Shortcuts

determined

determined.Categorical(vals: List[Any]) → Dict

A hyperparameter configuration for a discrete uniform distribution over a list of values.

Parameters

vals – A list of JSON-serializable values (int, float, str, or a some combination of those types in nested lists or dictionaries)

Returns

A dictionary representing the configuration.

determined.Constant(value: Any) → Dict

A hyperparameter configuration for a constant value.

Parameters

value – A JSON-serializable value (int, float, str, or a some combination of those types in a list or dictionary)

Returns

A dictionary representing the configuration.

determined.Double(minval: float, maxval: float) → Dict

A hyperparameter configuration for a continuous uniform distribution over float values.

Parameters
  • minval – Minimum float value, inclusive.

  • maxval – Maximum float value, inclusive.

Returns

A dictionary representing the configuration.

determined.Integer(minval: int, maxval: int) → Dict

A hyperparameter configuration for a discrete uniform distribution over integers.

Parameters
  • minval – Minimum integer value, inclusive.

  • maxval – Maximum integer value, inclusive.

Returns

A dictionary representing the configuration.

determined.Log(minval: float, maxval: float, base: int = 10) → Any

A hyperparameter configuration for a log uniform distribution over float values.

Parameters
  • minval – The minimum exponent to be used in the distribution. The minimum value of the hyperparameter will be base ^ minval.

  • maxval – The minimum exponent to be used in the distribution. The maximum value of the hyperparameter will be base ^ maxval.

  • base – The logarithm base to use for the distribution (default: 10)

Returns

A dictionary representing the configuration.