Struct ConvexAutomatedStoppingSpec

#[non_exhaustive]
pub struct ConvexAutomatedStoppingSpec {
    pub max_step_count: i64,
    pub min_step_count: i64,
    pub min_measurement_count: i64,
    pub learning_rate_parameter_name: String,
    pub use_elapsed_duration: bool,
    pub update_all_stopped_trials: Option<bool>,
    /* private fields */
}

Configuration for ConvexAutomatedStoppingSpec. When there are enough completed trials (configured by min_measurement_count), for pending trials with enough measurements and steps, the policy first computes an overestimate of the objective value at max_num_steps according to the slope of the incomplete objective value curve. No prediction can be made if the curve is completely flat. If the overestimation is worse than the best objective value of the completed trials, this pending trial will be early-stopped, but a last measurement will be added to the pending trial with max_num_steps and predicted objective value from the autoregression model.

Fields (Non-exhaustive)

Non-exhaustive structs could have additional fields added in future. Therefore, non-exhaustive structs cannot be constructed in external crates using the traditional Struct { .. } syntax; cannot be matched against without a wildcard ..; and struct update syntax will not work.

max_step_count: i64

Steps used in predicting the final objective for early stopped trials. In general, it’s set to be the same as the defined steps in training / tuning. If not defined, it will learn it from the completed trials. When use_steps is false, this field is set to the maximum elapsed seconds.

min_step_count: i64

Minimum number of steps for a trial to complete. Trials which do not have a measurement with step_count > min_step_count won’t be considered for early stopping. It’s ok to set it to 0, and a trial can be early stopped at any stage. By default, min_step_count is set to be one-tenth of the max_step_count. When use_elapsed_duration is true, this field is set to the minimum elapsed seconds.

min_measurement_count: i64

The minimal number of measurements in a Trial. Early-stopping checks will not trigger if less than min_measurement_count+1 completed trials or pending trials with less than min_measurement_count measurements. If not defined, the default value is 5.

learning_rate_parameter_name: String

The hyper-parameter name used in the tuning job that stands for learning rate. Leave it blank if learning rate is not in a parameter in tuning. The learning_rate is used to estimate the objective value of the ongoing trial.

use_elapsed_duration: bool

This bool determines whether or not the rule is applied based on elapsed_secs or steps. If use_elapsed_duration==false, the early stopping decision is made according to the predicted objective values according to the target steps. If use_elapsed_duration==true, elapsed_secs is used instead of steps. Also, in this case, the parameters max_num_steps and min_num_steps are overloaded to contain max_elapsed_seconds and min_elapsed_seconds.

update_all_stopped_trials: Option<bool>

ConvexAutomatedStoppingSpec by default only updates the trials that needs to be early stopped using a newly trained auto-regressive model. When this flag is set to True, all stopped trials from the beginning are potentially updated in terms of their final_measurement. Also, note that the training logic of autoregressive models is different in this case. Enabling this option has shown better results and this may be the default option in the future.

Implementations

impl ConvexAutomatedStoppingSpec

pub fn new() -> Self

pub fn set_max_step_count<T: Into<i64>>(self, v: T) -> Self

Sets the value of max_step_count.

pub fn set_min_step_count<T: Into<i64>>(self, v: T) -> Self

Sets the value of min_step_count.

pub fn set_min_measurement_count<T: Into<i64>>(self, v: T) -> Self

Sets the value of min_measurement_count.

pub fn set_learning_rate_parameter_name<T: Into<String>>(self, v: T) -> Self

Sets the value of learning_rate_parameter_name.

pub fn set_use_elapsed_duration<T: Into<bool>>(self, v: T) -> Self

Sets the value of use_elapsed_duration.

pub fn set_update_all_stopped_trials<T: Into<Option<bool>>>(self, v: T) -> Self

Sets the value of update_all_stopped_trials.

Trait Implementations

impl Clone for ConvexAutomatedStoppingSpec

fn clone(&self) -> ConvexAutomatedStoppingSpec

Returns a copy of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ConvexAutomatedStoppingSpec

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for ConvexAutomatedStoppingSpec

fn default() -> ConvexAutomatedStoppingSpec

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for ConvexAutomatedStoppingSpec

where ConvexAutomatedStoppingSpec: Default,

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Message for ConvexAutomatedStoppingSpec

fn typename() -> &'static str

The typename of this message.

impl PartialEq for ConvexAutomatedStoppingSpec

fn eq(&self, other: &ConvexAutomatedStoppingSpec) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Serialize for ConvexAutomatedStoppingSpec

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for ConvexAutomatedStoppingSpec