extern crate alloc;
use burn::prelude::*;
use burn::nn::LayerNorm;
use burn::nn::LayerNormConfig;
use burn::nn::Linear;
use burn::nn::LinearConfig;
use burn::nn::PaddingConfig1d;
use burn::nn::PaddingConfig2d;
use burn::nn::conv::Conv1d;
use burn::nn::conv::Conv1dConfig;
use burn::nn::conv::Conv2d;
use burn::nn::conv::Conv2dConfig;
#[derive(Module, Debug)]
pub struct Submodule1 {
constant662: burn::module::Param<Tensor<1>>,
constant663: burn::module::Param<Tensor<1>>,
constant664: burn::module::Param<Tensor<1>>,
constant665: burn::module::Param<Tensor<1>>,
constant666: burn::module::Param<Tensor<1>>,
constant667: burn::module::Param<Tensor<1>>,
constant668: burn::module::Param<Tensor<1>>,
constant669: burn::module::Param<Tensor<1>>,
constant670: burn::module::Param<Tensor<1>>,
conv2d1: Conv2d,
conv2d2: Conv2d,
conv2d3: Conv2d,
conv2d4: Conv2d,
conv2d5: Conv2d,
linear1: Linear,
constant681: burn::module::Param<Tensor<3>>,
#[module(skip)]
device: Device,
}
impl Submodule1 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let constant662: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([-1f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let constant663: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([2f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let constant664: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([1f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let constant665: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([-1f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let constant666: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([2f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let constant667: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([1f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let constant668: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([-1f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let constant669: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([2f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let constant670: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([1f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let conv2d1 = Conv2dConfig::new([1, 256], [3, 3])
.with_stride([2, 2])
.with_padding(PaddingConfig2d::Explicit(1, 1, 1, 1))
.with_dilation([1, 1])
.with_groups(1)
.with_bias(true)
.init(device);
let conv2d2 = Conv2dConfig::new([256, 256], [3, 3])
.with_stride([2, 2])
.with_padding(PaddingConfig2d::Explicit(1, 1, 1, 1))
.with_dilation([1, 1])
.with_groups(256)
.with_bias(true)
.init(device);
let conv2d3 = Conv2dConfig::new([256, 256], [1, 1])
.with_stride([1, 1])
.with_padding(PaddingConfig2d::Valid)
.with_dilation([1, 1])
.with_groups(1)
.with_bias(true)
.init(device);
let conv2d4 = Conv2dConfig::new([256, 256], [3, 3])
.with_stride([2, 2])
.with_padding(PaddingConfig2d::Explicit(1, 1, 1, 1))
.with_dilation([1, 1])
.with_groups(256)
.with_bias(true)
.init(device);
let conv2d5 = Conv2dConfig::new([256, 256], [1, 1])
.with_stride([1, 1])
.with_padding(PaddingConfig2d::Valid)
.with_dilation([1, 1])
.with_groups(1)
.with_bias(true)
.init(device);
let linear1 = LinearConfig::new(4096, 1024).with_bias(true).init(device);
let constant681: burn::module::Param<Tensor<3>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
3,
>::zeros([1, 9999, 1024], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[1, 9999, 1024].into(),
);
Self {
constant662,
constant663,
constant664,
constant665,
constant666,
constant667,
constant668,
constant669,
constant670,
conv2d1,
conv2d2,
conv2d3,
conv2d4,
conv2d5,
linear1,
constant681,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
audio_signal: Tensor<3>,
length: Tensor<1, Int>,
) -> (Tensor<3, Bool>, Tensor<2, Bool>, Tensor<3>, Tensor<3>, Tensor<1, Int>) {
let transpose1_out1 = audio_signal.permute([0, 2, 1]);
let cast1_out1 = length.float().cast(burn::tensor::DType::F32);
let constant662_out1 = self.constant662.val();
let add1_out1 = cast1_out1.add(constant662_out1);
let constant663_out1 = self.constant663.val();
let div1_out1 = add1_out1.div(constant663_out1);
let constant664_out1 = self.constant664.val();
let add2_out1 = div1_out1.add(constant664_out1);
let floor1_out1 = add2_out1.floor();
let constant665_out1 = self.constant665.val();
let add3_out1 = floor1_out1.add(constant665_out1);
let constant666_out1 = self.constant666.val();
let div2_out1 = add3_out1.div(constant666_out1);
let constant667_out1 = self.constant667.val();
let add4_out1 = div2_out1.add(constant667_out1);
let floor2_out1 = add4_out1.floor();
let constant668_out1 = self.constant668.val();
let add5_out1 = floor2_out1.add(constant668_out1);
let constant669_out1 = self.constant669.val();
let div3_out1 = add5_out1.div(constant669_out1);
let constant670_out1 = self.constant670.val();
let add6_out1 = div3_out1.add(constant670_out1);
let floor3_out1 = add6_out1.floor();
let cast4_out1 = floor3_out1.int().cast(burn::tensor::DType::I32);
let unsqueeze1_out1: Tensor<4> = transpose1_out1.unsqueeze_dims::<4>(&[1]);
let conv2d1_out1 = self.conv2d1.forward(unsqueeze1_out1);
let relu1_out1 = burn::tensor::activation::relu(conv2d1_out1);
let conv2d2_out1 = self.conv2d2.forward(relu1_out1);
let conv2d3_out1 = self.conv2d3.forward(conv2d2_out1);
let relu2_out1 = burn::tensor::activation::relu(conv2d3_out1);
let conv2d4_out1 = self.conv2d4.forward(relu2_out1);
let conv2d5_out1 = self.conv2d5.forward(conv2d4_out1);
let relu3_out1 = burn::tensor::activation::relu(conv2d5_out1);
let shape1_out1: [i64; 4] = {
let axes = &relu3_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather1_out1 = shape1_out1[0] as i64;
let gather2_out1 = shape1_out1[2] as i64;
let transpose2_out1 = relu3_out1.permute([0, 2, 1, 3]);
let unsqueeze2_out1 = [gather1_out1 as i64];
let unsqueeze3_out1 = [gather2_out1 as i64];
let constant676_out1: [i64; 1] = [-1i64];
let concat1_out1: [i64; 3usize] = [
&unsqueeze2_out1[..],
&unsqueeze3_out1[..],
&constant676_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape1_out1 = transpose2_out1.reshape(concat1_out1);
let linear1_out1 = self.linear1.forward(reshape1_out1);
let cast5_out1 = cast4_out1.cast(burn::tensor::DType::I64);
let shape3_out1: [i64; 3] = {
let axes = &linear1_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather3_out1 = shape3_out1[1] as i64;
let constant678_out1 = 5000i64;
let sub1_out1 = constant678_out1 - gather3_out1;
let constant679_out1 = 5000i64;
let add7_out1 = constant679_out1 + gather3_out1;
let constant680_out1 = 1i64;
let sub2_out1 = add7_out1 - constant680_out1;
let unsqueeze4_out1 = [sub1_out1 as i64];
let unsqueeze5_out1 = [sub2_out1 as i64];
let constant681_out1 = self.constant681.val();
let slice1_out1 = constant681_out1
.slice(s![.., unsqueeze4_out1[0]..unsqueeze5_out1[0], ..]);
let unsqueeze6_out1 = [gather3_out1 as i64];
let unsqueeze7_out1 = [gather3_out1 as i64];
let constant686_out1: [i64; 1] = [1i64];
let concat2_out1: [i64; 3usize] = [
&constant686_out1[..],
&unsqueeze6_out1[..],
&unsqueeze7_out1[..],
]
.concat()
.try_into()
.unwrap();
let constantofshape1_out1 = Tensor::<3, Int>::ones(concat2_out1, &self.device)
.bool();
let range1_out1 = {
let start = 0i64;
let delta = 1i64;
assert!(delta != 0);
let n = (((gather3_out1 as i128 - start as i128) as f64) / delta as f64)
.ceil()
.max(0.0) as i64;
Tensor::arange(0..n, &self.device)
.cast(burn::tensor::DType::I64)
.mul_scalar(delta)
.add_scalar(start)
};
let shape4_out1: [i64; 1] = {
let axes = &cast5_out1.clone().dims()[0..1];
let mut output = [0i64; 1];
for i in 0..1 {
output[i] = axes[i] as i64;
}
output
};
let gather4_out1 = shape4_out1[0] as i64;
let unsqueeze8_out1 = [gather4_out1 as i64];
let constant693_out1: [i64; 1] = [-1i64];
let concat3_out1: [i64; 2usize] = [&unsqueeze8_out1[..], &constant693_out1[..]]
.concat()
.try_into()
.unwrap();
let reshape2_out1 = concat3_out1;
let constant696_out1 = -1i64;
let mul2_out1 = {
let mut result = reshape2_out1;
let scalar = constant696_out1;
for result_item in result.iter_mut() {
*result_item = result_item.saturating_mul(scalar);
}
result
};
let max1_out1 = {
let __lhs = reshape2_out1;
let __rhs = mul2_out1;
core::array::from_fn::<i64, 2usize, _>(|__i| (__lhs[__i]).max(__rhs[__i]))
};
let expand1_out1 = {
let onnx_shape: [i64; 2usize] = max1_out1;
let input_dims = range1_out1.dims();
let mut shape = onnx_shape;
#[allow(clippy::needless_range_loop)]
for i in 0..1usize {
let dim_offset = 1usize + i;
if shape[dim_offset] == 1 && input_dims[i] > 1 {
shape[dim_offset] = input_dims[i] as i64;
}
}
range1_out1.expand(shape)
};
let unsqueeze9_out1: Tensor<2, Int> = cast5_out1
.clone()
.unsqueeze_dims::<2>(&[-1]);
let less1_out1 = expand1_out1.lower(unsqueeze9_out1);
let unsqueeze10_out1: Tensor<3, Bool> = less1_out1
.clone()
.unsqueeze_dims::<3>(&[1]);
let unsqueeze12_out1 = [gather3_out1 as i64];
let constant704_out1: [i64; 1] = [1i64];
let constant702_out1: [i64; 1] = [1i64];
let concat5_out1: [i64; 3usize] = [
&constant702_out1[..],
&unsqueeze12_out1[..],
&constant704_out1[..],
]
.concat()
.try_into()
.unwrap();
let tile1_out1 = {
let repeats: alloc::vec::Vec<usize> = concat5_out1
.iter()
.map(|&v| v as usize)
.collect();
unsqueeze10_out1.repeat(&repeats)
};
let transpose3_out1 = tile1_out1.clone().permute([0, 2, 1]);
let and1_out1 = tile1_out1.bool_and(transpose3_out1);
let unsqueeze13_out1 = [gather3_out1 as i64];
let slice2_out1 = constantofshape1_out1
.slice(s![.., 0..unsqueeze13_out1[0], ..]);
let unsqueeze14_out1 = [gather3_out1 as i64];
let slice3_out1 = slice2_out1.slice(s![.., .., 0..unsqueeze14_out1[0]]);
let and2_out1 = and1_out1.bool_and(slice3_out1);
(and2_out1, less1_out1, linear1_out1, slice1_out1, cast5_out1)
}
}
#[derive(Module, Debug)]
pub struct Submodule2 {
layernormalization1: LayerNorm,
linear2: Linear,
linear3: Linear,
constant713: burn::module::Param<Tensor<1>>,
layernormalization2: LayerNorm,
linear4: Linear,
linear5: Linear,
linear6: Linear,
linear7: Linear,
constant20: burn::module::Param<Tensor<2>>,
constant21: burn::module::Param<Tensor<2>>,
constant856: burn::module::Param<Tensor<1>>,
linear8: Linear,
#[module(skip)]
device: Device,
}
impl Submodule2 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization1 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear2 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear3 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant713: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization2 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear4 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear5 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear6 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear7 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant20: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant21: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant856: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear8 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization1,
linear2,
linear3,
constant713,
layernormalization2,
linear4,
linear5,
linear6,
linear7,
constant20,
constant21,
constant856,
linear8,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
and2_out1: Tensor<3, Bool>,
less1_out1: Tensor<2, Bool>,
linear1_out1: Tensor<3>,
slice1_out1: Tensor<3>,
) -> (Tensor<3>, Tensor<2, Bool>, Tensor<4, Bool>) {
let not1_out1 = and2_out1.bool_not();
let not2_out1 = less1_out1.bool_not();
let layernormalization1_out1 = {
let dtype = linear1_out1.clone().dtype();
self.layernormalization1
.forward(linear1_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear2_out1 = self.linear2.forward(layernormalization1_out1);
let sigmoid1_out1 = burn::tensor::activation::sigmoid(linear2_out1.clone());
let mul3_out1 = linear2_out1.mul(sigmoid1_out1);
let linear3_out1 = self.linear3.forward(mul3_out1);
let constant713_out1 = self.constant713.val();
let mul4_out1 = linear3_out1
.mul((constant713_out1).unsqueeze_dims(&[0isize, 1isize]));
let add8_out1 = linear1_out1.add(mul4_out1);
let layernormalization2_out1 = {
let dtype = add8_out1.clone().dtype();
self.layernormalization2
.forward(add8_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape7_out1: [i64; 3] = {
let axes = &layernormalization2_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather5_out1 = shape7_out1[0] as i64;
let linear4_out1 = self.linear4.forward(layernormalization2_out1.clone());
let unsqueeze15_out1 = [gather5_out1 as i64];
let constant718_out1: [i64; 1] = [128i64];
let constant716_out1: [i64; 1] = [-1i64];
let constant717_out1: [i64; 1] = [8i64];
let concat6_out1: [i64; 4usize] = [
&unsqueeze15_out1[..],
&constant716_out1[..],
&constant717_out1[..],
&constant718_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze16_out1 = [gather5_out1 as i64];
let constant722_out1: [i64; 1] = [128i64];
let constant720_out1: [i64; 1] = [-1i64];
let constant721_out1: [i64; 1] = [8i64];
let concat7_out1: [i64; 4usize] = [
&unsqueeze16_out1[..],
&constant720_out1[..],
&constant721_out1[..],
&constant722_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze17_out1 = [gather5_out1 as i64];
let constant726_out1: [i64; 1] = [128i64];
let constant724_out1: [i64; 1] = [-1i64];
let constant725_out1: [i64; 1] = [8i64];
let concat8_out1: [i64; 4usize] = [
&unsqueeze17_out1[..],
&constant724_out1[..],
&constant725_out1[..],
&constant726_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape3_out1 = linear4_out1.reshape(concat6_out1);
let linear5_out1 = self.linear5.forward(layernormalization2_out1.clone());
let reshape4_out1 = linear5_out1.reshape(concat7_out1);
let linear6_out1 = self.linear6.forward(layernormalization2_out1);
let reshape5_out1 = linear6_out1.reshape(concat8_out1);
let transpose4_out1 = reshape5_out1.permute([0, 2, 1, 3]);
let linear7_out1 = self.linear7.forward(slice1_out1);
let reshape6_out1 = linear7_out1.reshape([1, -1, 8, 128]);
let constant20_out1 = self.constant20.val();
let add9_out1 = reshape3_out1
.clone()
.add((constant20_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose5_out1 = add9_out1.permute([0, 2, 1, 3]);
let constant21_out1 = self.constant21.val();
let add10_out1 = reshape3_out1
.add((constant21_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose6_out1 = add10_out1.permute([0, 2, 1, 3]);
let transpose7_out1 = reshape6_out1.permute([0, 2, 3, 1]);
let matmul8_out1 = transpose6_out1.matmul(transpose7_out1);
let shape9_out1: [i64; 4] = {
let axes = &matmul8_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather7_out1 = shape9_out1[0] as i64;
let gather8_out1 = shape9_out1[1] as i64;
let gather9_out1 = shape9_out1[2] as i64;
let gather10_out1 = shape9_out1[3] as i64;
let pad1_out1 = matmul8_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze42_out1 = [gather7_out1 as i64];
let unsqueeze43_out1 = [gather8_out1 as i64];
let unsqueeze44_out1 = [gather9_out1 as i64];
let constant838_out1: [i64; 1] = [-1i64];
let concat34_out1: [i64; 4usize] = [
&unsqueeze42_out1[..],
&unsqueeze43_out1[..],
&constant838_out1[..],
&unsqueeze44_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape9_out1 = pad1_out1.reshape(concat34_out1);
let slice5_out1 = reshape9_out1.slice(s![.., .., 1.., ..]);
let unsqueeze45_out1 = [gather7_out1 as i64];
let unsqueeze46_out1 = [gather8_out1 as i64];
let unsqueeze47_out1 = [gather9_out1 as i64];
let unsqueeze48_out1 = [gather10_out1 as i64];
let concat35_out1: [i64; 4usize] = [
&unsqueeze45_out1[..],
&unsqueeze46_out1[..],
&unsqueeze47_out1[..],
&unsqueeze48_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape10_out1 = slice5_out1.reshape(concat35_out1);
let transpose9_out1 = reshape4_out1.permute([0, 2, 3, 1]);
let matmul9_out1 = transpose5_out1.matmul(transpose9_out1);
let shape13_out1: [i64; 4] = {
let axes = &matmul9_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice6_out1: [i64; 1] = shape13_out1[3..4].try_into().unwrap();
let squeeze1_out1 = slice6_out1[0] as i64;
let unsqueeze49_out1 = [squeeze1_out1 as i64];
let slice7_out1 = reshape10_out1.slice(s![.., .., .., 0..unsqueeze49_out1[0]]);
let add11_out1 = matmul9_out1.add(slice7_out1);
let constant856_out1 = self.constant856.val();
let div4_out1 = add11_out1
.div((constant856_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape14_out1: [i64; 4] = {
let axes = &transpose4_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather11_out1 = shape14_out1[0] as i64;
let unsqueeze50_out1: Tensor<4, Bool> = not1_out1.unsqueeze_dims::<4>(&[1]);
let constant859_out1 = -10000f32;
let where2_out1 = div4_out1
.mask_fill(unsqueeze50_out1.clone(), constant859_out1);
let softmax1_out1 = burn::tensor::activation::softmax(where2_out1, 3);
let constant860_out1 = 0f32;
let where3_out1 = softmax1_out1
.mask_fill(unsqueeze50_out1.clone(), constant860_out1);
let matmul10_out1 = where3_out1.matmul(transpose4_out1);
let transpose10_out1 = matmul10_out1.permute([0, 2, 1, 3]);
let unsqueeze51_out1 = [gather11_out1 as i64];
let constant863_out1: [i64; 1] = [1024i64];
let constant862_out1: [i64; 1] = [-1i64];
let concat36_out1: [i64; 3usize] = [
&unsqueeze51_out1[..],
&constant862_out1[..],
&constant863_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape11_out1 = transpose10_out1.reshape(concat36_out1);
let linear8_out1 = self.linear8.forward(reshape11_out1);
let add12_out1 = add8_out1.add(linear8_out1);
(add12_out1, not2_out1, unsqueeze50_out1)
}
}
#[derive(Module, Debug)]
pub struct Submodule3 {
layernormalization3: LayerNorm,
conv1d1: Conv1d,
conv1d2: Conv1d,
conv1d3: Conv1d,
layernormalization4: LayerNorm,
linear9: Linear,
linear10: Linear,
constant874: burn::module::Param<Tensor<1>>,
layernormalization5: LayerNorm,
layernormalization6: LayerNorm,
linear11: Linear,
linear12: Linear,
constant875: burn::module::Param<Tensor<1>>,
layernormalization7: LayerNorm,
linear13: Linear,
linear14: Linear,
linear15: Linear,
linear16: Linear,
constant34: burn::module::Param<Tensor<2>>,
constant35: burn::module::Param<Tensor<2>>,
constant921: burn::module::Param<Tensor<1>>,
linear17: Linear,
#[module(skip)]
device: Device,
}
impl Submodule3 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization3 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d1 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d2 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d3 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization4 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear9 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear10 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant874: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization5 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization6 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear11 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear12 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant875: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization7 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear13 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear14 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear15 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear16 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant34: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant35: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant921: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear17 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization3,
conv1d1,
conv1d2,
conv1d3,
layernormalization4,
linear9,
linear10,
constant874,
layernormalization5,
layernormalization6,
linear11,
linear12,
constant875,
layernormalization7,
linear13,
linear14,
linear15,
linear16,
constant34,
constant35,
constant921,
linear17,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add12_out1: Tensor<3>,
not2_out1: Tensor<2, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> (Tensor<3>, Tensor<3, Bool>) {
let layernormalization3_out1 = {
let dtype = add12_out1.clone().dtype();
self.layernormalization3
.forward(add12_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose11_out1 = layernormalization3_out1.permute([0, 2, 1]);
let conv1d1_out1 = self.conv1d1.forward(transpose11_out1);
let split_tensors = conv1d1_out1.split_with_sizes([1024, 1024].into(), 1);
let [split1_out1, split1_out2] = split_tensors.try_into().unwrap();
let sigmoid2_out1 = burn::tensor::activation::sigmoid(split1_out2);
let mul5_out1 = split1_out1.mul(sigmoid2_out1);
let unsqueeze52_out1: Tensor<3, Bool> = not2_out1.unsqueeze_dims::<3>(&[1]);
let constant865_out1 = 0f32;
let where4_out1 = mul5_out1
.mask_fill(unsqueeze52_out1.clone(), constant865_out1);
let pad2_out1 = where4_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d2_out1 = self.conv1d2.forward(pad2_out1);
let sigmoid3_out1 = burn::tensor::activation::sigmoid(conv1d2_out1.clone());
let mul6_out1 = conv1d2_out1.mul(sigmoid3_out1);
let conv1d3_out1 = self.conv1d3.forward(mul6_out1);
let transpose13_out1 = conv1d3_out1.permute([0, 2, 1]);
let add13_out1 = add12_out1.add(transpose13_out1);
let layernormalization4_out1 = {
let dtype = add13_out1.clone().dtype();
self.layernormalization4
.forward(add13_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear9_out1 = self.linear9.forward(layernormalization4_out1);
let sigmoid4_out1 = burn::tensor::activation::sigmoid(linear9_out1.clone());
let mul7_out1 = linear9_out1.mul(sigmoid4_out1);
let linear10_out1 = self.linear10.forward(mul7_out1);
let constant874_out1 = self.constant874.val();
let mul8_out1 = linear10_out1
.mul((constant874_out1).unsqueeze_dims(&[0isize, 1isize]));
let add14_out1 = add13_out1.add(mul8_out1);
let layernormalization5_out1 = {
let dtype = add14_out1.dtype();
self.layernormalization5
.forward(add14_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization6_out1 = {
let dtype = layernormalization5_out1.clone().dtype();
self.layernormalization6
.forward(layernormalization5_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear11_out1 = self.linear11.forward(layernormalization6_out1);
let sigmoid5_out1 = burn::tensor::activation::sigmoid(linear11_out1.clone());
let mul9_out1 = linear11_out1.mul(sigmoid5_out1);
let linear12_out1 = self.linear12.forward(mul9_out1);
let constant875_out1 = self.constant875.val();
let mul10_out1 = linear12_out1
.mul((constant875_out1).unsqueeze_dims(&[0isize, 1isize]));
let add15_out1 = layernormalization5_out1.add(mul10_out1);
let layernormalization7_out1 = {
let dtype = add15_out1.clone().dtype();
self.layernormalization7
.forward(add15_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape15_out1: [i64; 3] = {
let axes = &layernormalization7_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather12_out1 = shape15_out1[0] as i64;
let linear13_out1 = self.linear13.forward(layernormalization7_out1.clone());
let unsqueeze53_out1 = [gather12_out1 as i64];
let constant880_out1: [i64; 1] = [128i64];
let constant878_out1: [i64; 1] = [-1i64];
let constant879_out1: [i64; 1] = [8i64];
let concat38_out1: [i64; 4usize] = [
&unsqueeze53_out1[..],
&constant878_out1[..],
&constant879_out1[..],
&constant880_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze54_out1 = [gather12_out1 as i64];
let constant884_out1: [i64; 1] = [128i64];
let constant882_out1: [i64; 1] = [-1i64];
let constant883_out1: [i64; 1] = [8i64];
let concat39_out1: [i64; 4usize] = [
&unsqueeze54_out1[..],
&constant882_out1[..],
&constant883_out1[..],
&constant884_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze55_out1 = [gather12_out1 as i64];
let constant888_out1: [i64; 1] = [128i64];
let constant886_out1: [i64; 1] = [-1i64];
let constant887_out1: [i64; 1] = [8i64];
let concat40_out1: [i64; 4usize] = [
&unsqueeze55_out1[..],
&constant886_out1[..],
&constant887_out1[..],
&constant888_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape14_out1 = linear13_out1.reshape(concat38_out1);
let linear14_out1 = self.linear14.forward(layernormalization7_out1.clone());
let reshape15_out1 = linear14_out1.reshape(concat39_out1);
let linear15_out1 = self.linear15.forward(layernormalization7_out1);
let reshape16_out1 = linear15_out1.reshape(concat40_out1);
let transpose14_out1 = reshape16_out1.permute([0, 2, 1, 3]);
let linear16_out1 = self.linear16.forward(slice1_out1);
let reshape17_out1 = linear16_out1.reshape([1, -1, 8, 128]);
let constant34_out1 = self.constant34.val();
let add16_out1 = reshape14_out1
.clone()
.add((constant34_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose15_out1 = add16_out1.permute([0, 2, 1, 3]);
let constant35_out1 = self.constant35.val();
let add17_out1 = reshape14_out1
.add((constant35_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose16_out1 = add17_out1.permute([0, 2, 1, 3]);
let transpose17_out1 = reshape17_out1.permute([0, 2, 3, 1]);
let matmul20_out1 = transpose16_out1.matmul(transpose17_out1);
let shape16_out1: [i64; 4] = {
let axes = &matmul20_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather13_out1 = shape16_out1[0] as i64;
let gather14_out1 = shape16_out1[1] as i64;
let gather15_out1 = shape16_out1[2] as i64;
let gather16_out1 = shape16_out1[3] as i64;
let pad3_out1 = matmul20_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze56_out1 = [gather13_out1 as i64];
let unsqueeze57_out1 = [gather14_out1 as i64];
let unsqueeze58_out1 = [gather15_out1 as i64];
let constant903_out1: [i64; 1] = [-1i64];
let concat42_out1: [i64; 4usize] = [
&unsqueeze56_out1[..],
&unsqueeze57_out1[..],
&constant903_out1[..],
&unsqueeze58_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape20_out1 = pad3_out1.reshape(concat42_out1);
let slice10_out1 = reshape20_out1.slice(s![.., .., 1.., ..]);
let unsqueeze59_out1 = [gather13_out1 as i64];
let unsqueeze60_out1 = [gather14_out1 as i64];
let unsqueeze61_out1 = [gather15_out1 as i64];
let unsqueeze62_out1 = [gather16_out1 as i64];
let concat43_out1: [i64; 4usize] = [
&unsqueeze59_out1[..],
&unsqueeze60_out1[..],
&unsqueeze61_out1[..],
&unsqueeze62_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape21_out1 = slice10_out1.reshape(concat43_out1);
let transpose19_out1 = reshape15_out1.permute([0, 2, 3, 1]);
let matmul21_out1 = transpose15_out1.matmul(transpose19_out1);
let shape20_out1: [i64; 4] = {
let axes = &matmul21_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice11_out1: [i64; 1] = shape20_out1[3..4].try_into().unwrap();
let squeeze2_out1 = slice11_out1[0] as i64;
let unsqueeze63_out1 = [squeeze2_out1 as i64];
let slice12_out1 = reshape21_out1.slice(s![.., .., .., 0..unsqueeze63_out1[0]]);
let add18_out1 = matmul21_out1.add(slice12_out1);
let constant921_out1 = self.constant921.val();
let div5_out1 = add18_out1
.div((constant921_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape21_out1: [i64; 4] = {
let axes = &transpose14_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather17_out1 = shape21_out1[0] as i64;
let constant923_out1 = -10000f32;
let where5_out1 = div5_out1
.mask_fill(unsqueeze50_out1.clone(), constant923_out1);
let softmax2_out1 = burn::tensor::activation::softmax(where5_out1, 3);
let constant924_out1 = 0f32;
let where6_out1 = softmax2_out1.mask_fill(unsqueeze50_out1, constant924_out1);
let matmul22_out1 = where6_out1.matmul(transpose14_out1);
let transpose20_out1 = matmul22_out1.permute([0, 2, 1, 3]);
let unsqueeze64_out1 = [gather17_out1 as i64];
let constant927_out1: [i64; 1] = [1024i64];
let constant926_out1: [i64; 1] = [-1i64];
let concat44_out1: [i64; 3usize] = [
&unsqueeze64_out1[..],
&constant926_out1[..],
&constant927_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape22_out1 = transpose20_out1.reshape(concat44_out1);
let linear17_out1 = self.linear17.forward(reshape22_out1);
let add19_out1 = add15_out1.add(linear17_out1);
(add19_out1, unsqueeze52_out1)
}
}
#[derive(Module, Debug)]
pub struct Submodule4 {
layernormalization8: LayerNorm,
conv1d4: Conv1d,
conv1d5: Conv1d,
conv1d6: Conv1d,
layernormalization9: LayerNorm,
linear18: Linear,
linear19: Linear,
constant937: burn::module::Param<Tensor<1>>,
layernormalization10: LayerNorm,
layernormalization11: LayerNorm,
linear20: Linear,
linear21: Linear,
constant938: burn::module::Param<Tensor<1>>,
layernormalization12: LayerNorm,
linear22: Linear,
linear23: Linear,
linear24: Linear,
linear25: Linear,
constant48: burn::module::Param<Tensor<2>>,
constant49: burn::module::Param<Tensor<2>>,
constant984: burn::module::Param<Tensor<1>>,
linear26: Linear,
#[module(skip)]
device: Device,
}
impl Submodule4 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization8 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d4 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d5 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d6 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization9 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear18 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear19 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant937: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization10 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization11 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear20 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear21 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant938: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization12 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear22 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear23 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear24 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear25 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant48: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant49: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant984: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear26 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization8,
conv1d4,
conv1d5,
conv1d6,
layernormalization9,
linear18,
linear19,
constant937,
layernormalization10,
layernormalization11,
linear20,
linear21,
constant938,
layernormalization12,
linear22,
linear23,
linear24,
linear25,
constant48,
constant49,
constant984,
linear26,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add19_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> Tensor<3> {
let layernormalization8_out1 = {
let dtype = add19_out1.clone().dtype();
self.layernormalization8
.forward(add19_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose21_out1 = layernormalization8_out1.permute([0, 2, 1]);
let conv1d4_out1 = self.conv1d4.forward(transpose21_out1);
let split_tensors = conv1d4_out1.split_with_sizes([1024, 1024].into(), 1);
let [split2_out1, split2_out2] = split_tensors.try_into().unwrap();
let sigmoid6_out1 = burn::tensor::activation::sigmoid(split2_out2);
let mul11_out1 = split2_out1.mul(sigmoid6_out1);
let constant928_out1 = 0f32;
let where7_out1 = mul11_out1.mask_fill(unsqueeze52_out1, constant928_out1);
let pad4_out1 = where7_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d5_out1 = self.conv1d5.forward(pad4_out1);
let sigmoid7_out1 = burn::tensor::activation::sigmoid(conv1d5_out1.clone());
let mul12_out1 = conv1d5_out1.mul(sigmoid7_out1);
let conv1d6_out1 = self.conv1d6.forward(mul12_out1);
let transpose23_out1 = conv1d6_out1.permute([0, 2, 1]);
let add20_out1 = add19_out1.add(transpose23_out1);
let layernormalization9_out1 = {
let dtype = add20_out1.clone().dtype();
self.layernormalization9
.forward(add20_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear18_out1 = self.linear18.forward(layernormalization9_out1);
let sigmoid8_out1 = burn::tensor::activation::sigmoid(linear18_out1.clone());
let mul13_out1 = linear18_out1.mul(sigmoid8_out1);
let linear19_out1 = self.linear19.forward(mul13_out1);
let constant937_out1 = self.constant937.val();
let mul14_out1 = linear19_out1
.mul((constant937_out1).unsqueeze_dims(&[0isize, 1isize]));
let add21_out1 = add20_out1.add(mul14_out1);
let layernormalization10_out1 = {
let dtype = add21_out1.dtype();
self.layernormalization10
.forward(add21_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization11_out1 = {
let dtype = layernormalization10_out1.clone().dtype();
self.layernormalization11
.forward(
layernormalization10_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear20_out1 = self.linear20.forward(layernormalization11_out1);
let sigmoid9_out1 = burn::tensor::activation::sigmoid(linear20_out1.clone());
let mul15_out1 = linear20_out1.mul(sigmoid9_out1);
let linear21_out1 = self.linear21.forward(mul15_out1);
let constant938_out1 = self.constant938.val();
let mul16_out1 = linear21_out1
.mul((constant938_out1).unsqueeze_dims(&[0isize, 1isize]));
let add22_out1 = layernormalization10_out1.add(mul16_out1);
let layernormalization12_out1 = {
let dtype = add22_out1.clone().dtype();
self.layernormalization12
.forward(add22_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape22_out1: [i64; 3] = {
let axes = &layernormalization12_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather18_out1 = shape22_out1[0] as i64;
let linear22_out1 = self.linear22.forward(layernormalization12_out1.clone());
let unsqueeze65_out1 = [gather18_out1 as i64];
let constant943_out1: [i64; 1] = [128i64];
let constant941_out1: [i64; 1] = [-1i64];
let constant942_out1: [i64; 1] = [8i64];
let concat46_out1: [i64; 4usize] = [
&unsqueeze65_out1[..],
&constant941_out1[..],
&constant942_out1[..],
&constant943_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze66_out1 = [gather18_out1 as i64];
let constant947_out1: [i64; 1] = [128i64];
let constant945_out1: [i64; 1] = [-1i64];
let constant946_out1: [i64; 1] = [8i64];
let concat47_out1: [i64; 4usize] = [
&unsqueeze66_out1[..],
&constant945_out1[..],
