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.

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

Make normalizeVector(v) in place #81

Closed
Show file tree
Hide file tree
Changes from all commits
Commits
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
2 changes: 1 addition & 1 deletion collection.go
Original file line number Diff line number Diff line change
Expand Up @@ -225,7 +225,7 @@ func (c *Collection) AddDocument(ctx context.Context, doc Document) error {
doc.Embedding = embedding
} else {
if !isNormalized(doc.Embedding) {
doc.Embedding = normalizeVector(doc.Embedding)
normalizeVectorInPlace(doc.Embedding)
}
}

Expand Down
4 changes: 2 additions & 2 deletions collection_test.go
Original file line number Diff line number Diff line change
Expand Up @@ -567,7 +567,7 @@ func benchmarkCollection_Query(b *testing.B, n int, withContent bool) {
qv[j] = r.Float32()
}
// The document embeddings are normalized, so the query must be normalized too.
qv = normalizeVector(qv)
normalizeVectorInPlace(qv)

// Create collection
db := NewDB()
Expand All @@ -590,7 +590,7 @@ func benchmarkCollection_Query(b *testing.B, n int, withContent bool) {
for j := 0; j < d; j++ {
v[j] = r.Float32()
}
v = normalizeVector(v)
normalizeVectorInPlace(v)

// Add document with some metadata and content depending on parameter.
// When providing embeddings, the embedding func is not called.
Expand Down
2 changes: 1 addition & 1 deletion embed_cohere.go
Original file line number Diff line number Diff line change
Expand Up @@ -159,7 +159,7 @@ func NewEmbeddingFuncCohere(apiKey string, model EmbeddingModelCohere) Embedding
}
})
if !checkedNormalized {
v = normalizeVector(v)
normalizeVectorInPlace(v)
}

return v, nil
Expand Down
2 changes: 1 addition & 1 deletion embed_ollama.go
Original file line number Diff line number Diff line change
Expand Up @@ -91,7 +91,7 @@ func NewEmbeddingFuncOllama(model string, baseURLOllama string) EmbeddingFunc {
}
})
if !checkedNormalized {
v = normalizeVector(v)
normalizeVectorInPlace(v)
}

return v, nil
Expand Down
5 changes: 3 additions & 2 deletions embed_openai.go
Original file line number Diff line number Diff line change
Expand Up @@ -142,7 +142,8 @@ func newEmbeddingFuncOpenAICompat(baseURL, apiKey, model string, normalized *boo
if *normalized {
return v, nil
}
return normalizeVector(v), nil
normalizeVectorInPlace(v)
return v, nil
}
checkNormalized.Do(func() {
if isNormalized(v) {
Expand All @@ -152,7 +153,7 @@ func newEmbeddingFuncOpenAICompat(baseURL, apiKey, model string, normalized *boo
}
})
if !checkedNormalized {
v = normalizeVector(v)
normalizeVectorInPlace(v)
}

return v, nil
Expand Down
15 changes: 10 additions & 5 deletions vector.go
Original file line number Diff line number Diff line change
Expand Up @@ -19,7 +19,8 @@ func cosineSimilarity(a, b []float32) (float32, error) {
}

if !isNormalized(a) || !isNormalized(b) {
a, b = normalizeVector(a), normalizeVector(b)
a = normalizeVector(a)
b = normalizeVector(b)
}
dotProduct, err := dotProduct(a, b)
if err != nil {
Expand Down Expand Up @@ -49,19 +50,23 @@ func dotProduct(a, b []float32) (float32, error) {
return dotProduct, nil
}

func normalizeVector(v []float32) []float32 {
func normalizeVectorInPlace(v []float32) {
var norm float32
for _, val := range v {
norm += val * val
}
norm = float32(math.Sqrt(float64(norm)))

res := make([]float32, len(v))
for i, val := range v {
res[i] = val / norm
v[i] = val / norm
}
}

return res
func normalizeVector(v []float32) []float32 {
r := make([]float32, len(v))
copy(r, v)
normalizeVectorInPlace(r)
return r
}

// isNormalized checks if the vector is normalized.
Expand Down