seekrai 0.5.16__tar.gz → 0.5.17__tar.gz

{seekrai-0.5.16 → seekrai-0.5.17}/PKG-INFO

@@ -1,7 +1,8 @@
-Metadata-Version: 2.3
+Metadata-Version: 2.1
 Name: seekrai
-Version: 0.5.16
+Version: 0.5.17
 Summary: Python client for SeekrAI
+Home-page: https://gitlab.cb.ntent.com/ml/seekr-py
 License: Apache-2.0
 Author: SeekrFlow
 Author-email: support@seekr.com
@@ -12,8 +13,6 @@ Classifier: Programming Language :: Python :: 3
 Classifier: Programming Language :: Python :: 3.9
 Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
-Classifier: Programming Language :: Python :: 3.12
-Classifier: Programming Language :: Python :: 3.13
 Requires-Dist: click (>=8.1.7,<9.0.0)
 Requires-Dist: eval-type-backport (>=0.1.3,<0.3.0)
 Requires-Dist: filelock (>=3.13.1,<4.0.0)

{seekrai-0.5.16 → seekrai-0.5.17}/pyproject.toml

@@ -14,7 +14,7 @@ build-backend = "poetry.core.masonry.api"
 
 [tool.poetry]
 name = "seekrai"
-version = "0.5.16"
+version = "0.5.17"
 authors = [
     "SeekrFlow <support@seekr.com>"
 ]

{seekrai-0.5.16 → seekrai-0.5.17}/src/seekrai/client.py

@@ -25,6 +25,7 @@ class SeekrFlow:
     vector_database: resources.VectorDatabase
     agents: resources.Agents
     observability: resources.AgentObservability
+    explainability: resources.Explainability
 
     # client options
     client: SeekrFlowClient
@@ -91,6 +92,7 @@ class SeekrFlow:
         self.vector_database = resources.VectorDatabase(self.client)
         self.agents = resources.Agents(self.client)
         self.observability = resources.AgentObservability(self.client)
+        self.explainability = resources.Explainability(self.client)
 
 
 class AsyncSeekrFlow:
@@ -108,6 +110,7 @@ class AsyncSeekrFlow:
     vector_database: resources.AsyncVectorDatabase
     agents: resources.AsyncAgents
     observability: resources.AsyncAgentObservability
+    explainability: resources.AsyncExplainability
 
     # client options
     client: SeekrFlowClient
@@ -174,6 +177,7 @@ class AsyncSeekrFlow:
         self.vector_database = resources.AsyncVectorDatabase(self.client)
         self.agents = resources.AsyncAgents(self.client)
         self.observability = resources.AsyncAgentObservability(self.client)
+        self.explainability = resources.AsyncExplainability(self.client)
 
 
 Client = SeekrFlow

{seekrai-0.5.16 → seekrai-0.5.17}/src/seekrai/resources/explainability.py

@@ -2,7 +2,6 @@ from typing import Optional
 
 from seekrai.abstract import api_requestor
 from seekrai.resources.resource_base import ResourceBase
-from seekrai.seekrflow_response import SeekrFlowResponse
 from seekrai.types import (
     SeekrFlowRequest,
 )
@@ -14,15 +13,19 @@ from seekrai.types.explainability import (
 
 class Explainability(ResourceBase):
     def get_influential_finetuning_data(
-        self, model_id: str, question: str, answer: Optional[str], k: int = 5
+        self,
+        model_id: str,
+        question: str,
+        system_prompt: Optional[str] = None,
+        answer: Optional[str] = None,
     ) -> InfluentialFinetuningDataResponse:
         """
         Retrieve influential QA pair fine tuning data for a specific model.
         Args:
             - model_id (str): ID of the model to explain.
             - question (str): question from user,
+            - system_prompt (str | None): System prompt for the user's question.
             - answer (str | None): answer of the finetuned model to the question; if None, the answer is retrieved from the finetuned model specified by model_id,
-            - k (int): the number of results to be retrieved (5 by default)
         Returns:
             InfluentialFinetuningDataResponse: Object containing the influential fine tuning data.
         """
@@ -31,7 +34,7 @@ class Explainability(ResourceBase):
         )
         # Create query parameters dictionary
         parameter_payload = InfluentialFinetuningDataRequest(
-            question=question, answer=answer, k=k
+            question=question, system_prompt=system_prompt, answer=answer
         ).model_dump()
 
