diff --git a/pyproject.toml b/pyproject.toml index cd1405841..973e48de1 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -1,6 +1,6 @@ [tool.poetry] name = "cohere" -version = "5.9.3" +version = "5.9.4" description = "" readme = "README.md" authors = [] diff --git a/reference.md b/reference.md index c91398c36..95ebe505b 100644 --- a/reference.md +++ b/reference.md @@ -2858,17 +2858,14 @@ If you want to learn more how to use the embedding model, have a look at the [Se
```python -from cohere import Client, ImageEmbedRequestV2 +from cohere import Client client = Client( client_name="YOUR_CLIENT_NAME", token="YOUR_TOKEN", ) client.v2.embed( - request=ImageEmbedRequestV2( - images=["string"], - model="string", - ), + model="model", ) ``` @@ -2885,7 +2882,80 @@ client.v2.embed(
-**request:** `EmbedRequestV2` +**model:** `str` + +Defaults to embed-english-v2.0 + +The identifier of the model. Smaller "light" models are faster, while larger models will perform better. [Custom models](/docs/training-custom-models) can also be supplied with their full ID. + +Available models and corresponding embedding dimensions: + +* `embed-english-v3.0` 1024 +* `embed-multilingual-v3.0` 1024 +* `embed-english-light-v3.0` 384 +* `embed-multilingual-light-v3.0` 384 + +* `embed-english-v2.0` 4096 +* `embed-english-light-v2.0` 1024 +* `embed-multilingual-v2.0` 768 + +
+
+ +
+
+ +**texts:** `typing.Optional[typing.Sequence[str]]` — An array of strings for the model to embed. Maximum number of texts per call is `96`. We recommend reducing the length of each text to be under `512` tokens for optimal quality. + +
+
+ +
+
+ +**images:** `typing.Optional[typing.Sequence[str]]` + +An array of image data URIs for the model to embed. Maximum number of images per call is `1`. + +The image must be a valid [data URI](https://developer.mozilla.org/en-US/docs/Web/URI/Schemes/data). The image must be in either `image/jpeg` or `image/png` format and has a maximum size of 5MB. + +
+
+ +
+
+ +**input_type:** `typing.Optional[EmbedInputType]` + +
+
+ +
+
+ +**embedding_types:** `typing.Optional[typing.Sequence[EmbeddingType]]` + +Specifies the types of embeddings you want to get back. Not required and default is None, which returns the Embed Floats response type. Can be one or more of the following types. + +* `"float"`: Use this when you want to get back the default float embeddings. Valid for all models. +* `"int8"`: Use this when you want to get back signed int8 embeddings. Valid for only v3 models. +* `"uint8"`: Use this when you want to get back unsigned int8 embeddings. Valid for only v3 models. +* `"binary"`: Use this when you want to get back signed binary embeddings. Valid for only v3 models. +* `"ubinary"`: Use this when you want to get back unsigned binary embeddings. Valid for only v3 models. + +
+
+ +
+
+ +**truncate:** `typing.Optional[V2EmbedRequestTruncate]` + +One of `NONE|START|END` to specify how the API will handle inputs longer than the maximum token length. + +Passing `START` will discard the start of the input. `END` will discard the end of the input. In both cases, input is discarded until the remaining input is exactly the maximum input token length for the model. + +If `NONE` is selected, when the input exceeds the maximum input token length an error will be returned.