&constant946_out1[..],
&constant947_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze67_out1 = [gather18_out1 as i64];
let constant951_out1: [i64; 1] = [128i64];
let constant949_out1: [i64; 1] = [-1i64];
let constant950_out1: [i64; 1] = [8i64];
let concat48_out1: [i64; 4usize] = [
&unsqueeze67_out1[..],
&constant949_out1[..],
&constant950_out1[..],
&constant951_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape25_out1 = linear22_out1.reshape(concat46_out1);
let linear23_out1 = self.linear23.forward(layernormalization12_out1.clone());
let reshape26_out1 = linear23_out1.reshape(concat47_out1);
let linear24_out1 = self.linear24.forward(layernormalization12_out1);
let reshape27_out1 = linear24_out1.reshape(concat48_out1);
let transpose24_out1 = reshape27_out1.permute([0, 2, 1, 3]);
let linear25_out1 = self.linear25.forward(slice1_out1);
let reshape28_out1 = linear25_out1.reshape([1, -1, 8, 128]);
let constant48_out1 = self.constant48.val();
let add23_out1 = reshape25_out1
.clone()
.add((constant48_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose25_out1 = add23_out1.permute([0, 2, 1, 3]);
let constant49_out1 = self.constant49.val();
let add24_out1 = reshape25_out1
.add((constant49_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose26_out1 = add24_out1.permute([0, 2, 1, 3]);
let transpose27_out1 = reshape28_out1.permute([0, 2, 3, 1]);
let matmul32_out1 = transpose26_out1.matmul(transpose27_out1);
let shape23_out1: [i64; 4] = {
let axes = &matmul32_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather19_out1 = shape23_out1[0] as i64;
let gather20_out1 = shape23_out1[1] as i64;
let gather21_out1 = shape23_out1[2] as i64;
let gather22_out1 = shape23_out1[3] as i64;
let pad5_out1 = matmul32_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze68_out1 = [gather19_out1 as i64];
let unsqueeze69_out1 = [gather20_out1 as i64];
let unsqueeze70_out1 = [gather21_out1 as i64];
let constant966_out1: [i64; 1] = [-1i64];
let concat50_out1: [i64; 4usize] = [
&unsqueeze68_out1[..],
&unsqueeze69_out1[..],
&constant966_out1[..],
&unsqueeze70_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape31_out1 = pad5_out1.reshape(concat50_out1);
let slice15_out1 = reshape31_out1.slice(s![.., .., 1.., ..]);
let unsqueeze71_out1 = [gather19_out1 as i64];
let unsqueeze72_out1 = [gather20_out1 as i64];
let unsqueeze73_out1 = [gather21_out1 as i64];
let unsqueeze74_out1 = [gather22_out1 as i64];
let concat51_out1: [i64; 4usize] = [
&unsqueeze71_out1[..],
&unsqueeze72_out1[..],
&unsqueeze73_out1[..],
&unsqueeze74_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape32_out1 = slice15_out1.reshape(concat51_out1);
let transpose29_out1 = reshape26_out1.permute([0, 2, 3, 1]);
let matmul33_out1 = transpose25_out1.matmul(transpose29_out1);
let shape27_out1: [i64; 4] = {
let axes = &matmul33_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice16_out1: [i64; 1] = shape27_out1[3..4].try_into().unwrap();
let squeeze3_out1 = slice16_out1[0] as i64;
let unsqueeze75_out1 = [squeeze3_out1 as i64];
let slice17_out1 = reshape32_out1.slice(s![.., .., .., 0..unsqueeze75_out1[0]]);
let add25_out1 = matmul33_out1.add(slice17_out1);
let constant984_out1 = self.constant984.val();
let div6_out1 = add25_out1
.div((constant984_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape28_out1: [i64; 4] = {
let axes = &transpose24_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather23_out1 = shape28_out1[0] as i64;
let constant986_out1 = -10000f32;
let where8_out1 = div6_out1
.mask_fill(unsqueeze50_out1.clone(), constant986_out1);
let softmax3_out1 = burn::tensor::activation::softmax(where8_out1, 3);
let constant987_out1 = 0f32;
let where9_out1 = softmax3_out1.mask_fill(unsqueeze50_out1, constant987_out1);
let matmul34_out1 = where9_out1.matmul(transpose24_out1);
let transpose30_out1 = matmul34_out1.permute([0, 2, 1, 3]);
let unsqueeze76_out1 = [gather23_out1 as i64];
let constant990_out1: [i64; 1] = [1024i64];
let constant989_out1: [i64; 1] = [-1i64];
let concat52_out1: [i64; 3usize] = [
&unsqueeze76_out1[..],
&constant989_out1[..],
&constant990_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape33_out1 = transpose30_out1.reshape(concat52_out1);
let linear26_out1 = self.linear26.forward(reshape33_out1);
let add26_out1 = add22_out1.add(linear26_out1);
add26_out1
}
}
#[derive(Module, Debug)]
pub struct Submodule5 {
layernormalization13: LayerNorm,
conv1d7: Conv1d,
conv1d8: Conv1d,
conv1d9: Conv1d,
layernormalization14: LayerNorm,
linear27: Linear,
linear28: Linear,
constant1000: burn::module::Param<Tensor<1>>,
layernormalization15: LayerNorm,
layernormalization16: LayerNorm,
linear29: Linear,
linear30: Linear,
constant1001: burn::module::Param<Tensor<1>>,
layernormalization17: LayerNorm,
linear31: Linear,
linear32: Linear,
linear33: Linear,
linear34: Linear,
constant62: burn::module::Param<Tensor<2>>,
constant63: burn::module::Param<Tensor<2>>,
constant1047: burn::module::Param<Tensor<1>>,
linear35: Linear,
#[module(skip)]
device: Device,
}
impl Submodule5 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization13 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d7 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d8 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d9 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization14 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear27 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear28 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1000: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization15 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization16 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear29 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear30 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1001: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization17 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear31 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear32 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear33 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear34 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant62: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant63: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant1047: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear35 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization13,
conv1d7,
conv1d8,
conv1d9,
layernormalization14,
linear27,
linear28,
constant1000,
layernormalization15,
layernormalization16,
linear29,
linear30,
constant1001,
layernormalization17,
linear31,
linear32,
linear33,
linear34,
constant62,
constant63,
constant1047,
linear35,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add26_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> Tensor<3> {
let layernormalization13_out1 = {
let dtype = add26_out1.clone().dtype();
self.layernormalization13
.forward(add26_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose31_out1 = layernormalization13_out1.permute([0, 2, 1]);
let conv1d7_out1 = self.conv1d7.forward(transpose31_out1);
let split_tensors = conv1d7_out1.split_with_sizes([1024, 1024].into(), 1);
let [split3_out1, split3_out2] = split_tensors.try_into().unwrap();
let sigmoid10_out1 = burn::tensor::activation::sigmoid(split3_out2);
let mul17_out1 = split3_out1.mul(sigmoid10_out1);
let constant991_out1 = 0f32;
let where10_out1 = mul17_out1.mask_fill(unsqueeze52_out1, constant991_out1);
let pad6_out1 = where10_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d8_out1 = self.conv1d8.forward(pad6_out1);
let sigmoid11_out1 = burn::tensor::activation::sigmoid(conv1d8_out1.clone());
let mul18_out1 = conv1d8_out1.mul(sigmoid11_out1);
let conv1d9_out1 = self.conv1d9.forward(mul18_out1);
let transpose33_out1 = conv1d9_out1.permute([0, 2, 1]);
let add27_out1 = add26_out1.add(transpose33_out1);
let layernormalization14_out1 = {
let dtype = add27_out1.clone().dtype();
self.layernormalization14
.forward(add27_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear27_out1 = self.linear27.forward(layernormalization14_out1);
let sigmoid12_out1 = burn::tensor::activation::sigmoid(linear27_out1.clone());
let mul19_out1 = linear27_out1.mul(sigmoid12_out1);
let linear28_out1 = self.linear28.forward(mul19_out1);
let constant1000_out1 = self.constant1000.val();
let mul20_out1 = linear28_out1
.mul((constant1000_out1).unsqueeze_dims(&[0isize, 1isize]));
let add28_out1 = add27_out1.add(mul20_out1);
let layernormalization15_out1 = {
let dtype = add28_out1.dtype();
self.layernormalization15
.forward(add28_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization16_out1 = {
let dtype = layernormalization15_out1.clone().dtype();
self.layernormalization16
.forward(
layernormalization15_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear29_out1 = self.linear29.forward(layernormalization16_out1);
let sigmoid13_out1 = burn::tensor::activation::sigmoid(linear29_out1.clone());
let mul21_out1 = linear29_out1.mul(sigmoid13_out1);
let linear30_out1 = self.linear30.forward(mul21_out1);
let constant1001_out1 = self.constant1001.val();
let mul22_out1 = linear30_out1
.mul((constant1001_out1).unsqueeze_dims(&[0isize, 1isize]));
let add29_out1 = layernormalization15_out1.add(mul22_out1);
let layernormalization17_out1 = {
let dtype = add29_out1.clone().dtype();
self.layernormalization17
.forward(add29_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape29_out1: [i64; 3] = {
let axes = &layernormalization17_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather24_out1 = shape29_out1[0] as i64;
let linear31_out1 = self.linear31.forward(layernormalization17_out1.clone());
let unsqueeze77_out1 = [gather24_out1 as i64];
let constant1006_out1: [i64; 1] = [128i64];
let constant1004_out1: [i64; 1] = [-1i64];
let constant1005_out1: [i64; 1] = [8i64];
let concat54_out1: [i64; 4usize] = [
&unsqueeze77_out1[..],
&constant1004_out1[..],
&constant1005_out1[..],
&constant1006_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze78_out1 = [gather24_out1 as i64];
let constant1010_out1: [i64; 1] = [128i64];
let constant1008_out1: [i64; 1] = [-1i64];
let constant1009_out1: [i64; 1] = [8i64];
let concat55_out1: [i64; 4usize] = [
&unsqueeze78_out1[..],
&constant1008_out1[..],
&constant1009_out1[..],
&constant1010_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze79_out1 = [gather24_out1 as i64];
let constant1014_out1: [i64; 1] = [128i64];
let constant1012_out1: [i64; 1] = [-1i64];
let constant1013_out1: [i64; 1] = [8i64];
let concat56_out1: [i64; 4usize] = [
&unsqueeze79_out1[..],
&constant1012_out1[..],
&constant1013_out1[..],
&constant1014_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape36_out1 = linear31_out1.reshape(concat54_out1);
let linear32_out1 = self.linear32.forward(layernormalization17_out1.clone());
let reshape37_out1 = linear32_out1.reshape(concat55_out1);
let linear33_out1 = self.linear33.forward(layernormalization17_out1);
let reshape38_out1 = linear33_out1.reshape(concat56_out1);
let transpose34_out1 = reshape38_out1.permute([0, 2, 1, 3]);
let linear34_out1 = self.linear34.forward(slice1_out1);
let reshape39_out1 = linear34_out1.reshape([1, -1, 8, 128]);
let constant62_out1 = self.constant62.val();
let add30_out1 = reshape36_out1
.clone()
.add((constant62_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose35_out1 = add30_out1.permute([0, 2, 1, 3]);
let constant63_out1 = self.constant63.val();
let add31_out1 = reshape36_out1
.add((constant63_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose36_out1 = add31_out1.permute([0, 2, 1, 3]);
let transpose37_out1 = reshape39_out1.permute([0, 2, 3, 1]);
let matmul44_out1 = transpose36_out1.matmul(transpose37_out1);
let shape30_out1: [i64; 4] = {
let axes = &matmul44_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather25_out1 = shape30_out1[0] as i64;
let gather26_out1 = shape30_out1[1] as i64;
let gather27_out1 = shape30_out1[2] as i64;
let gather28_out1 = shape30_out1[3] as i64;
let pad7_out1 = matmul44_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze80_out1 = [gather25_out1 as i64];
let unsqueeze81_out1 = [gather26_out1 as i64];
let unsqueeze82_out1 = [gather27_out1 as i64];
let constant1029_out1: [i64; 1] = [-1i64];
let concat58_out1: [i64; 4usize] = [
&unsqueeze80_out1[..],
&unsqueeze81_out1[..],
&constant1029_out1[..],
&unsqueeze82_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape42_out1 = pad7_out1.reshape(concat58_out1);
let slice20_out1 = reshape42_out1.slice(s![.., .., 1.., ..]);
let unsqueeze83_out1 = [gather25_out1 as i64];
let unsqueeze84_out1 = [gather26_out1 as i64];
let unsqueeze85_out1 = [gather27_out1 as i64];
let unsqueeze86_out1 = [gather28_out1 as i64];
let concat59_out1: [i64; 4usize] = [
&unsqueeze83_out1[..],
&unsqueeze84_out1[..],
&unsqueeze85_out1[..],
&unsqueeze86_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape43_out1 = slice20_out1.reshape(concat59_out1);
let transpose39_out1 = reshape37_out1.permute([0, 2, 3, 1]);
let matmul45_out1 = transpose35_out1.matmul(transpose39_out1);
let shape34_out1: [i64; 4] = {
let axes = &matmul45_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice21_out1: [i64; 1] = shape34_out1[3..4].try_into().unwrap();
let squeeze4_out1 = slice21_out1[0] as i64;
let unsqueeze87_out1 = [squeeze4_out1 as i64];
let slice22_out1 = reshape43_out1.slice(s![.., .., .., 0..unsqueeze87_out1[0]]);
let add32_out1 = matmul45_out1.add(slice22_out1);
let constant1047_out1 = self.constant1047.val();
let div7_out1 = add32_out1
.div((constant1047_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape35_out1: [i64; 4] = {
let axes = &transpose34_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather29_out1 = shape35_out1[0] as i64;
let constant1049_out1 = -10000f32;
let where11_out1 = div7_out1
.mask_fill(unsqueeze50_out1.clone(), constant1049_out1);
let softmax4_out1 = burn::tensor::activation::softmax(where11_out1, 3);
let constant1050_out1 = 0f32;
let where12_out1 = softmax4_out1.mask_fill(unsqueeze50_out1, constant1050_out1);
let matmul46_out1 = where12_out1.matmul(transpose34_out1);
let transpose40_out1 = matmul46_out1.permute([0, 2, 1, 3]);
let unsqueeze88_out1 = [gather29_out1 as i64];
let constant1053_out1: [i64; 1] = [1024i64];
let constant1052_out1: [i64; 1] = [-1i64];
let concat60_out1: [i64; 3usize] = [
&unsqueeze88_out1[..],
&constant1052_out1[..],
&constant1053_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape44_out1 = transpose40_out1.reshape(concat60_out1);
let linear35_out1 = self.linear35.forward(reshape44_out1);
let add33_out1 = add29_out1.add(linear35_out1);
add33_out1
}
}
#[derive(Module, Debug)]
pub struct Submodule6 {
layernormalization18: LayerNorm,
conv1d10: Conv1d,
conv1d11: Conv1d,
conv1d12: Conv1d,
layernormalization19: LayerNorm,
linear36: Linear,
linear37: Linear,
constant1063: burn::module::Param<Tensor<1>>,
layernormalization20: LayerNorm,
layernormalization21: LayerNorm,
linear38: Linear,
linear39: Linear,
constant1064: burn::module::Param<Tensor<1>>,
layernormalization22: LayerNorm,
linear40: Linear,
linear41: Linear,
linear42: Linear,
linear43: Linear,
constant76: burn::module::Param<Tensor<2>>,
constant77: burn::module::Param<Tensor<2>>,
constant1110: burn::module::Param<Tensor<1>>,
linear44: Linear,
#[module(skip)]
device: Device,
}
impl Submodule6 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization18 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d10 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d11 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d12 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization19 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear36 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear37 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1063: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization20 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization21 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear38 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear39 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1064: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization22 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear40 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear41 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear42 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear43 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant76: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant77: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant1110: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear44 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization18,
conv1d10,
conv1d11,
conv1d12,
layernormalization19,
linear36,
linear37,
constant1063,
layernormalization20,
layernormalization21,
linear38,
linear39,
constant1064,
layernormalization22,
linear40,
linear41,
linear42,
linear43,
constant76,
constant77,
constant1110,
linear44,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add33_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> Tensor<3> {
let layernormalization18_out1 = {
let dtype = add33_out1.clone().dtype();
self.layernormalization18
.forward(add33_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose41_out1 = layernormalization18_out1.permute([0, 2, 1]);
let conv1d10_out1 = self.conv1d10.forward(transpose41_out1);
let split_tensors = conv1d10_out1.split_with_sizes([1024, 1024].into(), 1);
let [split4_out1, split4_out2] = split_tensors.try_into().unwrap();
let sigmoid14_out1 = burn::tensor::activation::sigmoid(split4_out2);
let mul23_out1 = split4_out1.mul(sigmoid14_out1);
let constant1054_out1 = 0f32;
let where13_out1 = mul23_out1.mask_fill(unsqueeze52_out1, constant1054_out1);
let pad8_out1 = where13_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d11_out1 = self.conv1d11.forward(pad8_out1);
let sigmoid15_out1 = burn::tensor::activation::sigmoid(conv1d11_out1.clone());
let mul24_out1 = conv1d11_out1.mul(sigmoid15_out1);
let conv1d12_out1 = self.conv1d12.forward(mul24_out1);
let transpose43_out1 = conv1d12_out1.permute([0, 2, 1]);
let add34_out1 = add33_out1.add(transpose43_out1);
let layernormalization19_out1 = {
let dtype = add34_out1.clone().dtype();
self.layernormalization19
.forward(add34_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear36_out1 = self.linear36.forward(layernormalization19_out1);
let sigmoid16_out1 = burn::tensor::activation::sigmoid(linear36_out1.clone());
let mul25_out1 = linear36_out1.mul(sigmoid16_out1);
let linear37_out1 = self.linear37.forward(mul25_out1);
let constant1063_out1 = self.constant1063.val();
let mul26_out1 = linear37_out1
.mul((constant1063_out1).unsqueeze_dims(&[0isize, 1isize]));
let add35_out1 = add34_out1.add(mul26_out1);
let layernormalization20_out1 = {
let dtype = add35_out1.dtype();
self.layernormalization20
.forward(add35_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization21_out1 = {
let dtype = layernormalization20_out1.clone().dtype();
self.layernormalization21
.forward(
layernormalization20_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear38_out1 = self.linear38.forward(layernormalization21_out1);
let sigmoid17_out1 = burn::tensor::activation::sigmoid(linear38_out1.clone());
let mul27_out1 = linear38_out1.mul(sigmoid17_out1);
let linear39_out1 = self.linear39.forward(mul27_out1);
let constant1064_out1 = self.constant1064.val();
let mul28_out1 = linear39_out1
.mul((constant1064_out1).unsqueeze_dims(&[0isize, 1isize]));
let add36_out1 = layernormalization20_out1.add(mul28_out1);
let layernormalization22_out1 = {
let dtype = add36_out1.clone().dtype();
self.layernormalization22
.forward(add36_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape36_out1: [i64; 3] = {
let axes = &layernormalization22_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather30_out1 = shape36_out1[0] as i64;
let linear40_out1 = self.linear40.forward(layernormalization22_out1.clone());
let unsqueeze89_out1 = [gather30_out1 as i64];
let constant1069_out1: [i64; 1] = [128i64];
let constant1067_out1: [i64; 1] = [-1i64];
let constant1068_out1: [i64; 1] = [8i64];
let concat62_out1: [i64; 4usize] = [
&unsqueeze89_out1[..],
&constant1067_out1[..],
&constant1068_out1[..],
&constant1069_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze90_out1 = [gather30_out1 as i64];
let constant1073_out1: [i64; 1] = [128i64];
let constant1071_out1: [i64; 1] = [-1i64];
let constant1072_out1: [i64; 1] = [8i64];
let concat63_out1: [i64; 4usize] = [
&unsqueeze90_out1[..],
&constant1071_out1[..],
&constant1072_out1[..],
&constant1073_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze91_out1 = [gather30_out1 as i64];
let constant1077_out1: [i64; 1] = [128i64];
let constant1075_out1: [i64; 1] = [-1i64];
let constant1076_out1: [i64; 1] = [8i64];
let concat64_out1: [i64; 4usize] = [
&unsqueeze91_out1[..],
&constant1075_out1[..],
&constant1076_out1[..],
&constant1077_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape47_out1 = linear40_out1.reshape(concat62_out1);
let linear41_out1 = self.linear41.forward(layernormalization22_out1.clone());
let reshape48_out1 = linear41_out1.reshape(concat63_out1);
let linear42_out1 = self.linear42.forward(layernormalization22_out1);
let reshape49_out1 = linear42_out1.reshape(concat64_out1);
let transpose44_out1 = reshape49_out1.permute([0, 2, 1, 3]);
let linear43_out1 = self.linear43.forward(slice1_out1);
let reshape50_out1 = linear43_out1.reshape([1, -1, 8, 128]);
let constant76_out1 = self.constant76.val();
let add37_out1 = reshape47_out1
.clone()
.add((constant76_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose45_out1 = add37_out1.permute([0, 2, 1, 3]);
let constant77_out1 = self.constant77.val();
let add38_out1 = reshape47_out1
.add((constant77_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose46_out1 = add38_out1.permute([0, 2, 1, 3]);
let transpose47_out1 = reshape50_out1.permute([0, 2, 3, 1]);
let matmul56_out1 = transpose46_out1.matmul(transpose47_out1);
let shape37_out1: [i64; 4] = {
let axes = &matmul56_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather31_out1 = shape37_out1[0] as i64;
let gather32_out1 = shape37_out1[1] as i64;
let gather33_out1 = shape37_out1[2] as i64;
let gather34_out1 = shape37_out1[3] as i64;
let pad9_out1 = matmul56_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze92_out1 = [gather31_out1 as i64];
let unsqueeze93_out1 = [gather32_out1 as i64];
let unsqueeze94_out1 = [gather33_out1 as i64];
let constant1092_out1: [i64; 1] = [-1i64];
let concat66_out1: [i64; 4usize] = [
&unsqueeze92_out1[..],
&unsqueeze93_out1[..],
&constant1092_out1[..],
&unsqueeze94_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape53_out1 = pad9_out1.reshape(concat66_out1);
let slice25_out1 = reshape53_out1.slice(s![.., .., 1.., ..]);
let unsqueeze95_out1 = [gather31_out1 as i64];
let unsqueeze96_out1 = [gather32_out1 as i64];
let unsqueeze97_out1 = [gather33_out1 as i64];
let unsqueeze98_out1 = [gather34_out1 as i64];
let concat67_out1: [i64; 4usize] = [
&unsqueeze95_out1[..],
&unsqueeze96_out1[..],
&unsqueeze97_out1[..],
&unsqueeze98_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape54_out1 = slice25_out1.reshape(concat67_out1);
let transpose49_out1 = reshape48_out1.permute([0, 2, 3, 1]);
let matmul57_out1 = transpose45_out1.matmul(transpose49_out1);
let shape41_out1: [i64; 4] = {
let axes = &matmul57_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice26_out1: [i64; 1] = shape41_out1[3..4].try_into().unwrap();
let squeeze5_out1 = slice26_out1[0] as i64;
let unsqueeze99_out1 = [squeeze5_out1 as i64];
let slice27_out1 = reshape54_out1.slice(s![.., .., .., 0..unsqueeze99_out1[0]]);
let add39_out1 = matmul57_out1.add(slice27_out1);
let constant1110_out1 = self.constant1110.val();
let div8_out1 = add39_out1
.div((constant1110_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape42_out1: [i64; 4] = {
let axes = &transpose44_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather35_out1 = shape42_out1[0] as i64;
let constant1112_out1 = -10000f32;
let where14_out1 = div8_out1
.mask_fill(unsqueeze50_out1.clone(), constant1112_out1);
let softmax5_out1 = burn::tensor::activation::softmax(where14_out1, 3);
let constant1113_out1 = 0f32;
let where15_out1 = softmax5_out1.mask_fill(unsqueeze50_out1, constant1113_out1);
let matmul58_out1 = where15_out1.matmul(transpose44_out1);
let transpose50_out1 = matmul58_out1.permute([0, 2, 1, 3]);
let unsqueeze100_out1 = [gather35_out1 as i64];
let constant1116_out1: [i64; 1] = [1024i64];
let constant1115_out1: [i64; 1] = [-1i64];
let concat68_out1: [i64; 3usize] = [
&unsqueeze100_out1[..],
&constant1115_out1[..],
&constant1116_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape55_out1 = transpose50_out1.reshape(concat68_out1);
let linear44_out1 = self.linear44.forward(reshape55_out1);
let add40_out1 = add36_out1.add(linear44_out1);
add40_out1
}
}
#[derive(Module, Debug)]
pub struct Submodule7 {
layernormalization23: LayerNorm,
conv1d13: Conv1d,
conv1d14: Conv1d,
conv1d15: Conv1d,
layernormalization24: LayerNorm,
linear45: Linear,
linear46: Linear,
constant1126: burn::module::Param<Tensor<1>>,
layernormalization25: LayerNorm,
layernormalization26: LayerNorm,
linear47: Linear,
linear48: Linear,
constant1127: burn::module::Param<Tensor<1>>,
layernormalization27: LayerNorm,
linear49: Linear,
linear50: Linear,
linear51: Linear,
linear52: Linear,
constant90: burn::module::Param<Tensor<2>>,
constant91: burn::module::Param<Tensor<2>>,
constant1173: burn::module::Param<Tensor<1>>,
linear53: Linear,
#[module(skip)]
device: Device,
}
impl Submodule7 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization23 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d13 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d14 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d15 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization24 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear45 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear46 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1126: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization25 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization26 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear47 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear48 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1127: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization27 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear49 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear50 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear51 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear52 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant90: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant91: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant1173: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear53 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization23,
conv1d13,
conv1d14,
conv1d15,
layernormalization24,
linear45,
linear46,
constant1126,
layernormalization25,
layernormalization26,
linear47,
linear48,
constant1127,
layernormalization27,
linear49,
linear50,
linear51,
linear52,
constant90,
constant91,
constant1173,
linear53,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add40_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> Tensor<3> {
let layernormalization23_out1 = {
let dtype = add40_out1.clone().dtype();
self.layernormalization23
.forward(add40_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose51_out1 = layernormalization23_out1.permute([0, 2, 1]);
let conv1d13_out1 = self.conv1d13.forward(transpose51_out1);
let split_tensors = conv1d13_out1.split_with_sizes([1024, 1024].into(), 1);
let [split5_out1, split5_out2] = split_tensors.try_into().unwrap();
let sigmoid18_out1 = burn::tensor::activation::sigmoid(split5_out2);
let mul29_out1 = split5_out1.mul(sigmoid18_out1);
let constant1117_out1 = 0f32;
let where16_out1 = mul29_out1.mask_fill(unsqueeze52_out1, constant1117_out1);
let pad10_out1 = where16_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d14_out1 = self.conv1d14.forward(pad10_out1);
let sigmoid19_out1 = burn::tensor::activation::sigmoid(conv1d14_out1.clone());
let mul30_out1 = conv1d14_out1.mul(sigmoid19_out1);
let conv1d15_out1 = self.conv1d15.forward(mul30_out1);
let transpose53_out1 = conv1d15_out1.permute([0, 2, 1]);
let add41_out1 = add40_out1.add(transpose53_out1);
let layernormalization24_out1 = {
let dtype = add41_out1.clone().dtype();
self.layernormalization24
.forward(add41_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear45_out1 = self.linear45.forward(layernormalization24_out1);
let sigmoid20_out1 = burn::tensor::activation::sigmoid(linear45_out1.clone());
let mul31_out1 = linear45_out1.mul(sigmoid20_out1);
let linear46_out1 = self.linear46.forward(mul31_out1);
let constant1126_out1 = self.constant1126.val();
let mul32_out1 = linear46_out1
.mul((constant1126_out1).unsqueeze_dims(&[0isize, 1isize]));
let add42_out1 = add41_out1.add(mul32_out1);
let layernormalization25_out1 = {
let dtype = add42_out1.dtype();
self.layernormalization25
.forward(add42_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization26_out1 = {
let dtype = layernormalization25_out1.clone().dtype();
self.layernormalization26
.forward(
layernormalization25_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear47_out1 = self.linear47.forward(layernormalization26_out1);
let sigmoid21_out1 = burn::tensor::activation::sigmoid(linear47_out1.clone());
let mul33_out1 = linear47_out1.mul(sigmoid21_out1);
let linear48_out1 = self.linear48.forward(mul33_out1);
let constant1127_out1 = self.constant1127.val();
let mul34_out1 = linear48_out1
.mul((constant1127_out1).unsqueeze_dims(&[0isize, 1isize]));
let add43_out1 = layernormalization25_out1.add(mul34_out1);
let layernormalization27_out1 = {
let dtype = add43_out1.clone().dtype();
self.layernormalization27
.forward(add43_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape43_out1: [i64; 3] = {
let axes = &layernormalization27_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather36_out1 = shape43_out1[0] as i64;
let linear49_out1 = self.linear49.forward(layernormalization27_out1.clone());
let unsqueeze101_out1 = [gather36_out1 as i64];
let constant1132_out1: [i64; 1] = [128i64];
let constant1130_out1: [i64; 1] = [-1i64];
let constant1131_out1: [i64; 1] = [8i64];
let concat70_out1: [i64; 4usize] = [
&unsqueeze101_out1[..],
&constant1130_out1[..],
&constant1131_out1[..],
&constant1132_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze102_out1 = [gather36_out1 as i64];
let constant1136_out1: [i64; 1] = [128i64];
let constant1134_out1: [i64; 1] = [-1i64];
let constant1135_out1: [i64; 1] = [8i64];
let concat71_out1: [i64; 4usize] = [
&unsqueeze102_out1[..],
&constant1134_out1[..],
&constant1135_out1[..],
&constant1136_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze103_out1 = [gather36_out1 as i64];
let constant1140_out1: [i64; 1] = [128i64];
let constant1138_out1: [i64; 1] = [-1i64];
let constant1139_out1: [i64; 1] = [8i64];
let concat72_out1: [i64; 4usize] = [
&unsqueeze103_out1[..],
&constant1138_out1[..],
&constant1139_out1[..],
&constant1140_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape58_out1 = linear49_out1.reshape(concat70_out1);
let linear50_out1 = self.linear50.forward(layernormalization27_out1.clone());
let reshape59_out1 = linear50_out1.reshape(concat71_out1);
let linear51_out1 = self.linear51.forward(layernormalization27_out1);
let reshape60_out1 = linear51_out1.reshape(concat72_out1);
let transpose54_out1 = reshape60_out1.permute([0, 2, 1, 3]);
let linear52_out1 = self.linear52.forward(slice1_out1);
let reshape61_out1 = linear52_out1.reshape([1, -1, 8, 128]);
let constant90_out1 = self.constant90.val();
let add44_out1 = reshape58_out1
.clone()
.add((constant90_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose55_out1 = add44_out1.permute([0, 2, 1, 3]);
let constant91_out1 = self.constant91.val();
let add45_out1 = reshape58_out1
.add((constant91_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose56_out1 = add45_out1.permute([0, 2, 1, 3]);
let transpose57_out1 = reshape61_out1.permute([0, 2, 3, 1]);
let matmul68_out1 = transpose56_out1.matmul(transpose57_out1);
let shape44_out1: [i64; 4] = {
let axes = &matmul68_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather37_out1 = shape44_out1[0] as i64;
let gather38_out1 = shape44_out1[1] as i64;
let gather39_out1 = shape44_out1[2] as i64;
let gather40_out1 = shape44_out1[3] as i64;
let pad11_out1 = matmul68_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze104_out1 = [gather37_out1 as i64];
let unsqueeze105_out1 = [gather38_out1 as i64];
let unsqueeze106_out1 = [gather39_out1 as i64];
let constant1155_out1: [i64; 1] = [-1i64];
let concat74_out1: [i64; 4usize] = [
&unsqueeze104_out1[..],
&unsqueeze105_out1[..],
&constant1155_out1[..],
&unsqueeze106_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape64_out1 = pad11_out1.reshape(concat74_out1);
let slice30_out1 = reshape64_out1.slice(s![.., .., 1.., ..]);
let unsqueeze107_out1 = [gather37_out1 as i64];
let unsqueeze108_out1 = [gather38_out1 as i64];
let unsqueeze109_out1 = [gather39_out1 as i64];
let unsqueeze110_out1 = [gather40_out1 as i64];
let concat75_out1: [i64; 4usize] = [
&unsqueeze107_out1[..],
&unsqueeze108_out1[..],
&unsqueeze109_out1[..],
&unsqueeze110_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape65_out1 = slice30_out1.reshape(concat75_out1);