         # if limit is not None:
@@ -41,26 +44,29 @@ class Explainability(ResourceBase):
         response, _, _ = requestor.request(
             options=SeekrFlowRequest(
                 method="GET",
-                url=f"v1/flow/explain/models/{model_id}/influential-finetuning-data",
+                url=f"flow/explain/models/{model_id}/influential-finetuning-data",
                 params=parameter_payload,
             ),
             stream=False,
         )
-        assert isinstance(response, SeekrFlowResponse)
         return InfluentialFinetuningDataResponse(**response.data)
 
 
 class AsyncExplainability(ResourceBase):
     async def get_influential_finetuning_data(
-        self, model_id: str, question: str, answer: Optional[str], k: int = 5
+        self,
+        model_id: str,
+        question: str,
+        system_prompt: Optional[str] = None,
+        answer: Optional[str] = None,
     ) -> InfluentialFinetuningDataResponse:
         """
         Retrieve influential QA pair finetuning data for a specific model asynchronously.
         Args:
             - model_id (str): ID of the model to explain.
             - question (str): question from user,
+            - system_prompt (str | None): System prompt for the user's question.
             - answer (str | None): answer of the finetuned model to the question; if None, the answer is retrieved from the finetuned model specified by model_id,
-            - k (int): the number of results to be retrieved (5 by default),
         Returns:
             InfluentialFinetuningDataResponse: Object containing the influential finetuning data.
         """
@@ -69,16 +75,18 @@ class AsyncExplainability(ResourceBase):
         )
         # Create query parameters dictionary
         parameter_payload = InfluentialFinetuningDataRequest(
-            model_id=model_id, question=question, answer=answer, k=k
+            model_id=model_id,
+            question=question,
+            system_prompt=system_prompt,
+            answer=answer,
         ).model_dump()
 
         response, _, _ = await requestor.arequest(
             options=SeekrFlowRequest(
                 method="GET",
-                url=f"v1/flow/explain/models/{model_id}/influential-finetuning-data",
+                url=f"flow/explain/models/{model_id}/influential-finetuning-data",
                 params=parameter_payload,
             ),
             stream=False,
         )
-        assert isinstance(response, SeekrFlowResponse)
         return InfluentialFinetuningDataResponse(**response.data)

{seekrai-0.5.16 → seekrai-0.5.17}/src/seekrai/resources/ingestion.py

@@ -1,4 +1,4 @@
-from typing import List
+from typing import List, Optional
 
 from seekrai.abstract import api_requestor
 from seekrai.resources.resource_base import ResourceBase
@@ -13,6 +13,7 @@ class Ingestion(ResourceBase):
     def ingest(
         self,
         files: List[str],
+        method: Optional[str] = "accuracy-optimized",
     ) -> IngestionResponse:
         """
         Start an ingestion job for the specified files.
@@ -27,9 +28,7 @@ class Ingestion(ResourceBase):
             client=self._client,
         )
 
-        parameter_payload = IngestionRequest(
-            files=files,
-        ).model_dump()
+        parameter_payload = IngestionRequest(files=files, method=method).model_dump()
 
         response, _, _ = requestor.request(
             options=SeekrFlowRequest(
@@ -95,6 +94,7 @@ class AsyncIngestion(ResourceBase):
     async def ingest(
         self,
         files: List[str],
+        method: Optional[str] = "accuracy-optimized",
     ) -> IngestionResponse:
         """
         Start an ingestion job for the specified files asynchronously.
@@ -109,9 +109,7 @@ class AsyncIngestion(ResourceBase):
             client=self._client,
         )
 
-        parameter_payload = IngestionRequest(
-            files=files,
-        ).model_dump()
+        parameter_payload = IngestionRequest(files=files, method=method).model_dump()
 
         response, _, _ = await requestor.arequest(
             options=SeekrFlowRequest(

