autoagents-burn 0.3.0

Agent Framework for Building Autonomous Agents
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

mod elems {
    cfg_if::cfg_if! {
        // NOTE: f16/bf16 is not always supported on wgpu depending on the hardware
        // https://github.com/gfx-rs/wgpu/issues/7468
        if #[cfg(all(feature = "f16", any(feature = "cuda", feature = "webgpu", feature = "wgpu", feature = "vulkan", feature = "metal", feature = "rocm", feature = "libtorch", feature = "candle-cuda")))]{
            pub type ElemType = burn::tensor::f16;
            pub const DTYPE_NAME: &str = "f16";
        }
        else if #[cfg(all(feature = "f16", any(feature = "cuda", feature = "webgpu", feature = "wgpu", feature = "vulkan", feature = "metal", feature = "rocm", feature = "libtorch", feature = "candle-cuda")))]{
            pub type ElemType = burn::tensor::bf16;
            pub const DTYPE_NAME: &str = "bf16";
        } else {
            pub type ElemType = f32;
            pub const DTYPE_NAME: &str = "f32";
        }
    }
}

pub use elems::*;

// Candle --------------------------------------------------------------------
#[cfg(any(feature = "candle-accelerate", feature = "candle-cpu"))]
pub mod burn_backend_types {
    use super::*;
    use burn::backend::candle::{Candle, CandleDevice};
    pub type InferenceBackend = Candle<ElemType>;
    pub type InferenceDevice = CandleDevice;
    pub const INFERENCE_DEVICE: std::sync::LazyLock<CandleDevice> =
        std::sync::LazyLock::new(|| CandleDevice::Cpu);
    pub const NAME: &str = "candle-cpu";
}

#[cfg(feature = "candle-cuda")]
pub mod burn_backend_types {
    use super::*;
    use burn::backend::candle::{Candle, CandleDevice};
    pub type InferenceBackend = Candle<ElemType>;
    pub type InferenceDevice = CandleDevice;
    pub const INFERENCE_DEVICE: std::sync::LazyLock<CandleDevice> =
        std::sync::LazyLock::new(|| CandleDevice::cuda(0));
    pub const NAME: &str = "candle-cuda";
}

#[cfg(feature = "candle-metal")]
pub mod burn_backend_types {
    use super::*;
    use burn::backend::candle::{Candle, CandleDevice};
    pub type InferenceBackend = Candle<ElemType>;
    pub type InferenceDevice = CandleDevice;
    pub const INFERENCE_DEVICE: std::sync::LazyLock<CandleDevice> =
        std::sync::LazyLock::new(|| CandleDevice::metal(0));
    pub const NAME: &str = "candle-metal";
}

// Cuda ----------------------------------------------------------------------

#[cfg(feature = "cuda")]
pub mod burn_backend_types {
    use super::*;
    use burn::backend::cuda::{Cuda, CudaDevice};
    pub type InferenceBackend = Cuda<ElemType>;
    pub type InferenceDevice = CudaDevice;
    pub const INFERENCE_DEVICE: std::sync::LazyLock<CudaDevice> =
        std::sync::LazyLock::new(|| CudaDevice::default());
    pub const NAME: &str = "cuda";
}

// ROCm ----------------------------------------------------------------------

#[cfg(feature = "rocm")]
pub mod burn_backend_types {
    use super::*;
    use burn::backend::rocm::{Rocm, RocmDevice};
    pub type InferenceBackend = Rocm<ElemType>;
    pub type InferenceDevice = RocmDevice;
    pub const INFERENCE_DEVICE: std::sync::LazyLock<RocmDevice> =
        std::sync::LazyLock::new(|| RocmDevice::default());
    pub const NAME: &str = "rocm";
}

// ndarray -------------------------------------------------------------------
// This backend is used for testing and by default when no backend is selected.

#[cfg(any(feature = "ndarray", not(feature = "selected-backend")))]
pub mod burn_backend_types {
    use super::*;
    use burn::backend::ndarray::{NdArray, NdArrayDevice};

    pub type InferenceBackend = NdArray<ElemType>;
    pub type InferenceDevice = NdArrayDevice;
    pub const INFERENCE_DEVICE: InferenceDevice = NdArrayDevice::Cpu;
    pub const NAME: &str = "ndarray";
}

// LibTorch ------------------------------------------------------------------

#[cfg(feature = "libtorch")]
pub mod burn_backend_types {
    use super::*;
    use burn::backend::libtorch::{LibTorch, LibTorchDevice};

    pub type InferenceBackend = LibTorch<ElemType>;
    pub type InferenceDevice = LibTorchDevice;

    #[cfg(not(target_os = "macos"))]
    pub const INFERENCE_DEVICE: std::sync::LazyLock<LibTorchDevice> =
        std::sync::LazyLock::new(|| LibTorchDevice::Cuda(0));
    #[cfg(target_os = "macos")]
    pub const INFERENCE_DEVICE: InferenceDevice = LibTorchDevice::Mps;
    pub const NAME: &str = "libtorch";
}

#[cfg(feature = "libtorch-cpu")]
pub mod burn_backend_types {
    use super::*;
    use burn::backend::libtorch::{LibTorch, LibTorchDevice};

    pub type InferenceBackend = LibTorch<ElemType>;
    pub type InferenceDevice = LibTorchDevice;
    pub const INFERENCE_DEVICE: InferenceDevice = LibTorchDevice::Cpu;
    pub const NAME: &str = "libtorch-cpu";
}

// WebGPU --------------------------------------------------------------------

#[cfg(any(feature = "wgpu", feature = "vulkan", feature = "metal"))]
pub mod burn_backend_types {
    use super::*;
    use burn::backend::wgpu::{Wgpu, WgpuDevice};
    pub type InferenceBackend = Wgpu<ElemType>;
    pub type InferenceDevice = WgpuDevice;
    pub const INFERENCE_DEVICE: InferenceDevice = WgpuDevice::DefaultDevice;
    #[cfg(all(feature = "wgpu", not(feature = "vulkan"), not(feature = "metal")))]
    pub const NAME: &str = "wgpu";
    #[cfg(feature = "vulkan")]
    pub const NAME: &str = "vulkan";
    #[cfg(feature = "metal")]
    pub const NAME: &str = "metal";
}

#[cfg(feature = "webgpu")]
pub mod burn_backend_types {
    use super::*;
    use burn::backend::wgpu::WebGpu;
    use burn::backend::wgpu::{graphics::AutoGraphicsApi, init_setup_async, WgpuDevice};

    pub type InferenceBackend = WebGpu<ElemType>;
    pub type InferenceDevice = WgpuDevice;
    pub const INFERENCE_DEVICE: InferenceDevice = WgpuDevice::DefaultDevice;
    pub const NAME: &str = "webgpu";

    pub async fn init_setup() {
        init_setup_async::<AutoGraphicsApi>(&WgpuDevice::default(), Default::default()).await;
    }
}

#[cfg(feature = "wgpu-cpu")]
pub mod burn_backend_types {
    use super::*;
    use burn::backend::wgpu::{Wgpu, WgpuDevice};
    pub type InferenceBackend = Wgpu<ElemType>;
    pub type InferenceDevice = WgpuDevice;
    pub const INFERENCE_DEVICE: InferenceDevice = WgpuDevice::Cpu;
    pub const NAME: &str = "wgpu-cpu";
}