let transpose59_out1 = reshape59_out1.permute([0, 2, 3, 1]);
let matmul69_out1 = transpose55_out1.matmul(transpose59_out1);
let shape48_out1: [i64; 4] = {
let axes = &matmul69_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice31_out1: [i64; 1] = shape48_out1[3..4].try_into().unwrap();
let squeeze6_out1 = slice31_out1[0] as i64;
let unsqueeze111_out1 = [squeeze6_out1 as i64];
let slice32_out1 = reshape65_out1.slice(s![.., .., .., 0..unsqueeze111_out1[0]]);
let add46_out1 = matmul69_out1.add(slice32_out1);
let constant1173_out1 = self.constant1173.val();
let div9_out1 = add46_out1
.div((constant1173_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape49_out1: [i64; 4] = {
let axes = &transpose54_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather41_out1 = shape49_out1[0] as i64;
let constant1175_out1 = -10000f32;
let where17_out1 = div9_out1
.mask_fill(unsqueeze50_out1.clone(), constant1175_out1);
let softmax6_out1 = burn::tensor::activation::softmax(where17_out1, 3);
let constant1176_out1 = 0f32;
let where18_out1 = softmax6_out1.mask_fill(unsqueeze50_out1, constant1176_out1);
let matmul70_out1 = where18_out1.matmul(transpose54_out1);
let transpose60_out1 = matmul70_out1.permute([0, 2, 1, 3]);
let unsqueeze112_out1 = [gather41_out1 as i64];
let constant1179_out1: [i64; 1] = [1024i64];
let constant1178_out1: [i64; 1] = [-1i64];
let concat76_out1: [i64; 3usize] = [
&unsqueeze112_out1[..],
&constant1178_out1[..],
&constant1179_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape66_out1 = transpose60_out1.reshape(concat76_out1);
let linear53_out1 = self.linear53.forward(reshape66_out1);
let add47_out1 = add43_out1.add(linear53_out1);
add47_out1
}
}
#[derive(Module, Debug)]
pub struct Submodule8 {
layernormalization28: LayerNorm,
conv1d16: Conv1d,
conv1d17: Conv1d,
conv1d18: Conv1d,
layernormalization29: LayerNorm,
linear54: Linear,
linear55: Linear,
constant1189: burn::module::Param<Tensor<1>>,
layernormalization30: LayerNorm,
layernormalization31: LayerNorm,
linear56: Linear,
linear57: Linear,
constant1190: burn::module::Param<Tensor<1>>,
layernormalization32: LayerNorm,
linear58: Linear,
linear59: Linear,
linear60: Linear,
linear61: Linear,
constant104: burn::module::Param<Tensor<2>>,
constant105: burn::module::Param<Tensor<2>>,
constant1236: burn::module::Param<Tensor<1>>,
linear62: Linear,
#[module(skip)]
device: Device,
}
impl Submodule8 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization28 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d16 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d17 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d18 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization29 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear54 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear55 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1189: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization30 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization31 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear56 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear57 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1190: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization32 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear58 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear59 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear60 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear61 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant104: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant105: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant1236: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear62 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization28,
conv1d16,
conv1d17,
conv1d18,
layernormalization29,
linear54,
linear55,
constant1189,
layernormalization30,
layernormalization31,
linear56,
linear57,
constant1190,
layernormalization32,
linear58,
linear59,
linear60,
linear61,
constant104,
constant105,
constant1236,
linear62,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add47_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> Tensor<3> {
let layernormalization28_out1 = {
let dtype = add47_out1.clone().dtype();
self.layernormalization28
.forward(add47_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose61_out1 = layernormalization28_out1.permute([0, 2, 1]);
let conv1d16_out1 = self.conv1d16.forward(transpose61_out1);
let split_tensors = conv1d16_out1.split_with_sizes([1024, 1024].into(), 1);
let [split6_out1, split6_out2] = split_tensors.try_into().unwrap();
let sigmoid22_out1 = burn::tensor::activation::sigmoid(split6_out2);
let mul35_out1 = split6_out1.mul(sigmoid22_out1);
let constant1180_out1 = 0f32;
let where19_out1 = mul35_out1.mask_fill(unsqueeze52_out1, constant1180_out1);
let pad12_out1 = where19_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d17_out1 = self.conv1d17.forward(pad12_out1);
let sigmoid23_out1 = burn::tensor::activation::sigmoid(conv1d17_out1.clone());
let mul36_out1 = conv1d17_out1.mul(sigmoid23_out1);
let conv1d18_out1 = self.conv1d18.forward(mul36_out1);
let transpose63_out1 = conv1d18_out1.permute([0, 2, 1]);
let add48_out1 = add47_out1.add(transpose63_out1);
let layernormalization29_out1 = {
let dtype = add48_out1.clone().dtype();
self.layernormalization29
.forward(add48_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear54_out1 = self.linear54.forward(layernormalization29_out1);
let sigmoid24_out1 = burn::tensor::activation::sigmoid(linear54_out1.clone());
let mul37_out1 = linear54_out1.mul(sigmoid24_out1);
let linear55_out1 = self.linear55.forward(mul37_out1);
let constant1189_out1 = self.constant1189.val();
let mul38_out1 = linear55_out1
.mul((constant1189_out1).unsqueeze_dims(&[0isize, 1isize]));
let add49_out1 = add48_out1.add(mul38_out1);
let layernormalization30_out1 = {
let dtype = add49_out1.dtype();
self.layernormalization30
.forward(add49_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization31_out1 = {
let dtype = layernormalization30_out1.clone().dtype();
self.layernormalization31
.forward(
layernormalization30_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear56_out1 = self.linear56.forward(layernormalization31_out1);
let sigmoid25_out1 = burn::tensor::activation::sigmoid(linear56_out1.clone());
let mul39_out1 = linear56_out1.mul(sigmoid25_out1);
let linear57_out1 = self.linear57.forward(mul39_out1);
let constant1190_out1 = self.constant1190.val();
let mul40_out1 = linear57_out1
.mul((constant1190_out1).unsqueeze_dims(&[0isize, 1isize]));
let add50_out1 = layernormalization30_out1.add(mul40_out1);
let layernormalization32_out1 = {
let dtype = add50_out1.clone().dtype();
self.layernormalization32
.forward(add50_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape50_out1: [i64; 3] = {
let axes = &layernormalization32_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather42_out1 = shape50_out1[0] as i64;
let linear58_out1 = self.linear58.forward(layernormalization32_out1.clone());
let unsqueeze113_out1 = [gather42_out1 as i64];
let constant1195_out1: [i64; 1] = [128i64];
let constant1193_out1: [i64; 1] = [-1i64];
let constant1194_out1: [i64; 1] = [8i64];
let concat78_out1: [i64; 4usize] = [
&unsqueeze113_out1[..],
&constant1193_out1[..],
&constant1194_out1[..],
&constant1195_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze114_out1 = [gather42_out1 as i64];
let constant1199_out1: [i64; 1] = [128i64];
let constant1197_out1: [i64; 1] = [-1i64];
let constant1198_out1: [i64; 1] = [8i64];
let concat79_out1: [i64; 4usize] = [
&unsqueeze114_out1[..],
&constant1197_out1[..],
&constant1198_out1[..],
&constant1199_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze115_out1 = [gather42_out1 as i64];
let constant1203_out1: [i64; 1] = [128i64];
let constant1201_out1: [i64; 1] = [-1i64];
let constant1202_out1: [i64; 1] = [8i64];
let concat80_out1: [i64; 4usize] = [
&unsqueeze115_out1[..],
&constant1201_out1[..],
&constant1202_out1[..],
&constant1203_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape69_out1 = linear58_out1.reshape(concat78_out1);
let linear59_out1 = self.linear59.forward(layernormalization32_out1.clone());
let reshape70_out1 = linear59_out1.reshape(concat79_out1);
let linear60_out1 = self.linear60.forward(layernormalization32_out1);
let reshape71_out1 = linear60_out1.reshape(concat80_out1);
let transpose64_out1 = reshape71_out1.permute([0, 2, 1, 3]);
let linear61_out1 = self.linear61.forward(slice1_out1);
let reshape72_out1 = linear61_out1.reshape([1, -1, 8, 128]);
let constant104_out1 = self.constant104.val();
let add51_out1 = reshape69_out1
.clone()
.add((constant104_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose65_out1 = add51_out1.permute([0, 2, 1, 3]);
let constant105_out1 = self.constant105.val();
let add52_out1 = reshape69_out1
.add((constant105_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose66_out1 = add52_out1.permute([0, 2, 1, 3]);
let transpose67_out1 = reshape72_out1.permute([0, 2, 3, 1]);
let matmul80_out1 = transpose66_out1.matmul(transpose67_out1);
let shape51_out1: [i64; 4] = {
let axes = &matmul80_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather43_out1 = shape51_out1[0] as i64;
let gather44_out1 = shape51_out1[1] as i64;
let gather45_out1 = shape51_out1[2] as i64;
let gather46_out1 = shape51_out1[3] as i64;
let pad13_out1 = matmul80_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze116_out1 = [gather43_out1 as i64];
let unsqueeze117_out1 = [gather44_out1 as i64];
let unsqueeze118_out1 = [gather45_out1 as i64];
let constant1218_out1: [i64; 1] = [-1i64];
let concat82_out1: [i64; 4usize] = [
&unsqueeze116_out1[..],
&unsqueeze117_out1[..],
&constant1218_out1[..],
&unsqueeze118_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape75_out1 = pad13_out1.reshape(concat82_out1);
let slice35_out1 = reshape75_out1.slice(s![.., .., 1.., ..]);
let unsqueeze119_out1 = [gather43_out1 as i64];
let unsqueeze120_out1 = [gather44_out1 as i64];
let unsqueeze121_out1 = [gather45_out1 as i64];
let unsqueeze122_out1 = [gather46_out1 as i64];
let concat83_out1: [i64; 4usize] = [
&unsqueeze119_out1[..],
&unsqueeze120_out1[..],
&unsqueeze121_out1[..],
&unsqueeze122_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape76_out1 = slice35_out1.reshape(concat83_out1);
let transpose69_out1 = reshape70_out1.permute([0, 2, 3, 1]);
let matmul81_out1 = transpose65_out1.matmul(transpose69_out1);
let shape55_out1: [i64; 4] = {
let axes = &matmul81_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice36_out1: [i64; 1] = shape55_out1[3..4].try_into().unwrap();
let squeeze7_out1 = slice36_out1[0] as i64;
let unsqueeze123_out1 = [squeeze7_out1 as i64];
let slice37_out1 = reshape76_out1.slice(s![.., .., .., 0..unsqueeze123_out1[0]]);
let add53_out1 = matmul81_out1.add(slice37_out1);
let constant1236_out1 = self.constant1236.val();
let div10_out1 = add53_out1
.div((constant1236_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape56_out1: [i64; 4] = {
let axes = &transpose64_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather47_out1 = shape56_out1[0] as i64;
let constant1238_out1 = -10000f32;
let where20_out1 = div10_out1
.mask_fill(unsqueeze50_out1.clone(), constant1238_out1);
let softmax7_out1 = burn::tensor::activation::softmax(where20_out1, 3);
let constant1239_out1 = 0f32;
let where21_out1 = softmax7_out1.mask_fill(unsqueeze50_out1, constant1239_out1);
let matmul82_out1 = where21_out1.matmul(transpose64_out1);
let transpose70_out1 = matmul82_out1.permute([0, 2, 1, 3]);
let unsqueeze124_out1 = [gather47_out1 as i64];
let constant1242_out1: [i64; 1] = [1024i64];
let constant1241_out1: [i64; 1] = [-1i64];
let concat84_out1: [i64; 3usize] = [
&unsqueeze124_out1[..],
&constant1241_out1[..],
&constant1242_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape77_out1 = transpose70_out1.reshape(concat84_out1);
let linear62_out1 = self.linear62.forward(reshape77_out1);
let add54_out1 = add50_out1.add(linear62_out1);
add54_out1
}
}
#[derive(Module, Debug)]
pub struct Submodule9 {
layernormalization33: LayerNorm,
conv1d19: Conv1d,
conv1d20: Conv1d,
conv1d21: Conv1d,
layernormalization34: LayerNorm,
linear63: Linear,
linear64: Linear,
constant1252: burn::module::Param<Tensor<1>>,
layernormalization35: LayerNorm,
layernormalization36: LayerNorm,
linear65: Linear,
linear66: Linear,
constant1253: burn::module::Param<Tensor<1>>,
layernormalization37: LayerNorm,
linear67: Linear,
linear68: Linear,
linear69: Linear,
linear70: Linear,
constant118: burn::module::Param<Tensor<2>>,
constant119: burn::module::Param<Tensor<2>>,
constant1299: burn::module::Param<Tensor<1>>,
linear71: Linear,
#[module(skip)]
device: Device,
}
impl Submodule9 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization33 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d19 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d20 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d21 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization34 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear63 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear64 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1252: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization35 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization36 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear65 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear66 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1253: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization37 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear67 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear68 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear69 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear70 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant118: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant119: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant1299: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear71 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization33,
conv1d19,
conv1d20,
conv1d21,
layernormalization34,
linear63,
linear64,
constant1252,
layernormalization35,
layernormalization36,
linear65,
linear66,
constant1253,
layernormalization37,
linear67,
linear68,
linear69,
linear70,
constant118,
constant119,
constant1299,
linear71,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add54_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> Tensor<3> {
let layernormalization33_out1 = {
let dtype = add54_out1.clone().dtype();
self.layernormalization33
.forward(add54_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose71_out1 = layernormalization33_out1.permute([0, 2, 1]);
let conv1d19_out1 = self.conv1d19.forward(transpose71_out1);
let split_tensors = conv1d19_out1.split_with_sizes([1024, 1024].into(), 1);
let [split7_out1, split7_out2] = split_tensors.try_into().unwrap();
let sigmoid26_out1 = burn::tensor::activation::sigmoid(split7_out2);
let mul41_out1 = split7_out1.mul(sigmoid26_out1);
let constant1243_out1 = 0f32;
let where22_out1 = mul41_out1.mask_fill(unsqueeze52_out1, constant1243_out1);
let pad14_out1 = where22_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d20_out1 = self.conv1d20.forward(pad14_out1);
let sigmoid27_out1 = burn::tensor::activation::sigmoid(conv1d20_out1.clone());
let mul42_out1 = conv1d20_out1.mul(sigmoid27_out1);
let conv1d21_out1 = self.conv1d21.forward(mul42_out1);
let transpose73_out1 = conv1d21_out1.permute([0, 2, 1]);
let add55_out1 = add54_out1.add(transpose73_out1);
let layernormalization34_out1 = {
let dtype = add55_out1.clone().dtype();
self.layernormalization34
.forward(add55_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear63_out1 = self.linear63.forward(layernormalization34_out1);
let sigmoid28_out1 = burn::tensor::activation::sigmoid(linear63_out1.clone());
let mul43_out1 = linear63_out1.mul(sigmoid28_out1);
let linear64_out1 = self.linear64.forward(mul43_out1);
let constant1252_out1 = self.constant1252.val();
let mul44_out1 = linear64_out1
.mul((constant1252_out1).unsqueeze_dims(&[0isize, 1isize]));
let add56_out1 = add55_out1.add(mul44_out1);
let layernormalization35_out1 = {
let dtype = add56_out1.dtype();
self.layernormalization35
.forward(add56_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization36_out1 = {
let dtype = layernormalization35_out1.clone().dtype();
self.layernormalization36
.forward(
layernormalization35_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear65_out1 = self.linear65.forward(layernormalization36_out1);
let sigmoid29_out1 = burn::tensor::activation::sigmoid(linear65_out1.clone());
let mul45_out1 = linear65_out1.mul(sigmoid29_out1);
let linear66_out1 = self.linear66.forward(mul45_out1);
let constant1253_out1 = self.constant1253.val();
let mul46_out1 = linear66_out1
.mul((constant1253_out1).unsqueeze_dims(&[0isize, 1isize]));
let add57_out1 = layernormalization35_out1.add(mul46_out1);
let layernormalization37_out1 = {
let dtype = add57_out1.clone().dtype();
self.layernormalization37
.forward(add57_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape57_out1: [i64; 3] = {
let axes = &layernormalization37_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather48_out1 = shape57_out1[0] as i64;
let linear67_out1 = self.linear67.forward(layernormalization37_out1.clone());
let unsqueeze125_out1 = [gather48_out1 as i64];
let constant1258_out1: [i64; 1] = [128i64];
let constant1256_out1: [i64; 1] = [-1i64];
let constant1257_out1: [i64; 1] = [8i64];
let concat86_out1: [i64; 4usize] = [
&unsqueeze125_out1[..],
&constant1256_out1[..],
&constant1257_out1[..],
&constant1258_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze126_out1 = [gather48_out1 as i64];
let constant1262_out1: [i64; 1] = [128i64];
let constant1260_out1: [i64; 1] = [-1i64];
let constant1261_out1: [i64; 1] = [8i64];
let concat87_out1: [i64; 4usize] = [
&unsqueeze126_out1[..],
&constant1260_out1[..],
&constant1261_out1[..],
&constant1262_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze127_out1 = [gather48_out1 as i64];
let constant1266_out1: [i64; 1] = [128i64];
let constant1264_out1: [i64; 1] = [-1i64];
let constant1265_out1: [i64; 1] = [8i64];
let concat88_out1: [i64; 4usize] = [
&unsqueeze127_out1[..],
&constant1264_out1[..],
&constant1265_out1[..],
&constant1266_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape80_out1 = linear67_out1.reshape(concat86_out1);
let linear68_out1 = self.linear68.forward(layernormalization37_out1.clone());
let reshape81_out1 = linear68_out1.reshape(concat87_out1);
let linear69_out1 = self.linear69.forward(layernormalization37_out1);
let reshape82_out1 = linear69_out1.reshape(concat88_out1);
let transpose74_out1 = reshape82_out1.permute([0, 2, 1, 3]);
let linear70_out1 = self.linear70.forward(slice1_out1);
let reshape83_out1 = linear70_out1.reshape([1, -1, 8, 128]);
let constant118_out1 = self.constant118.val();
let add58_out1 = reshape80_out1
.clone()
.add((constant118_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose75_out1 = add58_out1.permute([0, 2, 1, 3]);
let constant119_out1 = self.constant119.val();
let add59_out1 = reshape80_out1
.add((constant119_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose76_out1 = add59_out1.permute([0, 2, 1, 3]);
let transpose77_out1 = reshape83_out1.permute([0, 2, 3, 1]);
let matmul92_out1 = transpose76_out1.matmul(transpose77_out1);
let shape58_out1: [i64; 4] = {
let axes = &matmul92_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather49_out1 = shape58_out1[0] as i64;
let gather50_out1 = shape58_out1[1] as i64;
let gather51_out1 = shape58_out1[2] as i64;
let gather52_out1 = shape58_out1[3] as i64;
let pad15_out1 = matmul92_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze128_out1 = [gather49_out1 as i64];
let unsqueeze129_out1 = [gather50_out1 as i64];
let unsqueeze130_out1 = [gather51_out1 as i64];
let constant1281_out1: [i64; 1] = [-1i64];
let concat90_out1: [i64; 4usize] = [
&unsqueeze128_out1[..],
&unsqueeze129_out1[..],
&constant1281_out1[..],
&unsqueeze130_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape86_out1 = pad15_out1.reshape(concat90_out1);
let slice40_out1 = reshape86_out1.slice(s![.., .., 1.., ..]);
let unsqueeze131_out1 = [gather49_out1 as i64];
let unsqueeze132_out1 = [gather50_out1 as i64];
let unsqueeze133_out1 = [gather51_out1 as i64];
let unsqueeze134_out1 = [gather52_out1 as i64];
let concat91_out1: [i64; 4usize] = [
&unsqueeze131_out1[..],
&unsqueeze132_out1[..],
&unsqueeze133_out1[..],
&unsqueeze134_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape87_out1 = slice40_out1.reshape(concat91_out1);
let transpose79_out1 = reshape81_out1.permute([0, 2, 3, 1]);
let matmul93_out1 = transpose75_out1.matmul(transpose79_out1);
let shape62_out1: [i64; 4] = {
let axes = &matmul93_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice41_out1: [i64; 1] = shape62_out1[3..4].try_into().unwrap();
let squeeze8_out1 = slice41_out1[0] as i64;
let unsqueeze135_out1 = [squeeze8_out1 as i64];
let slice42_out1 = reshape87_out1.slice(s![.., .., .., 0..unsqueeze135_out1[0]]);
let add60_out1 = matmul93_out1.add(slice42_out1);
let constant1299_out1 = self.constant1299.val();
let div11_out1 = add60_out1
.div((constant1299_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape63_out1: [i64; 4] = {
let axes = &transpose74_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather53_out1 = shape63_out1[0] as i64;
let constant1301_out1 = -10000f32;
let where23_out1 = div11_out1
.mask_fill(unsqueeze50_out1.clone(), constant1301_out1);
let softmax8_out1 = burn::tensor::activation::softmax(where23_out1, 3);
let constant1302_out1 = 0f32;
let where24_out1 = softmax8_out1.mask_fill(unsqueeze50_out1, constant1302_out1);
let matmul94_out1 = where24_out1.matmul(transpose74_out1);
let transpose80_out1 = matmul94_out1.permute([0, 2, 1, 3]);
let unsqueeze136_out1 = [gather53_out1 as i64];
let constant1305_out1: [i64; 1] = [1024i64];
let constant1304_out1: [i64; 1] = [-1i64];
let concat92_out1: [i64; 3usize] = [
&unsqueeze136_out1[..],
&constant1304_out1[..],
&constant1305_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape88_out1 = transpose80_out1.reshape(concat92_out1);
let linear71_out1 = self.linear71.forward(reshape88_out1);
let add61_out1 = add57_out1.add(linear71_out1);
add61_out1
}
}
#[derive(Module, Debug)]
pub struct Submodule10 {
layernormalization38: LayerNorm,
conv1d22: Conv1d,
conv1d23: Conv1d,
conv1d24: Conv1d,
layernormalization39: LayerNorm,
linear72: Linear,
linear73: Linear,
constant1315: burn::module::Param<Tensor<1>>,
layernormalization40: LayerNorm,
layernormalization41: LayerNorm,
linear74: Linear,
linear75: Linear,
constant1316: burn::module::Param<Tensor<1>>,
layernormalization42: LayerNorm,
linear76: Linear,
linear77: Linear,
linear78: Linear,
linear79: Linear,
constant132: burn::module::Param<Tensor<2>>,
constant133: burn::module::Param<Tensor<2>>,
constant1362: burn::module::Param<Tensor<1>>,
linear80: Linear,
#[module(skip)]
device: Device,
}
impl Submodule10 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization38 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d22 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d23 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d24 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization39 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear72 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear73 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1315: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization40 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization41 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear74 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear75 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1316: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization42 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear76 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear77 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear78 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear79 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant132: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant133: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant1362: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear80 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization38,
conv1d22,
conv1d23,
conv1d24,
layernormalization39,
linear72,
linear73,
constant1315,
layernormalization40,
layernormalization41,
linear74,
linear75,
constant1316,
layernormalization42,
linear76,
linear77,
linear78,
linear79,
constant132,
constant133,
constant1362,
linear80,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add61_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> Tensor<3> {
let layernormalization38_out1 = {
let dtype = add61_out1.clone().dtype();
self.layernormalization38
.forward(add61_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose81_out1 = layernormalization38_out1.permute([0, 2, 1]);
let conv1d22_out1 = self.conv1d22.forward(transpose81_out1);
let split_tensors = conv1d22_out1.split_with_sizes([1024, 1024].into(), 1);
let [split8_out1, split8_out2] = split_tensors.try_into().unwrap();
let sigmoid30_out1 = burn::tensor::activation::sigmoid(split8_out2);
let mul47_out1 = split8_out1.mul(sigmoid30_out1);
let constant1306_out1 = 0f32;
let where25_out1 = mul47_out1.mask_fill(unsqueeze52_out1, constant1306_out1);
let pad16_out1 = where25_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d23_out1 = self.conv1d23.forward(pad16_out1);
let sigmoid31_out1 = burn::tensor::activation::sigmoid(conv1d23_out1.clone());
let mul48_out1 = conv1d23_out1.mul(sigmoid31_out1);
let conv1d24_out1 = self.conv1d24.forward(mul48_out1);
let transpose83_out1 = conv1d24_out1.permute([0, 2, 1]);
let add62_out1 = add61_out1.add(transpose83_out1);
let layernormalization39_out1 = {
let dtype = add62_out1.clone().dtype();
self.layernormalization39
.forward(add62_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear72_out1 = self.linear72.forward(layernormalization39_out1);
let sigmoid32_out1 = burn::tensor::activation::sigmoid(linear72_out1.clone());
let mul49_out1 = linear72_out1.mul(sigmoid32_out1);
let linear73_out1 = self.linear73.forward(mul49_out1);
let constant1315_out1 = self.constant1315.val();
let mul50_out1 = linear73_out1
.mul((constant1315_out1).unsqueeze_dims(&[0isize, 1isize]));
let add63_out1 = add62_out1.add(mul50_out1);
let layernormalization40_out1 = {
let dtype = add63_out1.dtype();
self.layernormalization40
.forward(add63_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization41_out1 = {
let dtype = layernormalization40_out1.clone().dtype();
self.layernormalization41
.forward(
layernormalization40_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear74_out1 = self.linear74.forward(layernormalization41_out1);
let sigmoid33_out1 = burn::tensor::activation::sigmoid(linear74_out1.clone());
let mul51_out1 = linear74_out1.mul(sigmoid33_out1);
let linear75_out1 = self.linear75.forward(mul51_out1);
let constant1316_out1 = self.constant1316.val();
let mul52_out1 = linear75_out1
.mul((constant1316_out1).unsqueeze_dims(&[0isize, 1isize]));
let add64_out1 = layernormalization40_out1.add(mul52_out1);
let layernormalization42_out1 = {
let dtype = add64_out1.clone().dtype();
self.layernormalization42
.forward(add64_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape64_out1: [i64; 3] = {
let axes = &layernormalization42_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather54_out1 = shape64_out1[0] as i64;
let linear76_out1 = self.linear76.forward(layernormalization42_out1.clone());
let unsqueeze137_out1 = [gather54_out1 as i64];
let constant1321_out1: [i64; 1] = [128i64];
let constant1319_out1: [i64; 1] = [-1i64];
let constant1320_out1: [i64; 1] = [8i64];
let concat94_out1: [i64; 4usize] = [
&unsqueeze137_out1[..],
&constant1319_out1[..],
&constant1320_out1[..],
&constant1321_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze138_out1 = [gather54_out1 as i64];
let constant1325_out1: [i64; 1] = [128i64];
let constant1323_out1: [i64; 1] = [-1i64];
let constant1324_out1: [i64; 1] = [8i64];
let concat95_out1: [i64; 4usize] = [
&unsqueeze138_out1[..],
&constant1323_out1[..],
&constant1324_out1[..],
&constant1325_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze139_out1 = [gather54_out1 as i64];
let constant1329_out1: [i64; 1] = [128i64];
let constant1327_out1: [i64; 1] = [-1i64];
let constant1328_out1: [i64; 1] = [8i64];
let concat96_out1: [i64; 4usize] = [
&unsqueeze139_out1[..],
&constant1327_out1[..],
&constant1328_out1[..],
&constant1329_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape91_out1 = linear76_out1.reshape(concat94_out1);
let linear77_out1 = self.linear77.forward(layernormalization42_out1.clone());
let reshape92_out1 = linear77_out1.reshape(concat95_out1);
let linear78_out1 = self.linear78.forward(layernormalization42_out1);
let reshape93_out1 = linear78_out1.reshape(concat96_out1);
let transpose84_out1 = reshape93_out1.permute([0, 2, 1, 3]);
let linear79_out1 = self.linear79.forward(slice1_out1);
let reshape94_out1 = linear79_out1.reshape([1, -1, 8, 128]);
let constant132_out1 = self.constant132.val();
let add65_out1 = reshape91_out1
.clone()
.add((constant132_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose85_out1 = add65_out1.permute([0, 2, 1, 3]);
let constant133_out1 = self.constant133.val();
let add66_out1 = reshape91_out1
.add((constant133_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose86_out1 = add66_out1.permute([0, 2, 1, 3]);
let transpose87_out1 = reshape94_out1.permute([0, 2, 3, 1]);
let matmul104_out1 = transpose86_out1.matmul(transpose87_out1);
let shape65_out1: [i64; 4] = {
let axes = &matmul104_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather55_out1 = shape65_out1[0] as i64;
let gather56_out1 = shape65_out1[1] as i64;
let gather57_out1 = shape65_out1[2] as i64;
let gather58_out1 = shape65_out1[3] as i64;
let pad17_out1 = matmul104_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze140_out1 = [gather55_out1 as i64];
let unsqueeze141_out1 = [gather56_out1 as i64];
let unsqueeze142_out1 = [gather57_out1 as i64];
let constant1344_out1: [i64; 1] = [-1i64];
let concat98_out1: [i64; 4usize] = [
&unsqueeze140_out1[..],
&unsqueeze141_out1[..],
&constant1344_out1[..],
&unsqueeze142_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape97_out1 = pad17_out1.reshape(concat98_out1);
let slice45_out1 = reshape97_out1.slice(s![.., .., 1.., ..]);
let unsqueeze143_out1 = [gather55_out1 as i64];
let unsqueeze144_out1 = [gather56_out1 as i64];
let unsqueeze145_out1 = [gather57_out1 as i64];
let unsqueeze146_out1 = [gather58_out1 as i64];
let concat99_out1: [i64; 4usize] = [
&unsqueeze143_out1[..],
&unsqueeze144_out1[..],
&unsqueeze145_out1[..],
&unsqueeze146_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape98_out1 = slice45_out1.reshape(concat99_out1);
let transpose89_out1 = reshape92_out1.permute([0, 2, 3, 1]);
let matmul105_out1 = transpose85_out1.matmul(transpose89_out1);
let shape69_out1: [i64; 4] = {
let axes = &matmul105_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice46_out1: [i64; 1] = shape69_out1[3..4].try_into().unwrap();
let squeeze9_out1 = slice46_out1[0] as i64;
let unsqueeze147_out1 = [squeeze9_out1 as i64];
let slice47_out1 = reshape98_out1.slice(s![.., .., .., 0..unsqueeze147_out1[0]]);
let add67_out1 = matmul105_out1.add(slice47_out1);
let constant1362_out1 = self.constant1362.val();
let div12_out1 = add67_out1
.div((constant1362_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape70_out1: [i64; 4] = {
let axes = &transpose84_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather59_out1 = shape70_out1[0] as i64;
let constant1364_out1 = -10000f32;
let where26_out1 = div12_out1
.mask_fill(unsqueeze50_out1.clone(), constant1364_out1);
let softmax9_out1 = burn::tensor::activation::softmax(where26_out1, 3);
let constant1365_out1 = 0f32;
let where27_out1 = softmax9_out1.mask_fill(unsqueeze50_out1, constant1365_out1);
let matmul106_out1 = where27_out1.matmul(transpose84_out1);
let transpose90_out1 = matmul106_out1.permute([0, 2, 1, 3]);
let unsqueeze148_out1 = [gather59_out1 as i64];
let constant1368_out1: [i64; 1] = [1024i64];
let constant1367_out1: [i64; 1] = [-1i64];
let concat100_out1: [i64; 3usize] = [
&unsqueeze148_out1[..],
&constant1367_out1[..],
&constant1368_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape99_out1 = transpose90_out1.reshape(concat100_out1);
let linear80_out1 = self.linear80.forward(reshape99_out1);
let add68_out1 = add64_out1.add(linear80_out1);
add68_out1
}
}
#[derive(Module, Debug)]
pub struct Submodule11 {
layernormalization43: LayerNorm,
conv1d25: Conv1d,
conv1d26: Conv1d,
conv1d27: Conv1d,
layernormalization44: LayerNorm,
linear81: Linear,
linear82: Linear,
constant1378: burn::module::Param<Tensor<1>>,
layernormalization45: LayerNorm,
layernormalization46: LayerNorm,
linear83: Linear,
linear84: Linear,
constant1379: burn::module::Param<Tensor<1>>,
layernormalization47: LayerNorm,
linear85: Linear,
linear86: Linear,
linear87: Linear,
linear88: Linear,
constant146: burn::module::Param<Tensor<2>>,
constant147: burn::module::Param<Tensor<2>>,
constant1425: burn::module::Param<Tensor<1>>,
linear89: Linear,
#[module(skip)]
device: Device,
}
