Module metric

Expand description

The metric module.

Modules§

state
State module.
store
Module responsible to save and exposes data collected during training.
vision
Metrics module for vision tasks.

Structs§

AccuracyInput
The accuracy metric input type.
AccuracyMetric
The accuracy metric.
AurocInput
The AUROC metric input type.
AurocMetric
The Area Under the Receiver Operating Characteristic Curve (AUROC, also referred to as ROC AUC) for binary classification.
CerInput
The character error rate metric input type.
CharErrorRate
Character error rate (CER) is defined as the edit distance (e.g. Levenshtein distance) between the predicted and reference character sequences, divided by the total number of characters in the reference. This metric is commonly used in tasks such as speech recognition, OCR, or text generation to quantify how closely the predicted output matches the ground truth at a character level.
ConfusionStatsInput
Input for confusion statistics error types.
CpuMemory
Memory information
CpuTemperature
CPU Temperature in celsius degrees
CpuUse
General CPU Usage metric
CudaMetric
Track basic cuda infos.
FBetaScoreMetric
The F-beta score metric.
HammingScore
The hamming score, sometimes referred to as multi-label or label-based accuracy.
HammingScoreInput
The hamming score input type.
IterationSpeedMetric
The loss metric.
LearningRateMetric
Track the learning rate across iterations.
LossInput
The loss metric input type.
LossMetric
The loss metric.
MetricEntry
Data type that contains the current state of a metric at a given time.
MetricMetadata
Metric metadata that can be used when computing metrics.
PerplexityInput
The perplexity metric input type.
PerplexityMetric
The perplexity metric.
PrecisionMetric
The Precision Metric
RecallMetric
The Recall Metric
TopKAccuracyInput
The top-k accuracy metric input type.
TopKAccuracyMetric
The Top-K accuracy metric.
WerInput
The word error rate metric input type.
WordErrorRate
The word error rate (WER) metric, similar to the CER, is defined as the edit distance (e.g. Levenshtein distance) between the predicted and reference word sequences, divided by the total number of words in the reference. Here, the “units” within the sequences are words.

Enums§

ClassReduction
The reduction strategy for classification metrics.
NumericEntry
Numeric metric entry.

Traits§

Adaptor
Adaptor are used to transform types so that they can be used by metrics.
ItemLazy
Items that are lazy are not ready to be processed by metrics.
Metric
Metric trait.
Numeric
Declare a metric to be numeric.

Functions§

format_float
Format a float with the given precision. Will use scientific notation if necessary.

Type Aliases§

MetricName
Type used to store metric names efficiently.