seekrai 0.3.2__py3-none-any.whl → 0.4.1__py3-none-any.whl

seekrai/abstract/api_requestor.py

@@ -524,6 +524,8 @@ class APIRequestor:
                 if rbody.strip().endswith("[DONE]"):
                     # TODO
                     rbody = rbody.replace("data: [DONE]", "")
+                if rbody.startswith("data: "):
+                    rbody = rbody[len("data: ") :]
                 data = json.loads(rbody)
         except (JSONDecodeError, UnicodeDecodeError) as e:
             raise error.APIError(

seekrai/filemanager.py

@@ -91,10 +91,9 @@ def _prepare_output(
 
     content_type = str(headers.get("content-type"))
 
-    assert remote_name, (
-        "No model name found in fine_tune object. "
-        "Please specify an `output` file name."
-    )
+    assert (
+        remote_name
+    ), "No model name found in fine_tune object. Please specify an `output` file name."
 
     if step > 0:
         remote_name += f"-checkpoint-{step}"

seekrai/resources/alignment.py

@@ -3,9 +3,12 @@ from typing import List
 from seekrai.abstract import api_requestor
 from seekrai.seekrflow_response import SeekrFlowResponse
 from seekrai.types import (
+    AlignmentEstimationRequest,
+    AlignmentEstimationResponse,
     AlignmentList,
     AlignmentRequest,
     AlignmentResponse,
+    AlignmentType,
     SeekrFlowClient,
     SeekrFlowRequest,
 )
@@ -19,14 +22,14 @@ class Alignment:
         self,
         instructions: str,
         files: List[str],
+        type: AlignmentType = AlignmentType.PRINCIPLE,
     ) -> AlignmentResponse:
         requestor = api_requestor.APIRequestor(
             client=self._client,
         )
 
         parameter_payload = AlignmentRequest(
-            instructions=instructions,
-            files=files,
+            instructions=instructions, files=files, type=type
         ).model_dump()
 
         response, _, _ = requestor.request(
@@ -92,6 +95,27 @@ class Alignment:
 
         return AlignmentResponse(**response.data)
 
+    def estimate(self, files: List[str]) -> AlignmentEstimationResponse:
+        requestor = api_requestor.APIRequestor(
+            client=self._client,
+        )
+
+        parameter_payload = AlignmentEstimationRequest(
+            files=files,
+        ).model_dump()
+
+        response, _, _ = requestor.request(
+            options=SeekrFlowRequest(
+                method="POST",
+                url="flow/alignment/estimate",
+                params=parameter_payload,
+            ),
+            stream=False,
+        )
+
+        assert isinstance(response, SeekrFlowResponse)
+        return AlignmentEstimationResponse(**response.data)
+
 
 class AsyncAlignment:
     def __init__(self, client: SeekrFlowClient) -> None:
@@ -101,14 +125,14 @@ class AsyncAlignment:
         self,
         instructions: str,
         files: List[str],
+        type: AlignmentType = AlignmentType.PRINCIPLE,
     ) -> AlignmentResponse:
         requestor = api_requestor.APIRequestor(
             client=self._client,
         )
 
         parameter_payload = AlignmentRequest(
-            instructions=instructions,
-            files=files,
+            instructions=instructions, files=files, type=type
         ).model_dump()
 
         response, _, _ = await requestor.arequest(
@@ -173,3 +197,24 @@ class AsyncAlignment:
         assert isinstance(response, SeekrFlowResponse)
 
         return AlignmentResponse(**response.data)
+
+    async def estimate(self, files: List[str]) -> AlignmentEstimationResponse:
+        requestor = api_requestor.APIRequestor(
+            client=self._client,
+        )
+
+        parameter_payload = AlignmentEstimationRequest(
+            files=files,
+        ).model_dump()
+
+        response, _, _ = await requestor.arequest(
+            options=SeekrFlowRequest(
+                method="POST",
+                url="flow/alignment/estimate",
+                params=parameter_payload,
+            ),
+            stream=False,
+        )
+
+        assert isinstance(response, SeekrFlowResponse)
+        return AlignmentEstimationResponse(**response.data)