diff --git a/src/cohere/__init__.py b/src/cohere/__init__.py index dd374d7d6..eef186d88 100644 --- a/src/cohere/__init__.py +++ b/src/cohere/__init__.py @@ -80,7 +80,6 @@ CitationStartEventDelta, CitationStartEventDeltaMessage, CitationStartStreamedChatResponseV2, - ClassificationEmbedRequestV2, ClassifyDataMetrics, ClassifyExample, ClassifyRequestTruncate, @@ -89,7 +88,6 @@ ClassifyResponseClassificationsItemClassificationType, ClassifyResponseClassificationsItemLabelsValue, ClientClosedRequestErrorBody, - ClusteringEmbedRequestV2, CompatibleEndpoint, Connector, ConnectorAuthStatus, @@ -120,7 +118,6 @@ EmbedJobStatus, EmbedJobTruncate, EmbedRequestTruncate, - EmbedRequestV2, EmbedResponse, EmbeddingType, EmbeddingsByTypeEmbedResponse, @@ -141,8 +138,6 @@ Generation, GetConnectorResponse, GetModelResponse, - ImageEmbedRequestV2, - Images, JsonObjectResponseFormat, JsonObjectResponseFormatV2, JsonResponseFormat, @@ -169,9 +164,7 @@ RerankerDataMetrics, ResponseFormat, ResponseFormatV2, - SearchDocumentEmbedRequestV2, SearchQueriesGenerationStreamedChatResponse, - SearchQueryEmbedRequestV2, SearchResultsStreamedChatResponse, SingleGeneration, SingleGenerationInStream, @@ -200,8 +193,6 @@ TextResponseFormatV2, TextSystemMessageContentItem, TextToolContent, - Texts, - TextsTruncate, TokenizeResponse, TooManyRequestsErrorBody, Tool, @@ -267,6 +258,7 @@ V2ChatRequestSafetyMode, V2ChatStreamRequestDocumentsItem, V2ChatStreamRequestSafetyMode, + V2EmbedRequestTruncate, V2RerankRequestDocumentsItem, V2RerankResponse, V2RerankResponseResultsItem, @@ -359,7 +351,6 @@ "CitationStartEventDelta", "CitationStartEventDeltaMessage", "CitationStartStreamedChatResponseV2", - "ClassificationEmbedRequestV2", "ClassifyDataMetrics", "ClassifyExample", "ClassifyRequestTruncate", @@ -372,7 +363,6 @@ "ClientClosedRequestErrorBody", "ClientEnvironment", "ClientV2", - "ClusteringEmbedRequestV2", "CompatibleEndpoint", "Connector", "ConnectorAuthStatus", @@ -409,7 +399,6 @@ "EmbedJobStatus", "EmbedJobTruncate", "EmbedRequestTruncate", - "EmbedRequestV2", "EmbedResponse", "EmbeddingType", "EmbeddingsByTypeEmbedResponse", @@ -432,8 +421,6 @@ "Generation", "GetConnectorResponse", "GetModelResponse", - "ImageEmbedRequestV2", - "Images", "InternalServerError", "JsonObjectResponseFormat", "JsonObjectResponseFormatV2", @@ -464,9 +451,7 @@ "ResponseFormat", "ResponseFormatV2", "SagemakerClient", - "SearchDocumentEmbedRequestV2", "SearchQueriesGenerationStreamedChatResponse", - "SearchQueryEmbedRequestV2", "SearchResultsStreamedChatResponse", "ServiceUnavailableError", "SingleGeneration", @@ -496,8 +481,6 @@ "TextResponseFormatV2", "TextSystemMessageContentItem", "TextToolContent", - "Texts", - "TextsTruncate", "TokenizeResponse", "TooManyRequestsError", "TooManyRequestsErrorBody", @@ -536,6 +519,7 @@ "V2ChatRequestSafetyMode", "V2ChatStreamRequestDocumentsItem", "V2ChatStreamRequestSafetyMode", + "V2EmbedRequestTruncate", "V2RerankRequestDocumentsItem", "V2RerankResponse", "V2RerankResponseResultsItem", diff --git a/src/cohere/core/client_wrapper.py b/src/cohere/core/client_wrapper.py index 10e94aea9..7d323ab41 100644 --- a/src/cohere/core/client_wrapper.py +++ b/src/cohere/core/client_wrapper.py @@ -24,7 +24,7 @@ def get_headers(self) -> typing.Dict[str, str]: headers: typing.Dict[str, str] = { "X-Fern-Language": "Python", "X-Fern-SDK-Name": "cohere", - "X-Fern-SDK-Version": "5.9.3", + "X-Fern-SDK-Version": "5.9.4", } if self._client_name is not None: headers["X-Client-Name"] = self._client_name diff --git a/src/cohere/finetuning/__init__.py b/src/cohere/finetuning/__init__.py index ea532e8b2..29eed472f 100644 --- a/src/cohere/finetuning/__init__.py +++ b/src/cohere/finetuning/__init__.py @@ -13,6 +13,7 @@ ListEventsResponse, ListFinetunedModelsResponse, ListTrainingStepMetricsResponse, + LoraTargetModules, Settings, Status, Strategy, @@ -33,6 +34,7 @@ "ListEventsResponse", "ListFinetunedModelsResponse", "ListTrainingStepMetricsResponse", + "LoraTargetModules", "Settings", "Status", "Strategy", diff --git a/src/cohere/finetuning/finetuning/__init__.py b/src/cohere/finetuning/finetuning/__init__.py index 71e235731..a8c6b7817 100644 --- a/src/cohere/finetuning/finetuning/__init__.py +++ b/src/cohere/finetuning/finetuning/__init__.py @@ -12,6 +12,7 @@ ListEventsResponse, ListFinetunedModelsResponse, ListTrainingStepMetricsResponse, + LoraTargetModules, Settings, Status, Strategy, @@ -32,6 +33,7 @@ "ListEventsResponse", "ListFinetunedModelsResponse", "ListTrainingStepMetricsResponse", + "LoraTargetModules", "Settings", "Status", "Strategy", diff --git a/src/cohere/finetuning/finetuning/types/__init__.py b/src/cohere/finetuning/finetuning/types/__init__.py index d573f6518..72f975c71 100644 --- a/src/cohere/finetuning/finetuning/types/__init__.py +++ b/src/cohere/finetuning/finetuning/types/__init__.py @@ -11,6 +11,7 @@ from .list_events_response import ListEventsResponse from .list_finetuned_models_response import ListFinetunedModelsResponse from .list_training_step_metrics_response import ListTrainingStepMetricsResponse +from .lora_target_modules import LoraTargetModules from .settings import Settings from .status import Status from .strategy import Strategy @@ -30,6 +31,7 @@ "ListEventsResponse", "ListFinetunedModelsResponse", "ListTrainingStepMetricsResponse", + "LoraTargetModules", "Settings", "Status", "Strategy", diff --git a/src/cohere/finetuning/finetuning/types/base_model.py b/src/cohere/finetuning/finetuning/types/base_model.py index 44ad40c31..fc993edb6 100644 --- a/src/cohere/finetuning/finetuning/types/base_model.py +++ b/src/cohere/finetuning/finetuning/types/base_model.py @@ -30,7 +30,7 @@ class BaseModel(UncheckedBaseModel): strategy: typing.Optional[Strategy] = pydantic.Field(default=None) """ - The fine-tuning strategy. + Deprecated: The fine-tuning strategy. """ if IS_PYDANTIC_V2: diff --git a/src/cohere/finetuning/finetuning/types/hyperparameters.py b/src/cohere/finetuning/finetuning/types/hyperparameters.py index 229549b46..2185f4c20 100644 --- a/src/cohere/finetuning/finetuning/types/hyperparameters.py +++ b/src/cohere/finetuning/finetuning/types/hyperparameters.py @@ -3,6 +3,7 @@ from ....core.unchecked_base_model import UncheckedBaseModel import typing import pydantic +from .lora_target_modules import LoraTargetModules from ....core.pydantic_utilities import IS_PYDANTIC_V2 @@ -38,6 +39,23 @@ class Hyperparameters(UncheckedBaseModel): The learning rate to be used during training. """ + lora_alpha: typing.Optional[int] = pydantic.Field(default=None) + """ + Controls the scaling factor for LoRA updates. Higher values make the + updates more impactful. + """ + + lora_rank: typing.Optional[int] = pydantic.Field(default=None) + """ + Specifies the rank for low-rank matrices. Lower ranks reduce parameters + but may limit model flexibility. + """ + + lora_target_modules: typing.Optional[LoraTargetModules] = pydantic.Field(default=None) + """ + The combination of LoRA modules to target. + """ + if IS_PYDANTIC_V2: model_config: typing.ClassVar[pydantic.ConfigDict] = pydantic.ConfigDict(extra="allow", frozen=True) # type: ignore # Pydantic v2 else: diff --git a/src/cohere/finetuning/finetuning/types/lora_target_modules.py b/src/cohere/finetuning/finetuning/types/lora_target_modules.py new file mode 100644 index 000000000..773ec7a5c --- /dev/null +++ b/src/cohere/finetuning/finetuning/types/lora_target_modules.py @@ -0,0 +1,13 @@ +# This file was auto-generated by Fern from our API Definition. + +import typing + +LoraTargetModules = typing.Union[ + typing.Literal[ + "LORA_TARGET_MODULES_UNSPECIFIED", + "LORA_TARGET_MODULES_QV", + "LORA_TARGET_MODULES_QKVO", + "LORA_TARGET_MODULES_QKVO_FFN", + ], + typing.Any, +] diff --git a/src/cohere/types/__init__.py b/src/cohere/types/__init__.py index 106547ce1..c52476d30 100644 --- a/src/cohere/types/__init__.py +++ b/src/cohere/types/__init__.py @@ -118,14 +118,6 @@ from .embed_job_status import EmbedJobStatus from .embed_job_truncate import EmbedJobTruncate from .embed_request_truncate import EmbedRequestTruncate -from .embed_request_v2 import ( - ClassificationEmbedRequestV2, - ClusteringEmbedRequestV2, - EmbedRequestV2, - ImageEmbedRequestV2, - SearchDocumentEmbedRequestV2, - SearchQueryEmbedRequestV2, -) from .embed_response import EmbedResponse, EmbeddingsByTypeEmbedResponse, EmbeddingsFloatsEmbedResponse from .embedding_type import EmbeddingType from .finetune_dataset_metrics import FinetuneDatasetMetrics @@ -149,7 +141,6 @@ from .generation import Generation from .get_connector_response import GetConnectorResponse from .get_model_response import GetModelResponse -from .images import Images from .json_response_format import JsonResponseFormat from .json_response_format_v2 import JsonResponseFormatV2 from .label_metric