Previous page Next page

On the page:


General info

Method

PUT /IP address:port/<api_ver>/task

DescriptionThe query is used to update the training parameters of a specific task. When sending data, you can use two content types: multipart/form-data or application/json

Sample request

PUT http://localhost:5000/v1.3/task
Content-Type: application/json
 
{
  "task_id": "6407bfd7144f37e141b786e2acf4faf2822cc48f296f5047e58c9e536874f4ec",
  "limit_by_epochs": {
    "enabled": true,
    "value": 50
  },
  "batch_size": 32,
  "lr": 0.001,
  "limit_by_time": {
    "enabled": true,
    "value": 3600
  },
  "limit_by_accuracy": {
    "enabled": true,
    "value": 95
  }
}

where:

ParameterDescription
task_id

A unique identifier of the task (uid), which can be obtained using a task list query (see Get the list of tasks)

limit_by_epochs


Limitation by the number of epochs that the neural network must complete during training

enabled

Set to true if it is necessary to stop training when the number of epochs is reached

value

The desired number of epochs

batch_size

Batch size (number of images transferred to the neural network at one time)

lr

Neural network training speed

limit_by_time


Time limit for training the neural network

enabled

Set to true if it is necessary to limit the training time

value

Maximum training duration in seconds

limit_by_accuracy


Desired accuracy of the trained neural network

enabled

Set to true if it is necessary to stop training when the specified accuracy is reached

value

The value of the desired accuracy of neural network training

Sample response

{ "message": "Done" }

Response in case of error:

{ "error": "error description" }


  • No labels