-    @cached_property
-    def with_raw_response(self) -> AsyncEvaluateResourceWithRawResponse:
-        """
-        This property can be used as a prefix for any HTTP method call to return
-        the raw response object instead of the parsed content.
-
-        For more information, see https://www.github.com/deeprails/deeprails-sdk-python#accessing-raw-response-data-eg-headers
-        """
-        return AsyncEvaluateResourceWithRawResponse(self)
-
-    @cached_property
-    def with_streaming_response(self) -> AsyncEvaluateResourceWithStreamingResponse:
-        """
-        An alternative to `.with_raw_response` that doesn't eagerly read the response body.
-
-        For more information, see https://www.github.com/deeprails/deeprails-sdk-python#with_streaming_response
-        """
-        return AsyncEvaluateResourceWithStreamingResponse(self)
-
-    async def create(
-        self,
-        *,
-        model_input: evaluate_create_params.ModelInput,
-        model_output: str,
-        run_mode: Literal["precision_plus", "precision", "smart", "economy"],
-        guardrail_metrics: List[
-            Literal[
-                "correctness",
-                "completeness",
-                "instruction_adherence",
-                "context_adherence",
-                "ground_truth_adherence",
-                "comprehensive_safety",
-            ]
-        ]
-        | Omit = omit,
-        model_used: str | Omit = omit,
-        nametag: str | Omit = omit,
-        # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.
-        # The extra values given here take precedence over values defined on the client or passed to this method.
-        extra_headers: Headers | None = None,
-        extra_query: Query | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = not_given,
-    ) -> Evaluation:
-        """
-        Use this endpoint to evaluate a model's input and output pair against selected
-        guardrail metrics
-
-        Args:
-          model_input: A dictionary of inputs sent to the LLM to generate output. The dictionary must
-              contain at least a `user_prompt` field or a `system_prompt` field. For
-              ground_truth_adherence guardrail metric, `ground_truth` should be provided.
-
-          model_output: Output generated by the LLM to be evaluated.
-
-          run_mode: Run mode for the evaluation. The run mode allows the user to optimize for speed,
-              accuracy, and cost by determining which models are used to evaluate the event.
-              Available run modes include `precision_plus`, `precision`, `smart`, and
-              `economy`. Defaults to `smart`.
-
-          guardrail_metrics: An array of guardrail metrics that the model input and output pair will be
-              evaluated on. For non-enterprise users, these will be limited to the allowed
-              guardrail metrics.
-
-          model_used: Model ID used to generate the output, like `gpt-4o` or `o3`.
-
-          nametag: An optional, user-defined tag for the evaluation.
-
-          extra_headers: Send extra headers
-
-          extra_query: Add additional query parameters to the request
-
-          extra_body: Add additional JSON properties to the request
-
-          timeout: Override the client-level default timeout for this request, in seconds
-        """
-        return await self._post(
-            "/evaluate",
-            body=await async_maybe_transform(
-                {
-                    "model_input": model_input,
-                    "model_output": model_output,
-                    "run_mode": run_mode,
-                    "guardrail_metrics": guardrail_metrics,
-                    "model_used": model_used,
-                    "nametag": nametag,
-                },
-                evaluate_create_params.EvaluateCreateParams,
-            ),
-            options=make_request_options(
-                extra_headers=extra_headers, extra_query=extra_query, extra_body=extra_body, timeout=timeout
-            ),
-            cast_to=Evaluation,
-        )
-
-    async def retrieve(
-        self,
-        eval_id: str,
-        *,
-        # Use the following arguments if you need to pass additional parameters to the API that aren't available via kwargs.
-        # The extra values given here take precedence over values defined on the client or passed to this method.
-        extra_headers: Headers | None = None,
-        extra_query: Query | None = None,
-        extra_body: Body | None = None,
-        timeout: float | httpx.Timeout | None | NotGiven = not_given,
-    ) -> Evaluation:
-        """
-        Use this endpoint to retrieve the evaluation record for a given evaluation ID
-
-        Args:
-          extra_headers: Send extra headers
-
-          extra_query: Add additional query parameters to the request
-
-          extra_body: Add additional JSON properties to the request
-
-          timeout: Override the client-level default timeout for this request, in seconds
-        """
-        if not eval_id:
-            raise ValueError(f"Expected a non-empty value for `eval_id` but received {eval_id!r}")
-        return await self._get(
-            f"/evaluate/{eval_id}",
-            options=make_request_options(
-                extra_headers=extra_headers, extra_query=extra_query, extra_body=extra_body, timeout=timeout
-            ),
-            cast_to=Evaluation,
-        )
-
-
-class EvaluateResourceWithRawResponse:
-    def __init__(self, evaluate: EvaluateResource) -> None:
-        self._evaluate = evaluate
-
-        self.create = to_raw_response_wrapper(
-            evaluate.create,
-        )
-        self.retrieve = to_raw_response_wrapper(
-            evaluate.retrieve,
-        )
-
-
-class AsyncEvaluateResourceWithRawResponse:
-    def __init__(self, evaluate: AsyncEvaluateResource) -> None:
-        self._evaluate = evaluate
-
-        self.create = async_to_raw_response_wrapper(
-            evaluate.create,
-        )
-        self.retrieve = async_to_raw_response_wrapper(
-            evaluate.retrieve,
-        )
-
-
-class EvaluateResourceWithStreamingResponse:
-    def __init__(self, evaluate: EvaluateResource) -> None:
-        self._evaluate = evaluate
-
-        self.create = to_streamed_response_wrapper(
-            evaluate.create,
-        )
-        self.retrieve = to_streamed_response_wrapper(
-            evaluate.retrieve,
-        )
-
-
-class AsyncEvaluateResourceWithStreamingResponse:
-    def __init__(self, evaluate: AsyncEvaluateResource) -> None:
-        self._evaluate = evaluate
-
-        self.create = async_to_streamed_response_wrapper(
-            evaluate.create,
-        )
-        self.retrieve = async_to_streamed_response_wrapper(
-            evaluate.retrieve,
-        )