seekrai/resources/chat/completions.py

@@ -30,13 +30,13 @@ class ChatCompletions:
         repetition_penalty: float = 1,
         stream: bool = False,
         logprobs: bool | None = False,
-        top_logprobs: int | None = 0,
+        top_logprobs: int | None = None,
         echo: bool = False,
         n: int = 1,
         safety_model: str | None = None,
         response_format: Dict[str, str | Dict[str, Any]] | None = None,
         tools: Dict[str, str | Dict[str, Any]] | None = None,
-        tool_choice: str | Dict[str, str | Dict[str, str]] | None = None,
+        tool_choice: str | Dict[str, str | Dict[str, str]] | None = "auto",
     ) -> ChatCompletionResponse | Iterator[ChatCompletionChunk]:
         """
         Method to generate completions based on a given prompt using a specified model.
@@ -89,7 +89,6 @@ class ChatCompletions:
             ChatCompletionResponse | Iterator[ChatCompletionChunk]: Object containing the completions
             or an iterator over completion chunks.
         """
-
         requestor = api_requestor.APIRequestor(
             client=self._client,
         )
@@ -110,7 +109,7 @@ class ChatCompletions:
             n=n,
             safety_model=safety_model,
             response_format=response_format,
-            tools=tools,
+            tools=tools or [],
             tool_choice=tool_choice,
         ).model_dump()
 

seekrai/types/__init__.py

@@ -1,9 +1,12 @@
 from seekrai.types.abstract import SeekrFlowClient
 from seekrai.types.alignment import (
+    AlignmentEstimationRequest,
+    AlignmentEstimationResponse,
     AlignmentJobStatus,
     AlignmentList,
     AlignmentRequest,
     AlignmentResponse,
+    AlignmentType,
 )
 from seekrai.types.chat_completions import (
     ChatCompletionChunk,
@@ -87,9 +90,12 @@ __all__ = [
     "ModelResponse",
     "ModelList",
     "AlignmentRequest",
+    "AlignmentEstimationRequest",
+    "AlignmentEstimationResponse",
     "AlignmentResponse",
     "AlignmentJobStatus",
     "AlignmentList",
+    "AlignmentType",
     "Project",
     "ProjectWithRuns",
     "GetProjectsResponse",

seekrai/types/alignment.py

@@ -9,6 +9,12 @@ from pydantic import Field
 from seekrai.types.abstract import BaseModel
 
 
+class AlignmentType(str, Enum):
+    PRINCIPLE = "principle"
+    CHAIN_OF_THOUGHT = "chain_of_thought"
+    RAFT = "raft"
+
+
 class AlignmentRequest(BaseModel):
     instructions: str = Field(
         default=..., description="Task description/instructions for the alignment task"
@@ -16,6 +22,28 @@ class AlignmentRequest(BaseModel):
     files: List[str] = Field(
         default=..., description="List of file ids to use for alignment"
     )
+    type: AlignmentType = Field(
+        default=AlignmentType.PRINCIPLE,
+        description="Type of alignment task (principle, chain_of_thought, or raft)",
+    )
+
+
+class AlignmentEstimationRequest(BaseModel):
+    files: List[str] = Field(
+        default=...,
+        description="List of file ids to use to generate an alignment estimate",
+    )
+
+
+class AlignmentEstimationResponse(BaseModel):
+    input_tokens: int = Field(
+        default=...,
+        description="Estimated number of input tokens for the given files",
+    )
+    output_tokens: int = Field(
+        default=...,
+        description="Estimated number of output tokens for the given files",
+    )
 
 
 class AlignmentJobStatus(str, Enum):

seekrai/types/finetune.py

@@ -172,6 +172,7 @@ class FinetuneResponse(BaseModel):
     inference_available: bool = False
     project_id: Optional[int] = None  # TODO - fix this
     completed_at: datetime | None = None
+    description: str | None = None
 
     # dataset token count
     # TODO

seekrai/types/ingestion.py

@@ -0,0 +1,8 @@
+from enum import Enum
+
+
+class IngestionJobStatus(Enum):
+    QUEUED = "queued"
+    RUNNING = "running"
+    COMPLETED = "completed"
+    FAILED = "failed"

seekrai/utils/files.py

@@ -47,7 +47,7 @@ def check_file(
 
     if file_size > MAX_FILE_SIZE_GB * NUM_BYTES_IN_GB:
         report_dict["message"] = (
-            f"Maximum supported file size is {MAX_FILE_SIZE_GB} GB. Found file with size of {round(file_size / NUM_BYTES_IN_GB ,3)} GB."
+            f"Maximum supported file size is {MAX_FILE_SIZE_GB} GB. Found file with size of {round(file_size / NUM_BYTES_IN_GB, 3)} GB."
         )
         report_dict["is_check_passed"] = False
     elif file_size == 0:
@@ -117,7 +117,7 @@ def _check_jsonl(file: Path) -> Dict[str, Any]:
                         report_dict["key_value"] = False
                         report_dict["message"] = (
                             f'Invalid value type for "text" key on line {idx + 1}. '
-                            f'Expected string. Found {type(json_line["text"])}.'
+                            f"Expected string. Found {type(json_line['text'])}."
                         )
 