impl Submodule11 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization43 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d25 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d26 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d27 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization44 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear81 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear82 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1378: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization45 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization46 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear83 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear84 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1379: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization47 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear85 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear86 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear87 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear88 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant146: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant147: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant1425: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear89 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization43,
conv1d25,
conv1d26,
conv1d27,
layernormalization44,
linear81,
linear82,
constant1378,
layernormalization45,
layernormalization46,
linear83,
linear84,
constant1379,
layernormalization47,
linear85,
linear86,
linear87,
linear88,
constant146,
constant147,
constant1425,
linear89,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add68_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> Tensor<3> {
let layernormalization43_out1 = {
let dtype = add68_out1.clone().dtype();
self.layernormalization43
.forward(add68_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose91_out1 = layernormalization43_out1.permute([0, 2, 1]);
let conv1d25_out1 = self.conv1d25.forward(transpose91_out1);
let split_tensors = conv1d25_out1.split_with_sizes([1024, 1024].into(), 1);
let [split9_out1, split9_out2] = split_tensors.try_into().unwrap();
let sigmoid34_out1 = burn::tensor::activation::sigmoid(split9_out2);
let mul53_out1 = split9_out1.mul(sigmoid34_out1);
let constant1369_out1 = 0f32;
let where28_out1 = mul53_out1.mask_fill(unsqueeze52_out1, constant1369_out1);
let pad18_out1 = where28_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d26_out1 = self.conv1d26.forward(pad18_out1);
let sigmoid35_out1 = burn::tensor::activation::sigmoid(conv1d26_out1.clone());
let mul54_out1 = conv1d26_out1.mul(sigmoid35_out1);
let conv1d27_out1 = self.conv1d27.forward(mul54_out1);
let transpose93_out1 = conv1d27_out1.permute([0, 2, 1]);
let add69_out1 = add68_out1.add(transpose93_out1);
let layernormalization44_out1 = {
let dtype = add69_out1.clone().dtype();
self.layernormalization44
.forward(add69_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear81_out1 = self.linear81.forward(layernormalization44_out1);
let sigmoid36_out1 = burn::tensor::activation::sigmoid(linear81_out1.clone());
let mul55_out1 = linear81_out1.mul(sigmoid36_out1);
let linear82_out1 = self.linear82.forward(mul55_out1);
let constant1378_out1 = self.constant1378.val();
let mul56_out1 = linear82_out1
.mul((constant1378_out1).unsqueeze_dims(&[0isize, 1isize]));
let add70_out1 = add69_out1.add(mul56_out1);
let layernormalization45_out1 = {
let dtype = add70_out1.dtype();
self.layernormalization45
.forward(add70_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization46_out1 = {
let dtype = layernormalization45_out1.clone().dtype();
self.layernormalization46
.forward(
layernormalization45_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear83_out1 = self.linear83.forward(layernormalization46_out1);
let sigmoid37_out1 = burn::tensor::activation::sigmoid(linear83_out1.clone());
let mul57_out1 = linear83_out1.mul(sigmoid37_out1);
let linear84_out1 = self.linear84.forward(mul57_out1);
let constant1379_out1 = self.constant1379.val();
let mul58_out1 = linear84_out1
.mul((constant1379_out1).unsqueeze_dims(&[0isize, 1isize]));
let add71_out1 = layernormalization45_out1.add(mul58_out1);
let layernormalization47_out1 = {
let dtype = add71_out1.clone().dtype();
self.layernormalization47
.forward(add71_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape71_out1: [i64; 3] = {
let axes = &layernormalization47_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather60_out1 = shape71_out1[0] as i64;
let linear85_out1 = self.linear85.forward(layernormalization47_out1.clone());
let unsqueeze149_out1 = [gather60_out1 as i64];
let constant1384_out1: [i64; 1] = [128i64];
let constant1382_out1: [i64; 1] = [-1i64];
let constant1383_out1: [i64; 1] = [8i64];
let concat102_out1: [i64; 4usize] = [
&unsqueeze149_out1[..],
&constant1382_out1[..],
&constant1383_out1[..],
&constant1384_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze150_out1 = [gather60_out1 as i64];
let constant1388_out1: [i64; 1] = [128i64];
let constant1386_out1: [i64; 1] = [-1i64];
let constant1387_out1: [i64; 1] = [8i64];
let concat103_out1: [i64; 4usize] = [
&unsqueeze150_out1[..],
&constant1386_out1[..],
&constant1387_out1[..],
&constant1388_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze151_out1 = [gather60_out1 as i64];
let constant1392_out1: [i64; 1] = [128i64];
let constant1390_out1: [i64; 1] = [-1i64];
let constant1391_out1: [i64; 1] = [8i64];
let concat104_out1: [i64; 4usize] = [
&unsqueeze151_out1[..],
&constant1390_out1[..],
&constant1391_out1[..],
&constant1392_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape102_out1 = linear85_out1.reshape(concat102_out1);
let linear86_out1 = self.linear86.forward(layernormalization47_out1.clone());
let reshape103_out1 = linear86_out1.reshape(concat103_out1);
let linear87_out1 = self.linear87.forward(layernormalization47_out1);
let reshape104_out1 = linear87_out1.reshape(concat104_out1);
let transpose94_out1 = reshape104_out1.permute([0, 2, 1, 3]);
let linear88_out1 = self.linear88.forward(slice1_out1);
let reshape105_out1 = linear88_out1.reshape([1, -1, 8, 128]);
let constant146_out1 = self.constant146.val();
let add72_out1 = reshape102_out1
.clone()
.add((constant146_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose95_out1 = add72_out1.permute([0, 2, 1, 3]);
let constant147_out1 = self.constant147.val();
let add73_out1 = reshape102_out1
.add((constant147_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose96_out1 = add73_out1.permute([0, 2, 1, 3]);
let transpose97_out1 = reshape105_out1.permute([0, 2, 3, 1]);
let matmul116_out1 = transpose96_out1.matmul(transpose97_out1);
let shape72_out1: [i64; 4] = {
let axes = &matmul116_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather61_out1 = shape72_out1[0] as i64;
let gather62_out1 = shape72_out1[1] as i64;
let gather63_out1 = shape72_out1[2] as i64;
let gather64_out1 = shape72_out1[3] as i64;
let pad19_out1 = matmul116_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze152_out1 = [gather61_out1 as i64];
let unsqueeze153_out1 = [gather62_out1 as i64];
let unsqueeze154_out1 = [gather63_out1 as i64];
let constant1407_out1: [i64; 1] = [-1i64];
let concat106_out1: [i64; 4usize] = [
&unsqueeze152_out1[..],
&unsqueeze153_out1[..],
&constant1407_out1[..],
&unsqueeze154_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape108_out1 = pad19_out1.reshape(concat106_out1);
let slice50_out1 = reshape108_out1.slice(s![.., .., 1.., ..]);
let unsqueeze155_out1 = [gather61_out1 as i64];
let unsqueeze156_out1 = [gather62_out1 as i64];
let unsqueeze157_out1 = [gather63_out1 as i64];
let unsqueeze158_out1 = [gather64_out1 as i64];
let concat107_out1: [i64; 4usize] = [
&unsqueeze155_out1[..],
&unsqueeze156_out1[..],
&unsqueeze157_out1[..],
&unsqueeze158_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape109_out1 = slice50_out1.reshape(concat107_out1);
let transpose99_out1 = reshape103_out1.permute([0, 2, 3, 1]);
let matmul117_out1 = transpose95_out1.matmul(transpose99_out1);
let shape76_out1: [i64; 4] = {
let axes = &matmul117_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice51_out1: [i64; 1] = shape76_out1[3..4].try_into().unwrap();
let squeeze10_out1 = slice51_out1[0] as i64;
let unsqueeze159_out1 = [squeeze10_out1 as i64];
let slice52_out1 = reshape109_out1
.slice(s![.., .., .., 0..unsqueeze159_out1[0]]);
let add74_out1 = matmul117_out1.add(slice52_out1);
let constant1425_out1 = self.constant1425.val();
let div13_out1 = add74_out1
.div((constant1425_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape77_out1: [i64; 4] = {
let axes = &transpose94_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather65_out1 = shape77_out1[0] as i64;
let constant1427_out1 = -10000f32;
let where29_out1 = div13_out1
.mask_fill(unsqueeze50_out1.clone(), constant1427_out1);
let softmax10_out1 = burn::tensor::activation::softmax(where29_out1, 3);
let constant1428_out1 = 0f32;
let where30_out1 = softmax10_out1.mask_fill(unsqueeze50_out1, constant1428_out1);
let matmul118_out1 = where30_out1.matmul(transpose94_out1);
let transpose100_out1 = matmul118_out1.permute([0, 2, 1, 3]);
let unsqueeze160_out1 = [gather65_out1 as i64];
let constant1431_out1: [i64; 1] = [1024i64];
let constant1430_out1: [i64; 1] = [-1i64];
let concat108_out1: [i64; 3usize] = [
&unsqueeze160_out1[..],
&constant1430_out1[..],
&constant1431_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape110_out1 = transpose100_out1.reshape(concat108_out1);
let linear89_out1 = self.linear89.forward(reshape110_out1);
let add75_out1 = add71_out1.add(linear89_out1);
add75_out1
}
}
#[derive(Module, Debug)]
pub struct Submodule12 {
layernormalization48: LayerNorm,
conv1d28: Conv1d,
conv1d29: Conv1d,
conv1d30: Conv1d,
layernormalization49: LayerNorm,
linear90: Linear,
linear91: Linear,
constant1441: burn::module::Param<Tensor<1>>,
layernormalization50: LayerNorm,
layernormalization51: LayerNorm,
linear92: Linear,
linear93: Linear,
constant1442: burn::module::Param<Tensor<1>>,
layernormalization52: LayerNorm,
linear94: Linear,
linear95: Linear,
linear96: Linear,
linear97: Linear,
constant160: burn::module::Param<Tensor<2>>,
constant161: burn::module::Param<Tensor<2>>,
constant1488: burn::module::Param<Tensor<1>>,
linear98: Linear,
#[module(skip)]
device: Device,
}
impl Submodule12 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization48 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d28 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d29 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d30 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization49 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear90 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear91 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1441: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization50 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization51 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear92 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear93 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1442: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization52 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear94 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear95 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear96 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear97 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant160: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant161: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant1488: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear98 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization48,
conv1d28,
conv1d29,
conv1d30,
layernormalization49,
linear90,
linear91,
constant1441,
layernormalization50,
layernormalization51,
linear92,
linear93,
constant1442,
layernormalization52,
linear94,
linear95,
linear96,
linear97,
constant160,
constant161,
constant1488,
linear98,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add75_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> Tensor<3> {
let layernormalization48_out1 = {
let dtype = add75_out1.clone().dtype();
self.layernormalization48
.forward(add75_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose101_out1 = layernormalization48_out1.permute([0, 2, 1]);
let conv1d28_out1 = self.conv1d28.forward(transpose101_out1);
let split_tensors = conv1d28_out1.split_with_sizes([1024, 1024].into(), 1);
let [split10_out1, split10_out2] = split_tensors.try_into().unwrap();
let sigmoid38_out1 = burn::tensor::activation::sigmoid(split10_out2);
let mul59_out1 = split10_out1.mul(sigmoid38_out1);
let constant1432_out1 = 0f32;
let where31_out1 = mul59_out1.mask_fill(unsqueeze52_out1, constant1432_out1);
let pad20_out1 = where31_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d29_out1 = self.conv1d29.forward(pad20_out1);
let sigmoid39_out1 = burn::tensor::activation::sigmoid(conv1d29_out1.clone());
let mul60_out1 = conv1d29_out1.mul(sigmoid39_out1);
let conv1d30_out1 = self.conv1d30.forward(mul60_out1);
let transpose103_out1 = conv1d30_out1.permute([0, 2, 1]);
let add76_out1 = add75_out1.add(transpose103_out1);
let layernormalization49_out1 = {
let dtype = add76_out1.clone().dtype();
self.layernormalization49
.forward(add76_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear90_out1 = self.linear90.forward(layernormalization49_out1);
let sigmoid40_out1 = burn::tensor::activation::sigmoid(linear90_out1.clone());
let mul61_out1 = linear90_out1.mul(sigmoid40_out1);
let linear91_out1 = self.linear91.forward(mul61_out1);
let constant1441_out1 = self.constant1441.val();
let mul62_out1 = linear91_out1
.mul((constant1441_out1).unsqueeze_dims(&[0isize, 1isize]));
let add77_out1 = add76_out1.add(mul62_out1);
let layernormalization50_out1 = {
let dtype = add77_out1.dtype();
self.layernormalization50
.forward(add77_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization51_out1 = {
let dtype = layernormalization50_out1.clone().dtype();
self.layernormalization51
.forward(
layernormalization50_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear92_out1 = self.linear92.forward(layernormalization51_out1);
let sigmoid41_out1 = burn::tensor::activation::sigmoid(linear92_out1.clone());
let mul63_out1 = linear92_out1.mul(sigmoid41_out1);
let linear93_out1 = self.linear93.forward(mul63_out1);
let constant1442_out1 = self.constant1442.val();
let mul64_out1 = linear93_out1
.mul((constant1442_out1).unsqueeze_dims(&[0isize, 1isize]));
let add78_out1 = layernormalization50_out1.add(mul64_out1);
let layernormalization52_out1 = {
let dtype = add78_out1.clone().dtype();
self.layernormalization52
.forward(add78_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape78_out1: [i64; 3] = {
let axes = &layernormalization52_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather66_out1 = shape78_out1[0] as i64;
let linear94_out1 = self.linear94.forward(layernormalization52_out1.clone());
let unsqueeze161_out1 = [gather66_out1 as i64];
let constant1447_out1: [i64; 1] = [128i64];
let constant1445_out1: [i64; 1] = [-1i64];
let constant1446_out1: [i64; 1] = [8i64];
let concat110_out1: [i64; 4usize] = [
&unsqueeze161_out1[..],
&constant1445_out1[..],
&constant1446_out1[..],
&constant1447_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze162_out1 = [gather66_out1 as i64];
let constant1451_out1: [i64; 1] = [128i64];
let constant1449_out1: [i64; 1] = [-1i64];
let constant1450_out1: [i64; 1] = [8i64];
let concat111_out1: [i64; 4usize] = [
&unsqueeze162_out1[..],
&constant1449_out1[..],
&constant1450_out1[..],
&constant1451_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze163_out1 = [gather66_out1 as i64];
let constant1455_out1: [i64; 1] = [128i64];
let constant1453_out1: [i64; 1] = [-1i64];
let constant1454_out1: [i64; 1] = [8i64];
let concat112_out1: [i64; 4usize] = [
&unsqueeze163_out1[..],
&constant1453_out1[..],
&constant1454_out1[..],
&constant1455_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape113_out1 = linear94_out1.reshape(concat110_out1);
let linear95_out1 = self.linear95.forward(layernormalization52_out1.clone());
let reshape114_out1 = linear95_out1.reshape(concat111_out1);
let linear96_out1 = self.linear96.forward(layernormalization52_out1);
let reshape115_out1 = linear96_out1.reshape(concat112_out1);
let transpose104_out1 = reshape115_out1.permute([0, 2, 1, 3]);
let linear97_out1 = self.linear97.forward(slice1_out1);
let reshape116_out1 = linear97_out1.reshape([1, -1, 8, 128]);
let constant160_out1 = self.constant160.val();
let add79_out1 = reshape113_out1
.clone()
.add((constant160_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose105_out1 = add79_out1.permute([0, 2, 1, 3]);
let constant161_out1 = self.constant161.val();
let add80_out1 = reshape113_out1
.add((constant161_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose106_out1 = add80_out1.permute([0, 2, 1, 3]);
let transpose107_out1 = reshape116_out1.permute([0, 2, 3, 1]);
let matmul128_out1 = transpose106_out1.matmul(transpose107_out1);
let shape79_out1: [i64; 4] = {
let axes = &matmul128_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather67_out1 = shape79_out1[0] as i64;
let gather68_out1 = shape79_out1[1] as i64;
let gather69_out1 = shape79_out1[2] as i64;
let gather70_out1 = shape79_out1[3] as i64;
let pad21_out1 = matmul128_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze164_out1 = [gather67_out1 as i64];
let unsqueeze165_out1 = [gather68_out1 as i64];
let unsqueeze166_out1 = [gather69_out1 as i64];
let constant1470_out1: [i64; 1] = [-1i64];
let concat114_out1: [i64; 4usize] = [
&unsqueeze164_out1[..],
&unsqueeze165_out1[..],
&constant1470_out1[..],
&unsqueeze166_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape119_out1 = pad21_out1.reshape(concat114_out1);
let slice55_out1 = reshape119_out1.slice(s![.., .., 1.., ..]);
let unsqueeze167_out1 = [gather67_out1 as i64];
let unsqueeze168_out1 = [gather68_out1 as i64];
let unsqueeze169_out1 = [gather69_out1 as i64];
let unsqueeze170_out1 = [gather70_out1 as i64];
let concat115_out1: [i64; 4usize] = [
&unsqueeze167_out1[..],
&unsqueeze168_out1[..],
&unsqueeze169_out1[..],
&unsqueeze170_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape120_out1 = slice55_out1.reshape(concat115_out1);
let transpose109_out1 = reshape114_out1.permute([0, 2, 3, 1]);
let matmul129_out1 = transpose105_out1.matmul(transpose109_out1);
let shape83_out1: [i64; 4] = {
let axes = &matmul129_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice56_out1: [i64; 1] = shape83_out1[3..4].try_into().unwrap();
let squeeze11_out1 = slice56_out1[0] as i64;
let unsqueeze171_out1 = [squeeze11_out1 as i64];
let slice57_out1 = reshape120_out1
.slice(s![.., .., .., 0..unsqueeze171_out1[0]]);
let add81_out1 = matmul129_out1.add(slice57_out1);
let constant1488_out1 = self.constant1488.val();
let div14_out1 = add81_out1
.div((constant1488_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape84_out1: [i64; 4] = {
let axes = &transpose104_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather71_out1 = shape84_out1[0] as i64;
let constant1490_out1 = -10000f32;
let where32_out1 = div14_out1
.mask_fill(unsqueeze50_out1.clone(), constant1490_out1);
let softmax11_out1 = burn::tensor::activation::softmax(where32_out1, 3);
let constant1491_out1 = 0f32;
let where33_out1 = softmax11_out1.mask_fill(unsqueeze50_out1, constant1491_out1);
let matmul130_out1 = where33_out1.matmul(transpose104_out1);
let transpose110_out1 = matmul130_out1.permute([0, 2, 1, 3]);
let unsqueeze172_out1 = [gather71_out1 as i64];
let constant1494_out1: [i64; 1] = [1024i64];
let constant1493_out1: [i64; 1] = [-1i64];
let concat116_out1: [i64; 3usize] = [
&unsqueeze172_out1[..],
&constant1493_out1[..],
&constant1494_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape121_out1 = transpose110_out1.reshape(concat116_out1);
let linear98_out1 = self.linear98.forward(reshape121_out1);
let add82_out1 = add78_out1.add(linear98_out1);
add82_out1
}
}
#[derive(Module, Debug)]
pub struct Submodule13 {
layernormalization53: LayerNorm,
conv1d31: Conv1d,
conv1d32: Conv1d,
conv1d33: Conv1d,
layernormalization54: LayerNorm,
linear99: Linear,
linear100: Linear,
constant1504: burn::module::Param<Tensor<1>>,
layernormalization55: LayerNorm,
layernormalization56: LayerNorm,
linear101: Linear,
linear102: Linear,
constant1505: burn::module::Param<Tensor<1>>,
layernormalization57: LayerNorm,
linear103: Linear,
linear104: Linear,
linear105: Linear,
linear106: Linear,
constant174: burn::module::Param<Tensor<2>>,
constant175: burn::module::Param<Tensor<2>>,
constant1551: burn::module::Param<Tensor<1>>,
linear107: Linear,
#[module(skip)]
device: Device,
}
impl Submodule13 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization53 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d31 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d32 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d33 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization54 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear99 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear100 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1504: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization55 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization56 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear101 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear102 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1505: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization57 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear103 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear104 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear105 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear106 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant174: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant175: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant1551: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear107 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization53,
conv1d31,
conv1d32,
conv1d33,
layernormalization54,
linear99,
linear100,
constant1504,
layernormalization55,
layernormalization56,
linear101,
linear102,
constant1505,
layernormalization57,
linear103,
linear104,
linear105,
linear106,
constant174,
constant175,
constant1551,
linear107,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add82_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> Tensor<3> {
let layernormalization53_out1 = {
let dtype = add82_out1.clone().dtype();
self.layernormalization53
.forward(add82_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose111_out1 = layernormalization53_out1.permute([0, 2, 1]);
let conv1d31_out1 = self.conv1d31.forward(transpose111_out1);
let split_tensors = conv1d31_out1.split_with_sizes([1024, 1024].into(), 1);
let [split11_out1, split11_out2] = split_tensors.try_into().unwrap();
let sigmoid42_out1 = burn::tensor::activation::sigmoid(split11_out2);
let mul65_out1 = split11_out1.mul(sigmoid42_out1);
let constant1495_out1 = 0f32;
let where34_out1 = mul65_out1.mask_fill(unsqueeze52_out1, constant1495_out1);
let pad22_out1 = where34_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d32_out1 = self.conv1d32.forward(pad22_out1);
let sigmoid43_out1 = burn::tensor::activation::sigmoid(conv1d32_out1.clone());
let mul66_out1 = conv1d32_out1.mul(sigmoid43_out1);
let conv1d33_out1 = self.conv1d33.forward(mul66_out1);
let transpose113_out1 = conv1d33_out1.permute([0, 2, 1]);
let add83_out1 = add82_out1.add(transpose113_out1);
let layernormalization54_out1 = {
let dtype = add83_out1.clone().dtype();
self.layernormalization54
.forward(add83_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear99_out1 = self.linear99.forward(layernormalization54_out1);
let sigmoid44_out1 = burn::tensor::activation::sigmoid(linear99_out1.clone());
let mul67_out1 = linear99_out1.mul(sigmoid44_out1);
let linear100_out1 = self.linear100.forward(mul67_out1);
let constant1504_out1 = self.constant1504.val();
let mul68_out1 = linear100_out1
.mul((constant1504_out1).unsqueeze_dims(&[0isize, 1isize]));
let add84_out1 = add83_out1.add(mul68_out1);
let layernormalization55_out1 = {
let dtype = add84_out1.dtype();
self.layernormalization55
.forward(add84_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization56_out1 = {
let dtype = layernormalization55_out1.clone().dtype();
self.layernormalization56
.forward(
layernormalization55_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear101_out1 = self.linear101.forward(layernormalization56_out1);
let sigmoid45_out1 = burn::tensor::activation::sigmoid(linear101_out1.clone());
let mul69_out1 = linear101_out1.mul(sigmoid45_out1);
let linear102_out1 = self.linear102.forward(mul69_out1);
let constant1505_out1 = self.constant1505.val();
let mul70_out1 = linear102_out1
.mul((constant1505_out1).unsqueeze_dims(&[0isize, 1isize]));
let add85_out1 = layernormalization55_out1.add(mul70_out1);
let layernormalization57_out1 = {
let dtype = add85_out1.clone().dtype();
self.layernormalization57
.forward(add85_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape85_out1: [i64; 3] = {
let axes = &layernormalization57_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather72_out1 = shape85_out1[0] as i64;
let linear103_out1 = self.linear103.forward(layernormalization57_out1.clone());
let unsqueeze173_out1 = [gather72_out1 as i64];
let constant1510_out1: [i64; 1] = [128i64];
let constant1508_out1: [i64; 1] = [-1i64];
let constant1509_out1: [i64; 1] = [8i64];
let concat118_out1: [i64; 4usize] = [
&unsqueeze173_out1[..],
&constant1508_out1[..],
&constant1509_out1[..],
&constant1510_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze174_out1 = [gather72_out1 as i64];
let constant1514_out1: [i64; 1] = [128i64];
let constant1512_out1: [i64; 1] = [-1i64];
let constant1513_out1: [i64; 1] = [8i64];
let concat119_out1: [i64; 4usize] = [
&unsqueeze174_out1[..],
&constant1512_out1[..],
&constant1513_out1[..],
&constant1514_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze175_out1 = [gather72_out1 as i64];
let constant1518_out1: [i64; 1] = [128i64];
let constant1516_out1: [i64; 1] = [-1i64];
let constant1517_out1: [i64; 1] = [8i64];
let concat120_out1: [i64; 4usize] = [
&unsqueeze175_out1[..],
&constant1516_out1[..],
&constant1517_out1[..],
&constant1518_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape124_out1 = linear103_out1.reshape(concat118_out1);
let linear104_out1 = self.linear104.forward(layernormalization57_out1.clone());
let reshape125_out1 = linear104_out1.reshape(concat119_out1);
let linear105_out1 = self.linear105.forward(layernormalization57_out1);
let reshape126_out1 = linear105_out1.reshape(concat120_out1);
let transpose114_out1 = reshape126_out1.permute([0, 2, 1, 3]);
let linear106_out1 = self.linear106.forward(slice1_out1);
let reshape127_out1 = linear106_out1.reshape([1, -1, 8, 128]);
let constant174_out1 = self.constant174.val();
let add86_out1 = reshape124_out1
.clone()
.add((constant174_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose115_out1 = add86_out1.permute([0, 2, 1, 3]);
let constant175_out1 = self.constant175.val();
let add87_out1 = reshape124_out1
.add((constant175_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose116_out1 = add87_out1.permute([0, 2, 1, 3]);
let transpose117_out1 = reshape127_out1.permute([0, 2, 3, 1]);
let matmul140_out1 = transpose116_out1.matmul(transpose117_out1);
let shape86_out1: [i64; 4] = {
let axes = &matmul140_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather73_out1 = shape86_out1[0] as i64;
let gather74_out1 = shape86_out1[1] as i64;
let gather75_out1 = shape86_out1[2] as i64;
let gather76_out1 = shape86_out1[3] as i64;
let pad23_out1 = matmul140_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze176_out1 = [gather73_out1 as i64];
let unsqueeze177_out1 = [gather74_out1 as i64];
let unsqueeze178_out1 = [gather75_out1 as i64];
let constant1533_out1: [i64; 1] = [-1i64];
let concat122_out1: [i64; 4usize] = [
&unsqueeze176_out1[..],
&unsqueeze177_out1[..],
&constant1533_out1[..],
&unsqueeze178_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape130_out1 = pad23_out1.reshape(concat122_out1);
let slice60_out1 = reshape130_out1.slice(s![.., .., 1.., ..]);
let unsqueeze179_out1 = [gather73_out1 as i64];
let unsqueeze180_out1 = [gather74_out1 as i64];
let unsqueeze181_out1 = [gather75_out1 as i64];
let unsqueeze182_out1 = [gather76_out1 as i64];
let concat123_out1: [i64; 4usize] = [
&unsqueeze179_out1[..],
&unsqueeze180_out1[..],
&unsqueeze181_out1[..],
&unsqueeze182_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape131_out1 = slice60_out1.reshape(concat123_out1);
let transpose119_out1 = reshape125_out1.permute([0, 2, 3, 1]);
let matmul141_out1 = transpose115_out1.matmul(transpose119_out1);
let shape90_out1: [i64; 4] = {
let axes = &matmul141_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice61_out1: [i64; 1] = shape90_out1[3..4].try_into().unwrap();
let squeeze12_out1 = slice61_out1[0] as i64;
let unsqueeze183_out1 = [squeeze12_out1 as i64];
let slice62_out1 = reshape131_out1
.slice(s![.., .., .., 0..unsqueeze183_out1[0]]);
let add88_out1 = matmul141_out1.add(slice62_out1);
let constant1551_out1 = self.constant1551.val();
let div15_out1 = add88_out1
.div((constant1551_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape91_out1: [i64; 4] = {
let axes = &transpose114_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather77_out1 = shape91_out1[0] as i64;
let constant1553_out1 = -10000f32;
let where35_out1 = div15_out1
.mask_fill(unsqueeze50_out1.clone(), constant1553_out1);
let softmax12_out1 = burn::tensor::activation::softmax(where35_out1, 3);
let constant1554_out1 = 0f32;
let where36_out1 = softmax12_out1.mask_fill(unsqueeze50_out1, constant1554_out1);
let matmul142_out1 = where36_out1.matmul(transpose114_out1);
let transpose120_out1 = matmul142_out1.permute([0, 2, 1, 3]);
let unsqueeze184_out1 = [gather77_out1 as i64];
let constant1557_out1: [i64; 1] = [1024i64];
let constant1556_out1: [i64; 1] = [-1i64];
let concat124_out1: [i64; 3usize] = [
&unsqueeze184_out1[..],
&constant1556_out1[..],
&constant1557_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape132_out1 = transpose120_out1.reshape(concat124_out1);
let linear107_out1 = self.linear107.forward(reshape132_out1);
let add89_out1 = add85_out1.add(linear107_out1);
add89_out1
}
}
#[derive(Module, Debug)]
pub struct Submodule14 {
layernormalization58: LayerNorm,
conv1d34: Conv1d,
conv1d35: Conv1d,
conv1d36: Conv1d,
layernormalization59: LayerNorm,
linear108: Linear,
linear109: Linear,
constant1567: burn::module::Param<Tensor<1>>,
layernormalization60: LayerNorm,
layernormalization61: LayerNorm,
linear110: Linear,
linear111: Linear,
constant1568: burn::module::Param<Tensor<1>>,
layernormalization62: LayerNorm,
linear112: Linear,
linear113: Linear,
linear114: Linear,
linear115: Linear,
constant188: burn::module::Param<Tensor<2>>,
constant189: burn::module::Param<Tensor<2>>,
constant1614: burn::module::Param<Tensor<1>>,
linear116: Linear,
#[module(skip)]
device: Device,
}
impl Submodule14 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization58 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d34 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d35 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d36 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization59 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear108 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear109 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1567: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization60 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization61 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear110 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear111 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1568: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization62 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear112 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear113 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear114 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear115 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant188: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant189: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant1614: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear116 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization58,
conv1d34,
conv1d35,
conv1d36,
layernormalization59,
linear108,
linear109,
constant1567,
layernormalization60,
layernormalization61,
linear110,
linear111,
constant1568,
layernormalization62,
linear112,
linear113,
linear114,
linear115,
constant188,
constant189,
constant1614,
linear116,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add89_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> Tensor<3> {
let layernormalization58_out1 = {
let dtype = add89_out1.clone().dtype();
self.layernormalization58
.forward(add89_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose121_out1 = layernormalization58_out1.permute([0, 2, 1]);
let conv1d34_out1 = self.conv1d34.forward(transpose121_out1);
let split_tensors = conv1d34_out1.split_with_sizes([1024, 1024].into(), 1);
let [split12_out1, split12_out2] = split_tensors.try_into().unwrap();
let sigmoid46_out1 = burn::tensor::activation::sigmoid(split12_out2);
let mul71_out1 = split12_out1.mul(sigmoid46_out1);