deeprails/types/evaluate_create_params.py

@@ -1,63 +0,0 @@
-# File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
-
-from __future__ import annotations
-
-from typing import List
-from typing_extensions import Literal, Required, TypedDict
-
-__all__ = ["EvaluateCreateParams", "ModelInput"]
-
-
-class EvaluateCreateParams(TypedDict, total=False):
-    model_input: Required[ModelInput]
-    """A dictionary of inputs sent to the LLM to generate output.
-
-    The dictionary must contain at least a `user_prompt` field or a `system_prompt`
-    field. For ground_truth_adherence guardrail metric, `ground_truth` should be
-    provided.
-    """
-
-    model_output: Required[str]
-    """Output generated by the LLM to be evaluated."""
-
-    run_mode: Required[Literal["precision_plus", "precision", "smart", "economy"]]
-    """Run mode for the evaluation.
-
-    The run mode allows the user to optimize for speed, accuracy, and cost by
-    determining which models are used to evaluate the event. Available run modes
-    include `precision_plus`, `precision`, `smart`, and `economy`. Defaults to
-    `smart`.
-    """
-
-    guardrail_metrics: List[
-        Literal[
-            "correctness",
-            "completeness",
-            "instruction_adherence",
-            "context_adherence",
-            "ground_truth_adherence",
-            "comprehensive_safety",
-        ]
-    ]
-    """
-    An array of guardrail metrics that the model input and output pair will be
-    evaluated on. For non-enterprise users, these will be limited to the allowed
-    guardrail metrics.
-    """
-
-    model_used: str
-    """Model ID used to generate the output, like `gpt-4o` or `o3`."""
-
-    nametag: str
-    """An optional, user-defined tag for the evaluation."""
-
-
-class ModelInput(TypedDict, total=False):
-    ground_truth: str
-    """The ground truth for evaluating Ground Truth Adherence guardrail."""
-
-    system_prompt: str
-    """The system prompt used to generate the output."""
-
-    user_prompt: str
-    """The user prompt used to generate the output."""

deeprails/types/evaluation.py

@@ -1,105 +0,0 @@
-# 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 pydantic import Field as FieldInfo
-
-from .._models import BaseModel
-
-__all__ = ["Evaluation", "ModelInput"]
-
-
-class ModelInput(BaseModel):
-    ground_truth: Optional[str] = None
-    """The ground truth for evaluating Ground Truth Adherence guardrail."""
-
-    system_prompt: Optional[str] = None
-    """The system prompt used to generate the output."""
-
-    user_prompt: Optional[str] = None
-    """The user prompt used to generate the output."""
-
-
-class Evaluation(BaseModel):
-    eval_id: str
-    """A unique evaluation ID."""
-
-    evaluation_status: Literal["in_progress", "completed", "canceled", "queued", "failed"]
-    """Status of the evaluation."""
-
-    api_model_input: ModelInput = FieldInfo(alias="model_input")
-    """A dictionary of inputs sent to the LLM to generate output.
-
-    The dictionary must contain at least a `user_prompt` field or a `system_prompt`
-    field. For ground_truth_adherence guardrail metric, `ground_truth` should be
-    provided.
-    """
-
-    api_model_output: str = FieldInfo(alias="model_output")
-    """Output generated by the LLM to be evaluated."""
-
-    run_mode: Literal["precision_plus", "precision", "smart", "economy"]
-    """Run mode for the evaluation.
-
-    The run mode allows the user to optimize for speed, accuracy, and cost by
-    determining which models are used to evaluate the event.
-    """
-
-    created_at: Optional[datetime] = None
-    """The time the evaluation was created in UTC."""
-
-    end_timestamp: Optional[datetime] = None
-    """The time the evaluation completed in UTC."""
-
-    error_message: Optional[str] = None
-    """Description of the error causing the evaluation to fail, if any."""
-
-    error_timestamp: Optional[datetime] = None
-    """The time the error causing the evaluation to fail was recorded."""
-
-    evaluation_result: Optional[Dict[str, object]] = None
-    """
-    Evaluation result consisting of average scores and rationales for each of the
-    evaluated guardrail metrics.
-    """
-
-    evaluation_total_cost: Optional[float] = None
-    """Total cost of the evaluation."""
-
-    guardrail_metrics: Optional[
-        List[
-            Literal[
-                "correctness",
-                "completeness",
-                "instruction_adherence",
-                "context_adherence",
-                "ground_truth_adherence",
-                "comprehensive_safety",
-            ]
-        ]
-    ] = None
-    """
-    An array of guardrail metrics that the model input and output pair will be
-    evaluated on.
-    """
-
-    api_model_used: Optional[str] = FieldInfo(alias="model_used", default=None)
-    """Model ID used to generate the output, like `gpt-4o` or `o3`."""
-
-    modified_at: Optional[datetime] = None
-    """The most recent time the evaluation was modified in UTC."""
-
-    nametag: Optional[str] = None
-    """An optional, user-defined tag for the evaluation."""
-
-    progress: Optional[int] = None
-    """Evaluation progress.
-
-    Values range between 0 and 100; 100 corresponds to a completed
-    `evaluation_status`.
-    """
-
-    start_timestamp: Optional[datetime] = None
-    """The time the evaluation started in UTC."""