gcloud alpha ai-platform jobs submit prediction - start an AI Platform batch prediction job
gcloud alpha ai-platform jobs submit prediction JOB --data-format=DATA_FORMAT --input-paths=INPUT_PATH,[INPUT_PATH,...] --output-path=OUTPUT_PATH --region=REGION (--model=MODEL | --model-dir=MODEL_DIR) [--batch-size=BATCH_SIZE] [--labels=[KEY=VALUE,...]] [--max-worker-count=MAX_WORKER_COUNT] [--runtime-version=RUNTIME_VERSION] [--signature-name=SIGNATURE_NAME] [--version=VERSION] [--accelerator-count=ACCELERATOR_COUNT --accelerator-type=ACCELERATOR_TYPE] [GCLOUD_WIDE_FLAG ...]
(ALPHA) Start an AI Platform batch prediction job.
- JOB
Name of the batch prediction job.
- --data-format=DATA_FORMAT
Data format of the input files. DATA_FORMAT must be one of:
- text
Text and JSON files; for text files, see https://www.tensorflow.org/guide/datasets#consuming_text_data, for JSON files, see https://cloud.google.com/ai-platform/prediction/docs/overview#batch_prediction_input_data
- tf-record
TFRecord files; see https://www.tensorflow.org/guide/datasets#consuming_tfrecord_data
- tf-record-gzip
GZIP-compressed TFRecord files.
- --input-paths=INPUT_PATH,[INPUT_PATH,...]
Cloud Storage paths to the instances to run prediction on.
Wildcards (*) accepted at the end of a path. More than one path can be specified if multiple file patterns are needed. For example,
gs://my-bucket/instances*,gs://my-bucket/other-instances1
will match any objects whose names start with instances in my-bucket as well as the other-instances1 bucket, while
gs://my-bucket/instance-dir/*
will match any objects in the instance-dir "directory" (since directories aren't a first-class Cloud Storage concept) of my-bucket.
- --output-path=OUTPUT_PATH
Cloud Storage path to which to save the output. Example: gs://my-bucket/output.
- --region=REGION
The Compute Engine region to run the job in.
- Exactly one of these must be specified:
- --model=MODEL
Name of the model to use for prediction.
- --model-dir=MODEL_DIR
Cloud Storage location where the model files are located.
- --batch-size=BATCH_SIZE
The number of records per batch. The service will buffer batch_size number of records in memory before invoking TensorFlow. Defaults to 64 if not specified.
- --labels=[KEY=VALUE,...]
List of label KEY=VALUE pairs to add.
Keys must start with a lowercase character and contain only hyphens (-), underscores (_), lowercase characters, and numbers. Values must contain only hyphens (-), underscores (_), lowercase characters, and numbers.
- --max-worker-count=MAX_WORKER_COUNT
The maximum number of workers to be used for parallel processing. Defaults to 10 if not specified.
- --runtime-version=RUNTIME_VERSION
AI Platform runtime version for this job. Must be specified unless --master-image-uri is specified instead. It is defined in documentation along with the list of supported versions: https://cloud.google.com/ai-platform/prediction/docs/runtime-version-list
- --signature-name=SIGNATURE_NAME
Name of the signature defined in the SavedModel to use for this job. Defaults to DEFAULT_SERVING_SIGNATURE_DEF_KEY in https://www.tensorflow.org/api_docs/python/tf/compat/v1/saved_model/signature_constants, which is "serving_default". Only applies to TensorFlow models.
- --version=VERSION
Model version to be used.
This flag may only be given if --model is specified. If unspecified, the default version of the model will be used. To list versions for a model, run
$ gcloud ai-platform versions list
- Accelerator Configuration.
- --accelerator-count=ACCELERATOR_COUNT
The number of accelerators to attach to the machines. Must be >= 1.
This flag argument must be specified if any of the other arguments in this group are specified.
- --accelerator-type=ACCELERATOR_TYPE
The available types of accelerators. ACCELERATOR_TYPE must be one of:
- nvidia-tesla-k80
NVIDIA Tesla K80 GPU
- nvidia-tesla-p100
NVIDIA Tesla P100 GPU.
This flag argument must be specified if any of the other arguments in this group are specified.
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.
This command is currently in alpha and might change without notice. If this command fails with API permission errors despite specifying the correct project, you might be trying to access an API with an invitation-only early access allowlist. These variants are also available:
$ gcloud ai-platform jobs submit prediction
$ gcloud beta ai-platform jobs submit prediction