boostr 0.1.0

ML framework built on numr - attention, quantization, model architectures
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
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
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218

//! Standard split-half RoPE implementation.

use super::rope_common::validate_and_prepare;
use crate::error::{Error, Result};
use numr::autograd::{Var, var_add, var_cat, var_mul, var_narrow, var_sub};
use numr::ops::{ScalarOps, ShapeOps, TensorOps, TypeConversionOps};
use numr::runtime::{Runtime, RuntimeClient};

/// Apply Rotary Position Embedding (RoPE) — split-half variant.
///
/// For each pair of dimensions (x[..., d], x[..., d+D/2]):
///   x_rot[..., d]     = x[..., d] * cos - x[..., d+D/2] * sin
///   x_rot[..., d+D/2] = x[..., d] * sin + x[..., d+D/2] * cos
///
/// # Arguments
///
/// - `x`: `[B, H, S, D]` — input tensor
/// - `cos_cache`: `[S, D/2]` — precomputed cosines (broadcast to [1, 1, S, D/2])
/// - `sin_cache`: `[S, D/2]` — precomputed sines (broadcast to [1, 1, S, D/2])
pub fn apply_rope_impl<R, C>(
    client: &C,
    x: &Var<R>,
    cos_cache: &Var<R>,
    sin_cache: &Var<R>,
) -> Result<Var<R>>
where
    R: Runtime<DType = numr::dtype::DType>,
    C: RuntimeClient<R> + ScalarOps<R> + ShapeOps<R> + TypeConversionOps<R>,
    R::Client: TensorOps<R> + ShapeOps<R> + TypeConversionOps<R>,
{
    let (_shape, _seq_len, half_d, cos_reshaped, sin_reshaped) =
        validate_and_prepare::<R, C>(client, x, cos_cache, sin_cache)?;

    // Split x into two halves along last dim: x1 = x[..., :D/2], x2 = x[..., D/2:]
    let x1 = var_narrow(x, -1, 0, half_d).map_err(Error::Numr)?;
    let x2 = var_narrow(x, -1, half_d, half_d).map_err(Error::Numr)?;

    // out1 = x1 * cos - x2 * sin
    let x1_cos = var_mul(&x1, &cos_reshaped, client).map_err(Error::Numr)?;
    let x2_sin = var_mul(&x2, &sin_reshaped, client).map_err(Error::Numr)?;
    let out1 = var_sub(&x1_cos, &x2_sin, client).map_err(Error::Numr)?;

    // out2 = x1 * sin + x2 * cos
    let x1_sin = var_mul(&x1, &sin_reshaped, client).map_err(Error::Numr)?;
    let x2_cos = var_mul(&x2, &cos_reshaped, client).map_err(Error::Numr)?;
    let out2 = var_add(&x1_sin, &x2_cos, client).map_err(Error::Numr)?;

    // Concatenate back: [out1, out2] along last dim
    var_cat(&[&out1, &out2], -1, client).map_err(Error::Numr)
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::test_utils::cpu_setup;
    use numr::runtime::cpu::CpuRuntime;
    use numr::tensor::Tensor;

    #[test]
    fn test_rope_output_shape() {
        let (client, device) = cpu_setup();
        let b = 1;
        let h = 2;
        let s = 4;
        let d = 8;

        let x = Var::new(
            Tensor::<CpuRuntime>::from_slice(&vec![1.0f32; b * h * s * d], &[b, h, s, d], &device),
            false,
        );
        // cos/sin cache: [S, D/2]
        let cos = Var::new(
            Tensor::<CpuRuntime>::from_slice(&vec![1.0f32; s * d / 2], &[s, d / 2], &device),
            false,
        );
        let sin = Var::new(
            Tensor::<CpuRuntime>::from_slice(&vec![0.0f32; s * d / 2], &[s, d / 2], &device),
            false,
        );

        let out = apply_rope_impl(&client, &x, &cos, &sin).unwrap();
        assert_eq!(out.tensor().shape(), &[b, h, s, d]);
    }

    #[test]
    fn test_rope_identity_with_zero_angle() {
        let (client, device) = cpu_setup();
        // cos=1, sin=0 → RoPE is identity
        let x_data: Vec<f32> = (0..16).map(|i| i as f32).collect();
        let x = Var::new(
            Tensor::<CpuRuntime>::from_slice(&x_data, &[1, 1, 2, 8], &device),
            false,
        );
        let cos = Var::new(
            Tensor::<CpuRuntime>::from_slice(&[1.0f32; 8], &[2, 4], &device),
            false,
        );
        let sin = Var::new(
            Tensor::<CpuRuntime>::from_slice(&[0.0f32; 8], &[2, 4], &device),
            false,
        );

        let out = apply_rope_impl(&client, &x, &cos, &sin).unwrap();
        let out_data: Vec<f32> = out.tensor().contiguous().to_vec();

