ferrotorch-core 0.3.0

Core tensor and autograd engine for ferrotorch — PyTorch in Rust
Documentation 
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
32
33
34

pub mod anomaly;
pub mod autocast;
pub mod autocast_ops;
pub mod checkpoint;
pub mod cond_scan;
pub mod fixed_point;
pub mod forward_ad;
pub mod grad_penalty;
pub mod gradcheck;
pub mod graph;
pub mod higher_order;
pub mod hooks;
pub mod no_grad;
pub mod saved_tensors;

pub use autocast::{
    AutocastDtype, autocast, autocast_dtype, is_autocast_debug, is_autocast_enabled,
    set_autocast_debug,
};
pub use autocast_ops::{
    AutocastCategory, AutocastEvent, autocast_category, autocast_guard, autocast_log,
    drain_autocast_events, should_cast_to_reduced, should_keep_full_precision,
};
pub use cond_scan::{cond, scan, validate_cond_branches};
pub use fixed_point::fixed_point;
pub use forward_ad::{
    DualTensor, dual_add, dual_cos, dual_div, dual_exp, dual_log, dual_matmul, dual_mul, dual_neg,
    dual_relu, dual_sigmoid, dual_sin, dual_sub, dual_tanh, jacfwd, jvp_exact,
};
pub use grad_penalty::{grad_norm, gradient_penalty, jvp, vjp};
pub use gradcheck::gradcheck;
pub use graph::{backward, backward_with_grad};
pub use higher_order::{grad, hessian, jacobian};
pub use no_grad::{enable_grad, inference_mode, is_grad_enabled, is_inference_mode, no_grad, set_grad_enabled};