Skip to content
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

embedding/openai: fix method-dependent embedding discrepancy #357

Merged
merged 1 commit into from
Nov 18, 2023
Merged

embedding/openai: fix method-dependent embedding discrepancy #357

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
6 changes: 6 additions & 0 deletions embeddings/openai/openai.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -42,6 +42,12 @@ func (e OpenAI) EmbedDocuments(ctx context.Context, texts []string) ([][]float32
return nil, err
}

// If the size of this batch is 1, don't average/combine the vectors.
if len(texts) == 1 {
emb = append(emb, curTextEmbeddings[0])
continue
}

textLengths := make([]int, 0, len(texts))
for _, text := range texts {
textLengths = append(textLengths, len(text))
Expand Down
24 changes: 24 additions & 0 deletions embeddings/openai/openai_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,6 +27,30 @@ func TestOpenaiEmbeddings(t *testing.T) {
assert.Len(t, embeddings, 3)
}

func TestOpenaiEmbeddingsQueryVsDocuments(t *testing.T) {
// Verifies that we get the same embedding for the same string, regardless
// of which method we call.
t.Parallel()

if openaiKey := os.Getenv("OPENAI_API_KEY"); openaiKey == "" {
t.Skip("OPENAI_API_KEY not set")
}
e, err := NewOpenAI()
require.NoError(t, err)

text := "hi there"

eq, err := e.EmbedQuery(context.Background(), text)
require.NoError(t, err)

eb, err := e.EmbedDocuments(context.Background(), []string{text})
require.NoError(t, err)

// Using strict equality should be OK here because we expect the same values
// for the same string, deterministically.
assert.Equal(t, eq, eb[0])
}

func TestOpenaiEmbeddingsWithOptions(t *testing.T) {
t.Parallel()

Expand Down