entrenar 0.7.12

Training & Optimization library with autograd, LoRA, quantization, and model merging
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31

//! Optimizers for training neural networks

mod adam;
mod adamw;
mod clip;
mod convergence_tests; // Tests split into convergence_tests/ directory
pub mod dp;
pub mod hpo;
mod optimizer;
mod scheduler;
mod sgd;
mod simd;

pub use adam::Adam;
pub use adamw::AdamW;
pub use clip::{clip_grad_norm, clip_grad_norm_refs};
pub use dp::{
    add_gaussian_noise, clip_gradient, estimate_noise_multiplier, grad_norm, privacy_cost_per_step,
    DpError, DpSgd, DpSgdConfig, PrivacyBudget, RdpAccountant,
};
pub use hpo::{
    AcquisitionFunction, GridSearch, HPOError, HyperbandScheduler, HyperparameterSpace,
    ParameterDomain, ParameterValue, SearchStrategy, SurrogateModel, TPEOptimizer, Trial,
    TrialStatus,
};
pub use optimizer::Optimizer;
pub use scheduler::{
    CosineAnnealingLR, LRScheduler, LinearWarmupLR, StepDecayLR, WarmupCosineDecayLR,
};
pub use sgd::SGD;
pub use simd::{simd_adam_update, simd_adamw_update, simd_axpy};