The default remote batch environment is "WolframBatch":

Submit a job using the default "WolframBatch" environment. RemoteBatchSubmit returns a RemoteBatchJobObject that allows you to manage and view the state of the job:

Query the job status:

Query the job status again, after it has completed:

Retrieve the job result:

Run a batch job on a computer with a large memory capacity:

Run a batch job with a given service credit budget:

Run a batch job with a given maximum runtime:

Receive SMS text notifications for job state change events:

Receive text notifications when the job starts and finishes, and emails every 2 hours while it runs:

Do not produce any notifications:

Use a default set of notifications:

Verify the detailed specification used for the default notifications:

Receive a notification when a job either succeeds or terminates:

Verify the detailed notification specification:

Jobs that exceed their memory limit are terminated:

Neural network operations use the CPU by default, even if a job’s machine class has GPU hardware available:

Specify the TargetDevice option in order to use a GPU for training or inference:

NetTrain[…,TargetDevice"GPU"] uses a single GPU, regardless of the number of GPUs available to the job:

Specify TargetDevice{"GPU",All} in order to use all available GPUs for training:

Multi-GPU inference is not supported:

Multi-GPU training fails if a batch size is not specified:

Specify the BatchSize option when performing training on multiple GPUs: