1
0
Fork 0
mirror of https://github.com/imjasonh/gcloud-help synced 2026-07-13 00:18:35 +00:00
gcloud-help/gcloud/beta/vector-search/collections/data-objects/search
2026-04-08 12:08:12 +00:00

237 lines
8.4 KiB
Text

NAME
gcloud beta vector-search collections data-objects search - search data
objects from a Vector Search collection
SYNOPSIS
gcloud beta vector-search collections data-objects search
--collection=COLLECTION --location=LOCATION
(--semantic-search-field=SEMANTIC_SEARCH_FIELD
--semantic-search-text=SEMANTIC_SEARCH_TEXT
--semantic-task-type=SEMANTIC_TASK_TYPE
| --text-search-data-fields=[DATA_FIELD_NAME,...]
--text-search-text=TEXT_SEARCH_TEXT
| [--vector-from-file=VECTOR_FROM_FILE
--vector-search-field=VECTOR_SEARCH_FIELD
: --distance-metric=DISTANCE_METRIC]) [--json-filter=JSON_FILTER]
[--top-k=TOP_K]
[--output-data-fields=[DATA_OUTPUT_FIELD,...]
--output-metadata-fields=[METADATA_OUTPUT_FIELD,...]
--output-vector-fields=[VECTOR_OUTPUT_FIELD,...]]
[--use-knn | [--use-index=INDEX
: --dense-scann-initial-candidate-count=CANDIDATE_COUNT
--dense-scann-search-leaves-pct=PERCENTAGE]] [GCLOUD_WIDE_FLAG ...]
DESCRIPTION
(BETA) Search data objects from a Vector Search collection.
EXAMPLES
To search data objects from collection my-collection in location
us-central1 using text search and return 10 results, run:
$ gcloud beta vector-search collections data-objects search \
--collection=my-collection --location=us-central1 \
--text-search-text="test" \
--text-search-data-fields="text_field" --top-k=10
To search data objects from collection my-collection in location
us-central1 using semantic search and return 5 results, run:
$ gcloud beta vector-search collections data-objects search \
--collection=my-collection --location=us-central1 \
--semantic-search-text="sci-fi" \
--semantic-search-field="plot_embedding" \
--semantic-task-type="retrieval-query" --top-k=5
To search data objects from collection my-collection in location
us-central1 using vector search with an index hint and return 7 results,
run:
$ gcloud beta vector-search collections data-objects search \
--collection=my-collection --location=us-central1 \
--vector-search-field="genre_embedding" \
--vector-from-file="vector.json" --use-index="my-index" \
--top-k=7
To search data objects from collection my-collection in location
us-central1 using vector search with KNN for exact results, run:
$ gcloud beta vector-search collections data-objects search \
--collection=my-collection --location=us-central1 \
--vector-search-field="genre_embedding" \
--vector-from-file="vector.json" --use-knn --top-k=7
REQUIRED FLAGS
--collection=COLLECTION
The collection to search data objects from.
--location=LOCATION
Location of the collection.
Search type
Exactly one of these must be specified:
Semantic Search
--semantic-search-field=SEMANTIC_SEARCH_FIELD
The vector field to search.
This flag argument must be specified if any of the other arguments in
this group are specified.
--semantic-search-text=SEMANTIC_SEARCH_TEXT
The query text for semantic search.
This flag argument must be specified if any of the other arguments in
this group are specified.
--semantic-task-type=SEMANTIC_TASK_TYPE
The task type of the query embedding for semantic search.
SEMANTIC_TASK_TYPE must be one of:
classification
Specifies that the given text will be classified.
clustering
Specifies that the embeddings will be used for clustering.
code-retrieval-query
Specifies that the embeddings will be used for code retrieval.
fact-verification
Specifies that the embeddings will be used for fact verification.
question-answering
Specifies that the embeddings will be used for question
answering.
retrieval-document
Specifies the given text is a document from the corpus being
searched.
retrieval-query
Specifies the given text is a query in a search/retrieval
setting.
semantic-similarity
Specifies the given text will be used for STS.
This flag argument must be specified if any of the other arguments in
this group are specified.
Text Search
--text-search-data-fields=[DATA_FIELD_NAME,...]
The data field names to search.
This flag argument must be specified if any of the other arguments in
this group are specified.
--text-search-text=TEXT_SEARCH_TEXT
The query text for text search.
This flag argument must be specified if any of the other arguments in
this group are specified.
Vector Search
--vector-from-file=VECTOR_FROM_FILE
Path to a JSON file containing dense or sparse vector to search with.
▸ Example file content for dense vector:
{
"dense": {
"values": [
0.7,
0.6,
0.5,
0.4
]
}
}
▸ Example file content for sparse vector:
{
"sparse": {
"indices": [1, 5, 10],
"values": [0.1, 0.5, 0.21]
}
}
This flag argument must be specified if any of the other arguments in
this group are specified.
--vector-search-field=VECTOR_SEARCH_FIELD
The vector field to search.
This flag argument must be specified if any of the other arguments in
this group are specified.
--distance-metric=DISTANCE_METRIC
The distance metric to use for the KNN search. If not specified,
dot-product will be used as the default. DISTANCE_METRIC must be one
of:
cosine-distance
Cosine distance metric.
dot-product
Dot product distance metric.
OPTIONAL FLAGS
--json-filter=JSON_FILTER
A filter expression in JSON format to apply to the search, e.g.
'{"genre": {"$eq": "sci-fi"}}'.
--top-k=TOP_K
The number of nearest neighbors to return. Default is 10.
Output fields
--output-data-fields=[DATA_OUTPUT_FIELD,...]
List of data fields to include in the output. Use * to include all data
fields.
--output-metadata-fields=[METADATA_OUTPUT_FIELD,...]
List of metadata fields to include in the output. Use * to include all
metadata fields.
--output-vector-fields=[VECTOR_OUTPUT_FIELD,...]
List of vector fields to include in the output. Use * to include all
vector fields.
Search Hint
At most one of these can be specified:
--use-knn
If set to true, the search will use the system's default K-Nearest
Neighbor (KNN) index engine. This flag is compatible only with
Semantic Search and Vector Search.
Or at least one of these can be specified:
Use Index Options
--use-index=INDEX
Full resource name or ID of the index to use for the search. This
flag is compatible only with Semantic Search and Vector Search.
This flag argument must be specified if any of the other arguments
in this group are specified.
--dense-scann-initial-candidate-count=CANDIDATE_COUNT
The number of initial candidates for dense ScaNN.
--dense-scann-search-leaves-pct=PERCENTAGE
The percentage of leaves to search for dense ScaNN, in the range
[0, 100].
GCLOUD WIDE FLAGS
These flags are available to all commands: --access-token-file, --account,
--billing-project, --configuration, --flags-file, --flatten, --format,
--help, --impersonate-service-account, --log-http, --project, --quiet,
--trace-token, --user-output-enabled, --verbosity.
Run $ gcloud help for details.
NOTES
This command is currently in beta and might change without notice. This
variant is also available:
$ gcloud vector-search collections data-objects search