{seekrai-0.5.16 → seekrai-0.5.17}/src/seekrai/resources/vectordb.py

@@ -107,7 +107,8 @@ class VectorDatabase(ResourceBase):
         self,
         database_id: str,
         files: List[str],
-        method: str,
+        method: Optional[str] = "accuracy-optimized",
+        chunking_method: Optional[str] = "markdown",
         token_count: int = 800,
         overlap_tokens: int = 100,
     ) -> VectorDatabaseIngestionResponse:
@@ -129,6 +130,7 @@ class VectorDatabase(ResourceBase):
         parameter_payload = VectorDatabaseIngestionRequest(
             file_ids=files,
             method=method,
+            chunking_method=chunking_method,
             token_count=token_count,
             overlap_tokens=overlap_tokens,
         ).model_dump()
@@ -338,7 +340,8 @@ class AsyncVectorDatabase(ResourceBase):
         self,
         database_id: str,
         files: List[str],
-        method: str,
+        method: Optional[str] = "accuracy-optimized",
+        chunking_method: Optional[str] = "markdown",
         token_count: int = 800,
         overlap_tokens: int = 100,
     ) -> VectorDatabaseIngestionResponse:
@@ -360,6 +363,7 @@ class AsyncVectorDatabase(ResourceBase):
         parameter_payload = VectorDatabaseIngestionRequest(
             file_ids=files,
             method=method,
+            chunking_method=chunking_method,
             token_count=token_count,
             overlap_tokens=overlap_tokens,
         ).model_dump()

seekrai-0.5.17/src/seekrai/types/explainability.py

@@ -0,0 +1,26 @@
+from __future__ import annotations
+
+from typing import Any, Dict, List, Optional
+
+from pydantic import Field
+
+from seekrai.types.abstract import BaseModel
+
+
+class InfluentialFinetuningDataResponse(BaseModel):
+    results: List[Dict[str, Any]] = Field(
+        ..., description="List of influential training data results"
+    )
+    version: str = Field(..., description="Version of the explainability service")
+
+
+class InfluentialFinetuningDataRequest(BaseModel):
+    question: str = Field(..., description="Question from user")
+    system_prompt: Optional[str] = Field(
+        None,
+        description="System prompt for the user's question.",
+    )
+    answer: Optional[str] = Field(
+        None,
+        description="Answer of the finetuned model to the question; if None, the answer is retrieved from the finetuned model",
+    )

{seekrai-0.5.16 → seekrai-0.5.17}/src/seekrai/types/vectordb.py

@@ -37,7 +37,12 @@ class VectorDatabaseIngestionRequest(BaseModel):
     """Request model for creating a new vector database ingestion job."""
 
     file_ids: List[str] = Field(..., description="List of file IDs to ingest")
-    method: str = Field(..., description="Method to use for ingestion")
+    method: Optional[str] = Field(
+        default="accuracy-optimized", description="Method to use for ingestion"
+    )
+    chunking_method: Optional[str] = Field(
+        default="markdown", description="Configure how your content will be segmented"
+    )
     token_count: int = Field(default=800, description="Token count for ingestion")
     overlap_tokens: int = Field(default=100, description="Overlap tokens for ingestion")
 

seekrai-0.5.16/src/seekrai/types/explainability.py

@@ -1,57 +0,0 @@
-from __future__ import annotations
-
-from datetime import datetime
-from enum import Enum
-from typing import Any, Dict, List, Literal
-
-from pydantic import Field
-
-from seekrai.types.abstract import BaseModel
-
-
-class InfluentialFinetuningDataResponse(BaseModel):
-    results: List[Dict[str, Any]]
-    version: str
-
-
-class InfluentialFinetuningDataRequest(BaseModel):
-    question: str
-    answer: str = Field(
-        default="",
-        description="Response could be generated or given",
-    )
-    k: int
-
-
-class ExplainabilityJobStatus(Enum):
-    QUEUED = "queued"
-    RUNNING = "running"
-    COMPLETED = "completed"
-    FAILED = "failed"
-
-    # TODO should titles along the following get added:
-    # create_index
-    # populate_index
-    # delete_index
-    # influential-finetuning-data
-
-
-class ExplainabilityRequest(BaseModel):
-    files: List[str] = Field(
-        default=..., description="List of file ids to use for fine tuning"
-    )
-    method: str = Field(default="best", description="Method to use for explainability")
-
-
-class ExplainabilityResponse(BaseModel):
-    id: str = Field(default=..., description="Explainability job ID")
-    created_at: datetime
-    status: ExplainabilityJobStatus
-    output_files: List[str]
-
-
-class ExplainabilityList(BaseModel):
-    # object type
-    object: Literal["list"] | None = None
-    # list of fine-tune job objects
-    data: List[ExplainabilityResponse] | None = None