import LabelMetric @@ -207,8 +198,6 @@ from .summarize_response import SummarizeResponse from .system_message_content import SystemMessageContent from .system_message_content_item import SystemMessageContentItem, TextSystemMessageContentItem -from .texts import Texts -from .texts_truncate import TextsTruncate from .tokenize_response import TokenizeResponse from .too_many_requests_error_body import TooManyRequestsErrorBody from .tool import Tool @@ -310,7 +299,6 @@ "CitationStartEventDelta", "CitationStartEventDeltaMessage", "CitationStartStreamedChatResponseV2", - "ClassificationEmbedRequestV2", "ClassifyDataMetrics", "ClassifyExample", "ClassifyRequestTruncate", @@ -319,7 +307,6 @@ "ClassifyResponseClassificationsItemClassificationType", "ClassifyResponseClassificationsItemLabelsValue", "ClientClosedRequestErrorBody", - "ClusteringEmbedRequestV2", "CompatibleEndpoint", "Connector", "ConnectorAuthStatus", @@ -350,7 +337,6 @@ "EmbedJobStatus", "EmbedJobTruncate", "EmbedRequestTruncate", - "EmbedRequestV2", "EmbedResponse", "EmbeddingType", "EmbeddingsByTypeEmbedResponse", @@ -371,8 +357,6 @@ "Generation", "GetConnectorResponse", "GetModelResponse", - "ImageEmbedRequestV2", - "Images", "JsonObjectResponseFormat", "JsonObjectResponseFormatV2", "JsonResponseFormat", @@ -399,9 +383,7 @@ "RerankerDataMetrics", "ResponseFormat", "ResponseFormatV2", - "SearchDocumentEmbedRequestV2", "SearchQueriesGenerationStreamedChatResponse", - "SearchQueryEmbedRequestV2", "SearchResultsStreamedChatResponse", "SingleGeneration", "SingleGenerationInStream", @@ -430,8 +412,6 @@ "TextResponseFormatV2", "TextSystemMessageContentItem", "TextToolContent", - "Texts", - "TextsTruncate", "TokenizeResponse", "TooManyRequestsErrorBody", "Tool", diff --git a/src/cohere/types/embed_request_v2.py b/src/cohere/types/embed_request_v2.py deleted file mode 100644 index d5c36dbb5..000000000 --- a/src/cohere/types/embed_request_v2.py +++ /dev/null @@ -1,107 +0,0 @@ -# This file was auto-generated by Fern from our API Definition. - -from __future__ import annotations -from ..core.unchecked_base_model import UncheckedBaseModel -import typing -from .embedding_type import EmbeddingType -from .texts_truncate import TextsTruncate -from ..core.pydantic_utilities import IS_PYDANTIC_V2 -import pydantic -import typing_extensions -from ..core.unchecked_base_model import UnionMetadata - - -class SearchDocumentEmbedRequestV2(UncheckedBaseModel): - input_type: typing.Literal["search_document"] = "search_document" - texts: typing.List[str] - model: str - embedding_types: typing.Optional[typing.List[EmbeddingType]] = None - truncate: typing.Optional[TextsTruncate] = None - - if IS_PYDANTIC_V2: - model_config: typing.ClassVar[pydantic.ConfigDict] = pydantic.ConfigDict(extra="allow", frozen=True) # type: ignore # Pydantic v2 - else: - - class Config: - frozen = True - smart_union = True - extra = pydantic.Extra.allow - - -class SearchQueryEmbedRequestV2(UncheckedBaseModel): - input_type: typing.Literal["search_query"] = "search_query" - texts: typing.List[str] - model: str - embedding_types: typing.Optional[typing.List[EmbeddingType]] = None - truncate: typing.Optional[TextsTruncate] = None - - if IS_PYDANTIC_V2: - model_config: typing.ClassVar[pydantic.ConfigDict] = pydantic.ConfigDict(extra="allow", frozen=True) # type: ignore # Pydantic v2 - else: - - class Config: - frozen = True - smart_union = True - extra = pydantic.Extra.allow - - -class ClassificationEmbedRequestV2(UncheckedBaseModel): - input_type: typing.Literal["classification"] = "classification" - texts: typing.List[str] - model: str - embedding_types: typing.Optional[typing.List[EmbeddingType]] = None - truncate: typing.Optional[TextsTruncate] = None - - if IS_PYDANTIC_V2: - model_config: typing.ClassVar[pydantic.ConfigDict] = pydantic.ConfigDict(extra="allow", frozen=True) # type: ignore # Pydantic v2 - else: - - class Config: - frozen = True - smart_union = True - extra = pydantic.Extra.allow - - -class ClusteringEmbedRequestV2(UncheckedBaseModel): - input_type: typing.Literal["clustering"] = "clustering" - texts: typing.List[str] - model: str - embedding_types: typing.Optional[typing.List[EmbeddingType]] = None - truncate: typing.Optional[TextsTruncate] = None - - if IS_PYDANTIC_V2: - model_config: typing.ClassVar[pydantic.ConfigDict] = pydantic.ConfigDict(extra="allow", frozen=True) # type: ignore # Pydantic v2 - else: - - class Config: - frozen = True - smart_union = True - extra = pydantic.Extra.allow - - -class ImageEmbedRequestV2(UncheckedBaseModel): - input_type: typing.Literal["image"] = "image" - images: typing.List[str] - model: str - embedding_types: typing.Optional[typing.List[EmbeddingType]] = None - - if IS_PYDANTIC_V2: - model_config: typing.ClassVar[pydantic.ConfigDict] = pydantic.ConfigDict(extra="allow", frozen=True) # type: ignore # Pydantic v2 - else: - - class Config: - frozen = True - smart_union = True - extra = pydantic.Extra.allow - - -EmbedRequestV2 = typing_extensions.Annotated[ - typing.Union[ - SearchDocumentEmbedRequestV2, - SearchQueryEmbedRequestV2, - ClassificationEmbedRequestV2, - ClusteringEmbedRequestV2, - ImageEmbedRequestV2, - ], - UnionMetadata(discriminant="input_type"), -] diff --git a/src/cohere/types/images.py b/src/cohere/types/images.py deleted file mode 100644 index 5c829189a..000000000 --- a/src/cohere/types/images.py +++ /dev/null @@ -1,50 +0,0 @@ -# This file was auto-generated by Fern from our API Definition. - -from ..core.unchecked_base_model import UncheckedBaseModel -import typing -import pydantic -from .embedding_type import EmbeddingType -from ..core.pydantic_utilities import IS_PYDANTIC_V2 - - -class Images(UncheckedBaseModel): - images: typing.List[str] = pydantic.Field() - """ - An array of image data URIs for the model to embed. Maximum number of images per call is `1`. - The image must be a valid [data URI](https://developer.mozilla.org/en-US/docs/Web/URI/Schemes/data). The image must be in either `image/jpeg` or `image/png` format and has a maximum size of 5MB. - """ - - model: str = pydantic.Field() - """ - Defaults to embed-english-v2.0 - The identifier of the model. Smaller "light" models are faster, while larger models will perform better. [Custom models](/docs/training-custom-models) can also be supplied with their full ID. - Available models and corresponding embedding dimensions: - - - `embed-english-v3.0` 1024 - - `embed-multilingual-v3.0` 1024 - - `embed-english-light-v3.0` 384 - - `embed-multilingual-light-v3.0` 384 - - `embed-english-v2.0` 4096 - - `embed-english-light-v2.0` 1024 - - `embed-multilingual-v2.0` 768 - """ - - embedding_types: typing.Optional[typing.List[EmbeddingType]] = pydantic.Field(default=None) - """ - Specifies the types of embeddings you want to get back. Not required and default is None, which returns the Embed Floats response type. Can be one or more of the following types. - - - `"float"`: Use this when you want to get back the default float embeddings. Valid for all models. - - `"int8"`: Use this when you want to get back signed int8 embeddings. Valid for only v3 models. - - `"uint8"`: Use this when you want to get back unsigned int8 embeddings. Valid for only v3 models. - - `"binary"`: Use this when you want to get back signed binary embeddings. Valid for only v3 models. - - `"ubinary"`: Use this when you want to get back unsigned binary embeddings. Valid for only v3 models. - """ - - if IS_PYDANTIC_V2: - model_config: typing.ClassVar[pydantic.ConfigDict] = pydantic.ConfigDict(extra="allow", frozen=True) # type: ignore # Pydantic v2 - else: - - class Config: - frozen = True - smart_union = True - extra = pydantic.Extra.allow diff --git a/src/cohere/types/texts.py b/src/cohere/types/texts.py deleted file mode 100644 index 2f2d0978a..000000000 --- a/src/cohere/types/texts.py +++ /dev/null @@ -1,62 +0,0 @@ -# This file was auto-generated by Fern from our API Definition. - -from ..core.unchecked_base_model import UncheckedBaseModel -import typing -import pydantic -from .embedding_type import EmbeddingType -from .texts_truncate import TextsTruncate -from ..core.pydantic_utilities import IS_PYDANTIC_V2 - - -class Texts(UncheckedBaseModel): - texts: typing.List[str] = pydantic.Field() - """ - An array of strings for the model to embed. Maximum number of texts per call is `96`. We recommend reducing the length of each text to be under `512` tokens for optimal quality. - """ - - model: str = pydantic.Field() - """ - Defaults to embed-english-v2.0 - - The identifier of the model. Smaller "light" models are faster, while larger models will perform better. [Custom models](/docs/training-custom-models) can also be supplied with their full ID. - - Available models and corresponding embedding dimensions: - - - `embed-english-v3.0` 