embedding_type may be mandatory in some cases but is marked as optional.

[wj3171868]

Related to pydantic/pydantic-ai#4044

SDK Version (required)

Python: 3.11
Pydantic AI: 1.40.0
LLM provider SDK: cohere 5.20.1

Describe the bug
"embedding_type" field on the "embed" method may be mandatory in some cases but is marked as optional.

Exception when embedding type is not specified:
File cohere/core/unchecked_base_model.py line 304 in construct_type TypeError: cohere.core.unchecked_base_model.UncheckedBaseModel.model_construct() argument after ** must be a mapping, not a list

From issue with Pydantic AI:

Cohere embedding may require an "embedding_types" attribute when calling the "embed" method of their client in > some cases. It appears this value cannot be set in pydantic-ai interaction with the cohere client.
It was noted there that the "embedding_type" field is marked as optional

The fix may be set the "embedding_type" to default to "float". If "float" is an assumed default. The exception noted above could be caught earlier or avoided.

Example Code:
client.embed( model=self.model_name, texts=inputs, output_dimension=settings.get('dimensions'), input_type=cohere_input_type, max_tokens=settings.get('cohere_max_tokens'), truncate=settings.get('cohere_truncate', 'NONE'), request_options=request_options, embedding_types=["float"] # hack: added as workaround )

Screenshots
If applicable, add screenshots to help explain your problem.