                         report_dict["is_check_passed"] = False

{seekrai-0.3.2.dist-info → seekrai-0.4.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: seekrai
-Version: 0.3.2
+Version: 0.4.1
 Summary: Python client for SeekrAI
 Home-page: https://gitlab.cb.ntent.com/ml/seekr-py
 License: Apache-2.0

{seekrai-0.3.2.dist-info → seekrai-0.4.1.dist-info}/RECORD

@@ -1,14 +1,14 @@
 seekrai/__init__.py,sha256=HC6iy-IdwqecabH-6a80Lsy9qO2PBToAI0WqEErV41c,935
 seekrai/abstract/__init__.py,sha256=wNiOTW9TJpUgfCJCG-wAbhhWWH2PtoVpAuL3nxvQGps,56
-seekrai/abstract/api_requestor.py,sha256=tI43JdbUfNQHWjvRrenUM7C7Ea-zD-7yd2JNS-Lki04,18297
+seekrai/abstract/api_requestor.py,sha256=NT41M0_Utfw61bMqibP1PEdIIQmmKWfzj-Y-xk4DNY0,18395
 seekrai/client.py,sha256=Yhejl-2a-Uoc8nWi-XxETZT4a4Ou_t_TRLcp0e9APIY,5376
 seekrai/constants.py,sha256=hoR2iF5te5Ydjt_lxIOSGID4vESIakG4F-3xAWdwxaU,1854
 seekrai/error.py,sha256=rAYL8qEd8INwYMMKvhS-HKeC3QkWL4Wq-zfazFU-zBg,4861
-seekrai/filemanager.py,sha256=8RuSzJvELD-fCI2Wd_t0jSKeVrmFwF7E5AzXIgDFxNA,9572
+seekrai/filemanager.py,sha256=5tl7fjsLbNy-U1Ko_5iQSdkZ0Bca_bhvR9FhWpaG8FA,9561
 seekrai/resources/__init__.py,sha256=-OpvrZp0c_ro_e1G14gjOILtcSfpSOeRZeK8kt41WLo,1014
-seekrai/resources/alignment.py,sha256=6qQm9w0Em0q3zVeOzs8cX3wulr-B-wh4Pcr3pboIeTE,4468
+seekrai/resources/alignment.py,sha256=MMrADpXNZa0eQ-jqfLcP7-fisJXgpRl-03GPdkjkTcc,5951
 seekrai/resources/chat/__init__.py,sha256=KmtPupgECtEN80NyvcnSmieTAFXhwmVxhMHP0qhspA4,618
-seekrai/resources/chat/completions.py,sha256=pGBVrIvUfY-wr0ooiywzuhIjZ4GwiojH3ofmRE6bYlw,11643
+seekrai/resources/chat/completions.py,sha256=U75hKIDBAX8h5-4aFe20DowO1adqBD6ZRHjPK9EdqqM,11653
 seekrai/resources/completions.py,sha256=w3La3zPMlN00y-b-tJwLgvZVH-xK_dKC6ktI5Ggn1us,8564
 seekrai/resources/deployments.py,sha256=HmP7MuxAlLUoQl6z705_d1Y53MDvGgSQXhlgN3DKX2A,6078
 seekrai/resources/embeddings.py,sha256=3lohUrkdFqzSg8FgS7p4r87jwjE0NXU1PilWv278quk,2705
@@ -18,9 +18,9 @@ seekrai/resources/images.py,sha256=E48lAe7YsZ2WXBHR_qz4SF7P4Y-U7t61m_bWNS91pM0,4
 seekrai/resources/models.py,sha256=Pdd0S0gZdratWcHJPKNb7LkEdUGjr3xNR06W6GDiyxk,5000
 seekrai/resources/projects.py,sha256=AWJUeUDSzkbxBksHjJ4a3c83UR62TlMGHutm2NdV6Xk,3790
 seekrai/seekrflow_response.py,sha256=5RFEQzamDy7sTSDkxSsZQThZ3biNmeCPeHWdrFId5Go,1320
-seekrai/types/__init__.py,sha256=E41__vAq-tnm5aVYW0Ef1Ky-GMPnMpUKlyLr3pYhLSM,2358
+seekrai/types/__init__.py,sha256=34IvdhaeKDNqA03G0nwta8sptpURzbXeTlqr6mhagnw,2532
 seekrai/types/abstract.py,sha256=TqWFQV_6bPblywfCH-r8FCkXWvPkc9KlJ4QVgyrnaMc,642
