model.py

# pytorch_diffusion + derived encoder decoder
import math
import torch
import torch.nn as nn
import numpy as np
from einops import rearrange
from typing import Optional, Any

from ..attention import MemoryEfficientCrossAttention
from comfy import model_management
import comfy.ops

if model_management.xformers_enabled_vae():
    import xformers
    import xformers.ops

def get_timestep_embedding(timesteps, embedding_dim):
    """
    This matches the implementation in Denoising Diffusion Probabilistic Models:
    From Fairseq.
    Build sinusoidal embeddings.
    This matches the implementation in tensor2tensor, but differs slightly
    from the description in Section 3.5 of "Attention Is All You Need".
    """
    assert len(timesteps.shape) == 1

    half_dim = embedding_dim // 2
    emb = math.log(10000) / (half_dim - 1)
    emb = torch.exp(torch.arange(half_dim, dtype=torch.float32) * -emb)
    emb = emb.to(device=timesteps.device)
    emb = timesteps.float()[:, None] * emb[None, :]
    emb = torch.cat([torch.sin(emb), torch.cos(emb)], dim=1)
    if embedding_dim % 2 == 1:  # zero pad
        emb = torch.nn.functional.pad(emb, (0,1,0,0))
    return emb


def nonlinearity(x):
    # swish
    return x*torch.sigmoid(x)


def Normalize(in_channels, num_groups=32):
    return torch.nn.GroupNorm(num_groups=num_groups, num_channels=in_channels, eps=1e-6, affine=True)


class Upsample(nn.Module):
    def __init__(self, in_channels, with_conv):
        super().__init__()
        self.with_conv = with_conv
        if self.with_conv:
            self.conv = comfy.ops.Conv2d(in_channels,
                                        in_channels,
                                        kernel_size=3,
                                        stride=1,
                                        padding=1)

    def forward(self, x):
        try:
            x = torch.nn.functional.interpolate(x, scale_factor=2.0, mode="nearest")
        except: #operation not implemented for bf16
            b, c, h, w = x.shape
            out = torch.empty((b, c, h*2, w*2), dtype=x.dtype, layout=x.layout, device=x.device)
            split = 8
            l = out.shape[1] // split
            for i in range(0, out.shape[1], l):
                out[:,i:i+l] = torch.nn.functional.interpolate(x[:,i:i+l].to(torch.float32), scale_factor=2.0, mode="nearest").to(x.dtype)
            del x
            x = out

        if self.with_conv:
            x = self.conv(x)
        return x


class Downsample(nn.Module):
    def __init__(self, in_channels, with_conv):
        super().__init__()
        self.with_conv = with_conv
        if self.with_conv:
            # no asymmetric padding in torch conv, must do it ourselves
            self.conv = comfy.ops.Conv2d(in_channels,
                                        in_channels,
                                        kernel_size=3,
                                        stride=2,
                                        padding=0)

    def forward(self, x):
        if self.with_conv:
            pad = (0,1,0,1)
            x = torch.nn.functional.pad(x, pad, mode="constant", value=0)
            x = self.conv(x)
        else:
            x = torch.nn.functional.avg_pool2d(x, kernel_size=2, stride=2)
        return x


class ResnetBlock(nn.Module):
    def __init__(self, *, in_channels, out_channels=None, conv_shortcut=False,
                 dropout, temb_channels=512):
        super().__init__()
        self.in_channels = in_channels
        out_channels = in_channels if out_channels is None else out_channels
        self.out_channels = out_channels
        self.use_conv_shortcut = conv_shortcut

        self.swish = torch.nn.SiLU(inplace=True)
        self.norm1 = Normalize(in_channels)
        self.conv1 = comfy.ops.Conv2d(in_channels,
                                     out_channels,
                                     kernel_size=3,
                                     stride=1,
                                     padding=1)
        if temb_channels > 0:
            self.temb_proj = comfy.ops.Linear(temb_channels,
                                             out_channels)
        self.norm2 = Normalize(out_channels)
        self.dropout = torch.nn.Dropout(dropout, inplace=True)
        self.conv2 = comfy.ops.Conv2d(out_channels,
                                     out_channels,
                                     kernel_size=3,
                                     stride=1,
                                     padding=1)
        if self.in_channels != self.out_channels:
            if self.use_conv_shortcut:
                self.conv_shortcut = comfy.ops.Conv2d(in_channels,
                                                     out_channels,
                                                     kernel_size=3,
                                                     stride=1,
                                                     padding=1)
            else:
                self.nin_shortcut = comfy.ops.Conv2d(in_channels,
                                                    out_channels,
                                                    kernel_size=1,
                                                    stride=1,
                                                    padding=0)

    def forward(self, x, temb):
        h = x
        h = self.norm1(h)
        h = self.swish(h)
        h = self.conv1(h)

        if temb is not None:
            h = h + self.temb_proj(self.swish(temb))[:,:,None,None]

        h = self.norm2(h)
        h = self.swish(h)
        h = self.dropout(h)
        h = self.conv2(h)

        if self.in_channels != self.out_channels:
            if self.use_conv_shortcut:
                x = self.conv_shortcut(x)
            else:
                x = self.nin_shortcut(x)

        return x+h

def slice_attention(q, k, v):
    r1 = torch.zeros_like(k, device=q.device)
    scale = (int(q.shape[-1])**(-0.5))

    mem_free_total = model_management.get_free_memory(q.device)

    gb = 1024 ** 3
    tensor_size = q.shape[0] * q.shape[1] * k.shape[2] * q.element_size()
    modifier = 3 if q.element_size() == 2 else 2.5
    mem_required = tensor_size * modifier
    steps = 1

    if mem_required > mem_free_total:
        steps = 2**(math.ceil(math.log(mem_required / mem_free_total, 2)))

    while True:
        try:
            slice_size = q.shape[1] // steps if (q.shape[1] % steps) == 0 else q.shape[1]
            for i in range(0, q.shape[1], slice_size):
                end = i + slice_size
                s1 = torch.bmm(q[:, i:end], k) * scale

                s2 = torch.nn.functional.softmax(s1, dim=2).permute(0,2,1)
                del s1

                r1[:, :, i:end] = torch.bmm(v, s2)
                del s2
            break
        except model_management.OOM_EXCEPTION as e:
            steps *= 2
            if steps > 128:
                raise e
            print("out of memory error, increasing steps and trying again", steps)

    return r1

class AttnBlock(nn.Module):
    def __init__(self, in_channels):
        super().__init__()
        self.in_channels = in_channels

        self.norm = Normalize(in_channels)
        self.q = comfy.ops.Conv2d(in_channels,
                                 in_channels,
                                 kernel_size=1,
                                 stride=1,
                                 padding=0)
        self.k = comfy.ops.Conv2d(in_channels,
                                 in_channels,
                                 kernel_size=1,
                                 stride=1,
                                 padding=0)
        self.v = comfy.ops.Conv2d(in_channels,
                                 in_channels,
                                 kernel_size=1,
                                 stride=1,
                                 padding=0)
        self.proj_out = comfy.ops.Conv2d(in_channels,
                                        in_channels,
                                        kernel_size=1,
                                        stride=1,
                                        padding=0)

    def forward(self, x):
        h_ = x
        h_ = self.norm(h_)
        q = self.q(h_)
        k = self.k(h_)
        v = self.v(h_)

        # compute attention
        b,c,h,w = q.shape

        q = q.reshape(b,c,h*w)
        q = q.permute(0,2,1)   # b,hw,c
        k = k.reshape(b,c,h*w) # b,c,hw
        v = v.reshape(b,c,h*w)

        r1 = slice_attention(q, k, v)
        h_ = r1.reshape(b,c,h,w)
        del r1
        h_ = self.proj_out(h_)

        return x+h_

class MemoryEfficientAttnBlock(nn.Module):
    """
        Uses xformers efficient implementation,
        see https://github.com/MatthieuTPHR/diffusers/blob/d80b531ff8060ec1ea982b65a1b8df70f73aa67c/src/diffusers/models/attention.py#L223
        Note: this is a single-head self-attention operation
    """
    #
    def __init__(self, in_channels):
        super().__init__()
        self.in_channels = in_channels

        self.norm = Normalize(in_channels)
        self.q = comfy.ops.Conv2d(in_channels,
                                 in_channels,
                                 kernel_size=1,
                                 stride=1,
                                 padding=0)
        self.k = comfy.ops.Conv2d(in_channels,
                                 in_channels,
                                 kernel_size=1,
                                 stride=1,
                                 padding=0)
        self.v = comfy.ops.Conv2d(in_channels,
                                 in_channels,
                                 kernel_size=1,
                                 stride=1,
                                 padding=0)
        self.proj_out = comfy.ops.Conv2d(in_channels,
                                        in_channels,
                                        kernel_size=1,
                                        stride=1,
                                        padding=0)
        self.attention_op: Optional[Any] = None

    def forward(self, x):
        h_ = x
        h_ = self.norm(h_)
        q = self.q(h_)
        k = self.k(h_)
        v = self.v(h_)