let constant1558_out1 = 0f32;
let where37_out1 = mul71_out1.mask_fill(unsqueeze52_out1, constant1558_out1);
let pad24_out1 = where37_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d35_out1 = self.conv1d35.forward(pad24_out1);
let sigmoid47_out1 = burn::tensor::activation::sigmoid(conv1d35_out1.clone());
let mul72_out1 = conv1d35_out1.mul(sigmoid47_out1);
let conv1d36_out1 = self.conv1d36.forward(mul72_out1);
let transpose123_out1 = conv1d36_out1.permute([0, 2, 1]);
let add90_out1 = add89_out1.add(transpose123_out1);
let layernormalization59_out1 = {
let dtype = add90_out1.clone().dtype();
self.layernormalization59
.forward(add90_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear108_out1 = self.linear108.forward(layernormalization59_out1);
let sigmoid48_out1 = burn::tensor::activation::sigmoid(linear108_out1.clone());
let mul73_out1 = linear108_out1.mul(sigmoid48_out1);
let linear109_out1 = self.linear109.forward(mul73_out1);
let constant1567_out1 = self.constant1567.val();
let mul74_out1 = linear109_out1
.mul((constant1567_out1).unsqueeze_dims(&[0isize, 1isize]));
let add91_out1 = add90_out1.add(mul74_out1);
let layernormalization60_out1 = {
let dtype = add91_out1.dtype();
self.layernormalization60
.forward(add91_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization61_out1 = {
let dtype = layernormalization60_out1.clone().dtype();
self.layernormalization61
.forward(
layernormalization60_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear110_out1 = self.linear110.forward(layernormalization61_out1);
let sigmoid49_out1 = burn::tensor::activation::sigmoid(linear110_out1.clone());
let mul75_out1 = linear110_out1.mul(sigmoid49_out1);
let linear111_out1 = self.linear111.forward(mul75_out1);
let constant1568_out1 = self.constant1568.val();
let mul76_out1 = linear111_out1
.mul((constant1568_out1).unsqueeze_dims(&[0isize, 1isize]));
let add92_out1 = layernormalization60_out1.add(mul76_out1);
let layernormalization62_out1 = {
let dtype = add92_out1.clone().dtype();
self.layernormalization62
.forward(add92_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape92_out1: [i64; 3] = {
let axes = &layernormalization62_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather78_out1 = shape92_out1[0] as i64;
let linear112_out1 = self.linear112.forward(layernormalization62_out1.clone());
let unsqueeze185_out1 = [gather78_out1 as i64];
let constant1573_out1: [i64; 1] = [128i64];
let constant1571_out1: [i64; 1] = [-1i64];
let constant1572_out1: [i64; 1] = [8i64];
let concat126_out1: [i64; 4usize] = [
&unsqueeze185_out1[..],
&constant1571_out1[..],
&constant1572_out1[..],
&constant1573_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze186_out1 = [gather78_out1 as i64];
let constant1577_out1: [i64; 1] = [128i64];
let constant1575_out1: [i64; 1] = [-1i64];
let constant1576_out1: [i64; 1] = [8i64];
let concat127_out1: [i64; 4usize] = [
&unsqueeze186_out1[..],
&constant1575_out1[..],
&constant1576_out1[..],
&constant1577_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze187_out1 = [gather78_out1 as i64];
let constant1581_out1: [i64; 1] = [128i64];
let constant1579_out1: [i64; 1] = [-1i64];
let constant1580_out1: [i64; 1] = [8i64];
let concat128_out1: [i64; 4usize] = [
&unsqueeze187_out1[..],
&constant1579_out1[..],
&constant1580_out1[..],
&constant1581_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape135_out1 = linear112_out1.reshape(concat126_out1);
let linear113_out1 = self.linear113.forward(layernormalization62_out1.clone());
let reshape136_out1 = linear113_out1.reshape(concat127_out1);
let linear114_out1 = self.linear114.forward(layernormalization62_out1);
let reshape137_out1 = linear114_out1.reshape(concat128_out1);
let transpose124_out1 = reshape137_out1.permute([0, 2, 1, 3]);
let linear115_out1 = self.linear115.forward(slice1_out1);
let reshape138_out1 = linear115_out1.reshape([1, -1, 8, 128]);
let constant188_out1 = self.constant188.val();
let add93_out1 = reshape135_out1
.clone()
.add((constant188_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose125_out1 = add93_out1.permute([0, 2, 1, 3]);
let constant189_out1 = self.constant189.val();
let add94_out1 = reshape135_out1
.add((constant189_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose126_out1 = add94_out1.permute([0, 2, 1, 3]);
let transpose127_out1 = reshape138_out1.permute([0, 2, 3, 1]);
let matmul152_out1 = transpose126_out1.matmul(transpose127_out1);
let shape93_out1: [i64; 4] = {
let axes = &matmul152_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather79_out1 = shape93_out1[0] as i64;
let gather80_out1 = shape93_out1[1] as i64;
let gather81_out1 = shape93_out1[2] as i64;
let gather82_out1 = shape93_out1[3] as i64;
let pad25_out1 = matmul152_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze188_out1 = [gather79_out1 as i64];
let unsqueeze189_out1 = [gather80_out1 as i64];
let unsqueeze190_out1 = [gather81_out1 as i64];
let constant1596_out1: [i64; 1] = [-1i64];
let concat130_out1: [i64; 4usize] = [
&unsqueeze188_out1[..],
&unsqueeze189_out1[..],
&constant1596_out1[..],
&unsqueeze190_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape141_out1 = pad25_out1.reshape(concat130_out1);
let slice65_out1 = reshape141_out1.slice(s![.., .., 1.., ..]);
let unsqueeze191_out1 = [gather79_out1 as i64];
let unsqueeze192_out1 = [gather80_out1 as i64];
let unsqueeze193_out1 = [gather81_out1 as i64];
let unsqueeze194_out1 = [gather82_out1 as i64];
let concat131_out1: [i64; 4usize] = [
&unsqueeze191_out1[..],
&unsqueeze192_out1[..],
&unsqueeze193_out1[..],
&unsqueeze194_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape142_out1 = slice65_out1.reshape(concat131_out1);
let transpose129_out1 = reshape136_out1.permute([0, 2, 3, 1]);
let matmul153_out1 = transpose125_out1.matmul(transpose129_out1);
let shape97_out1: [i64; 4] = {
let axes = &matmul153_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice66_out1: [i64; 1] = shape97_out1[3..4].try_into().unwrap();
let squeeze13_out1 = slice66_out1[0] as i64;
let unsqueeze195_out1 = [squeeze13_out1 as i64];
let slice67_out1 = reshape142_out1
.slice(s![.., .., .., 0..unsqueeze195_out1[0]]);
let add95_out1 = matmul153_out1.add(slice67_out1);
let constant1614_out1 = self.constant1614.val();
let div16_out1 = add95_out1
.div((constant1614_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape98_out1: [i64; 4] = {
let axes = &transpose124_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather83_out1 = shape98_out1[0] as i64;
let constant1616_out1 = -10000f32;
let where38_out1 = div16_out1
.mask_fill(unsqueeze50_out1.clone(), constant1616_out1);
let softmax13_out1 = burn::tensor::activation::softmax(where38_out1, 3);
let constant1617_out1 = 0f32;
let where39_out1 = softmax13_out1.mask_fill(unsqueeze50_out1, constant1617_out1);
let matmul154_out1 = where39_out1.matmul(transpose124_out1);
let transpose130_out1 = matmul154_out1.permute([0, 2, 1, 3]);
let unsqueeze196_out1 = [gather83_out1 as i64];
let constant1620_out1: [i64; 1] = [1024i64];
let constant1619_out1: [i64; 1] = [-1i64];
let concat132_out1: [i64; 3usize] = [
&unsqueeze196_out1[..],
&constant1619_out1[..],
&constant1620_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape143_out1 = transpose130_out1.reshape(concat132_out1);
let linear116_out1 = self.linear116.forward(reshape143_out1);
let add96_out1 = add92_out1.add(linear116_out1);
add96_out1
}
}
#[derive(Module, Debug)]
pub struct Submodule15 {
layernormalization63: LayerNorm,
conv1d37: Conv1d,
conv1d38: Conv1d,
conv1d39: Conv1d,
layernormalization64: LayerNorm,
linear117: Linear,
linear118: Linear,
constant1630: burn::module::Param<Tensor<1>>,
layernormalization65: LayerNorm,
layernormalization66: LayerNorm,
linear119: Linear,
linear120: Linear,
constant1631: burn::module::Param<Tensor<1>>,
layernormalization67: LayerNorm,
linear121: Linear,
linear122: Linear,
linear123: Linear,
linear124: Linear,
constant202: burn::module::Param<Tensor<2>>,
constant203: burn::module::Param<Tensor<2>>,
constant1677: burn::module::Param<Tensor<1>>,
linear125: Linear,
#[module(skip)]
device: Device,
}
impl Submodule15 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization63 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d37 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d38 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d39 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization64 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear117 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear118 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1630: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization65 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization66 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear119 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear120 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1631: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization67 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear121 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear122 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear123 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear124 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant202: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant203: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant1677: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear125 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization63,
conv1d37,
conv1d38,
conv1d39,
layernormalization64,
linear117,
linear118,
constant1630,
layernormalization65,
layernormalization66,
linear119,
linear120,
constant1631,
layernormalization67,
linear121,
linear122,
linear123,
linear124,
constant202,
constant203,
constant1677,
linear125,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add96_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> Tensor<3> {
let layernormalization63_out1 = {
let dtype = add96_out1.clone().dtype();
self.layernormalization63
.forward(add96_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose131_out1 = layernormalization63_out1.permute([0, 2, 1]);
let conv1d37_out1 = self.conv1d37.forward(transpose131_out1);
let split_tensors = conv1d37_out1.split_with_sizes([1024, 1024].into(), 1);
let [split13_out1, split13_out2] = split_tensors.try_into().unwrap();
let sigmoid50_out1 = burn::tensor::activation::sigmoid(split13_out2);
let mul77_out1 = split13_out1.mul(sigmoid50_out1);
let constant1621_out1 = 0f32;
let where40_out1 = mul77_out1.mask_fill(unsqueeze52_out1, constant1621_out1);
let pad26_out1 = where40_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d38_out1 = self.conv1d38.forward(pad26_out1);
let sigmoid51_out1 = burn::tensor::activation::sigmoid(conv1d38_out1.clone());
let mul78_out1 = conv1d38_out1.mul(sigmoid51_out1);
let conv1d39_out1 = self.conv1d39.forward(mul78_out1);
let transpose133_out1 = conv1d39_out1.permute([0, 2, 1]);
let add97_out1 = add96_out1.add(transpose133_out1);
let layernormalization64_out1 = {
let dtype = add97_out1.clone().dtype();
self.layernormalization64
.forward(add97_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear117_out1 = self.linear117.forward(layernormalization64_out1);
let sigmoid52_out1 = burn::tensor::activation::sigmoid(linear117_out1.clone());
let mul79_out1 = linear117_out1.mul(sigmoid52_out1);
let linear118_out1 = self.linear118.forward(mul79_out1);
let constant1630_out1 = self.constant1630.val();
let mul80_out1 = linear118_out1
.mul((constant1630_out1).unsqueeze_dims(&[0isize, 1isize]));
let add98_out1 = add97_out1.add(mul80_out1);
let layernormalization65_out1 = {
let dtype = add98_out1.dtype();
self.layernormalization65
.forward(add98_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization66_out1 = {
let dtype = layernormalization65_out1.clone().dtype();
self.layernormalization66
.forward(
layernormalization65_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear119_out1 = self.linear119.forward(layernormalization66_out1);
let sigmoid53_out1 = burn::tensor::activation::sigmoid(linear119_out1.clone());
let mul81_out1 = linear119_out1.mul(sigmoid53_out1);
let linear120_out1 = self.linear120.forward(mul81_out1);
let constant1631_out1 = self.constant1631.val();
let mul82_out1 = linear120_out1
.mul((constant1631_out1).unsqueeze_dims(&[0isize, 1isize]));
let add99_out1 = layernormalization65_out1.add(mul82_out1);
let layernormalization67_out1 = {
let dtype = add99_out1.clone().dtype();
self.layernormalization67
.forward(add99_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape99_out1: [i64; 3] = {
let axes = &layernormalization67_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather84_out1 = shape99_out1[0] as i64;
let linear121_out1 = self.linear121.forward(layernormalization67_out1.clone());
let unsqueeze197_out1 = [gather84_out1 as i64];
let constant1636_out1: [i64; 1] = [128i64];
let constant1634_out1: [i64; 1] = [-1i64];
let constant1635_out1: [i64; 1] = [8i64];
let concat134_out1: [i64; 4usize] = [
&unsqueeze197_out1[..],
&constant1634_out1[..],
&constant1635_out1[..],
&constant1636_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze198_out1 = [gather84_out1 as i64];
let constant1640_out1: [i64; 1] = [128i64];
let constant1638_out1: [i64; 1] = [-1i64];
let constant1639_out1: [i64; 1] = [8i64];
let concat135_out1: [i64; 4usize] = [
&unsqueeze198_out1[..],
&constant1638_out1[..],
&constant1639_out1[..],
&constant1640_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze199_out1 = [gather84_out1 as i64];
let constant1644_out1: [i64; 1] = [128i64];
let constant1642_out1: [i64; 1] = [-1i64];
let constant1643_out1: [i64; 1] = [8i64];
let concat136_out1: [i64; 4usize] = [
&unsqueeze199_out1[..],
&constant1642_out1[..],
&constant1643_out1[..],
&constant1644_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape146_out1 = linear121_out1.reshape(concat134_out1);
let linear122_out1 = self.linear122.forward(layernormalization67_out1.clone());
let reshape147_out1 = linear122_out1.reshape(concat135_out1);
let linear123_out1 = self.linear123.forward(layernormalization67_out1);
let reshape148_out1 = linear123_out1.reshape(concat136_out1);
let transpose134_out1 = reshape148_out1.permute([0, 2, 1, 3]);
let linear124_out1 = self.linear124.forward(slice1_out1);
let reshape149_out1 = linear124_out1.reshape([1, -1, 8, 128]);
let constant202_out1 = self.constant202.val();
let add100_out1 = reshape146_out1
.clone()
.add((constant202_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose135_out1 = add100_out1.permute([0, 2, 1, 3]);
let constant203_out1 = self.constant203.val();
let add101_out1 = reshape146_out1
.add((constant203_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose136_out1 = add101_out1.permute([0, 2, 1, 3]);
let transpose137_out1 = reshape149_out1.permute([0, 2, 3, 1]);
let matmul164_out1 = transpose136_out1.matmul(transpose137_out1);
let shape100_out1: [i64; 4] = {
let axes = &matmul164_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather85_out1 = shape100_out1[0] as i64;
let gather86_out1 = shape100_out1[1] as i64;
let gather87_out1 = shape100_out1[2] as i64;
let gather88_out1 = shape100_out1[3] as i64;
let pad27_out1 = matmul164_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze200_out1 = [gather85_out1 as i64];
let unsqueeze201_out1 = [gather86_out1 as i64];
let unsqueeze202_out1 = [gather87_out1 as i64];
let constant1659_out1: [i64; 1] = [-1i64];
let concat138_out1: [i64; 4usize] = [
&unsqueeze200_out1[..],
&unsqueeze201_out1[..],
&constant1659_out1[..],
&unsqueeze202_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape152_out1 = pad27_out1.reshape(concat138_out1);
let slice70_out1 = reshape152_out1.slice(s![.., .., 1.., ..]);
let unsqueeze203_out1 = [gather85_out1 as i64];
let unsqueeze204_out1 = [gather86_out1 as i64];
let unsqueeze205_out1 = [gather87_out1 as i64];
let unsqueeze206_out1 = [gather88_out1 as i64];
let concat139_out1: [i64; 4usize] = [
&unsqueeze203_out1[..],
&unsqueeze204_out1[..],
&unsqueeze205_out1[..],
&unsqueeze206_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape153_out1 = slice70_out1.reshape(concat139_out1);
let transpose139_out1 = reshape147_out1.permute([0, 2, 3, 1]);
let matmul165_out1 = transpose135_out1.matmul(transpose139_out1);
let shape104_out1: [i64; 4] = {
let axes = &matmul165_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice71_out1: [i64; 1] = shape104_out1[3..4].try_into().unwrap();
let squeeze14_out1 = slice71_out1[0] as i64;
let unsqueeze207_out1 = [squeeze14_out1 as i64];
let slice72_out1 = reshape153_out1
.slice(s![.., .., .., 0..unsqueeze207_out1[0]]);
let add102_out1 = matmul165_out1.add(slice72_out1);
let constant1677_out1 = self.constant1677.val();
let div17_out1 = add102_out1
.div((constant1677_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape105_out1: [i64; 4] = {
let axes = &transpose134_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather89_out1 = shape105_out1[0] as i64;
let constant1679_out1 = -10000f32;
let where41_out1 = div17_out1
.mask_fill(unsqueeze50_out1.clone(), constant1679_out1);
let softmax14_out1 = burn::tensor::activation::softmax(where41_out1, 3);
let constant1680_out1 = 0f32;
let where42_out1 = softmax14_out1.mask_fill(unsqueeze50_out1, constant1680_out1);
let matmul166_out1 = where42_out1.matmul(transpose134_out1);
let transpose140_out1 = matmul166_out1.permute([0, 2, 1, 3]);
let unsqueeze208_out1 = [gather89_out1 as i64];
let constant1683_out1: [i64; 1] = [1024i64];
let constant1682_out1: [i64; 1] = [-1i64];
let concat140_out1: [i64; 3usize] = [
&unsqueeze208_out1[..],
&constant1682_out1[..],
&constant1683_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape154_out1 = transpose140_out1.reshape(concat140_out1);
let linear125_out1 = self.linear125.forward(reshape154_out1);
let add103_out1 = add99_out1.add(linear125_out1);
add103_out1
}
}
#[derive(Module, Debug)]
pub struct Submodule16 {
layernormalization68: LayerNorm,
conv1d40: Conv1d,
conv1d41: Conv1d,
conv1d42: Conv1d,
layernormalization69: LayerNorm,
linear126: Linear,
linear127: Linear,
constant1693: burn::module::Param<Tensor<1>>,
layernormalization70: LayerNorm,
layernormalization71: LayerNorm,
linear128: Linear,
linear129: Linear,
constant1694: burn::module::Param<Tensor<1>>,
layernormalization72: LayerNorm,
linear130: Linear,
linear131: Linear,
linear132: Linear,
linear133: Linear,
constant216: burn::module::Param<Tensor<2>>,
constant217: burn::module::Param<Tensor<2>>,
constant1740: burn::module::Param<Tensor<1>>,
linear134: Linear,
#[module(skip)]
device: Device,
}
impl Submodule16 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization68 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d40 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d41 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d42 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization69 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear126 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear127 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1693: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization70 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization71 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear128 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear129 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1694: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization72 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear130 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear131 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear132 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear133 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant216: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant217: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant1740: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear134 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization68,
conv1d40,
conv1d41,
conv1d42,
layernormalization69,
linear126,
linear127,
constant1693,
layernormalization70,
layernormalization71,
linear128,
linear129,
constant1694,
layernormalization72,
linear130,
linear131,
linear132,
linear133,
constant216,
constant217,
constant1740,
linear134,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add103_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> Tensor<3> {
let layernormalization68_out1 = {
let dtype = add103_out1.clone().dtype();
self.layernormalization68
.forward(add103_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose141_out1 = layernormalization68_out1.permute([0, 2, 1]);
let conv1d40_out1 = self.conv1d40.forward(transpose141_out1);
let split_tensors = conv1d40_out1.split_with_sizes([1024, 1024].into(), 1);
let [split14_out1, split14_out2] = split_tensors.try_into().unwrap();
let sigmoid54_out1 = burn::tensor::activation::sigmoid(split14_out2);
let mul83_out1 = split14_out1.mul(sigmoid54_out1);
let constant1684_out1 = 0f32;
let where43_out1 = mul83_out1.mask_fill(unsqueeze52_out1, constant1684_out1);
let pad28_out1 = where43_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d41_out1 = self.conv1d41.forward(pad28_out1);
let sigmoid55_out1 = burn::tensor::activation::sigmoid(conv1d41_out1.clone());
let mul84_out1 = conv1d41_out1.mul(sigmoid55_out1);
let conv1d42_out1 = self.conv1d42.forward(mul84_out1);
let transpose143_out1 = conv1d42_out1.permute([0, 2, 1]);
let add104_out1 = add103_out1.add(transpose143_out1);
let layernormalization69_out1 = {
let dtype = add104_out1.clone().dtype();
self.layernormalization69
.forward(add104_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear126_out1 = self.linear126.forward(layernormalization69_out1);
let sigmoid56_out1 = burn::tensor::activation::sigmoid(linear126_out1.clone());
let mul85_out1 = linear126_out1.mul(sigmoid56_out1);
let linear127_out1 = self.linear127.forward(mul85_out1);
let constant1693_out1 = self.constant1693.val();
let mul86_out1 = linear127_out1
.mul((constant1693_out1).unsqueeze_dims(&[0isize, 1isize]));
let add105_out1 = add104_out1.add(mul86_out1);
let layernormalization70_out1 = {
let dtype = add105_out1.dtype();
self.layernormalization70
.forward(add105_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization71_out1 = {
let dtype = layernormalization70_out1.clone().dtype();
self.layernormalization71
.forward(
layernormalization70_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear128_out1 = self.linear128.forward(layernormalization71_out1);
let sigmoid57_out1 = burn::tensor::activation::sigmoid(linear128_out1.clone());
let mul87_out1 = linear128_out1.mul(sigmoid57_out1);
let linear129_out1 = self.linear129.forward(mul87_out1);
let constant1694_out1 = self.constant1694.val();
let mul88_out1 = linear129_out1
.mul((constant1694_out1).unsqueeze_dims(&[0isize, 1isize]));
let add106_out1 = layernormalization70_out1.add(mul88_out1);
let layernormalization72_out1 = {
let dtype = add106_out1.clone().dtype();
self.layernormalization72
.forward(add106_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape106_out1: [i64; 3] = {
let axes = &layernormalization72_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather90_out1 = shape106_out1[0] as i64;
let linear130_out1 = self.linear130.forward(layernormalization72_out1.clone());
let unsqueeze209_out1 = [gather90_out1 as i64];
let constant1699_out1: [i64; 1] = [128i64];
let constant1697_out1: [i64; 1] = [-1i64];
let constant1698_out1: [i64; 1] = [8i64];
let concat142_out1: [i64; 4usize] = [
&unsqueeze209_out1[..],
&constant1697_out1[..],
&constant1698_out1[..],
&constant1699_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze210_out1 = [gather90_out1 as i64];
let constant1703_out1: [i64; 1] = [128i64];
let constant1701_out1: [i64; 1] = [-1i64];
let constant1702_out1: [i64; 1] = [8i64];
let concat143_out1: [i64; 4usize] = [
&unsqueeze210_out1[..],
&constant1701_out1[..],
&constant1702_out1[..],
&constant1703_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze211_out1 = [gather90_out1 as i64];
let constant1707_out1: [i64; 1] = [128i64];
let constant1705_out1: [i64; 1] = [-1i64];
let constant1706_out1: [i64; 1] = [8i64];
let concat144_out1: [i64; 4usize] = [
&unsqueeze211_out1[..],
&constant1705_out1[..],
&constant1706_out1[..],
&constant1707_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape157_out1 = linear130_out1.reshape(concat142_out1);
let linear131_out1 = self.linear131.forward(layernormalization72_out1.clone());
let reshape158_out1 = linear131_out1.reshape(concat143_out1);
let linear132_out1 = self.linear132.forward(layernormalization72_out1);
let reshape159_out1 = linear132_out1.reshape(concat144_out1);
let transpose144_out1 = reshape159_out1.permute([0, 2, 1, 3]);
let linear133_out1 = self.linear133.forward(slice1_out1);
let reshape160_out1 = linear133_out1.reshape([1, -1, 8, 128]);
let constant216_out1 = self.constant216.val();
let add107_out1 = reshape157_out1
.clone()
.add((constant216_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose145_out1 = add107_out1.permute([0, 2, 1, 3]);
let constant217_out1 = self.constant217.val();
let add108_out1 = reshape157_out1
.add((constant217_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose146_out1 = add108_out1.permute([0, 2, 1, 3]);
let transpose147_out1 = reshape160_out1.permute([0, 2, 3, 1]);
let matmul176_out1 = transpose146_out1.matmul(transpose147_out1);
let shape107_out1: [i64; 4] = {
let axes = &matmul176_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather91_out1 = shape107_out1[0] as i64;
let gather92_out1 = shape107_out1[1] as i64;
let gather93_out1 = shape107_out1[2] as i64;
let gather94_out1 = shape107_out1[3] as i64;
let pad29_out1 = matmul176_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze212_out1 = [gather91_out1 as i64];
let unsqueeze213_out1 = [gather92_out1 as i64];
let unsqueeze214_out1 = [gather93_out1 as i64];
let constant1722_out1: [i64; 1] = [-1i64];
let concat146_out1: [i64; 4usize] = [
&unsqueeze212_out1[..],
&unsqueeze213_out1[..],
&constant1722_out1[..],
&unsqueeze214_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape163_out1 = pad29_out1.reshape(concat146_out1);
let slice75_out1 = reshape163_out1.slice(s![.., .., 1.., ..]);
let unsqueeze215_out1 = [gather91_out1 as i64];
let unsqueeze216_out1 = [gather92_out1 as i64];
let unsqueeze217_out1 = [gather93_out1 as i64];
let unsqueeze218_out1 = [gather94_out1 as i64];
let concat147_out1: [i64; 4usize] = [
&unsqueeze215_out1[..],
&unsqueeze216_out1[..],
&unsqueeze217_out1[..],
&unsqueeze218_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape164_out1 = slice75_out1.reshape(concat147_out1);
let transpose149_out1 = reshape158_out1.permute([0, 2, 3, 1]);
let matmul177_out1 = transpose145_out1.matmul(transpose149_out1);
let shape111_out1: [i64; 4] = {
let axes = &matmul177_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice76_out1: [i64; 1] = shape111_out1[3..4].try_into().unwrap();
let squeeze15_out1 = slice76_out1[0] as i64;
let unsqueeze219_out1 = [squeeze15_out1 as i64];
let slice77_out1 = reshape164_out1
.slice(s![.., .., .., 0..unsqueeze219_out1[0]]);
let add109_out1 = matmul177_out1.add(slice77_out1);
let constant1740_out1 = self.constant1740.val();
let div18_out1 = add109_out1
.div((constant1740_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape112_out1: [i64; 4] = {
let axes = &transpose144_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather95_out1 = shape112_out1[0] as i64;
let constant1742_out1 = -10000f32;
let where44_out1 = div18_out1
.mask_fill(unsqueeze50_out1.clone(), constant1742_out1);
let softmax15_out1 = burn::tensor::activation::softmax(where44_out1, 3);
let constant1743_out1 = 0f32;
let where45_out1 = softmax15_out1.mask_fill(unsqueeze50_out1, constant1743_out1);
let matmul178_out1 = where45_out1.matmul(transpose144_out1);
let transpose150_out1 = matmul178_out1.permute([0, 2, 1, 3]);
let unsqueeze220_out1 = [gather95_out1 as i64];
let constant1746_out1: [i64; 1] = [1024i64];
let constant1745_out1: [i64; 1] = [-1i64];
let concat148_out1: [i64; 3usize] = [
&unsqueeze220_out1[..],
&constant1745_out1[..],
&constant1746_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape165_out1 = transpose150_out1.reshape(concat148_out1);
let linear134_out1 = self.linear134.forward(reshape165_out1);
let add110_out1 = add106_out1.add(linear134_out1);
add110_out1
}
}
#[derive(Module, Debug)]
pub struct Submodule17 {
layernormalization73: LayerNorm,
conv1d43: Conv1d,
conv1d44: Conv1d,
conv1d45: Conv1d,
layernormalization74: LayerNorm,
linear135: Linear,
linear136: Linear,
constant1756: burn::module::Param<Tensor<1>>,
layernormalization75: LayerNorm,
layernormalization76: LayerNorm,
linear137: Linear,
linear138: Linear,
constant1757: burn::module::Param<Tensor<1>>,
layernormalization77: LayerNorm,
linear139: Linear,
linear140: Linear,
linear141: Linear,
linear142: Linear,
constant230: burn::module::Param<Tensor<2>>,
constant231: burn::module::Param<Tensor<2>>,
constant1803: burn::module::Param<Tensor<1>>,
linear143: Linear,
#[module(skip)]
device: Device,
}
impl Submodule17 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization73 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d43 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d44 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d45 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization74 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear135 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear136 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1756: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization75 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization76 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear137 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear138 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1757: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization77 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear139 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear140 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear141 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear142 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant230: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant231: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant1803: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear143 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization73,
conv1d43,
conv1d44,
conv1d45,
layernormalization74,
linear135,
linear136,
constant1756,
layernormalization75,
layernormalization76,
linear137,
linear138,
constant1757,
layernormalization77,
linear139,
linear140,
linear141,
linear142,
constant230,
constant231,
constant1803,
linear143,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add110_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> Tensor<3> {
let layernormalization73_out1 = {
let dtype = add110_out1.clone().dtype();
self.layernormalization73
.forward(add110_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose151_out1 = layernormalization73_out1.permute([0, 2, 1]);
let conv1d43_out1 = self.conv1d43.forward(transpose151_out1);
let split_tensors = conv1d43_out1.split_with_sizes([1024, 1024].into(), 1);
let [split15_out1, split15_out2] = split_tensors.try_into().unwrap();
let sigmoid58_out1 = burn::tensor::activation::sigmoid(split15_out2);
let mul89_out1 = split15_out1.mul(sigmoid58_out1);
let constant1747_out1 = 0f32;
let where46_out1 = mul89_out1.mask_fill(unsqueeze52_out1, constant1747_out1);
let pad30_out1 = where46_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d44_out1 = self.conv1d44.forward(pad30_out1);
let sigmoid59_out1 = burn::tensor::activation::sigmoid(conv1d44_out1.clone());
let mul90_out1 = conv1d44_out1.mul(sigmoid59_out1);
let conv1d45_out1 = self.conv1d45.forward(mul90_out1);
let transpose153_out1 = conv1d45_out1.permute([0, 2, 1]);
let add111_out1 = add110_out1.add(transpose153_out1);
let layernormalization74_out1 = {
let dtype = add111_out1.clone().dtype();
self.layernormalization74
.forward(add111_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear135_out1 = self.linear135.forward(layernormalization74_out1);
let sigmoid60_out1 = burn::tensor::activation::sigmoid(linear135_out1.clone());
let mul91_out1 = linear135_out1.mul(sigmoid60_out1);
let linear136_out1 = self.linear136.forward(mul91_out1);
let constant1756_out1 = self.constant1756.val();
let mul92_out1 = linear136_out1
.mul((constant1756_out1).unsqueeze_dims(&[0isize, 1isize]));
let add112_out1 = add111_out1.add(mul92_out1);
let layernormalization75_out1 = {
let dtype = add112_out1.dtype();
self.layernormalization75
.forward(add112_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization76_out1 = {
let dtype = layernormalization75_out1.clone().dtype();
self.layernormalization76
.forward(
layernormalization75_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear137_out1 = self.linear137.forward(layernormalization76_out1);
let sigmoid61_out1 = burn::tensor::activation::sigmoid(linear137_out1.clone());
let mul93_out1 = linear137_out1.mul(sigmoid61_out1);
let linear138_out1 = self.linear138.forward(mul93_out1);
let constant1757_out1 = self.constant1757.val();
let mul94_out1 = linear138_out1
.mul((constant1757_out1).unsqueeze_dims(&[0isize, 1isize]));
let add113_out1 = layernormalization75_out1.add(mul94_out1);
let layernormalization77_out1 = {
let dtype = add113_out1.clone().dtype();
self.layernormalization77
.forward(add113_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape113_out1: [i64; 3] = {