        // With sin=0, cos=1: out1 = x1*1 - x2*0 = x1, out2 = x2*1 + x1*0 = x2
        // So cat(x1, x2) = x (identity)
        for (i, (&a, &b)) in out_data.iter().zip(x_data.iter()).enumerate() {
            assert!(
                (a - b).abs() < 1e-5,
                "mismatch at {}: got {}, expected {}",
                i,
                a,
                b
            );
        }
    }

    #[test]
    fn test_rope_90_degree_rotation() {
        let (client, device) = cpu_setup();
        // cos=0, sin=1 → 90° rotation
        // out1 = x1*0 - x2*1 = -x2
        // out2 = x2*0 + x1*1 = x1
        let x = Var::new(
            Tensor::<CpuRuntime>::from_slice(
                &[1.0f32, 2.0, 3.0, 4.0], // x1=[1,2], x2=[3,4]
                &[1, 1, 1, 4],
                &device,
            ),
            false,
        );
        let cos = Var::new(
            Tensor::<CpuRuntime>::from_slice(&[0.0f32, 0.0], &[1, 2], &device),
            false,
        );
        let sin = Var::new(
            Tensor::<CpuRuntime>::from_slice(&[1.0f32, 1.0], &[1, 2], &device),
            false,
        );

        let out = apply_rope_impl(&client, &x, &cos, &sin).unwrap();
        let out_data: Vec<f32> = out.tensor().contiguous().to_vec();

        // out1 = -x2 = [-3, -4], out2 = x1 = [1, 2] → [-3, -4, 1, 2]
        assert!((out_data[0] - (-3.0)).abs() < 1e-5);
        assert!((out_data[1] - (-4.0)).abs() < 1e-5);
        assert!((out_data[2] - 1.0).abs() < 1e-5);
        assert!((out_data[3] - 2.0).abs() < 1e-5);
    }

    #[test]
    fn test_rope_narrowing_matches_exact_cache() {
        // Verifies that a pre-allocated cache larger than seq_len produces
        // identical output to an exact-size cache (exercises the narrowing path).
        let (client, device) = cpu_setup();

        let x_data: Vec<f32> = (0..16).map(|i| i as f32 * 0.1).collect();
        let x = Var::new(
            Tensor::<CpuRuntime>::from_slice(&x_data, &[1, 1, 2, 8], &device),
            false,
        );

        // Exact-size caches: [S=2, D/2=4]
        let cos_exact = Var::new(
            Tensor::<CpuRuntime>::from_slice(&[1.0f32; 8], &[2, 4], &device),
            false,
        );
        let sin_exact = Var::new(
            Tensor::<CpuRuntime>::from_slice(&[0.0f32; 8], &[2, 4], &device),
            false,
        );

        // Pre-allocated caches: [max_S=8, D/2=4] — will be narrowed to S=2
        let cos_large = Var::new(
            Tensor::<CpuRuntime>::from_slice(&[1.0f32; 32], &[8, 4], &device),
            false,
        );
        let sin_large = Var::new(
            Tensor::<CpuRuntime>::from_slice(&[0.0f32; 32], &[8, 4], &device),
            false,
        );

        let out_exact = apply_rope_impl(&client, &x, &cos_exact, &sin_exact).unwrap();
        let out_narrowed = apply_rope_impl(&client, &x, &cos_large, &sin_large).unwrap();

        let exact_data: Vec<f32> = out_exact.tensor().contiguous().to_vec();
        let narrowed_data: Vec<f32> = out_narrowed.tensor().contiguous().to_vec();

        assert_eq!(exact_data.len(), narrowed_data.len());
        for (i, (&a, &b)) in exact_data.iter().zip(narrowed_data.iter()).enumerate() {
            assert!(
                (a - b).abs() < 1e-5,
                "mismatch at {i}: exact={a}, narrowed={b}"
            );
        }
    }

    #[test]
    fn test_rope_invalid_odd_dim() {
        let (client, device) = cpu_setup();
        let x = Var::new(
            Tensor::<CpuRuntime>::from_slice(&[1.0f32; 3], &[1, 1, 1, 3], &device),
            false,
        );
        let cos = Var::new(
            Tensor::<CpuRuntime>::from_slice(&[1.0f32], &[1, 1], &device),
            false,
        );
        let sin = Var::new(
            Tensor::<CpuRuntime>::from_slice(&[0.0f32], &[1, 1], &device),
            false,
        );

        let result = apply_rope_impl(&client, &x, &cos, &sin);
        assert!(result.is_err());
    }
}