teeny-kernels 0.2.0

Teenygrad is a high-performance, memory-safe Rust ML training and inference library. It targets devices from microcontrollers to traditional GPUs with statically-typed kernels and full async support.
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
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136

/*
 * Copyright (c) 2026 Teenygrad.
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *   http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

#![allow(non_snake_case)]

use teeny_macros::kernel;
use teeny_triton::triton::{
    types::{AddOffsets, Comparison},
    *,
};

/// Adamax step (infinity-norm variant of Adam).
///
/// ```text
/// exp_avg = beta1 * exp_avg + (1 - beta1) * g
/// exp_inf = max(beta2 * exp_inf, |g| + eps)
/// p       = p - clr * exp_avg / exp_inf
/// ```
///
/// `clr = lr / (1 - beta1^t)` is precomputed on the host.
///
/// Grid: `[ceil(n_elements / BLOCK_SIZE), 1, 1]`.
#[kernel]
pub fn adamax_step<T: Triton, const BLOCK_SIZE: i32>(
    params_ptr: T::Pointer<f32>,
    grad_ptr: T::Pointer<f32>,
    exp_avg_ptr: T::Pointer<f32>,
    exp_inf_ptr: T::Pointer<f32>,
    n_elements: i32,
    clr: f32,
    beta1: f32,
    beta2: f32,
    eps: f32,
    weight_decay: f32,
) where
    T::I32Tensor: types::Tensor<i32, 1>,
    T::I32Tensor: Comparison<i32, BoolTensor = T::BoolTensor>,
    T::Pointer<f32>: AddOffsets<i32, 1, T::I32Tensor, Output = T::Tensor<T::Pointer<f32>>>,
{
    let pid = T::program_id(Axis::X);
    let block_start = pid * BLOCK_SIZE;
    let offsets = T::arange(0, BLOCK_SIZE) + block_start;
    let mask = offsets.lt(n_elements);

    let p = T::load(
        params_ptr.add_offsets(offsets),
        Some(mask),
        None,
        &[],
        None,
        None,
        None,
        false,
    );
    let g = T::load(
        grad_ptr.add_offsets(offsets),
        Some(mask),
        None,
        &[],
        None,
        None,
        None,
        false,
    );
    let exp_avg = T::load(
        exp_avg_ptr.add_offsets(offsets),
        Some(mask),
        None,
        &[],
        None,
        None,
        None,
        false,
    );
    let exp_inf = T::load(
        exp_inf_ptr.add_offsets(offsets),
        Some(mask),
        None,
        &[],
        None,
        None,
        None,
        false,
    );

    let beta1_t = T::full(&[BLOCK_SIZE], beta1);
    let one_m_beta1 = T::full(&[BLOCK_SIZE], 1.0_f32 - beta1);
    let beta2_t = T::full(&[BLOCK_SIZE], beta2);
    let eps_t = T::full(&[BLOCK_SIZE], eps);
    let clr_t = T::full(&[BLOCK_SIZE], clr);
    let wd_t = T::full(&[BLOCK_SIZE], weight_decay);

    let g_eff = g + wd_t * p;
    let exp_avg_new = beta1_t * exp_avg + one_m_beta1 * g_eff;
    // exp_inf = max(beta2 * exp_inf, |g| + eps)
    let exp_inf_new = T::maximum(beta2_t * exp_inf, T::abs(g_eff) + eps_t);
    let p_new = p - clr_t * exp_avg_new / exp_inf_new;

    T::store(
        params_ptr.add_offsets(offsets),
        p_new,
        Some(mask),
        &[],
        None,
        None,
    );
    T::store(
        exp_avg_ptr.add_offsets(offsets),
        exp_avg_new,
        Some(mask),
        &[],
        None,
        None,
    );
    T::store(
        exp_inf_ptr.add_offsets(offsets),
        exp_inf_new,
        Some(mask),
        &[],
        None,
        None,
    );
}