let axes = &layernormalization77_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather96_out1 = shape113_out1[0] as i64;
let linear139_out1 = self.linear139.forward(layernormalization77_out1.clone());
let unsqueeze221_out1 = [gather96_out1 as i64];
let constant1762_out1: [i64; 1] = [128i64];
let constant1760_out1: [i64; 1] = [-1i64];
let constant1761_out1: [i64; 1] = [8i64];
let concat150_out1: [i64; 4usize] = [
&unsqueeze221_out1[..],
&constant1760_out1[..],
&constant1761_out1[..],
&constant1762_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze222_out1 = [gather96_out1 as i64];
let constant1766_out1: [i64; 1] = [128i64];
let constant1764_out1: [i64; 1] = [-1i64];
let constant1765_out1: [i64; 1] = [8i64];
let concat151_out1: [i64; 4usize] = [
&unsqueeze222_out1[..],
&constant1764_out1[..],
&constant1765_out1[..],
&constant1766_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze223_out1 = [gather96_out1 as i64];
let constant1770_out1: [i64; 1] = [128i64];
let constant1768_out1: [i64; 1] = [-1i64];
let constant1769_out1: [i64; 1] = [8i64];
let concat152_out1: [i64; 4usize] = [
&unsqueeze223_out1[..],
&constant1768_out1[..],
&constant1769_out1[..],
&constant1770_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape168_out1 = linear139_out1.reshape(concat150_out1);
let linear140_out1 = self.linear140.forward(layernormalization77_out1.clone());
let reshape169_out1 = linear140_out1.reshape(concat151_out1);
let linear141_out1 = self.linear141.forward(layernormalization77_out1);
let reshape170_out1 = linear141_out1.reshape(concat152_out1);
let transpose154_out1 = reshape170_out1.permute([0, 2, 1, 3]);
let linear142_out1 = self.linear142.forward(slice1_out1);
let reshape171_out1 = linear142_out1.reshape([1, -1, 8, 128]);
let constant230_out1 = self.constant230.val();
let add114_out1 = reshape168_out1
.clone()
.add((constant230_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose155_out1 = add114_out1.permute([0, 2, 1, 3]);
let constant231_out1 = self.constant231.val();
let add115_out1 = reshape168_out1
.add((constant231_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose156_out1 = add115_out1.permute([0, 2, 1, 3]);
let transpose157_out1 = reshape171_out1.permute([0, 2, 3, 1]);
let matmul188_out1 = transpose156_out1.matmul(transpose157_out1);
let shape114_out1: [i64; 4] = {
let axes = &matmul188_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather97_out1 = shape114_out1[0] as i64;
let gather98_out1 = shape114_out1[1] as i64;
let gather99_out1 = shape114_out1[2] as i64;
let gather100_out1 = shape114_out1[3] as i64;
let pad31_out1 = matmul188_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze224_out1 = [gather97_out1 as i64];
let unsqueeze225_out1 = [gather98_out1 as i64];
let unsqueeze226_out1 = [gather99_out1 as i64];
let constant1785_out1: [i64; 1] = [-1i64];
let concat154_out1: [i64; 4usize] = [
&unsqueeze224_out1[..],
&unsqueeze225_out1[..],
&constant1785_out1[..],
&unsqueeze226_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape174_out1 = pad31_out1.reshape(concat154_out1);
let slice80_out1 = reshape174_out1.slice(s![.., .., 1.., ..]);
let unsqueeze227_out1 = [gather97_out1 as i64];
let unsqueeze228_out1 = [gather98_out1 as i64];
let unsqueeze229_out1 = [gather99_out1 as i64];
let unsqueeze230_out1 = [gather100_out1 as i64];
let concat155_out1: [i64; 4usize] = [
&unsqueeze227_out1[..],
&unsqueeze228_out1[..],
&unsqueeze229_out1[..],
&unsqueeze230_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape175_out1 = slice80_out1.reshape(concat155_out1);
let transpose159_out1 = reshape169_out1.permute([0, 2, 3, 1]);
let matmul189_out1 = transpose155_out1.matmul(transpose159_out1);
let shape118_out1: [i64; 4] = {
let axes = &matmul189_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice81_out1: [i64; 1] = shape118_out1[3..4].try_into().unwrap();
let squeeze16_out1 = slice81_out1[0] as i64;
let unsqueeze231_out1 = [squeeze16_out1 as i64];
let slice82_out1 = reshape175_out1
.slice(s![.., .., .., 0..unsqueeze231_out1[0]]);
let add116_out1 = matmul189_out1.add(slice82_out1);
let constant1803_out1 = self.constant1803.val();
let div19_out1 = add116_out1
.div((constant1803_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape119_out1: [i64; 4] = {
let axes = &transpose154_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather101_out1 = shape119_out1[0] as i64;
let constant1805_out1 = -10000f32;
let where47_out1 = div19_out1
.mask_fill(unsqueeze50_out1.clone(), constant1805_out1);
let softmax16_out1 = burn::tensor::activation::softmax(where47_out1, 3);
let constant1806_out1 = 0f32;
let where48_out1 = softmax16_out1.mask_fill(unsqueeze50_out1, constant1806_out1);
let matmul190_out1 = where48_out1.matmul(transpose154_out1);
let transpose160_out1 = matmul190_out1.permute([0, 2, 1, 3]);
let unsqueeze232_out1 = [gather101_out1 as i64];
let constant1809_out1: [i64; 1] = [1024i64];
let constant1808_out1: [i64; 1] = [-1i64];
let concat156_out1: [i64; 3usize] = [
&unsqueeze232_out1[..],
&constant1808_out1[..],
&constant1809_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape176_out1 = transpose160_out1.reshape(concat156_out1);
let linear143_out1 = self.linear143.forward(reshape176_out1);
let add117_out1 = add113_out1.add(linear143_out1);
add117_out1
}
}
#[derive(Module, Debug)]
pub struct Submodule18 {
layernormalization78: LayerNorm,
conv1d46: Conv1d,
conv1d47: Conv1d,
conv1d48: Conv1d,
layernormalization79: LayerNorm,
linear144: Linear,
linear145: Linear,
constant1819: burn::module::Param<Tensor<1>>,
layernormalization80: LayerNorm,
layernormalization81: LayerNorm,
linear146: Linear,
linear147: Linear,
constant1820: burn::module::Param<Tensor<1>>,
layernormalization82: LayerNorm,
linear148: Linear,
linear149: Linear,
linear150: Linear,
linear151: Linear,
constant244: burn::module::Param<Tensor<2>>,
constant245: burn::module::Param<Tensor<2>>,
constant1866: burn::module::Param<Tensor<1>>,
linear152: Linear,
#[module(skip)]
device: Device,
}
impl Submodule18 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization78 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d46 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d47 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d48 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization79 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear144 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear145 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1819: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization80 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization81 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear146 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear147 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1820: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization82 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear148 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear149 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear150 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear151 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant244: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant245: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant1866: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear152 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization78,
conv1d46,
conv1d47,
conv1d48,
layernormalization79,
linear144,
linear145,
constant1819,
layernormalization80,
layernormalization81,
linear146,
linear147,
constant1820,
layernormalization82,
linear148,
linear149,
linear150,
linear151,
constant244,
constant245,
constant1866,
linear152,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add117_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> Tensor<3> {
let layernormalization78_out1 = {
let dtype = add117_out1.clone().dtype();
self.layernormalization78
.forward(add117_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose161_out1 = layernormalization78_out1.permute([0, 2, 1]);
let conv1d46_out1 = self.conv1d46.forward(transpose161_out1);
let split_tensors = conv1d46_out1.split_with_sizes([1024, 1024].into(), 1);
let [split16_out1, split16_out2] = split_tensors.try_into().unwrap();
let sigmoid62_out1 = burn::tensor::activation::sigmoid(split16_out2);
let mul95_out1 = split16_out1.mul(sigmoid62_out1);
let constant1810_out1 = 0f32;
let where49_out1 = mul95_out1.mask_fill(unsqueeze52_out1, constant1810_out1);
let pad32_out1 = where49_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d47_out1 = self.conv1d47.forward(pad32_out1);
let sigmoid63_out1 = burn::tensor::activation::sigmoid(conv1d47_out1.clone());
let mul96_out1 = conv1d47_out1.mul(sigmoid63_out1);
let conv1d48_out1 = self.conv1d48.forward(mul96_out1);
let transpose163_out1 = conv1d48_out1.permute([0, 2, 1]);
let add118_out1 = add117_out1.add(transpose163_out1);
let layernormalization79_out1 = {
let dtype = add118_out1.clone().dtype();
self.layernormalization79
.forward(add118_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear144_out1 = self.linear144.forward(layernormalization79_out1);
let sigmoid64_out1 = burn::tensor::activation::sigmoid(linear144_out1.clone());
let mul97_out1 = linear144_out1.mul(sigmoid64_out1);
let linear145_out1 = self.linear145.forward(mul97_out1);
let constant1819_out1 = self.constant1819.val();
let mul98_out1 = linear145_out1
.mul((constant1819_out1).unsqueeze_dims(&[0isize, 1isize]));
let add119_out1 = add118_out1.add(mul98_out1);
let layernormalization80_out1 = {
let dtype = add119_out1.dtype();
self.layernormalization80
.forward(add119_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization81_out1 = {
let dtype = layernormalization80_out1.clone().dtype();
self.layernormalization81
.forward(
layernormalization80_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear146_out1 = self.linear146.forward(layernormalization81_out1);
let sigmoid65_out1 = burn::tensor::activation::sigmoid(linear146_out1.clone());
let mul99_out1 = linear146_out1.mul(sigmoid65_out1);
let linear147_out1 = self.linear147.forward(mul99_out1);
let constant1820_out1 = self.constant1820.val();
let mul100_out1 = linear147_out1
.mul((constant1820_out1).unsqueeze_dims(&[0isize, 1isize]));
let add120_out1 = layernormalization80_out1.add(mul100_out1);
let layernormalization82_out1 = {
let dtype = add120_out1.clone().dtype();
self.layernormalization82
.forward(add120_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape120_out1: [i64; 3] = {
let axes = &layernormalization82_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather102_out1 = shape120_out1[0] as i64;
let linear148_out1 = self.linear148.forward(layernormalization82_out1.clone());
let unsqueeze233_out1 = [gather102_out1 as i64];
let constant1825_out1: [i64; 1] = [128i64];
let constant1823_out1: [i64; 1] = [-1i64];
let constant1824_out1: [i64; 1] = [8i64];
let concat158_out1: [i64; 4usize] = [
&unsqueeze233_out1[..],
&constant1823_out1[..],
&constant1824_out1[..],
&constant1825_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze234_out1 = [gather102_out1 as i64];
let constant1829_out1: [i64; 1] = [128i64];
let constant1827_out1: [i64; 1] = [-1i64];
let constant1828_out1: [i64; 1] = [8i64];
let concat159_out1: [i64; 4usize] = [
&unsqueeze234_out1[..],
&constant1827_out1[..],
&constant1828_out1[..],
&constant1829_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze235_out1 = [gather102_out1 as i64];
let constant1833_out1: [i64; 1] = [128i64];
let constant1831_out1: [i64; 1] = [-1i64];
let constant1832_out1: [i64; 1] = [8i64];
let concat160_out1: [i64; 4usize] = [
&unsqueeze235_out1[..],
&constant1831_out1[..],
&constant1832_out1[..],
&constant1833_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape179_out1 = linear148_out1.reshape(concat158_out1);
let linear149_out1 = self.linear149.forward(layernormalization82_out1.clone());
let reshape180_out1 = linear149_out1.reshape(concat159_out1);
let linear150_out1 = self.linear150.forward(layernormalization82_out1);
let reshape181_out1 = linear150_out1.reshape(concat160_out1);
let transpose164_out1 = reshape181_out1.permute([0, 2, 1, 3]);
let linear151_out1 = self.linear151.forward(slice1_out1);
let reshape182_out1 = linear151_out1.reshape([1, -1, 8, 128]);
let constant244_out1 = self.constant244.val();
let add121_out1 = reshape179_out1
.clone()
.add((constant244_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose165_out1 = add121_out1.permute([0, 2, 1, 3]);
let constant245_out1 = self.constant245.val();
let add122_out1 = reshape179_out1
.add((constant245_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose166_out1 = add122_out1.permute([0, 2, 1, 3]);
let transpose167_out1 = reshape182_out1.permute([0, 2, 3, 1]);
let matmul200_out1 = transpose166_out1.matmul(transpose167_out1);
let shape121_out1: [i64; 4] = {
let axes = &matmul200_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather103_out1 = shape121_out1[0] as i64;
let gather104_out1 = shape121_out1[1] as i64;
let gather105_out1 = shape121_out1[2] as i64;
let gather106_out1 = shape121_out1[3] as i64;
let pad33_out1 = matmul200_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze236_out1 = [gather103_out1 as i64];
let unsqueeze237_out1 = [gather104_out1 as i64];
let unsqueeze238_out1 = [gather105_out1 as i64];
let constant1848_out1: [i64; 1] = [-1i64];
let concat162_out1: [i64; 4usize] = [
&unsqueeze236_out1[..],
&unsqueeze237_out1[..],
&constant1848_out1[..],
&unsqueeze238_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape185_out1 = pad33_out1.reshape(concat162_out1);
let slice85_out1 = reshape185_out1.slice(s![.., .., 1.., ..]);
let unsqueeze239_out1 = [gather103_out1 as i64];
let unsqueeze240_out1 = [gather104_out1 as i64];
let unsqueeze241_out1 = [gather105_out1 as i64];
let unsqueeze242_out1 = [gather106_out1 as i64];
let concat163_out1: [i64; 4usize] = [
&unsqueeze239_out1[..],
&unsqueeze240_out1[..],
&unsqueeze241_out1[..],
&unsqueeze242_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape186_out1 = slice85_out1.reshape(concat163_out1);
let transpose169_out1 = reshape180_out1.permute([0, 2, 3, 1]);
let matmul201_out1 = transpose165_out1.matmul(transpose169_out1);
let shape125_out1: [i64; 4] = {
let axes = &matmul201_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice86_out1: [i64; 1] = shape125_out1[3..4].try_into().unwrap();
let squeeze17_out1 = slice86_out1[0] as i64;
let unsqueeze243_out1 = [squeeze17_out1 as i64];
let slice87_out1 = reshape186_out1
.slice(s![.., .., .., 0..unsqueeze243_out1[0]]);
let add123_out1 = matmul201_out1.add(slice87_out1);
let constant1866_out1 = self.constant1866.val();
let div20_out1 = add123_out1
.div((constant1866_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape126_out1: [i64; 4] = {
let axes = &transpose164_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather107_out1 = shape126_out1[0] as i64;
let constant1868_out1 = -10000f32;
let where50_out1 = div20_out1
.mask_fill(unsqueeze50_out1.clone(), constant1868_out1);
let softmax17_out1 = burn::tensor::activation::softmax(where50_out1, 3);
let constant1869_out1 = 0f32;
let where51_out1 = softmax17_out1.mask_fill(unsqueeze50_out1, constant1869_out1);
let matmul202_out1 = where51_out1.matmul(transpose164_out1);
let transpose170_out1 = matmul202_out1.permute([0, 2, 1, 3]);
let unsqueeze244_out1 = [gather107_out1 as i64];
let constant1872_out1: [i64; 1] = [1024i64];
let constant1871_out1: [i64; 1] = [-1i64];
let concat164_out1: [i64; 3usize] = [
&unsqueeze244_out1[..],
&constant1871_out1[..],
&constant1872_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape187_out1 = transpose170_out1.reshape(concat164_out1);
let linear152_out1 = self.linear152.forward(reshape187_out1);
let add124_out1 = add120_out1.add(linear152_out1);
add124_out1
}
}
#[derive(Module, Debug)]
pub struct Submodule19 {
layernormalization83: LayerNorm,
conv1d49: Conv1d,
conv1d50: Conv1d,
conv1d51: Conv1d,
layernormalization84: LayerNorm,
linear153: Linear,
linear154: Linear,
constant1882: burn::module::Param<Tensor<1>>,
layernormalization85: LayerNorm,
layernormalization86: LayerNorm,
linear155: Linear,
linear156: Linear,
constant1883: burn::module::Param<Tensor<1>>,
layernormalization87: LayerNorm,
linear157: Linear,
linear158: Linear,
linear159: Linear,
linear160: Linear,
constant258: burn::module::Param<Tensor<2>>,
constant259: burn::module::Param<Tensor<2>>,
constant1929: burn::module::Param<Tensor<1>>,
linear161: Linear,
#[module(skip)]
device: Device,
}
impl Submodule19 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization83 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d49 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d50 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d51 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization84 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear153 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear154 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1882: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization85 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization86 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear155 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear156 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1883: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization87 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear157 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear158 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear159 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear160 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant258: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant259: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant1929: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear161 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization83,
conv1d49,
conv1d50,
conv1d51,
layernormalization84,
linear153,
linear154,
constant1882,
layernormalization85,
layernormalization86,
linear155,
linear156,
constant1883,
layernormalization87,
linear157,
linear158,
linear159,
linear160,
constant258,
constant259,
constant1929,
linear161,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add124_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> Tensor<3> {
let layernormalization83_out1 = {
let dtype = add124_out1.clone().dtype();
self.layernormalization83
.forward(add124_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose171_out1 = layernormalization83_out1.permute([0, 2, 1]);
let conv1d49_out1 = self.conv1d49.forward(transpose171_out1);
let split_tensors = conv1d49_out1.split_with_sizes([1024, 1024].into(), 1);
let [split17_out1, split17_out2] = split_tensors.try_into().unwrap();
let sigmoid66_out1 = burn::tensor::activation::sigmoid(split17_out2);
let mul101_out1 = split17_out1.mul(sigmoid66_out1);
let constant1873_out1 = 0f32;
let where52_out1 = mul101_out1.mask_fill(unsqueeze52_out1, constant1873_out1);
let pad34_out1 = where52_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d50_out1 = self.conv1d50.forward(pad34_out1);
let sigmoid67_out1 = burn::tensor::activation::sigmoid(conv1d50_out1.clone());
let mul102_out1 = conv1d50_out1.mul(sigmoid67_out1);
let conv1d51_out1 = self.conv1d51.forward(mul102_out1);
let transpose173_out1 = conv1d51_out1.permute([0, 2, 1]);
let add125_out1 = add124_out1.add(transpose173_out1);
let layernormalization84_out1 = {
let dtype = add125_out1.clone().dtype();
self.layernormalization84
.forward(add125_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear153_out1 = self.linear153.forward(layernormalization84_out1);
let sigmoid68_out1 = burn::tensor::activation::sigmoid(linear153_out1.clone());
let mul103_out1 = linear153_out1.mul(sigmoid68_out1);
let linear154_out1 = self.linear154.forward(mul103_out1);
let constant1882_out1 = self.constant1882.val();
let mul104_out1 = linear154_out1
.mul((constant1882_out1).unsqueeze_dims(&[0isize, 1isize]));
let add126_out1 = add125_out1.add(mul104_out1);
let layernormalization85_out1 = {
let dtype = add126_out1.dtype();
self.layernormalization85
.forward(add126_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization86_out1 = {
let dtype = layernormalization85_out1.clone().dtype();
self.layernormalization86
.forward(
layernormalization85_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear155_out1 = self.linear155.forward(layernormalization86_out1);
let sigmoid69_out1 = burn::tensor::activation::sigmoid(linear155_out1.clone());
let mul105_out1 = linear155_out1.mul(sigmoid69_out1);
let linear156_out1 = self.linear156.forward(mul105_out1);
let constant1883_out1 = self.constant1883.val();
let mul106_out1 = linear156_out1
.mul((constant1883_out1).unsqueeze_dims(&[0isize, 1isize]));
let add127_out1 = layernormalization85_out1.add(mul106_out1);
let layernormalization87_out1 = {
let dtype = add127_out1.clone().dtype();
self.layernormalization87
.forward(add127_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape127_out1: [i64; 3] = {
let axes = &layernormalization87_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather108_out1 = shape127_out1[0] as i64;
let linear157_out1 = self.linear157.forward(layernormalization87_out1.clone());
let unsqueeze245_out1 = [gather108_out1 as i64];
let constant1888_out1: [i64; 1] = [128i64];
let constant1886_out1: [i64; 1] = [-1i64];
let constant1887_out1: [i64; 1] = [8i64];
let concat166_out1: [i64; 4usize] = [
&unsqueeze245_out1[..],
&constant1886_out1[..],
&constant1887_out1[..],
&constant1888_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze246_out1 = [gather108_out1 as i64];
let constant1892_out1: [i64; 1] = [128i64];
let constant1890_out1: [i64; 1] = [-1i64];
let constant1891_out1: [i64; 1] = [8i64];
let concat167_out1: [i64; 4usize] = [
&unsqueeze246_out1[..],
&constant1890_out1[..],
&constant1891_out1[..],
&constant1892_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze247_out1 = [gather108_out1 as i64];
let constant1896_out1: [i64; 1] = [128i64];
let constant1894_out1: [i64; 1] = [-1i64];
let constant1895_out1: [i64; 1] = [8i64];
let concat168_out1: [i64; 4usize] = [
&unsqueeze247_out1[..],
&constant1894_out1[..],
&constant1895_out1[..],
&constant1896_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape190_out1 = linear157_out1.reshape(concat166_out1);
let linear158_out1 = self.linear158.forward(layernormalization87_out1.clone());
let reshape191_out1 = linear158_out1.reshape(concat167_out1);
let linear159_out1 = self.linear159.forward(layernormalization87_out1);
let reshape192_out1 = linear159_out1.reshape(concat168_out1);
let transpose174_out1 = reshape192_out1.permute([0, 2, 1, 3]);
let linear160_out1 = self.linear160.forward(slice1_out1);
let reshape193_out1 = linear160_out1.reshape([1, -1, 8, 128]);
let constant258_out1 = self.constant258.val();
let add128_out1 = reshape190_out1
.clone()
.add((constant258_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose175_out1 = add128_out1.permute([0, 2, 1, 3]);
let constant259_out1 = self.constant259.val();
let add129_out1 = reshape190_out1
.add((constant259_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose176_out1 = add129_out1.permute([0, 2, 1, 3]);
let transpose177_out1 = reshape193_out1.permute([0, 2, 3, 1]);
let matmul212_out1 = transpose176_out1.matmul(transpose177_out1);
let shape128_out1: [i64; 4] = {
let axes = &matmul212_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather109_out1 = shape128_out1[0] as i64;
let gather110_out1 = shape128_out1[1] as i64;
let gather111_out1 = shape128_out1[2] as i64;
let gather112_out1 = shape128_out1[3] as i64;
let pad35_out1 = matmul212_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze248_out1 = [gather109_out1 as i64];
let unsqueeze249_out1 = [gather110_out1 as i64];
let unsqueeze250_out1 = [gather111_out1 as i64];
let constant1911_out1: [i64; 1] = [-1i64];
let concat170_out1: [i64; 4usize] = [
&unsqueeze248_out1[..],
&unsqueeze249_out1[..],
&constant1911_out1[..],
&unsqueeze250_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape196_out1 = pad35_out1.reshape(concat170_out1);
let slice90_out1 = reshape196_out1.slice(s![.., .., 1.., ..]);
let unsqueeze251_out1 = [gather109_out1 as i64];
let unsqueeze252_out1 = [gather110_out1 as i64];
let unsqueeze253_out1 = [gather111_out1 as i64];
let unsqueeze254_out1 = [gather112_out1 as i64];
let concat171_out1: [i64; 4usize] = [
&unsqueeze251_out1[..],
&unsqueeze252_out1[..],
&unsqueeze253_out1[..],
&unsqueeze254_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape197_out1 = slice90_out1.reshape(concat171_out1);
let transpose179_out1 = reshape191_out1.permute([0, 2, 3, 1]);
let matmul213_out1 = transpose175_out1.matmul(transpose179_out1);
let shape132_out1: [i64; 4] = {
let axes = &matmul213_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice91_out1: [i64; 1] = shape132_out1[3..4].try_into().unwrap();
let squeeze18_out1 = slice91_out1[0] as i64;
let unsqueeze255_out1 = [squeeze18_out1 as i64];
let slice92_out1 = reshape197_out1
.slice(s![.., .., .., 0..unsqueeze255_out1[0]]);
let add130_out1 = matmul213_out1.add(slice92_out1);
let constant1929_out1 = self.constant1929.val();
let div21_out1 = add130_out1
.div((constant1929_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape133_out1: [i64; 4] = {
let axes = &transpose174_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather113_out1 = shape133_out1[0] as i64;
let constant1931_out1 = -10000f32;
let where53_out1 = div21_out1
.mask_fill(unsqueeze50_out1.clone(), constant1931_out1);
let softmax18_out1 = burn::tensor::activation::softmax(where53_out1, 3);
let constant1932_out1 = 0f32;
let where54_out1 = softmax18_out1.mask_fill(unsqueeze50_out1, constant1932_out1);
let matmul214_out1 = where54_out1.matmul(transpose174_out1);
let transpose180_out1 = matmul214_out1.permute([0, 2, 1, 3]);
let unsqueeze256_out1 = [gather113_out1 as i64];
let constant1935_out1: [i64; 1] = [1024i64];
let constant1934_out1: [i64; 1] = [-1i64];
let concat172_out1: [i64; 3usize] = [
&unsqueeze256_out1[..],
&constant1934_out1[..],
&constant1935_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape198_out1 = transpose180_out1.reshape(concat172_out1);
let linear161_out1 = self.linear161.forward(reshape198_out1);
let add131_out1 = add127_out1.add(linear161_out1);
add131_out1
}
}
#[derive(Module, Debug)]
pub struct Submodule20 {
layernormalization88: LayerNorm,
conv1d52: Conv1d,
conv1d53: Conv1d,
conv1d54: Conv1d,
layernormalization89: LayerNorm,
linear162: Linear,
linear163: Linear,
constant1945: burn::module::Param<Tensor<1>>,
layernormalization90: LayerNorm,
layernormalization91: LayerNorm,
linear164: Linear,
linear165: Linear,
constant1946: burn::module::Param<Tensor<1>>,
layernormalization92: LayerNorm,
linear166: Linear,
linear167: Linear,
linear168: Linear,
linear169: Linear,
constant272: burn::module::Param<Tensor<2>>,
constant273: burn::module::Param<Tensor<2>>,
constant1992: burn::module::Param<Tensor<1>>,
linear170: Linear,
#[module(skip)]
device: Device,
}
impl Submodule20 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization88 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d52 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d53 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d54 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization89 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear162 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear163 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1945: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization90 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization91 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear164 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear165 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant1946: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization92 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear166 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear167 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear168 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear169 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant272: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant273: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant1992: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear170 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization88,
conv1d52,
conv1d53,
conv1d54,
layernormalization89,
linear162,
linear163,
constant1945,
layernormalization90,
layernormalization91,
linear164,
linear165,
constant1946,
layernormalization92,
linear166,
linear167,
linear168,
linear169,
constant272,
constant273,
constant1992,
linear170,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add131_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> Tensor<3> {
let layernormalization88_out1 = {
let dtype = add131_out1.clone().dtype();
self.layernormalization88
.forward(add131_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose181_out1 = layernormalization88_out1.permute([0, 2, 1]);
let conv1d52_out1 = self.conv1d52.forward(transpose181_out1);
let split_tensors = conv1d52_out1.split_with_sizes([1024, 1024].into(), 1);
let [split18_out1, split18_out2] = split_tensors.try_into().unwrap();
let sigmoid70_out1 = burn::tensor::activation::sigmoid(split18_out2);
let mul107_out1 = split18_out1.mul(sigmoid70_out1);
let constant1936_out1 = 0f32;
let where55_out1 = mul107_out1.mask_fill(unsqueeze52_out1, constant1936_out1);
let pad36_out1 = where55_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d53_out1 = self.conv1d53.forward(pad36_out1);
let sigmoid71_out1 = burn::tensor::activation::sigmoid(conv1d53_out1.clone());
let mul108_out1 = conv1d53_out1.mul(sigmoid71_out1);
let conv1d54_out1 = self.conv1d54.forward(mul108_out1);
let transpose183_out1 = conv1d54_out1.permute([0, 2, 1]);
let add132_out1 = add131_out1.add(transpose183_out1);
let layernormalization89_out1 = {
let dtype = add132_out1.clone().dtype();
self.layernormalization89
.forward(add132_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear162_out1 = self.linear162.forward(layernormalization89_out1);
let sigmoid72_out1 = burn::tensor::activation::sigmoid(linear162_out1.clone());
let mul109_out1 = linear162_out1.mul(sigmoid72_out1);
let linear163_out1 = self.linear163.forward(mul109_out1);
let constant1945_out1 = self.constant1945.val();
let mul110_out1 = linear163_out1
.mul((constant1945_out1).unsqueeze_dims(&[0isize, 1isize]));
let add133_out1 = add132_out1.add(mul110_out1);
let layernormalization90_out1 = {
let dtype = add133_out1.dtype();
self.layernormalization90
.forward(add133_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization91_out1 = {
let dtype = layernormalization90_out1.clone().dtype();
self.layernormalization91
.forward(
layernormalization90_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear164_out1 = self.linear164.forward(layernormalization91_out1);
let sigmoid73_out1 = burn::tensor::activation::sigmoid(linear164_out1.clone());
let mul111_out1 = linear164_out1.mul(sigmoid73_out1);
let linear165_out1 = self.linear165.forward(mul111_out1);
let constant1946_out1 = self.constant1946.val();
let mul112_out1 = linear165_out1
.mul((constant1946_out1).unsqueeze_dims(&[0isize, 1isize]));
let add134_out1 = layernormalization90_out1.add(mul112_out1);
let layernormalization92_out1 = {
let dtype = add134_out1.clone().dtype();
self.layernormalization92
.forward(add134_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape134_out1: [i64; 3] = {
let axes = &layernormalization92_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather114_out1 = shape134_out1[0] as i64;
let linear166_out1 = self.linear166.forward(layernormalization92_out1.clone());
let unsqueeze257_out1 = [gather114_out1 as i64];
let constant1951_out1: [i64; 1] = [128i64];
let constant1949_out1: [i64; 1] = [-1i64];
let constant1950_out1: [i64; 1] = [8i64];
let concat174_out1: [i64; 4usize] = [
&unsqueeze257_out1[..],
&constant1949_out1[..],
&constant1950_out1[..],
&constant1951_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze258_out1 = [gather114_out1 as i64];
let constant1955_out1: [i64; 1] = [128i64];
let constant1953_out1: [i64; 1] = [-1i64];
let constant1954_out1: [i64; 1] = [8i64];
let concat175_out1: [i64; 4usize] = [
&unsqueeze258_out1[..],
&constant1953_out1[..],
&constant1954_out1[..],
&constant1955_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze259_out1 = [gather114_out1 as i64];
let constant1959_out1: [i64; 1] = [128i64];
let constant1957_out1: [i64; 1] = [-1i64];
let constant1958_out1: [i64; 1] = [8i64];
let concat176_out1: [i64; 4usize] = [
&unsqueeze259_out1[..],
&constant1957_out1[..],
&constant1958_out1[..],
&constant1959_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape201_out1 = linear166_out1.reshape(concat174_out1);
let linear167_out1 = self.linear167.forward(layernormalization92_out1.clone());
let reshape202_out1 = linear167_out1.reshape(concat175_out1);
let linear168_out1 = self.linear168.forward(layernormalization92_out1);
let reshape203_out1 = linear168_out1.reshape(concat176_out1);
let transpose184_out1 = reshape203_out1.permute([0, 2, 1, 3]);
let linear169_out1 = self.linear169.forward(slice1_out1);
let reshape204_out1 = linear169_out1.reshape([1, -1, 8, 128]);
let constant272_out1 = self.constant272.val();
let add135_out1 = reshape201_out1
.clone()
.add((constant272_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose185_out1 = add135_out1.permute([0, 2, 1, 3]);
