Struct TrainingLog 

pub struct TrainingLog { /* private fields */ }

Records per-epoch training metrics.

Implementations

impl TrainingLog

pub fn new() -> Self

Create a new empty training log.

pub fn log_epoch(&mut self, metrics: HashMap<String, f32>)

Record one epoch’s metrics.

pub fn entries(&self) -> &[HashMap<String, f32>]

All logged entries.

pub fn get_metric_history(&self, name: &str) -> Vec<f32>

Extract the history of a single metric across all epochs.

Epochs that did not record the metric are skipped.

pub fn num_epochs(&self) -> usize

Number of epochs logged so far.

pub fn to_csv(&self) -> String

Export the log as a CSV string.

The header row contains the union of all metric names across every epoch, sorted alphabetically for deterministic output. Missing values are empty.

pub fn to_jsonl(&self) -> String

Export the log as JSONL (one JSON object per epoch).

Compatible with TensorBoard’s --logdir when saved as .jsonl. Each line: {"epoch": N, "loss": 0.5, "accuracy": 0.9, ...}

pub fn save_jsonl(&self, path: &Path) -> Result<()>

Write the log to a JSONL file (append-friendly).

pub fn append_epoch_jsonl( path: &Path, epoch: usize, metrics: &HashMap<String, f32>, ) -> Result<()>

Append a single epoch’s metrics to a JSONL file (streaming mode).

Trait Implementations

impl Clone for TrainingLog

fn clone(&self) -> TrainingLog

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for TrainingLog

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

Formats the value using the given formatter.

impl Default for TrainingLog

fn default() -> TrainingLog

Returns the “default value” for a type.