-seekrai/types/alignment.py,sha256=-lxF0Cyj3w4brMft8WAoc2EDMAlbX7zdRYD04ot98-Y,1129
+seekrai/types/alignment.py,sha256=H9rAYsr4rg7uZCgl2tDZjkeq-b_Ov27gljejH8fC26w,1919
 seekrai/types/chat_completions.py,sha256=xRTHBbDJDbz0HgW042WX3csQDolhjEuO81w0rzFSeBU,3691
 seekrai/types/common.py,sha256=OH3l3u-0_5oz1KYrcHMybFESzivDySocYlJAsLSLOWU,1940
 seekrai/types/completions.py,sha256=lm9AFdZR3Xg5AHPkV-qETHikkwMJmkHrLGr5GG-YR-M,2171
@@ -28,18 +28,19 @@ seekrai/types/deployments.py,sha256=n7_t7DEeBSC8cDJSjIfvWtcgUql1DaEn89zuGYN_RaI,
 seekrai/types/embeddings.py,sha256=OANoLNOs0aceS8NppVvvcNYQbF7-pAOAmcr30pw64OU,749
 seekrai/types/error.py,sha256=uTKISs9aRC4_6zwirtNkanxepN8KY-SqCq0kNbfZylQ,370
 seekrai/types/files.py,sha256=XmtiM6d9i3tnYS-Kii3QpxZJRqemJi2rvLJ32GsECXQ,2602
-seekrai/types/finetune.py,sha256=Izh7NfaW7pGzSN5GpWYSaZpkHV2TilE1CF62DrB8vE0,6089
+seekrai/types/finetune.py,sha256=VNuAZx-Lq4Q55hY47obWnFLDr9if93v5Lwr_qcyCtfU,6124
 seekrai/types/images.py,sha256=Fusj8OhVYFsT8kz636lRGGivLbPXo_ZNgakKwmzJi3U,914
+seekrai/types/ingestion.py,sha256=GqDzPjEA_rtiiDKpEmTOXOitZSSKG-JkH1siII5m8oA,152
 seekrai/types/models.py,sha256=1ZfW9WwayApkISRizDntjkWhYNv-wkbrRVIfHn2QuC4,1242
 seekrai/types/projects.py,sha256=JFgpZdovia8Orcnhp6QkIEAXzyPCfKT_bUiwjxUaHHQ,670
 seekrai/utils/__init__.py,sha256=dfbiYEc47EBVRkq6C4O9y6tTGuPuV3LbV3__v01Mbds,658
 seekrai/utils/_log.py,sha256=Cayw5B394H2WGVTXPXS2AN8znQdxsgrLqADXgqmokvU,1649
 seekrai/utils/api_helpers.py,sha256=0Y8BblNIr9h_R12zdmhkxgTlxgoRkbq84QNi4nNWGu8,2385
-seekrai/utils/files.py,sha256=B61Pwra49MVVWjPtdkx4hBtAuUe9UI63hdNus87Uq0o,7164
+seekrai/utils/files.py,sha256=7ixn_hgV-6pEhYqLyOp-EN0o8c1CzUwJzX9n3PQ5oqo,7164
 seekrai/utils/tools.py,sha256=jgJTL-dOIouDbEJLdQpQfpXhqaz_poQYS52adyUtBjo,1781
 seekrai/version.py,sha256=q6iGQVFor8zXiPP5F-3vy9TndOxKv5JXbaNJ2kdOQws,125
-seekrai-0.3.2.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-seekrai-0.3.2.dist-info/METADATA,sha256=M1kwCRaiObFvV3IpKSYII0kx3_FqGyPRZ8tU1ukz3e8,4748
-seekrai-0.3.2.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
-seekrai-0.3.2.dist-info/entry_points.txt,sha256=N49yOEGi1sK7Xr13F_rkkcOxQ88suyiMoOmRhUHTZ_U,48
-seekrai-0.3.2.dist-info/RECORD,,
+seekrai-0.4.1.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+seekrai-0.4.1.dist-info/METADATA,sha256=28ojGXBP-65YoZv0VfwQ__mN76eC_I53aSso7caRz_E,4748
+seekrai-0.4.1.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
+seekrai-0.4.1.dist-info/entry_points.txt,sha256=N49yOEGi1sK7Xr13F_rkkcOxQ88suyiMoOmRhUHTZ_U,48
+seekrai-0.4.1.dist-info/RECORD,,