let constant273_out1 = self.constant273.val();
let add136_out1 = reshape201_out1
.add((constant273_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose186_out1 = add136_out1.permute([0, 2, 1, 3]);
let transpose187_out1 = reshape204_out1.permute([0, 2, 3, 1]);
let matmul224_out1 = transpose186_out1.matmul(transpose187_out1);
let shape135_out1: [i64; 4] = {
let axes = &matmul224_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather115_out1 = shape135_out1[0] as i64;
let gather116_out1 = shape135_out1[1] as i64;
let gather117_out1 = shape135_out1[2] as i64;
let gather118_out1 = shape135_out1[3] as i64;
let pad37_out1 = matmul224_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze260_out1 = [gather115_out1 as i64];
let unsqueeze261_out1 = [gather116_out1 as i64];
let unsqueeze262_out1 = [gather117_out1 as i64];
let constant1974_out1: [i64; 1] = [-1i64];
let concat178_out1: [i64; 4usize] = [
&unsqueeze260_out1[..],
&unsqueeze261_out1[..],
&constant1974_out1[..],
&unsqueeze262_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape207_out1 = pad37_out1.reshape(concat178_out1);
let slice95_out1 = reshape207_out1.slice(s![.., .., 1.., ..]);
let unsqueeze263_out1 = [gather115_out1 as i64];
let unsqueeze264_out1 = [gather116_out1 as i64];
let unsqueeze265_out1 = [gather117_out1 as i64];
let unsqueeze266_out1 = [gather118_out1 as i64];
let concat179_out1: [i64; 4usize] = [
&unsqueeze263_out1[..],
&unsqueeze264_out1[..],
&unsqueeze265_out1[..],
&unsqueeze266_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape208_out1 = slice95_out1.reshape(concat179_out1);
let transpose189_out1 = reshape202_out1.permute([0, 2, 3, 1]);
let matmul225_out1 = transpose185_out1.matmul(transpose189_out1);
let shape139_out1: [i64; 4] = {
let axes = &matmul225_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice96_out1: [i64; 1] = shape139_out1[3..4].try_into().unwrap();
let squeeze19_out1 = slice96_out1[0] as i64;
let unsqueeze267_out1 = [squeeze19_out1 as i64];
let slice97_out1 = reshape208_out1
.slice(s![.., .., .., 0..unsqueeze267_out1[0]]);
let add137_out1 = matmul225_out1.add(slice97_out1);
let constant1992_out1 = self.constant1992.val();
let div22_out1 = add137_out1
.div((constant1992_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape140_out1: [i64; 4] = {
let axes = &transpose184_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather119_out1 = shape140_out1[0] as i64;
let constant1994_out1 = -10000f32;
let where56_out1 = div22_out1
.mask_fill(unsqueeze50_out1.clone(), constant1994_out1);
let softmax19_out1 = burn::tensor::activation::softmax(where56_out1, 3);
let constant1995_out1 = 0f32;
let where57_out1 = softmax19_out1.mask_fill(unsqueeze50_out1, constant1995_out1);
let matmul226_out1 = where57_out1.matmul(transpose184_out1);
let transpose190_out1 = matmul226_out1.permute([0, 2, 1, 3]);
let unsqueeze268_out1 = [gather119_out1 as i64];
let constant1998_out1: [i64; 1] = [1024i64];
let constant1997_out1: [i64; 1] = [-1i64];
let concat180_out1: [i64; 3usize] = [
&unsqueeze268_out1[..],
&constant1997_out1[..],
&constant1998_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape209_out1 = transpose190_out1.reshape(concat180_out1);
let linear170_out1 = self.linear170.forward(reshape209_out1);
let add138_out1 = add134_out1.add(linear170_out1);
add138_out1
}
}
#[derive(Module, Debug)]
pub struct Submodule21 {
layernormalization93: LayerNorm,
conv1d55: Conv1d,
conv1d56: Conv1d,
conv1d57: Conv1d,
layernormalization94: LayerNorm,
linear171: Linear,
linear172: Linear,
constant2008: burn::module::Param<Tensor<1>>,
layernormalization95: LayerNorm,
layernormalization96: LayerNorm,
linear173: Linear,
linear174: Linear,
constant2009: burn::module::Param<Tensor<1>>,
layernormalization97: LayerNorm,
linear175: Linear,
linear176: Linear,
linear177: Linear,
linear178: Linear,
constant286: burn::module::Param<Tensor<2>>,
constant287: burn::module::Param<Tensor<2>>,
constant2055: burn::module::Param<Tensor<1>>,
linear179: Linear,
#[module(skip)]
device: Device,
}
impl Submodule21 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization93 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d55 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d56 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d57 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization94 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear171 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear172 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant2008: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization95 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization96 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear173 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear174 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant2009: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization97 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear175 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear176 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear177 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear178 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant286: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant287: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant2055: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear179 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization93,
conv1d55,
conv1d56,
conv1d57,
layernormalization94,
linear171,
linear172,
constant2008,
layernormalization95,
layernormalization96,
linear173,
linear174,
constant2009,
layernormalization97,
linear175,
linear176,
linear177,
linear178,
constant286,
constant287,
constant2055,
linear179,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add138_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> Tensor<3> {
let layernormalization93_out1 = {
let dtype = add138_out1.clone().dtype();
self.layernormalization93
.forward(add138_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose191_out1 = layernormalization93_out1.permute([0, 2, 1]);
let conv1d55_out1 = self.conv1d55.forward(transpose191_out1);
let split_tensors = conv1d55_out1.split_with_sizes([1024, 1024].into(), 1);
let [split19_out1, split19_out2] = split_tensors.try_into().unwrap();
let sigmoid74_out1 = burn::tensor::activation::sigmoid(split19_out2);
let mul113_out1 = split19_out1.mul(sigmoid74_out1);
let constant1999_out1 = 0f32;
let where58_out1 = mul113_out1.mask_fill(unsqueeze52_out1, constant1999_out1);
let pad38_out1 = where58_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d56_out1 = self.conv1d56.forward(pad38_out1);
let sigmoid75_out1 = burn::tensor::activation::sigmoid(conv1d56_out1.clone());
let mul114_out1 = conv1d56_out1.mul(sigmoid75_out1);
let conv1d57_out1 = self.conv1d57.forward(mul114_out1);
let transpose193_out1 = conv1d57_out1.permute([0, 2, 1]);
let add139_out1 = add138_out1.add(transpose193_out1);
let layernormalization94_out1 = {
let dtype = add139_out1.clone().dtype();
self.layernormalization94
.forward(add139_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear171_out1 = self.linear171.forward(layernormalization94_out1);
let sigmoid76_out1 = burn::tensor::activation::sigmoid(linear171_out1.clone());
let mul115_out1 = linear171_out1.mul(sigmoid76_out1);
let linear172_out1 = self.linear172.forward(mul115_out1);
let constant2008_out1 = self.constant2008.val();
let mul116_out1 = linear172_out1
.mul((constant2008_out1).unsqueeze_dims(&[0isize, 1isize]));
let add140_out1 = add139_out1.add(mul116_out1);
let layernormalization95_out1 = {
let dtype = add140_out1.dtype();
self.layernormalization95
.forward(add140_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization96_out1 = {
let dtype = layernormalization95_out1.clone().dtype();
self.layernormalization96
.forward(
layernormalization95_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear173_out1 = self.linear173.forward(layernormalization96_out1);
let sigmoid77_out1 = burn::tensor::activation::sigmoid(linear173_out1.clone());
let mul117_out1 = linear173_out1.mul(sigmoid77_out1);
let linear174_out1 = self.linear174.forward(mul117_out1);
let constant2009_out1 = self.constant2009.val();
let mul118_out1 = linear174_out1
.mul((constant2009_out1).unsqueeze_dims(&[0isize, 1isize]));
let add141_out1 = layernormalization95_out1.add(mul118_out1);
let layernormalization97_out1 = {
let dtype = add141_out1.clone().dtype();
self.layernormalization97
.forward(add141_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape141_out1: [i64; 3] = {
let axes = &layernormalization97_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather120_out1 = shape141_out1[0] as i64;
let linear175_out1 = self.linear175.forward(layernormalization97_out1.clone());
let unsqueeze269_out1 = [gather120_out1 as i64];
let constant2014_out1: [i64; 1] = [128i64];
let constant2012_out1: [i64; 1] = [-1i64];
let constant2013_out1: [i64; 1] = [8i64];
let concat182_out1: [i64; 4usize] = [
&unsqueeze269_out1[..],
&constant2012_out1[..],
&constant2013_out1[..],
&constant2014_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze270_out1 = [gather120_out1 as i64];
let constant2018_out1: [i64; 1] = [128i64];
let constant2016_out1: [i64; 1] = [-1i64];
let constant2017_out1: [i64; 1] = [8i64];
let concat183_out1: [i64; 4usize] = [
&unsqueeze270_out1[..],
&constant2016_out1[..],
&constant2017_out1[..],
&constant2018_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze271_out1 = [gather120_out1 as i64];
let constant2022_out1: [i64; 1] = [128i64];
let constant2020_out1: [i64; 1] = [-1i64];
let constant2021_out1: [i64; 1] = [8i64];
let concat184_out1: [i64; 4usize] = [
&unsqueeze271_out1[..],
&constant2020_out1[..],
&constant2021_out1[..],
&constant2022_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape212_out1 = linear175_out1.reshape(concat182_out1);
let linear176_out1 = self.linear176.forward(layernormalization97_out1.clone());
let reshape213_out1 = linear176_out1.reshape(concat183_out1);
let linear177_out1 = self.linear177.forward(layernormalization97_out1);
let reshape214_out1 = linear177_out1.reshape(concat184_out1);
let transpose194_out1 = reshape214_out1.permute([0, 2, 1, 3]);
let linear178_out1 = self.linear178.forward(slice1_out1);
let reshape215_out1 = linear178_out1.reshape([1, -1, 8, 128]);
let constant286_out1 = self.constant286.val();
let add142_out1 = reshape212_out1
.clone()
.add((constant286_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose195_out1 = add142_out1.permute([0, 2, 1, 3]);
let constant287_out1 = self.constant287.val();
let add143_out1 = reshape212_out1
.add((constant287_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose196_out1 = add143_out1.permute([0, 2, 1, 3]);
let transpose197_out1 = reshape215_out1.permute([0, 2, 3, 1]);
let matmul236_out1 = transpose196_out1.matmul(transpose197_out1);
let shape142_out1: [i64; 4] = {
let axes = &matmul236_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather121_out1 = shape142_out1[0] as i64;
let gather122_out1 = shape142_out1[1] as i64;
let gather123_out1 = shape142_out1[2] as i64;
let gather124_out1 = shape142_out1[3] as i64;
let pad39_out1 = matmul236_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze272_out1 = [gather121_out1 as i64];
let unsqueeze273_out1 = [gather122_out1 as i64];
let unsqueeze274_out1 = [gather123_out1 as i64];
let constant2037_out1: [i64; 1] = [-1i64];
let concat186_out1: [i64; 4usize] = [
&unsqueeze272_out1[..],
&unsqueeze273_out1[..],
&constant2037_out1[..],
&unsqueeze274_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape218_out1 = pad39_out1.reshape(concat186_out1);
let slice100_out1 = reshape218_out1.slice(s![.., .., 1.., ..]);
let unsqueeze275_out1 = [gather121_out1 as i64];
let unsqueeze276_out1 = [gather122_out1 as i64];
let unsqueeze277_out1 = [gather123_out1 as i64];
let unsqueeze278_out1 = [gather124_out1 as i64];
let concat187_out1: [i64; 4usize] = [
&unsqueeze275_out1[..],
&unsqueeze276_out1[..],
&unsqueeze277_out1[..],
&unsqueeze278_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape219_out1 = slice100_out1.reshape(concat187_out1);
let transpose199_out1 = reshape213_out1.permute([0, 2, 3, 1]);
let matmul237_out1 = transpose195_out1.matmul(transpose199_out1);
let shape146_out1: [i64; 4] = {
let axes = &matmul237_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice101_out1: [i64; 1] = shape146_out1[3..4].try_into().unwrap();
let squeeze20_out1 = slice101_out1[0] as i64;
let unsqueeze279_out1 = [squeeze20_out1 as i64];
let slice102_out1 = reshape219_out1
.slice(s![.., .., .., 0..unsqueeze279_out1[0]]);
let add144_out1 = matmul237_out1.add(slice102_out1);
let constant2055_out1 = self.constant2055.val();
let div23_out1 = add144_out1
.div((constant2055_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape147_out1: [i64; 4] = {
let axes = &transpose194_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather125_out1 = shape147_out1[0] as i64;
let constant2057_out1 = -10000f32;
let where59_out1 = div23_out1
.mask_fill(unsqueeze50_out1.clone(), constant2057_out1);
let softmax20_out1 = burn::tensor::activation::softmax(where59_out1, 3);
let constant2058_out1 = 0f32;
let where60_out1 = softmax20_out1.mask_fill(unsqueeze50_out1, constant2058_out1);
let matmul238_out1 = where60_out1.matmul(transpose194_out1);
let transpose200_out1 = matmul238_out1.permute([0, 2, 1, 3]);
let unsqueeze280_out1 = [gather125_out1 as i64];
let constant2061_out1: [i64; 1] = [1024i64];
let constant2060_out1: [i64; 1] = [-1i64];
let concat188_out1: [i64; 3usize] = [
&unsqueeze280_out1[..],
&constant2060_out1[..],
&constant2061_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape220_out1 = transpose200_out1.reshape(concat188_out1);
let linear179_out1 = self.linear179.forward(reshape220_out1);
let add145_out1 = add141_out1.add(linear179_out1);
add145_out1
}
}
#[derive(Module, Debug)]
pub struct Submodule22 {
layernormalization98: LayerNorm,
conv1d58: Conv1d,
conv1d59: Conv1d,
conv1d60: Conv1d,
layernormalization99: LayerNorm,
linear180: Linear,
linear181: Linear,
constant2071: burn::module::Param<Tensor<1>>,
layernormalization100: LayerNorm,
layernormalization101: LayerNorm,
linear182: Linear,
linear183: Linear,
constant2072: burn::module::Param<Tensor<1>>,
layernormalization102: LayerNorm,
linear184: Linear,
linear185: Linear,
linear186: Linear,
linear187: Linear,
constant300: burn::module::Param<Tensor<2>>,
constant301: burn::module::Param<Tensor<2>>,
constant2118: burn::module::Param<Tensor<1>>,
linear188: Linear,
#[module(skip)]
device: Device,
}
impl Submodule22 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization98 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d58 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d59 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d60 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization99 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear180 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear181 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant2071: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization100 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization101 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear182 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear183 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant2072: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization102 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear184 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear185 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear186 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear187 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant300: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant301: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant2118: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear188 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization98,
conv1d58,
conv1d59,
conv1d60,
layernormalization99,
linear180,
linear181,
constant2071,
layernormalization100,
layernormalization101,
linear182,
linear183,
constant2072,
layernormalization102,
linear184,
linear185,
linear186,
linear187,
constant300,
constant301,
constant2118,
linear188,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add145_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> Tensor<3> {
let layernormalization98_out1 = {
let dtype = add145_out1.clone().dtype();
self.layernormalization98
.forward(add145_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose201_out1 = layernormalization98_out1.permute([0, 2, 1]);
let conv1d58_out1 = self.conv1d58.forward(transpose201_out1);
let split_tensors = conv1d58_out1.split_with_sizes([1024, 1024].into(), 1);
let [split20_out1, split20_out2] = split_tensors.try_into().unwrap();
let sigmoid78_out1 = burn::tensor::activation::sigmoid(split20_out2);
let mul119_out1 = split20_out1.mul(sigmoid78_out1);
let constant2062_out1 = 0f32;
let where61_out1 = mul119_out1.mask_fill(unsqueeze52_out1, constant2062_out1);
let pad40_out1 = where61_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d59_out1 = self.conv1d59.forward(pad40_out1);
let sigmoid79_out1 = burn::tensor::activation::sigmoid(conv1d59_out1.clone());
let mul120_out1 = conv1d59_out1.mul(sigmoid79_out1);
let conv1d60_out1 = self.conv1d60.forward(mul120_out1);
let transpose203_out1 = conv1d60_out1.permute([0, 2, 1]);
let add146_out1 = add145_out1.add(transpose203_out1);
let layernormalization99_out1 = {
let dtype = add146_out1.clone().dtype();
self.layernormalization99
.forward(add146_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear180_out1 = self.linear180.forward(layernormalization99_out1);
let sigmoid80_out1 = burn::tensor::activation::sigmoid(linear180_out1.clone());
let mul121_out1 = linear180_out1.mul(sigmoid80_out1);
let linear181_out1 = self.linear181.forward(mul121_out1);
let constant2071_out1 = self.constant2071.val();
let mul122_out1 = linear181_out1
.mul((constant2071_out1).unsqueeze_dims(&[0isize, 1isize]));
let add147_out1 = add146_out1.add(mul122_out1);
let layernormalization100_out1 = {
let dtype = add147_out1.dtype();
self.layernormalization100
.forward(add147_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization101_out1 = {
let dtype = layernormalization100_out1.clone().dtype();
self.layernormalization101
.forward(
layernormalization100_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear182_out1 = self.linear182.forward(layernormalization101_out1);
let sigmoid81_out1 = burn::tensor::activation::sigmoid(linear182_out1.clone());
let mul123_out1 = linear182_out1.mul(sigmoid81_out1);
let linear183_out1 = self.linear183.forward(mul123_out1);
let constant2072_out1 = self.constant2072.val();
let mul124_out1 = linear183_out1
.mul((constant2072_out1).unsqueeze_dims(&[0isize, 1isize]));
let add148_out1 = layernormalization100_out1.add(mul124_out1);
let layernormalization102_out1 = {
let dtype = add148_out1.clone().dtype();
self.layernormalization102
.forward(add148_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape148_out1: [i64; 3] = {
let axes = &layernormalization102_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather126_out1 = shape148_out1[0] as i64;
let linear184_out1 = self.linear184.forward(layernormalization102_out1.clone());
let unsqueeze281_out1 = [gather126_out1 as i64];
let constant2077_out1: [i64; 1] = [128i64];
let constant2075_out1: [i64; 1] = [-1i64];
let constant2076_out1: [i64; 1] = [8i64];
let concat190_out1: [i64; 4usize] = [
&unsqueeze281_out1[..],
&constant2075_out1[..],
&constant2076_out1[..],
&constant2077_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze282_out1 = [gather126_out1 as i64];
let constant2081_out1: [i64; 1] = [128i64];
let constant2079_out1: [i64; 1] = [-1i64];
let constant2080_out1: [i64; 1] = [8i64];
let concat191_out1: [i64; 4usize] = [
&unsqueeze282_out1[..],
&constant2079_out1[..],
&constant2080_out1[..],
&constant2081_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze283_out1 = [gather126_out1 as i64];
let constant2085_out1: [i64; 1] = [128i64];
let constant2083_out1: [i64; 1] = [-1i64];
let constant2084_out1: [i64; 1] = [8i64];
let concat192_out1: [i64; 4usize] = [
&unsqueeze283_out1[..],
&constant2083_out1[..],
&constant2084_out1[..],
&constant2085_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape223_out1 = linear184_out1.reshape(concat190_out1);
let linear185_out1 = self.linear185.forward(layernormalization102_out1.clone());
let reshape224_out1 = linear185_out1.reshape(concat191_out1);
let linear186_out1 = self.linear186.forward(layernormalization102_out1);
let reshape225_out1 = linear186_out1.reshape(concat192_out1);
let transpose204_out1 = reshape225_out1.permute([0, 2, 1, 3]);
let linear187_out1 = self.linear187.forward(slice1_out1);
let reshape226_out1 = linear187_out1.reshape([1, -1, 8, 128]);
let constant300_out1 = self.constant300.val();
let add149_out1 = reshape223_out1
.clone()
.add((constant300_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose205_out1 = add149_out1.permute([0, 2, 1, 3]);
let constant301_out1 = self.constant301.val();
let add150_out1 = reshape223_out1
.add((constant301_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose206_out1 = add150_out1.permute([0, 2, 1, 3]);
let transpose207_out1 = reshape226_out1.permute([0, 2, 3, 1]);
let matmul248_out1 = transpose206_out1.matmul(transpose207_out1);
let shape149_out1: [i64; 4] = {
let axes = &matmul248_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather127_out1 = shape149_out1[0] as i64;
let gather128_out1 = shape149_out1[1] as i64;
let gather129_out1 = shape149_out1[2] as i64;
let gather130_out1 = shape149_out1[3] as i64;
let pad41_out1 = matmul248_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze284_out1 = [gather127_out1 as i64];
let unsqueeze285_out1 = [gather128_out1 as i64];
let unsqueeze286_out1 = [gather129_out1 as i64];
let constant2100_out1: [i64; 1] = [-1i64];
let concat194_out1: [i64; 4usize] = [
&unsqueeze284_out1[..],
&unsqueeze285_out1[..],
&constant2100_out1[..],
&unsqueeze286_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape229_out1 = pad41_out1.reshape(concat194_out1);
let slice105_out1 = reshape229_out1.slice(s![.., .., 1.., ..]);
let unsqueeze287_out1 = [gather127_out1 as i64];
let unsqueeze288_out1 = [gather128_out1 as i64];
let unsqueeze289_out1 = [gather129_out1 as i64];
let unsqueeze290_out1 = [gather130_out1 as i64];
let concat195_out1: [i64; 4usize] = [
&unsqueeze287_out1[..],
&unsqueeze288_out1[..],
&unsqueeze289_out1[..],
&unsqueeze290_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape230_out1 = slice105_out1.reshape(concat195_out1);
let transpose209_out1 = reshape224_out1.permute([0, 2, 3, 1]);
let matmul249_out1 = transpose205_out1.matmul(transpose209_out1);
let shape153_out1: [i64; 4] = {
let axes = &matmul249_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice106_out1: [i64; 1] = shape153_out1[3..4].try_into().unwrap();
let squeeze21_out1 = slice106_out1[0] as i64;
let unsqueeze291_out1 = [squeeze21_out1 as i64];
let slice107_out1 = reshape230_out1
.slice(s![.., .., .., 0..unsqueeze291_out1[0]]);
let add151_out1 = matmul249_out1.add(slice107_out1);
let constant2118_out1 = self.constant2118.val();
let div24_out1 = add151_out1
.div((constant2118_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape154_out1: [i64; 4] = {
let axes = &transpose204_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather131_out1 = shape154_out1[0] as i64;
let constant2120_out1 = -10000f32;
let where62_out1 = div24_out1
.mask_fill(unsqueeze50_out1.clone(), constant2120_out1);
let softmax21_out1 = burn::tensor::activation::softmax(where62_out1, 3);
let constant2121_out1 = 0f32;
let where63_out1 = softmax21_out1.mask_fill(unsqueeze50_out1, constant2121_out1);
let matmul250_out1 = where63_out1.matmul(transpose204_out1);
let transpose210_out1 = matmul250_out1.permute([0, 2, 1, 3]);
let unsqueeze292_out1 = [gather131_out1 as i64];
let constant2124_out1: [i64; 1] = [1024i64];
let constant2123_out1: [i64; 1] = [-1i64];
let concat196_out1: [i64; 3usize] = [
&unsqueeze292_out1[..],
&constant2123_out1[..],
&constant2124_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape231_out1 = transpose210_out1.reshape(concat196_out1);
let linear188_out1 = self.linear188.forward(reshape231_out1);
let add152_out1 = add148_out1.add(linear188_out1);
add152_out1
}
}
#[derive(Module, Debug)]
pub struct Submodule23 {
layernormalization103: LayerNorm,
conv1d61: Conv1d,
conv1d62: Conv1d,
conv1d63: Conv1d,
layernormalization104: LayerNorm,
linear189: Linear,
linear190: Linear,
constant2134: burn::module::Param<Tensor<1>>,
layernormalization105: LayerNorm,
layernormalization106: LayerNorm,
linear191: Linear,
linear192: Linear,
constant2135: burn::module::Param<Tensor<1>>,
layernormalization107: LayerNorm,
linear193: Linear,
linear194: Linear,
linear195: Linear,
linear196: Linear,
constant314: burn::module::Param<Tensor<2>>,
constant315: burn::module::Param<Tensor<2>>,
constant2181: burn::module::Param<Tensor<1>>,
linear197: Linear,
#[module(skip)]
device: Device,
}
impl Submodule23 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization103 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d61 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d62 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d63 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization104 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear189 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear190 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant2134: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization105 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization106 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear191 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear192 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant2135: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization107 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear193 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear194 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear195 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear196 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant314: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant315: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant2181: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear197 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization103,
conv1d61,
conv1d62,
conv1d63,
layernormalization104,
linear189,
linear190,
constant2134,
layernormalization105,
layernormalization106,
linear191,
linear192,
constant2135,
layernormalization107,
linear193,
linear194,
linear195,
linear196,
constant314,
constant315,
constant2181,
linear197,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add152_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> Tensor<3> {
let layernormalization103_out1 = {
let dtype = add152_out1.clone().dtype();
self.layernormalization103
.forward(add152_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose211_out1 = layernormalization103_out1.permute([0, 2, 1]);
let conv1d61_out1 = self.conv1d61.forward(transpose211_out1);
let split_tensors = conv1d61_out1.split_with_sizes([1024, 1024].into(), 1);
let [split21_out1, split21_out2] = split_tensors.try_into().unwrap();
let sigmoid82_out1 = burn::tensor::activation::sigmoid(split21_out2);
let mul125_out1 = split21_out1.mul(sigmoid82_out1);
let constant2125_out1 = 0f32;
let where64_out1 = mul125_out1.mask_fill(unsqueeze52_out1, constant2125_out1);
let pad42_out1 = where64_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d62_out1 = self.conv1d62.forward(pad42_out1);
let sigmoid83_out1 = burn::tensor::activation::sigmoid(conv1d62_out1.clone());
let mul126_out1 = conv1d62_out1.mul(sigmoid83_out1);
let conv1d63_out1 = self.conv1d63.forward(mul126_out1);
let transpose213_out1 = conv1d63_out1.permute([0, 2, 1]);
let add153_out1 = add152_out1.add(transpose213_out1);
let layernormalization104_out1 = {
let dtype = add153_out1.clone().dtype();
self.layernormalization104
.forward(add153_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear189_out1 = self.linear189.forward(layernormalization104_out1);
let sigmoid84_out1 = burn::tensor::activation::sigmoid(linear189_out1.clone());
let mul127_out1 = linear189_out1.mul(sigmoid84_out1);
let linear190_out1 = self.linear190.forward(mul127_out1);
let constant2134_out1 = self.constant2134.val();
let mul128_out1 = linear190_out1
.mul((constant2134_out1).unsqueeze_dims(&[0isize, 1isize]));
let add154_out1 = add153_out1.add(mul128_out1);
let layernormalization105_out1 = {
let dtype = add154_out1.dtype();
self.layernormalization105
.forward(add154_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization106_out1 = {
let dtype = layernormalization105_out1.clone().dtype();
self.layernormalization106
.forward(
layernormalization105_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear191_out1 = self.linear191.forward(layernormalization106_out1);
let sigmoid85_out1 = burn::tensor::activation::sigmoid(linear191_out1.clone());
let mul129_out1 = linear191_out1.mul(sigmoid85_out1);
let linear192_out1 = self.linear192.forward(mul129_out1);
let constant2135_out1 = self.constant2135.val();
let mul130_out1 = linear192_out1
.mul((constant2135_out1).unsqueeze_dims(&[0isize, 1isize]));
let add155_out1 = layernormalization105_out1.add(mul130_out1);
let layernormalization107_out1 = {
let dtype = add155_out1.clone().dtype();
self.layernormalization107
.forward(add155_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape155_out1: [i64; 3] = {
let axes = &layernormalization107_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather132_out1 = shape155_out1[0] as i64;
let linear193_out1 = self.linear193.forward(layernormalization107_out1.clone());
let unsqueeze293_out1 = [gather132_out1 as i64];
let constant2140_out1: [i64; 1] = [128i64];
let constant2138_out1: [i64; 1] = [-1i64];
let constant2139_out1: [i64; 1] = [8i64];
let concat198_out1: [i64; 4usize] = [
&unsqueeze293_out1[..],
&constant2138_out1[..],
&constant2139_out1[..],
&constant2140_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze294_out1 = [gather132_out1 as i64];
let constant2144_out1: [i64; 1] = [128i64];
let constant2142_out1: [i64; 1] = [-1i64];
let constant2143_out1: [i64; 1] = [8i64];
let concat199_out1: [i64; 4usize] = [
&unsqueeze294_out1[..],
&constant2142_out1[..],
&constant2143_out1[..],
&constant2144_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze295_out1 = [gather132_out1 as i64];
let constant2148_out1: [i64; 1] = [128i64];
let constant2146_out1: [i64; 1] = [-1i64];
let constant2147_out1: [i64; 1] = [8i64];
let concat200_out1: [i64; 4usize] = [
&unsqueeze295_out1[..],
&constant2146_out1[..],
&constant2147_out1[..],
&constant2148_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape234_out1 = linear193_out1.reshape(concat198_out1);
let linear194_out1 = self.linear194.forward(layernormalization107_out1.clone());
let reshape235_out1 = linear194_out1.reshape(concat199_out1);
let linear195_out1 = self.linear195.forward(layernormalization107_out1);
let reshape236_out1 = linear195_out1.reshape(concat200_out1);
let transpose214_out1 = reshape236_out1.permute([0, 2, 1, 3]);
let linear196_out1 = self.linear196.forward(slice1_out1);
let reshape237_out1 = linear196_out1.reshape([1, -1, 8, 128]);
let constant314_out1 = self.constant314.val();
let add156_out1 = reshape234_out1
.clone()
.add((constant314_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose215_out1 = add156_out1.permute([0, 2, 1, 3]);
let constant315_out1 = self.constant315.val();
let add157_out1 = reshape234_out1
.add((constant315_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose216_out1 = add157_out1.permute([0, 2, 1, 3]);
let transpose217_out1 = reshape237_out1.permute([0, 2, 3, 1]);
let matmul260_out1 = transpose216_out1.matmul(transpose217_out1);
let shape156_out1: [i64; 4] = {
let axes = &matmul260_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather133_out1 = shape156_out1[0] as i64;
let gather134_out1 = shape156_out1[1] as i64;
let gather135_out1 = shape156_out1[2] as i64;
let gather136_out1 = shape156_out1[3] as i64;
let pad43_out1 = matmul260_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze296_out1 = [gather133_out1 as i64];
let unsqueeze297_out1 = [gather134_out1 as i64];
let unsqueeze298_out1 = [gather135_out1 as i64];
let constant2163_out1: [i64; 1] = [-1i64];
let concat202_out1: [i64; 4usize] = [
&unsqueeze296_out1[..],
&unsqueeze297_out1[..],
&constant2163_out1[..],
&unsqueeze298_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape240_out1 = pad43_out1.reshape(concat202_out1);
let slice110_out1 = reshape240_out1.slice(s![.., .., 1.., ..]);
let unsqueeze299_out1 = [gather133_out1 as i64];
let unsqueeze300_out1 = [gather134_out1 as i64];
let unsqueeze301_out1 = [gather135_out1 as i64];
let unsqueeze302_out1 = [gather136_out1 as i64];
let concat203_out1: [i64; 4usize] = [
&unsqueeze299_out1[..],
&unsqueeze300_out1[..],
&unsqueeze301_out1[..],
&unsqueeze302_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape241_out1 = slice110_out1.reshape(concat203_out1);
let transpose219_out1 = reshape235_out1.permute([0, 2, 3, 1]);
let matmul261_out1 = transpose215_out1.matmul(transpose219_out1);
let shape160_out1: [i64; 4] = {
let axes = &matmul261_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice111_out1: [i64; 1] = shape160_out1[3..4].try_into().unwrap();
