Specify Embedding dimension for multimodal input

import (
	"context"
	"encoding/json"
	"fmt"
	"io"

	aiplatform "cloud.google.com/go/aiplatform/apiv1beta1"
	aiplatformpb "cloud.google.com/go/aiplatform/apiv1beta1/aiplatformpb"
	"google.golang.org/api/option"
	"google.golang.org/protobuf/encoding/protojson"
	"google.golang.org/protobuf/types/known/structpb"
)

// generateWithLowerDimension shows how to generate lower-dimensional embeddings for text and image inputs.
func generateWithLowerDimension(w io.Writer, project, location string) error {
	// location = "us-central1"
	ctx := context.Background()
	apiEndpoint := fmt.Sprintf("%s-aiplatform.googleapis.com:443", location)
	client, err := aiplatform.NewPredictionClient(ctx, option.WithEndpoint(apiEndpoint))
	if err != nil {
		return fmt.Errorf("failed to construct API client: %w", err)
	}
	defer client.Close()

	model := "multimodalembedding@001"
	endpoint := fmt.Sprintf("projects/%s/locations/%s/publishers/google/models/%s", project, location, model)

	// This is the input to the model's prediction call. For schema, see:
	// https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/multimodal-embeddings-api#request_body
	instance, err := structpb.NewValue(map[string]any{
		"image": map[string]any{
			// Image input can be provided either as a Google Cloud Storage URI or as
			// base64-encoded bytes using the "bytesBase64Encoded" field.
			"gcsUri": "gs://cloud-samples-data/vertex-ai/llm/prompts/landmark1.png",
		},
		"text": "Colosseum",
	})
	if err != nil {
		return fmt.Errorf("failed to construct request payload: %w", err)
	}

	// TODO(developer): Try different dimenions: 128, 256, 512, 1408
	outputDimensionality := 128
	params, err := structpb.NewValue(map[string]any{
		"dimension": outputDimensionality,
	})
	if err != nil {
		return fmt.Errorf("failed to construct request params: %w", err)
	}

	req := &aiplatformpb.PredictRequest{
		Endpoint: endpoint,
		// The model supports only 1 instance per request.
		Instances:  []*structpb.Value{instance},
		Parameters: params,
	}

	resp, err := client.Predict(ctx, req)
	if err != nil {
		return fmt.Errorf("failed to generate embeddings: %w", err)
	}

	instanceEmbeddingsJson, err := protojson.Marshal(resp.GetPredictions()[0])
	if err != nil {
		return fmt.Errorf("failed to convert protobuf value to JSON: %w", err)
	}
	// For response schema, see:
	// https://cloud.google.com/vertex-ai/generative-ai/docs/model-reference/multimodal-embeddings-api#response-body
	var instanceEmbeddings struct {
		ImageEmbeddings []float32 `json:"imageEmbedding"`
		TextEmbeddings  []float32 `json:"textEmbedding"`
	}
	if err := json.Unmarshal(instanceEmbeddingsJson, &instanceEmbeddings); err != nil {
		return fmt.Errorf("failed to unmarshal JSON: %w", err)
	}

	imageEmbedding := instanceEmbeddings.ImageEmbeddings
	textEmbedding := instanceEmbeddings.TextEmbeddings

	fmt.Fprintf(w, "Text embedding (length=%d): %v\n", len(textEmbedding), textEmbedding)
	fmt.Fprintf(w, "Image embedding (length=%d): %v\n", len(imageEmbedding), imageEmbedding)
	// Example response:
	// Text Embedding (length=128): [0.27469793 -0.14625867 0.022280363 ... ]
	// Image Embedding (length=128): [0.06225733 -0.040650766 0.02604402 ... ]

	return nil
}

import vertexai

from vertexai.vision_models import Image, MultiModalEmbeddingModel

# TODO(developer): Update & uncomment line below
# PROJECT_ID = "your-project-id"
vertexai.init(project=PROJECT_ID, location="us-central1")

# TODO(developer): Try different dimenions: 128, 256, 512, 1408
embedding_dimension = 128

model = MultiModalEmbeddingModel.from_pretrained("multimodalembedding@001")
image = Image.load_from_file(
    "gs://cloud-samples-data/vertex-ai/llm/prompts/landmark1.png"
)

embeddings = model.get_embeddings(
    image=image,
    contextual_text="Colosseum",
    dimension=embedding_dimension,
)

print(f"Image Embedding: {embeddings.image_embedding}")
print(f"Text Embedding: {embeddings.text_embedding}")

# Example response:
# Image Embedding: [0.0622573346, -0.0406507477, 0.0260440577, ...]
# Text Embedding: [0.27469793, -0.146258667, 0.0222803634, ...]