1024 - - `embed-multilingual-v3.0` 1024 - - `embed-english-light-v3.0` 384 - - `embed-multilingual-light-v3.0` 384 - - - `embed-english-v2.0` 4096 - - `embed-english-light-v2.0` 1024 - - `embed-multilingual-v2.0` 768 - """ - - embedding_types: typing.Optional[typing.List[EmbeddingType]] = pydantic.Field(default=None) - """ - Specifies the types of embeddings you want to get back. Not required and default is None, which returns the Embed Floats response type. Can be one or more of the following types. - - - `"float"`: Use this when you want to get back the default float embeddings. Valid for all models. - - `"int8"`: Use this when you want to get back signed int8 embeddings. Valid for only v3 models. - - `"uint8"`: Use this when you want to get back unsigned int8 embeddings. Valid for only v3 models. - - `"binary"`: Use this when you want to get back signed binary embeddings. Valid for only v3 models. - - `"ubinary"`: Use this when you want to get back unsigned binary embeddings. Valid for only v3 models. - """ - - truncate: typing.Optional[TextsTruncate] = pydantic.Field(default=None) - """ - One of `NONE|START|END` to specify how the API will handle inputs longer than the maximum token length. - - Passing `START` will discard the start of the input. `END` will discard the end of the input. In both cases, input is discarded until the remaining input is exactly the maximum input token length for the model. - - If `NONE` is selected, when the input exceeds the maximum input token length an error will be returned. - """ - - if IS_PYDANTIC_V2: - model_config: typing.ClassVar[pydantic.ConfigDict] = pydantic.ConfigDict(extra="allow", frozen=True) # type: ignore # Pydantic v2 - else: - - class Config: - frozen = True - smart_union = True - extra = pydantic.Extra.allow diff --git a/src/cohere/types/texts_truncate.py b/src/cohere/types/texts_truncate.py deleted file mode 100644 index b0e2faf0e..000000000 --- a/src/cohere/types/texts_truncate.py +++ /dev/null @@ -1,5 +0,0 @@ -# This file was auto-generated by Fern from our API Definition. - -import typing - -TextsTruncate = typing.Union[typing.Literal["NONE", "START", "END"], typing.Any] diff --git a/src/cohere/v2/__init__.py b/src/cohere/v2/__init__.py index 4a904803f..f84be0915 100644 --- a/src/cohere/v2/__init__.py +++ b/src/cohere/v2/__init__.py @@ -5,6 +5,7 @@ V2ChatRequestSafetyMode, V2ChatStreamRequestDocumentsItem, V2ChatStreamRequestSafetyMode, + V2EmbedRequestTruncate, V2RerankRequestDocumentsItem, V2RerankResponse, V2RerankResponseResultsItem, @@ -16,6 +17,7 @@ "V2ChatRequestSafetyMode", "V2ChatStreamRequestDocumentsItem", "V2ChatStreamRequestSafetyMode", + "V2EmbedRequestTruncate", "V2RerankRequestDocumentsItem", "V2RerankResponse", "V2RerankResponseResultsItem", diff --git a/src/cohere/v2/client.py b/src/cohere/v2/client.py index b504acfb5..a462ca464 100644 --- a/src/cohere/v2/client.py +++ b/src/cohere/v2/client.py @@ -35,7 +35,9 @@ from .types.v2chat_request_documents_item import V2ChatRequestDocumentsItem from .types.v2chat_request_safety_mode import V2ChatRequestSafetyMode from ..types.chat_response import ChatResponse -from ..types.embed_request_v2 import EmbedRequestV2 +from ..types.embed_input_type import EmbedInputType +from ..types.embedding_type import EmbeddingType +from .types.v2embed_request_truncate import V2EmbedRequestTruncate from ..types.embed_by_type_response import EmbedByTypeResponse from .types.v2rerank_request_documents_item import V2RerankRequestDocumentsItem from .types.v2rerank_response import V2RerankResponse @@ -663,7 +665,15 @@ def chat( raise ApiError(status_code=_response.status_code, body=_response_json) def embed( - self, *, request: EmbedRequestV2, request_options: typing.Optional[RequestOptions] = None + self, + *, + model: str, + texts: typing.Optional[typing.Sequence[str]] = OMIT, + images: typing.Optional[typing.Sequence[str]] = OMIT, + input_type: typing.Optional[EmbedInputType] = OMIT, + embedding_types: typing.Optional[typing.Sequence[EmbeddingType]] = OMIT, + truncate: typing.Optional[V2EmbedRequestTruncate] = OMIT, + request_options: typing.Optional[RequestOptions] = None, ) -> EmbedByTypeResponse: """ This endpoint returns text embeddings. An embedding is a list of floating point numbers that captures semantic information about the text that it represents. @@ -674,7 +684,47 @@ def embed( Parameters ---------- - request : EmbedRequestV2 + model : str + Defaults to embed-english-v2.0 + + The identifier of the model. Smaller "light" models are faster, while larger models will perform better. [Custom models](/docs/training-custom-models) can also be supplied with their full ID. + + Available models and corresponding embedding dimensions: + + * `embed-english-v3.0` 1024 + * `embed-multilingual-v3.0` 1024 + * `embed-english-light-v3.0` 384 + * `embed-multilingual-light-v3.0` 384 + + * `embed-english-v2.0` 4096 + * `embed-english-light-v2.0` 1024 + * `embed-multilingual-v2.0` 768 + + texts : typing.Optional[typing.Sequence[str]] + An array of strings for the model to embed. Maximum number of texts per call is `96`. We recommend reducing the length of each text to be under `512` tokens for optimal quality. + + images : typing.Optional[typing.Sequence[str]] + An array of image data URIs for the model to embed. Maximum number of images per call is `1`. + + The image must be a valid [data URI](https://developer.mozilla.org/en-US/docs/Web/URI/Schemes/data). The image must be in either `image/jpeg` or `image/png` format and has a maximum size of 5MB. + + input_type : typing.Optional[EmbedInputType] + + embedding_types : typing.Optional[typing.Sequence[EmbeddingType]] + Specifies the types of embeddings you want to get back. Not required and default is None, which returns the Embed Floats response type. Can be one or more of the following types. + + * `"float"`: Use this when you want to get back the default float embeddings. Valid for all models. + * `"int8"`: Use this when you want to get back signed int8 embeddings. Valid for only v3 models. + * `"uint8"`: Use this when you want to get back unsigned int8 embeddings. Valid for only v3 models. + * `"binary"`: Use this when you want to get back signed binary embeddings. Valid for only v3 models. + * `"ubinary"`: Use this when you want to get back unsigned binary embeddings. Valid for only v3 models. + + truncate : typing.Optional[V2EmbedRequestTruncate] + One of `NONE|START|END` to specify how the API will handle inputs longer than the maximum token length. + + Passing `START` will discard the start of the input. `END` will discard the end of the input. In both cases, input is discarded until the remaining input is exactly the maximum input token length for the model. + + If `NONE` is selected, when the input exceeds the maximum input token length an error will be returned. request_options : typing.Optional[RequestOptions] Request-specific configuration. @@ -686,23 +736,27 @@ def embed( Examples -------- - from cohere import Client, ImageEmbedRequestV2 + from cohere import Client client = Client( client_name="YOUR_CLIENT_NAME", token="YOUR_TOKEN", ) client.v2.embed( - request=ImageEmbedRequestV2( - images=["string"], - model="string", - ), + model="model", ) """ _response = self._client_wrapper.httpx_client.request( "v2/embed", method="POST", - json=convert_and_respect_annotation_metadata(object_=request, annotation=EmbedRequestV2, direction="write"), + json={ + "texts": texts, + "images": images, + "model": model, + "input_type": input_type, + "embedding_types": embedding_types, + "truncate": truncate, + }, request_options=request_options, omit=OMIT, ) @@ -1672,7 +1726,15 @@ async def main() -> None: raise ApiError(status_code=_response.status_code, body=_response_json) async def embed( - self, *, request: EmbedRequestV2, request_options: typing.Optional[RequestOptions] = None + self, + *, + model: str, + texts: typing.Optional[typing.Sequence[str]] = OMIT, + images: typing.Optional[typing.Sequence[str]] = OMIT, + input_type: typing.Optional[EmbedInputType] = OMIT, + embedding_types: typing.Optional[typing.Sequence[EmbeddingType]] = OMIT, + truncate: typing.Optional[V2EmbedRequestTruncate] = OMIT, + request_options: typing.Optional[RequestOptions] = None, ) -> EmbedByTypeResponse: """ This endpoint returns text embeddings. An embedding is a list of floating point numbers that captures semantic information about the text that it represents. @@ -1683,7 +1745,47 @@ async def embed( Parameters ---------- - request : EmbedRequestV2 + model : str + Defaults to embed-english-v2.0 + + The identifier of the model. Smaller "light" models are faster, while larger models will perform better. [Custom models](/docs/training-custom-models) can also be supplied with their full ID. + + Available models and