src/components/convnext.py

import typing as tp

import numpy as np
import torch
import torch.nn as nn
import torch.nn.functional as F

from .conv import StreamingConv1d, TransposedLayerNorm
from .streaming import StreamingContainer, StreamingAdd
from .spectrogram import StreamingLogMelSpectrogram
from ..utils.compile import torch_compile_lazy


# DropPath copied from timm library
def drop_path(
    x, drop_prob: float = 0.0, training: bool = False, scale_by_keep: bool = True
):
    """Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks).

    This is the same as the DropConnect impl I created for EfficientNet, etc networks, however,
    the original name is misleading as 'Drop Connect' is a different form of dropout in a separate paper...
    See discussion: https://github.com/tensorflow/tpu/issues/494#issuecomment-532968956 ... I've opted for
    changing the layer and argument names to 'drop path' rather than mix DropConnect as a layer name and use
    'survival rate' as the argument.

    """  # noqa: E501

    if drop_prob == 0.0 or not training:
        return x
    keep_prob = 1 - drop_prob
    shape = (x.shape[0],) + (1,) * (
        x.ndim - 1
    )  # work with diff dim tensors, not just 2D ConvNets
    random_tensor = x.new_empty(shape).bernoulli_(keep_prob)
    if keep_prob > 0.0 and scale_by_keep:
        random_tensor.div_(keep_prob)
    return x * random_tensor


class DropPath(nn.Module):
    """Drop paths (Stochastic Depth) per sample  (when applied in main path of residual blocks)."""  # noqa: E501

    def __init__(self, drop_prob: float = 0.0, scale_by_keep: bool = True):
        super(DropPath, self).__init__()
        self.drop_prob = drop_prob
        self.scale_by_keep = scale_by_keep

    def forward(self, x):
        return drop_path(x, self.drop_prob, self.training, self.scale_by_keep)

    def extra_repr(self):
        return f"drop_prob={round(self.drop_prob,3):0.3f}"


class LayerNorm(nn.Module):
    r"""LayerNorm that supports two data formats: channels_last (default) or channels_first.
    The ordering of the dimensions in the inputs. channels_last corresponds to inputs with
    shape (batch_size, height, width, channels) while channels_first corresponds to inputs
    with shape (batch_size, channels, height, width).
    """  # noqa: E501

    def __init__(self, normalized_shape, eps=1e-6, data_format="channels_last"):
        super().__init__()
        self.weight = nn.Parameter(torch.ones(normalized_shape))
        self.bias = nn.Parameter(torch.zeros(normalized_shape))
        self.eps = eps
        self.data_format = data_format
        if self.data_format not in ["channels_last", "channels_first"]:
            raise NotImplementedError
        self.normalized_shape = (normalized_shape,)

    def forward(self, x):
        if self.data_format == "channels_last":
            return F.layer_norm(
                x, self.normalized_shape, self.weight, self.bias, self.eps
            )
        elif self.data_format == "channels_first":
            u = x.mean(1, keepdim=True)
            s = (x - u).pow(2).mean(1, keepdim=True)
            x = (x - u) / torch.sqrt(s + self.eps)
            x = self.weight[:, None] * x + self.bias[:, None]
            return x


# ConvNeXt Block copied from https://github.com/fishaudio/fish-diffusion/blob/main/fish_diffusion/modules/convnext.py
class ConvNeXtBlock(StreamingContainer):
    r"""ConvNeXt Block. There are two equivalent implementations:
    (1) DwConv -> LayerNorm (channels_first) -> 1x1 Conv -> GELU -> 1x1 Conv; all in (N, C, H, W)
    (2) DwConv -> Permute to (N, H, W, C); LayerNorm (channels_last) -> Linear -> GELU -> Linear; Permute back
    We use (2) as we find it slightly faster in PyTorch

