Struct TrainingCheckpoint 

pub struct TrainingCheckpoint {
    pub step: usize,
    pub epoch: usize,
    pub model_state: HashMap<String, Vec<f32>>,
    pub optimizer_state: HashMap<String, Vec<f32>>,
    pub scheduler_state: HashMap<String, f32>,
    pub rng_states: HashMap<String, Vec<u8>>,
    pub loss: f32,
    pub metrics: HashMap<String, f32>,
    pub config: HashMap<String, String>,
    pub timestamp: u64,
    pub version: String,
    pub distributed_meta: DistributedMetadata,
}

Training checkpoint containing all necessary state

Fields

step: usize

Step number when checkpoint was created

epoch: usize

Epoch number

model_state: HashMap<String, Vec<f32>>

Model parameters

optimizer_state: HashMap<String, Vec<f32>>

Optimizer state

scheduler_state: HashMap<String, f32>

Learning rate scheduler state

rng_states: HashMap<String, Vec<u8>>

Random number generator states

loss: f32

Loss value at checkpoint

metrics: HashMap<String, f32>

Metrics at checkpoint

config: HashMap<String, String>

Training configuration

timestamp: u64

Timestamp when checkpoint was created

version: String

Version identifier

distributed_meta: DistributedMetadata

Distributed training metadata

Trait Implementations

impl Clone for TrainingCheckpoint

fn clone(&self) -> TrainingCheckpoint

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for TrainingCheckpoint

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for TrainingCheckpoint

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for TrainingCheckpoint

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

where __S: Serializer,

Serialize this value into the given Serde serializer.