let squeeze22_out1 = slice111_out1[0] as i64;
let unsqueeze303_out1 = [squeeze22_out1 as i64];
let slice112_out1 = reshape241_out1
.slice(s![.., .., .., 0..unsqueeze303_out1[0]]);
let add158_out1 = matmul261_out1.add(slice112_out1);
let constant2181_out1 = self.constant2181.val();
let div25_out1 = add158_out1
.div((constant2181_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape161_out1: [i64; 4] = {
let axes = &transpose214_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather137_out1 = shape161_out1[0] as i64;
let constant2183_out1 = -10000f32;
let where65_out1 = div25_out1
.mask_fill(unsqueeze50_out1.clone(), constant2183_out1);
let softmax22_out1 = burn::tensor::activation::softmax(where65_out1, 3);
let constant2184_out1 = 0f32;
let where66_out1 = softmax22_out1.mask_fill(unsqueeze50_out1, constant2184_out1);
let matmul262_out1 = where66_out1.matmul(transpose214_out1);
let transpose220_out1 = matmul262_out1.permute([0, 2, 1, 3]);
let unsqueeze304_out1 = [gather137_out1 as i64];
let constant2187_out1: [i64; 1] = [1024i64];
let constant2186_out1: [i64; 1] = [-1i64];
let concat204_out1: [i64; 3usize] = [
&unsqueeze304_out1[..],
&constant2186_out1[..],
&constant2187_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape242_out1 = transpose220_out1.reshape(concat204_out1);
let linear197_out1 = self.linear197.forward(reshape242_out1);
let add159_out1 = add155_out1.add(linear197_out1);
add159_out1
}
}
#[derive(Module, Debug)]
pub struct Submodule24 {
layernormalization108: LayerNorm,
conv1d64: Conv1d,
conv1d65: Conv1d,
conv1d66: Conv1d,
layernormalization109: LayerNorm,
linear198: Linear,
linear199: Linear,
constant2197: burn::module::Param<Tensor<1>>,
layernormalization110: LayerNorm,
layernormalization111: LayerNorm,
linear200: Linear,
linear201: Linear,
constant2198: burn::module::Param<Tensor<1>>,
layernormalization112: LayerNorm,
linear202: Linear,
linear203: Linear,
linear204: Linear,
linear205: Linear,
constant328: burn::module::Param<Tensor<2>>,
constant329: burn::module::Param<Tensor<2>>,
constant2244: burn::module::Param<Tensor<1>>,
linear206: Linear,
#[module(skip)]
device: Device,
}
impl Submodule24 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization108 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d64 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d65 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d66 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization109 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear198 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear199 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant2197: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization110 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization111 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear200 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear201 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant2198: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization112 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear202 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear203 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear204 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear205 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant328: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant329: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant2244: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear206 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
Self {
layernormalization108,
conv1d64,
conv1d65,
conv1d66,
layernormalization109,
linear198,
linear199,
constant2197,
layernormalization110,
layernormalization111,
linear200,
linear201,
constant2198,
layernormalization112,
linear202,
linear203,
linear204,
linear205,
constant328,
constant329,
constant2244,
linear206,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add159_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
unsqueeze50_out1: Tensor<4, Bool>,
) -> Tensor<3> {
let layernormalization108_out1 = {
let dtype = add159_out1.clone().dtype();
self.layernormalization108
.forward(add159_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose221_out1 = layernormalization108_out1.permute([0, 2, 1]);
let conv1d64_out1 = self.conv1d64.forward(transpose221_out1);
let split_tensors = conv1d64_out1.split_with_sizes([1024, 1024].into(), 1);
let [split22_out1, split22_out2] = split_tensors.try_into().unwrap();
let sigmoid86_out1 = burn::tensor::activation::sigmoid(split22_out2);
let mul131_out1 = split22_out1.mul(sigmoid86_out1);
let constant2188_out1 = 0f32;
let where67_out1 = mul131_out1.mask_fill(unsqueeze52_out1, constant2188_out1);
let pad44_out1 = where67_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d65_out1 = self.conv1d65.forward(pad44_out1);
let sigmoid87_out1 = burn::tensor::activation::sigmoid(conv1d65_out1.clone());
let mul132_out1 = conv1d65_out1.mul(sigmoid87_out1);
let conv1d66_out1 = self.conv1d66.forward(mul132_out1);
let transpose223_out1 = conv1d66_out1.permute([0, 2, 1]);
let add160_out1 = add159_out1.add(transpose223_out1);
let layernormalization109_out1 = {
let dtype = add160_out1.clone().dtype();
self.layernormalization109
.forward(add160_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear198_out1 = self.linear198.forward(layernormalization109_out1);
let sigmoid88_out1 = burn::tensor::activation::sigmoid(linear198_out1.clone());
let mul133_out1 = linear198_out1.mul(sigmoid88_out1);
let linear199_out1 = self.linear199.forward(mul133_out1);
let constant2197_out1 = self.constant2197.val();
let mul134_out1 = linear199_out1
.mul((constant2197_out1).unsqueeze_dims(&[0isize, 1isize]));
let add161_out1 = add160_out1.add(mul134_out1);
let layernormalization110_out1 = {
let dtype = add161_out1.dtype();
self.layernormalization110
.forward(add161_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization111_out1 = {
let dtype = layernormalization110_out1.clone().dtype();
self.layernormalization111
.forward(
layernormalization110_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear200_out1 = self.linear200.forward(layernormalization111_out1);
let sigmoid89_out1 = burn::tensor::activation::sigmoid(linear200_out1.clone());
let mul135_out1 = linear200_out1.mul(sigmoid89_out1);
let linear201_out1 = self.linear201.forward(mul135_out1);
let constant2198_out1 = self.constant2198.val();
let mul136_out1 = linear201_out1
.mul((constant2198_out1).unsqueeze_dims(&[0isize, 1isize]));
let add162_out1 = layernormalization110_out1.add(mul136_out1);
let layernormalization112_out1 = {
let dtype = add162_out1.clone().dtype();
self.layernormalization112
.forward(add162_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape162_out1: [i64; 3] = {
let axes = &layernormalization112_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather138_out1 = shape162_out1[0] as i64;
let linear202_out1 = self.linear202.forward(layernormalization112_out1.clone());
let unsqueeze305_out1 = [gather138_out1 as i64];
let constant2203_out1: [i64; 1] = [128i64];
let constant2201_out1: [i64; 1] = [-1i64];
let constant2202_out1: [i64; 1] = [8i64];
let concat206_out1: [i64; 4usize] = [
&unsqueeze305_out1[..],
&constant2201_out1[..],
&constant2202_out1[..],
&constant2203_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze306_out1 = [gather138_out1 as i64];
let constant2207_out1: [i64; 1] = [128i64];
let constant2205_out1: [i64; 1] = [-1i64];
let constant2206_out1: [i64; 1] = [8i64];
let concat207_out1: [i64; 4usize] = [
&unsqueeze306_out1[..],
&constant2205_out1[..],
&constant2206_out1[..],
&constant2207_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze307_out1 = [gather138_out1 as i64];
let constant2211_out1: [i64; 1] = [128i64];
let constant2209_out1: [i64; 1] = [-1i64];
let constant2210_out1: [i64; 1] = [8i64];
let concat208_out1: [i64; 4usize] = [
&unsqueeze307_out1[..],
&constant2209_out1[..],
&constant2210_out1[..],
&constant2211_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape245_out1 = linear202_out1.reshape(concat206_out1);
let linear203_out1 = self.linear203.forward(layernormalization112_out1.clone());
let reshape246_out1 = linear203_out1.reshape(concat207_out1);
let linear204_out1 = self.linear204.forward(layernormalization112_out1);
let reshape247_out1 = linear204_out1.reshape(concat208_out1);
let transpose224_out1 = reshape247_out1.permute([0, 2, 1, 3]);
let linear205_out1 = self.linear205.forward(slice1_out1);
let reshape248_out1 = linear205_out1.reshape([1, -1, 8, 128]);
let constant328_out1 = self.constant328.val();
let add163_out1 = reshape245_out1
.clone()
.add((constant328_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose225_out1 = add163_out1.permute([0, 2, 1, 3]);
let constant329_out1 = self.constant329.val();
let add164_out1 = reshape245_out1
.add((constant329_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose226_out1 = add164_out1.permute([0, 2, 1, 3]);
let transpose227_out1 = reshape248_out1.permute([0, 2, 3, 1]);
let matmul272_out1 = transpose226_out1.matmul(transpose227_out1);
let shape163_out1: [i64; 4] = {
let axes = &matmul272_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather139_out1 = shape163_out1[0] as i64;
let gather140_out1 = shape163_out1[1] as i64;
let gather141_out1 = shape163_out1[2] as i64;
let gather142_out1 = shape163_out1[3] as i64;
let pad45_out1 = matmul272_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze308_out1 = [gather139_out1 as i64];
let unsqueeze309_out1 = [gather140_out1 as i64];
let unsqueeze310_out1 = [gather141_out1 as i64];
let constant2226_out1: [i64; 1] = [-1i64];
let concat210_out1: [i64; 4usize] = [
&unsqueeze308_out1[..],
&unsqueeze309_out1[..],
&constant2226_out1[..],
&unsqueeze310_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape251_out1 = pad45_out1.reshape(concat210_out1);
let slice115_out1 = reshape251_out1.slice(s![.., .., 1.., ..]);
let unsqueeze311_out1 = [gather139_out1 as i64];
let unsqueeze312_out1 = [gather140_out1 as i64];
let unsqueeze313_out1 = [gather141_out1 as i64];
let unsqueeze314_out1 = [gather142_out1 as i64];
let concat211_out1: [i64; 4usize] = [
&unsqueeze311_out1[..],
&unsqueeze312_out1[..],
&unsqueeze313_out1[..],
&unsqueeze314_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape252_out1 = slice115_out1.reshape(concat211_out1);
let transpose229_out1 = reshape246_out1.permute([0, 2, 3, 1]);
let matmul273_out1 = transpose225_out1.matmul(transpose229_out1);
let shape167_out1: [i64; 4] = {
let axes = &matmul273_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice116_out1: [i64; 1] = shape167_out1[3..4].try_into().unwrap();
let squeeze23_out1 = slice116_out1[0] as i64;
let unsqueeze315_out1 = [squeeze23_out1 as i64];
let slice117_out1 = reshape252_out1
.slice(s![.., .., .., 0..unsqueeze315_out1[0]]);
let add165_out1 = matmul273_out1.add(slice117_out1);
let constant2244_out1 = self.constant2244.val();
let div26_out1 = add165_out1
.div((constant2244_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape168_out1: [i64; 4] = {
let axes = &transpose224_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather143_out1 = shape168_out1[0] as i64;
let constant2246_out1 = -10000f32;
let where68_out1 = div26_out1
.mask_fill(unsqueeze50_out1.clone(), constant2246_out1);
let softmax23_out1 = burn::tensor::activation::softmax(where68_out1, 3);
let constant2247_out1 = 0f32;
let where69_out1 = softmax23_out1.mask_fill(unsqueeze50_out1, constant2247_out1);
let matmul274_out1 = where69_out1.matmul(transpose224_out1);
let transpose230_out1 = matmul274_out1.permute([0, 2, 1, 3]);
let unsqueeze316_out1 = [gather143_out1 as i64];
let constant2250_out1: [i64; 1] = [1024i64];
let constant2249_out1: [i64; 1] = [-1i64];
let concat212_out1: [i64; 3usize] = [
&unsqueeze316_out1[..],
&constant2249_out1[..],
&constant2250_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape253_out1 = transpose230_out1.reshape(concat212_out1);
let linear206_out1 = self.linear206.forward(reshape253_out1);
let add166_out1 = add162_out1.add(linear206_out1);
add166_out1
}
}
#[derive(Module, Debug)]
pub struct Submodule25 {
layernormalization113: LayerNorm,
conv1d67: Conv1d,
conv1d68: Conv1d,
conv1d69: Conv1d,
layernormalization114: LayerNorm,
linear207: Linear,
linear208: Linear,
constant2260: burn::module::Param<Tensor<1>>,
layernormalization115: LayerNorm,
layernormalization116: LayerNorm,
linear209: Linear,
linear210: Linear,
constant2261: burn::module::Param<Tensor<1>>,
layernormalization117: LayerNorm,
linear211: Linear,
linear212: Linear,
linear213: Linear,
linear214: Linear,
constant342: burn::module::Param<Tensor<2>>,
constant343: burn::module::Param<Tensor<2>>,
#[module(skip)]
device: Device,
}
impl Submodule25 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let layernormalization113 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d67 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d68 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d69 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization114 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear207 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear208 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant2260: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization115 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let layernormalization116 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear209 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear210 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant2261: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization117 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear211 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear212 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear213 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let linear214 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let constant342: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
let constant343: burn::module::Param<Tensor<2>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
2,
>::zeros([8, 128], (device, burn::tensor::DType::F32)),
device.clone(),
false,
[8, 128].into(),
);
Self {
layernormalization113,
conv1d67,
conv1d68,
conv1d69,
layernormalization114,
linear207,
linear208,
constant2260,
layernormalization115,
layernormalization116,
linear209,
linear210,
constant2261,
layernormalization117,
linear211,
linear212,
linear213,
linear214,
constant342,
constant343,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
add166_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
slice1_out1: Tensor<3>,
) -> (Tensor<4>, Tensor<4>, Tensor<4>, Tensor<4>, Tensor<3>) {
let layernormalization113_out1 = {
let dtype = add166_out1.clone().dtype();
self.layernormalization113
.forward(add166_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose231_out1 = layernormalization113_out1.permute([0, 2, 1]);
let conv1d67_out1 = self.conv1d67.forward(transpose231_out1);
let split_tensors = conv1d67_out1.split_with_sizes([1024, 1024].into(), 1);
let [split23_out1, split23_out2] = split_tensors.try_into().unwrap();
let sigmoid90_out1 = burn::tensor::activation::sigmoid(split23_out2);
let mul137_out1 = split23_out1.mul(sigmoid90_out1);
let constant2251_out1 = 0f32;
let where70_out1 = mul137_out1.mask_fill(unsqueeze52_out1, constant2251_out1);
let pad46_out1 = where70_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d68_out1 = self.conv1d68.forward(pad46_out1);
let sigmoid91_out1 = burn::tensor::activation::sigmoid(conv1d68_out1.clone());
let mul138_out1 = conv1d68_out1.mul(sigmoid91_out1);
let conv1d69_out1 = self.conv1d69.forward(mul138_out1);
let transpose233_out1 = conv1d69_out1.permute([0, 2, 1]);
let add167_out1 = add166_out1.add(transpose233_out1);
let layernormalization114_out1 = {
let dtype = add167_out1.clone().dtype();
self.layernormalization114
.forward(add167_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear207_out1 = self.linear207.forward(layernormalization114_out1);
let sigmoid92_out1 = burn::tensor::activation::sigmoid(linear207_out1.clone());
let mul139_out1 = linear207_out1.mul(sigmoid92_out1);
let linear208_out1 = self.linear208.forward(mul139_out1);
let constant2260_out1 = self.constant2260.val();
let mul140_out1 = linear208_out1
.mul((constant2260_out1).unsqueeze_dims(&[0isize, 1isize]));
let add168_out1 = add167_out1.add(mul140_out1);
let layernormalization115_out1 = {
let dtype = add168_out1.dtype();
self.layernormalization115
.forward(add168_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let layernormalization116_out1 = {
let dtype = layernormalization115_out1.clone().dtype();
self.layernormalization116
.forward(
layernormalization115_out1.clone().cast(burn::tensor::DType::F32),
)
.cast(dtype)
};
let linear209_out1 = self.linear209.forward(layernormalization116_out1);
let sigmoid93_out1 = burn::tensor::activation::sigmoid(linear209_out1.clone());
let mul141_out1 = linear209_out1.mul(sigmoid93_out1);
let linear210_out1 = self.linear210.forward(mul141_out1);
let constant2261_out1 = self.constant2261.val();
let mul142_out1 = linear210_out1
.mul((constant2261_out1).unsqueeze_dims(&[0isize, 1isize]));
let add169_out1 = layernormalization115_out1.add(mul142_out1);
let layernormalization117_out1 = {
let dtype = add169_out1.clone().dtype();
self.layernormalization117
.forward(add169_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let shape169_out1: [i64; 3] = {
let axes = &layernormalization117_out1.clone().dims()[0..3];
let mut output = [0i64; 3];
for i in 0..3 {
output[i] = axes[i] as i64;
}
output
};
let gather144_out1 = shape169_out1[0] as i64;
let linear211_out1 = self.linear211.forward(layernormalization117_out1.clone());
let unsqueeze317_out1 = [gather144_out1 as i64];
let constant2266_out1: [i64; 1] = [128i64];
let constant2264_out1: [i64; 1] = [-1i64];
let constant2265_out1: [i64; 1] = [8i64];
let concat214_out1: [i64; 4usize] = [
&unsqueeze317_out1[..],
&constant2264_out1[..],
&constant2265_out1[..],
&constant2266_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze318_out1 = [gather144_out1 as i64];
let constant2270_out1: [i64; 1] = [128i64];
let constant2268_out1: [i64; 1] = [-1i64];
let constant2269_out1: [i64; 1] = [8i64];
let concat215_out1: [i64; 4usize] = [
&unsqueeze318_out1[..],
&constant2268_out1[..],
&constant2269_out1[..],
&constant2270_out1[..],
]
.concat()
.try_into()
.unwrap();
let unsqueeze319_out1 = [gather144_out1 as i64];
let constant2274_out1: [i64; 1] = [128i64];
let constant2272_out1: [i64; 1] = [-1i64];
let constant2273_out1: [i64; 1] = [8i64];
let concat216_out1: [i64; 4usize] = [
&unsqueeze319_out1[..],
&constant2272_out1[..],
&constant2273_out1[..],
&constant2274_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape256_out1 = linear211_out1.reshape(concat214_out1);
let linear212_out1 = self.linear212.forward(layernormalization117_out1.clone());
let reshape257_out1 = linear212_out1.reshape(concat215_out1);
let linear213_out1 = self.linear213.forward(layernormalization117_out1);
let reshape258_out1 = linear213_out1.reshape(concat216_out1);
let transpose234_out1 = reshape258_out1.permute([0, 2, 1, 3]);
let linear214_out1 = self.linear214.forward(slice1_out1);
let reshape259_out1 = linear214_out1.reshape([1, -1, 8, 128]);
let constant342_out1 = self.constant342.val();
let add170_out1 = reshape256_out1
.clone()
.add((constant342_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose235_out1 = add170_out1.permute([0, 2, 1, 3]);
let constant343_out1 = self.constant343.val();
let add171_out1 = reshape256_out1
.add((constant343_out1).unsqueeze_dims(&[0isize, 1isize]));
let transpose236_out1 = add171_out1.permute([0, 2, 1, 3]);
let transpose237_out1 = reshape259_out1.permute([0, 2, 3, 1]);
let matmul284_out1 = transpose236_out1.matmul(transpose237_out1);
(
matmul284_out1,
reshape257_out1,
transpose235_out1,
transpose234_out1,
add169_out1,
)
}
}
#[derive(Module, Debug)]
pub struct Submodule26 {
constant2307: burn::module::Param<Tensor<1>>,
linear215: Linear,
layernormalization118: LayerNorm,
conv1d70: Conv1d,
conv1d71: Conv1d,
conv1d72: Conv1d,
layernormalization119: LayerNorm,
linear216: Linear,
linear217: Linear,
constant2323: burn::module::Param<Tensor<1>>,
layernormalization120: LayerNorm,
#[module(skip)]
device: Device,
}
impl Submodule26 {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let constant2307: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([11.313708305358887f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let linear215 = LinearConfig::new(1024, 1024).with_bias(false).init(device);
let layernormalization118 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let conv1d70 = Conv1dConfig::new(1024, 2048, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let conv1d71 = Conv1dConfig::new(1024, 1024, 9)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1024)
.with_bias(true)
.init(device);
let conv1d72 = Conv1dConfig::new(1024, 1024, 1)
.with_stride(1)
.with_padding(PaddingConfig1d::Valid)
.with_dilation(1)
.with_groups(1)
.with_bias(false)
.init(device);
let layernormalization119 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
let linear216 = LinearConfig::new(1024, 4096).with_bias(false).init(device);
let linear217 = LinearConfig::new(4096, 1024).with_bias(false).init(device);
let constant2323: burn::module::Param<Tensor<1>> = burn::module::Param::uninitialized(
burn::module::ParamId::new(),
move |device, _require_grad| Tensor::<
1,
>::from_data(
burn::tensor::TensorData::from([0.5f64]),
(device, burn::tensor::DType::F32),
),
device.clone(),
false,
[1].into(),
);
let layernormalization120 = LayerNormConfig::new(1024)
.with_epsilon(0.000009999999747378752f64)
.with_bias(true)
.init(device);
Self {
constant2307,
linear215,
layernormalization118,
conv1d70,
conv1d71,
conv1d72,
layernormalization119,
linear216,
linear217,
constant2323,
layernormalization120,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
matmul284_out1: Tensor<4>,
reshape257_out1: Tensor<4>,
transpose235_out1: Tensor<4>,
transpose234_out1: Tensor<4>,
unsqueeze50_out1: Tensor<4, Bool>,
add169_out1: Tensor<3>,
unsqueeze52_out1: Tensor<3, Bool>,
) -> Tensor<3> {
let shape170_out1: [i64; 4] = {
let axes = &matmul284_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather145_out1 = shape170_out1[0] as i64;
let gather146_out1 = shape170_out1[1] as i64;
let gather147_out1 = shape170_out1[2] as i64;
let gather148_out1 = shape170_out1[3] as i64;
let pad47_out1 = matmul284_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (0usize, 0usize), (1usize, 0usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let unsqueeze320_out1 = [gather145_out1 as i64];
let unsqueeze321_out1 = [gather146_out1 as i64];
let unsqueeze322_out1 = [gather147_out1 as i64];
let constant2289_out1: [i64; 1] = [-1i64];
let concat218_out1: [i64; 4usize] = [
&unsqueeze320_out1[..],
&unsqueeze321_out1[..],
&constant2289_out1[..],
&unsqueeze322_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape262_out1 = pad47_out1.reshape(concat218_out1);
let slice120_out1 = reshape262_out1.slice(s![.., .., 1.., ..]);
let unsqueeze323_out1 = [gather145_out1 as i64];
let unsqueeze324_out1 = [gather146_out1 as i64];
let unsqueeze325_out1 = [gather147_out1 as i64];
let unsqueeze326_out1 = [gather148_out1 as i64];
let concat219_out1: [i64; 4usize] = [
&unsqueeze323_out1[..],
&unsqueeze324_out1[..],
&unsqueeze325_out1[..],
&unsqueeze326_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape263_out1 = slice120_out1.reshape(concat219_out1);
let transpose239_out1 = reshape257_out1.permute([0, 2, 3, 1]);
let matmul285_out1 = transpose235_out1.matmul(transpose239_out1);
let shape174_out1: [i64; 4] = {
let axes = &matmul285_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let slice121_out1: [i64; 1] = shape174_out1[3..4].try_into().unwrap();
let squeeze24_out1 = slice121_out1[0] as i64;
let unsqueeze327_out1 = [squeeze24_out1 as i64];
let slice122_out1 = reshape263_out1
.slice(s![.., .., .., 0..unsqueeze327_out1[0]]);
let add172_out1 = matmul285_out1.add(slice122_out1);
let constant2307_out1 = self.constant2307.val();
let div27_out1 = add172_out1
.div((constant2307_out1).unsqueeze_dims(&[0isize, 1isize, 2isize]));
let shape175_out1: [i64; 4] = {
let axes = &transpose234_out1.clone().dims()[0..4];
let mut output = [0i64; 4];
for i in 0..4 {
output[i] = axes[i] as i64;
}
output
};
let gather149_out1 = shape175_out1[0] as i64;
let constant2309_out1 = -10000f32;
let where71_out1 = div27_out1
.mask_fill(unsqueeze50_out1.clone(), constant2309_out1);
let softmax24_out1 = burn::tensor::activation::softmax(where71_out1, 3);
let constant2310_out1 = 0f32;
let where72_out1 = softmax24_out1.mask_fill(unsqueeze50_out1, constant2310_out1);
let matmul286_out1 = where72_out1.matmul(transpose234_out1);
let transpose240_out1 = matmul286_out1.permute([0, 2, 1, 3]);
let unsqueeze328_out1 = [gather149_out1 as i64];
let constant2313_out1: [i64; 1] = [1024i64];
let constant2312_out1: [i64; 1] = [-1i64];
let concat220_out1: [i64; 3usize] = [
&unsqueeze328_out1[..],
&constant2312_out1[..],
&constant2313_out1[..],
]
.concat()
.try_into()
.unwrap();
let reshape264_out1 = transpose240_out1.reshape(concat220_out1);
let linear215_out1 = self.linear215.forward(reshape264_out1);
let add173_out1 = add169_out1.add(linear215_out1);
let layernormalization118_out1 = {
let dtype = add173_out1.clone().dtype();
self.layernormalization118
.forward(add173_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose241_out1 = layernormalization118_out1.permute([0, 2, 1]);
let conv1d70_out1 = self.conv1d70.forward(transpose241_out1);
let split_tensors = conv1d70_out1.split_with_sizes([1024, 1024].into(), 1);
let [split24_out1, split24_out2] = split_tensors.try_into().unwrap();
let sigmoid94_out1 = burn::tensor::activation::sigmoid(split24_out2);
let mul143_out1 = split24_out1.mul(sigmoid94_out1);
let constant2314_out1 = 0f32;
let where73_out1 = mul143_out1.mask_fill(unsqueeze52_out1, constant2314_out1);
let pad48_out1 = where73_out1
.pad(
[(0usize, 0usize), (0usize, 0usize), (4usize, 4usize)],
burn::tensor::ops::PadMode::Constant(0f32),
);
let conv1d71_out1 = self.conv1d71.forward(pad48_out1);
let sigmoid95_out1 = burn::tensor::activation::sigmoid(conv1d71_out1.clone());
let mul144_out1 = conv1d71_out1.mul(sigmoid95_out1);
let conv1d72_out1 = self.conv1d72.forward(mul144_out1);
let transpose243_out1 = conv1d72_out1.permute([0, 2, 1]);
let add174_out1 = add173_out1.add(transpose243_out1);
let layernormalization119_out1 = {
let dtype = add174_out1.clone().dtype();
self.layernormalization119
.forward(add174_out1.clone().cast(burn::tensor::DType::F32))
.cast(dtype)
};
let linear216_out1 = self.linear216.forward(layernormalization119_out1);
let sigmoid96_out1 = burn::tensor::activation::sigmoid(linear216_out1.clone());
let mul145_out1 = linear216_out1.mul(sigmoid96_out1);
let linear217_out1 = self.linear217.forward(mul145_out1);
let constant2323_out1 = self.constant2323.val();
let mul146_out1 = linear217_out1
.mul((constant2323_out1).unsqueeze_dims(&[0isize, 1isize]));
let add175_out1 = add174_out1.add(mul146_out1);
let layernormalization120_out1 = {
let dtype = add175_out1.dtype();
self.layernormalization120
.forward(add175_out1.cast(burn::tensor::DType::F32))
.cast(dtype)
};
let transpose244_out1 = layernormalization120_out1.permute([0, 2, 1]);
transpose244_out1
}
}
#[derive(Module, Debug)]
pub struct Model {
submodule1: Submodule1,
submodule2: Submodule2,
submodule3: Submodule3,
submodule4: Submodule4,
submodule5: Submodule5,
submodule6: Submodule6,
submodule7: Submodule7,
submodule8: Submodule8,
submodule9: Submodule9,
submodule10: Submodule10,
submodule11: Submodule11,
submodule12: Submodule12,
submodule13: Submodule13,
submodule14: Submodule14,
submodule15: Submodule15,
submodule16: Submodule16,
submodule17: Submodule17,
submodule18: Submodule18,
submodule19: Submodule19,
submodule20: Submodule20,
submodule21: Submodule21,
submodule22: Submodule22,
submodule23: Submodule23,
submodule24: Submodule24,
submodule25: Submodule25,
submodule26: Submodule26,
#[module(skip)]
device: Device,
}
impl Model {
#[allow(unused_variables)]
pub fn new(device: &Device) -> Self {
let submodule1 = Submodule1::new(device);
let submodule2 = Submodule2::new(device);
let submodule3 = Submodule3::new(device);
let submodule4 = Submodule4::new(device);
let submodule5 = Submodule5::new(device);
let submodule6 = Submodule6::new(device);
let submodule7 = Submodule7::new(device);
let submodule8 = Submodule8::new(device);
let submodule9 = Submodule9::new(device);
let submodule10 = Submodule10::new(device);
let submodule11 = Submodule11::new(device);
let submodule12 = Submodule12::new(device);
let submodule13 = Submodule13::new(device);
let submodule14 = Submodule14::new(device);
let submodule15 = Submodule15::new(device);
let submodule16 = Submodule16::new(device);
let submodule17 = Submodule17::new(device);
let submodule18 = Submodule18::new(device);
let submodule19 = Submodule19::new(device);
let submodule20 = Submodule20::new(device);
let submodule21 = Submodule21::new(device);
let submodule22 = Submodule22::new(device);
let submodule23 = Submodule23::new(device);
let submodule24 = Submodule24::new(device);
let submodule25 = Submodule25::new(device);
let submodule26 = Submodule26::new(device);
Self {
submodule1,
submodule2,
submodule3,
submodule4,
submodule5,
submodule6,
submodule7,
submodule8,
submodule9,
submodule10,
submodule11,
submodule12,
submodule13,
submodule14,
submodule15,
submodule16,
submodule17,
submodule18,
submodule19,
submodule20,
submodule21,
submodule22,
submodule23,
submodule24,
submodule25,
submodule26,
device: device.clone(),
}
}
#[allow(clippy::let_and_return, clippy::approx_constant)]
pub fn forward(
&self,
audio_signal: Tensor<3>,
length: Tensor<1, Int>,
) -> (Tensor<3>, Tensor<1, Int>) {
let (and2_out1, less1_out1, linear1_out1, slice1_out1, cast5_out1) = self
.submodule1
.forward(audio_signal, length);
let (add12_out1, not2_out1, unsqueeze50_out1) = self
.submodule2
.forward(and2_out1, less1_out1, linear1_out1, slice1_out1.clone());
let (add19_out1, unsqueeze52_out1) = self
.submodule3
.forward(
add12_out1,
not2_out1,
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let add26_out1 = self
.submodule4
.forward(
add19_out1,
unsqueeze52_out1.clone(),
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let add33_out1 = self
.submodule5
.forward(
add26_out1,
unsqueeze52_out1.clone(),
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let add40_out1 = self
.submodule6
.forward(
add33_out1,
unsqueeze52_out1.clone(),
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let add47_out1 = self
.submodule7
.forward(
add40_out1,
unsqueeze52_out1.clone(),
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let add54_out1 = self
.submodule8
.forward(
add47_out1,
unsqueeze52_out1.clone(),
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let add61_out1 = self
.submodule9
.forward(
add54_out1,
unsqueeze52_out1.clone(),
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let add68_out1 = self
.submodule10
.forward(
add61_out1,
unsqueeze52_out1.clone(),
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let add75_out1 = self
.submodule11
.forward(
add68_out1,
unsqueeze52_out1.clone(),
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let add82_out1 = self
.submodule12
.forward(
add75_out1,
unsqueeze52_out1.clone(),
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let add89_out1 = self
.submodule13
.forward(
add82_out1,
unsqueeze52_out1.clone(),
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let add96_out1 = self
.submodule14
.forward(
add89_out1,
unsqueeze52_out1.clone(),
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let add103_out1 = self
.submodule15
.forward(
add96_out1,
unsqueeze52_out1.clone(),
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let add110_out1 = self
.submodule16
.forward(
add103_out1,
unsqueeze52_out1.clone(),
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let add117_out1 = self
.submodule17
.forward(
add110_out1,
unsqueeze52_out1.clone(),
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let add124_out1 = self
.submodule18
.forward(
add117_out1,
unsqueeze52_out1.clone(),
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let add131_out1 = self
.submodule19
.forward(
add124_out1,
unsqueeze52_out1.clone(),
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let add138_out1 = self
.submodule20
.forward(
add131_out1,
unsqueeze52_out1.clone(),
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let add145_out1 = self
.submodule21
.forward(
add138_out1,
unsqueeze52_out1.clone(),
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let add152_out1 = self
.submodule22
.forward(
add145_out1,
unsqueeze52_out1.clone(),
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let add159_out1 = self
.submodule23
.forward(
add152_out1,
unsqueeze52_out1.clone(),
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let add166_out1 = self
.submodule24
.forward(
add159_out1,
unsqueeze52_out1.clone(),
slice1_out1.clone(),
unsqueeze50_out1.clone(),
);
let (
matmul284_out1,
reshape257_out1,
transpose235_out1,
transpose234_out1,
add169_out1,
) = self.submodule25.forward(add166_out1, unsqueeze52_out1.clone(), slice1_out1);
let transpose244_out1 = self
.submodule26
.forward(
matmul284_out1,
reshape257_out1,
transpose235_out1,
transpose234_out1,
unsqueeze50_out1,
add169_out1,
unsqueeze52_out1,
);
(transpose244_out1, cast5_out1)
}
}