    Args:
        dim (int): Number of input channels.
        drop_path (float): Stochastic depth rate. Default: 0.0
        layer_scale_init_value (float): Init value for Layer Scale. Default: 1e-6.
        mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. Default: 4.0.
        kernel_size (int): Kernel size for depthwise conv. Default: 7.
        dilation (int): Dilation for depthwise conv. Default: 1.
    """  # noqa: E501

    def __init__(
        self,
        dim: int,
        drop_path: float = 0.0,
        layer_scale_init_value: float = 1e-6,
        mlp_ratio: float = 4.0,
        kernel_size: int = 7,
        dilation: int = 1,
        norm: str = "none",
        norm_params: tp.Dict[str, tp.Any] = {},
        causal: bool = False,
        pad_mode: str = "reflect",
    ):
        super().__init__()

        self.dwconv = StreamingConv1d(
            dim,
            dim,
            kernel_size=kernel_size,
            dilation=dilation,
            norm="weight_norm",
            norm_kwargs=norm_params,
            causal=causal,
            pad_mode=pad_mode,
            groups=dim,
        )

        self.norm = LayerNorm(dim, eps=1e-6)
        self.pwconv1 = nn.Linear(
            dim, int(mlp_ratio * dim)
        )  # pointwise/1x1 convs, implemented with linear layers
        self.act = nn.GELU()
        self.pwconv2 = nn.Linear(int(mlp_ratio * dim), dim)
        self.gamma = (
            nn.Parameter(layer_scale_init_value * torch.ones((dim)), requires_grad=True)
            if layer_scale_init_value > 0
            else None
        )
        self.drop_path = DropPath(drop_path) if drop_path > 0.0 else nn.Identity()
        self.add = StreamingAdd()

    def forward(self, x, apply_residual: bool = True):
        input = x

        x = self.dwconv(x)
        x = x.permute(0, 2, 1)  # (N, C, L) -> (N, L, C)
        x = self.norm(x)
        x = self.pwconv1(x)
        x = self.act(x)
        x = self.pwconv2(x)

        if self.gamma is not None:
            x = self.gamma * x

        x = x.permute(0, 2, 1)  # (N, L, C) -> (N, C, L)
        x = self.drop_path(x)

        if apply_residual:
            # x = input + x
            x = self.add(input, x)

        return x


class VocosBackbone(StreamingContainer):
    """
    Vocos backbone module built with ConvNeXt blocks. Supports additional conditioning with Adaptive Layer Normalization

    Args:
        input_channels (int): Number of input features channels.
        dim (int): Hidden dimension of the model.
        intermediate_dim (int): Intermediate dimension used in ConvNeXtBlock.
        num_layers (int): Number of ConvNeXtBlock layers.
    """

    def __init__(
        self,
        input_channels: int = 80,
        dim: int = 512,
        mlp_ratio: float = 3.0,
        kernel_size: int = 7,
        dilation: int = 1,
        norm: str = "none",
        norm_params: tp.Dict[str, tp.Any] = {},
        causal: bool = False,
        pad_mode: str = "reflect",
        num_layers: int = 8,
        layer_scale_init_value: float = 1e-6
    ):
        super().__init__()
        self.input_channels = input_channels
        self.embed = StreamingConv1d(
                input_channels,
                dim,
                kernel_size=kernel_size,
                dilation=1,
                norm=norm,
                norm_kwargs=norm_params,
                causal=causal,
                pad_mode=pad_mode,
            )
        self.norm = nn.LayerNorm(dim, eps=1e-6)

        layer_scale_init_value = layer_scale_init_value or 1 / num_layers
        self.convnext = nn.ModuleList(
            [
                ConvNeXtBlock(
                    dim=dim,
                    mlp_ratio=mlp_ratio,
                    layer_scale_init_value=layer_scale_init_value,
                    kernel_size=kernel_size,
                    norm=norm,
                    norm_params=norm_params,
                    causal=causal,
                    pad_mode=pad_mode,
                )
                for _ in range(num_layers)
            ]
        )
        self.final_layer_norm = nn.LayerNorm(dim, eps=1e-6)
        self.apply(self._init_weights)

    def _init_weights(self, m):
        if isinstance(m, (nn.Conv1d, nn.Linear)):
            nn.init.trunc_normal_(m.weight, std=0.02)
            nn.init.constant_(m.bias, 0)

    def forward(self, x: torch.Tensor) -> torch.Tensor:
        x = self.embed(x)
        x = self.norm(x.transpose(1, 2))
        x = x.transpose(1, 2)
        for conv_block in self.convnext:
            x = conv_block(x)
        x = self.final_layer_norm(x.transpose(1, 2)).transpose(1, 2)
        return x