corresponding embedding dimensions: + + * `embed-english-v3.0` 1024 + * `embed-multilingual-v3.0` 1024 + * `embed-english-light-v3.0` 384 + * `embed-multilingual-light-v3.0` 384 + + * `embed-english-v2.0` 4096 + * `embed-english-light-v2.0` 1024 + * `embed-multilingual-v2.0` 768 + + texts : typing.Optional[typing.Sequence[str]] + An array of strings for the model to embed. Maximum number of texts per call is `96`. We recommend reducing the length of each text to be under `512` tokens for optimal quality. + + images : typing.Optional[typing.Sequence[str]] + An array of image data URIs for the model to embed. Maximum number of images per call is `1`. + + The image must be a valid [data URI](https://developer.mozilla.org/en-US/docs/Web/URI/Schemes/data). The image must be in either `image/jpeg` or `image/png` format and has a maximum size of 5MB. + + input_type : typing.Optional[EmbedInputType] + + embedding_types : typing.Optional[typing.Sequence[EmbeddingType]] + Specifies the types of embeddings you want to get back. Not required and default is None, which returns the Embed Floats response type. Can be one or more of the following types. + + * `"float"`: Use this when you want to get back the default float embeddings. Valid for all models. + * `"int8"`: Use this when you want to get back signed int8 embeddings. Valid for only v3 models. + * `"uint8"`: Use this when you want to get back unsigned int8 embeddings. Valid for only v3 models. + * `"binary"`: Use this when you want to get back signed binary embeddings. Valid for only v3 models. + * `"ubinary"`: Use this when you want to get back unsigned binary embeddings. Valid for only v3 models. + + truncate : typing.Optional[V2EmbedRequestTruncate] + One of `NONE|START|END` to specify how the API will handle inputs longer than the maximum token length. + + Passing `START` will discard the start of the input. `END` will discard the end of the input. In both cases, input is discarded until the remaining input is exactly the maximum input token length for the model. + + If `NONE` is selected, when the input exceeds the maximum input token length an error will be returned. request_options : typing.Optional[RequestOptions] Request-specific configuration. @@ -1697,7 +1799,7 @@ async def embed( -------- import asyncio - from cohere import AsyncClient, ImageEmbedRequestV2 + from cohere import AsyncClient client = AsyncClient( client_name="YOUR_CLIENT_NAME", @@ -1707,10 +1809,7 @@ async def embed( async def main() -> None: await client.v2.embed( - request=ImageEmbedRequestV2( - images=["string"], - model="string", - ), + model="model", ) @@ -1719,7 +1818,14 @@ async def main() -> None: _response = await self._client_wrapper.httpx_client.request( "v2/embed", method="POST", - json=convert_and_respect_annotation_metadata(object_=request, annotation=EmbedRequestV2, direction="write"), + json={ + "texts": texts, + "images": images, + "model": model, + "input_type": input_type, + "embedding_types": embedding_types, + "truncate": truncate, + }, request_options=request_options, omit=OMIT, ) diff --git a/src/cohere/v2/types/__init__.py b/src/cohere/v2/types/__init__.py index 529151db8..fff15903b 100644 --- a/src/cohere/v2/types/__init__.py +++ b/src/cohere/v2/types/__init__.py @@ -4,6 +4,7 @@ from .v2chat_request_safety_mode import V2ChatRequestSafetyMode from .v2chat_stream_request_documents_item import V2ChatStreamRequestDocumentsItem from .v2chat_stream_request_safety_mode import V2ChatStreamRequestSafetyMode +from .v2embed_request_truncate import V2EmbedRequestTruncate from .v2rerank_request_documents_item import V2RerankRequestDocumentsItem from .v2rerank_response import V2RerankResponse from .v2rerank_response_results_item import V2RerankResponseResultsItem @@ -14,6 +15,7 @@ "V2ChatRequestSafetyMode", "V2ChatStreamRequestDocumentsItem", "V2ChatStreamRequestSafetyMode", + "V2EmbedRequestTruncate", "V2RerankRequestDocumentsItem", "V2RerankResponse", "V2RerankResponseResultsItem", diff --git a/src/cohere/v2/types/v2embed_request_truncate.py b/src/cohere/v2/types/v2embed_request_truncate.py new file mode 100644 index 000000000..807e9939d --- /dev/null +++ b/src/cohere/v2/types/v2embed_request_truncate.py @@ -0,0 +1,5 @@ +# This file was auto-generated by Fern from our API Definition. + +import typing + +V2EmbedRequestTruncate = typing.Union[typing.Literal["NONE", "START", "END"], typing.Any]