Skip to content

fix(utils): preserve embedding types across batches #797

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
OllieinCanada wants to merge 1 commit into
base: main
Choose a base branch
from
+21 −2
Open

fix(utils): preserve embedding types across batches #797

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 7 additions & 2 deletions src/cohere/utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -224,8 +224,13 @@ def merge_embed_responses(responses: typing.List[EmbedResponse]) -> EmbedRespons
for response in embeddings_type
]

# only get set keys from the pydantic model (i.e. exclude fields that are set to 'None')
fields = [x for x in get_fields(embeddings_type[0].embeddings) if getattr(embeddings_type[0].embeddings, x) is not None]
# Include a type when any batch returned it. Looking only at the first
# response can silently discard valid embeddings from later batches.
fields = [
field
for field in get_fields(embeddings_type[0].embeddings)
if any(getattr(response.embeddings, field) is not None for response in embeddings_type)
]

merged_dicts = {
field: [
Expand Down
14 changes: 14 additions & 0 deletions tests/test_embed_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -205,6 +205,20 @@ def test_merge_embeddings_by_type_with_none_field_in_later_response(self) -> Non
result = merge_embed_responses([resp1, resp2])
self.assertEqual(result.embeddings.float_, [[1.0, 2.0]]) # type: ignore

def test_merge_embeddings_by_type_keeps_field_from_later_response(self) -> None:
resp1 = EmbeddingsByTypeEmbedResponse(
response_type="embeddings_by_type", id="1",
embeddings=EmbedByTypeResponseEmbeddings(float_=[[1.0, 2.0]]))
resp2 = EmbeddingsByTypeEmbedResponse(
response_type="embeddings_by_type", id="2",
embeddings=EmbedByTypeResponseEmbeddings(
float_=[[3.0, 4.0]], int8=[[3, 4]]))

result = merge_embed_responses([resp1, resp2])

self.assertEqual(result.embeddings.float_, [[1.0, 2.0], [3.0, 4.0]]) # type: ignore
self.assertEqual(result.embeddings.int8, [[3, 4]]) # type: ignore

def test_sum_fields_if_not_none_with_none_entries(self) -> None:
# billed_units list may contain None when ApiMeta.billed_units is unset;
# sum_fields_if_not_none must skip None objects without raising AttributeError
Expand Down