atomr-accel-train 0.1.0

Distributed training blueprints on atomr-accel-cuda: DataParallelTrainer, PipelineParallelTrainer, TensorParallelTrainer, AsyncParameterServer, optimizer + loss enums.
Documentation 
1
2
3
4
5
6
7
8
9
10

//! Loss kinds.

#[derive(Debug, Clone, Copy)]
pub enum LossKind {
    Mse,
    CrossEntropy,
    /// Categorical cross-entropy with per-class weights — common in
    /// imbalanced classification.
    WeightedCrossEntropy,
}