address https://github.com/lucidrains/audiolm-pytorch/issues/177

from functools import partial, wraps

from itertools import zip_longest

from typing import Optional

import torch

from torch import nn, einsum

        attn_xpos_scale_base = None,

        attn_dynamic_pos_bias = False,

        squeeze_excite = False,

        complex_stft_discr_logits_abs = True

        complex_stft_discr_logits_abs = True,

        stft_discriminator: Optional[nn.Module] = None  # can pass in own stft discriminator

    ):

        super().__init__()

        discr_rel_factors = [int(s1 / s2) for s1, s2 in zip(discr_multi_scales[:-1], discr_multi_scales[1:])]

        self.downsamples = nn.ModuleList([nn.Identity()] + [nn.AvgPool1d(2 * factor, stride = factor, padding = factor) for factor in discr_rel_factors])

        self.stft_discriminator = stft_discriminator

        if not exists(self.stft_discriminator):

            self.stft_discriminator = ComplexSTFTDiscriminator(

                stft_normalized = stft_normalized,

                logits_abs = complex_stft_discr_logits_abs  # whether to output as abs() or use view_as_real

__version__ = '0.28.0'

__version__ = '0.28.1'