Vertex AI Text Embedding

Imports the VertexTextEmbedding class and initializes an instance named embed_model with a specified project and location. Uses APPLICATION_DEFAULT_CREDENTIALS if no credentials is specified. The default model is textembedding-gecko@003 in document retrival mode.

python

from llama_index.embeddings.vertex import VertexTextEmbedding

embed_model = VertexTextEmbedding(project="speedy-atom-413006", location="us-central1")

python

embed_model.dict()

Document and Query Retrival

python

embed_text_result = embed_model.get_text_embedding("Hello World!")

python

embed_text_result[:5]

python

embed_query_result = embed_model.get_query_embedding("Hello World!")

python

embed_query_result[:5]

python

from llama_index.core.base.embeddings.base import SimilarityMode

embed_model.similarity(
    embed_text_result, embed_query_result, SimilarityMode.DOT_PRODUCT
)

Using the async interface

python

import nest_asyncio

nest_asyncio.apply()

result = await embed_model.aget_text_embedding("Hello World!")

python

result[:5]