        # compute attention
        B, C, H, W = q.shape
        q, k, v = map(
            lambda t: t.view(B, C, -1).transpose(1, 2).contiguous(),
            (q, k, v),
        )

        try:
            out = xformers.ops.memory_efficient_attention(q, k, v, attn_bias=None, op=self.attention_op)
            out = out.transpose(1, 2).reshape(B, C, H, W)
        except NotImplementedError as e:
            out = slice_attention(q.view(B, -1, C), k.view(B, -1, C).transpose(1, 2), v.view(B, -1, C).transpose(1, 2)).reshape(B, C, H, W)

        out = self.proj_out(out)
        return x+out

class MemoryEfficientAttnBlockPytorch(nn.Module):
    def __init__(self, in_channels):
        super().__init__()
        self.in_channels = in_channels

        self.norm = Normalize(in_channels)
        self.q = comfy.ops.Conv2d(in_channels,
                                 in_channels,
                                 kernel_size=1,
                                 stride=1,
                                 padding=0)
        self.k = comfy.ops.Conv2d(in_channels,
                                 in_channels,
                                 kernel_size=1,
                                 stride=1,
                                 padding=0)
        self.v = comfy.ops.Conv2d(in_channels,
                                 in_channels,
                                 kernel_size=1,
                                 stride=1,
                                 padding=0)
        self.proj_out = comfy.ops.Conv2d(in_channels,
                                        in_channels,
                                        kernel_size=1,
                                        stride=1,
                                        padding=0)
        self.attention_op: Optional[Any] = None

    def forward(self, x):
        h_ = x
        h_ = self.norm(h_)
        q = self.q(h_)
        k = self.k(h_)
        v = self.v(h_)

        # compute attention
        B, C, H, W = q.shape
        q, k, v = map(
            lambda t: t.view(B, 1, C, -1).transpose(2, 3).contiguous(),
            (q, k, v),
        )

        try:
            out = torch.nn.functional.scaled_dot_product_attention(q, k, v, attn_mask=None, dropout_p=0.0, is_causal=False)
            out = out.transpose(2, 3).reshape(B, C, H, W)
        except model_management.OOM_EXCEPTION as e:
            print("scaled_dot_product_attention OOMed: switched to slice attention")
            out = slice_attention(q.view(B, -1, C), k.view(B, -1, C).transpose(1, 2), v.view(B, -1, C).transpose(1, 2)).reshape(B, C, H, W)

        out = self.proj_out(out)
        return x+out

class MemoryEfficientCrossAttentionWrapper(MemoryEfficientCrossAttention):
    def forward(self, x, context=None, mask=None):
        b, c, h, w = x.shape
        x = rearrange(x, 'b c h w -> b (h w) c')
        out = super().forward(x, context=context, mask=mask)
        out = rearrange(out, 'b (h w) c -> b c h w', h=h, w=w, c=c)
        return x + out


def make_attn(in_channels, attn_type="vanilla", attn_kwargs=None):
    assert attn_type in ["vanilla", "vanilla-xformers", "memory-efficient-cross-attn", "linear", "none"], f'attn_type {attn_type} unknown'
    if model_management.xformers_enabled_vae() and attn_type == "vanilla":
        attn_type = "vanilla-xformers"
    if model_management.pytorch_attention_enabled() and attn_type == "vanilla":
        attn_type = "vanilla-pytorch"
    print(f"making attention of type '{attn_type}' with {in_channels} in_channels")
    if attn_type == "vanilla":
        assert attn_kwargs is None
        return AttnBlock(in_channels)
    elif attn_type == "vanilla-xformers":
        print(f"building MemoryEfficientAttnBlock with {in_channels} in_channels...")
        return MemoryEfficientAttnBlock(in_channels)
    elif attn_type == "vanilla-pytorch":
        return MemoryEfficientAttnBlockPytorch(in_channels)
    elif type == "memory-efficient-cross-attn":
        attn_kwargs["query_dim"] = in_channels
        return MemoryEfficientCrossAttentionWrapper(**attn_kwargs)
    elif attn_type == "none":
        return nn.Identity(in_channels)
    else:
        raise NotImplementedError()


class Model(nn.Module):
    def __init__(self, *, ch, out_ch, ch_mult=(1,2,4,8), num_res_blocks,
                 attn_resolutions, dropout=0.0, resamp_with_conv=True, in_channels,
                 resolution, use_timestep=True, use_linear_attn=False, attn_type="vanilla"):
        super().__init__()
        if use_linear_attn: attn_type = "linear"
        self.ch = ch
        self.temb_ch = self.ch*4
        self.num_resolutions = len(ch_mult)
        self.num_res_blocks = num_res_blocks
        self.resolution = resolution
        self.in_channels = in_channels

        self.use_timestep = use_timestep
        if self.use_timestep:
            # timestep embedding
            self.temb = nn.Module()
            self.temb.dense = nn.ModuleList([
                comfy.ops.Linear(self.ch,
                                self.temb_ch),
                comfy.ops.Linear(self.temb_ch,
                                self.temb_ch),
            ])

        # downsampling
        self.conv_in = comfy.ops.Conv2d(in_channels,
                                       self.ch,
                                       kernel_size=3,
                                       stride=1,
                                       padding=1)

        curr_res = resolution
        in_ch_mult = (1,)+tuple(ch_mult)
        self.down = nn.ModuleList()
        for i_level in range(self.num_resolutions):
            block = nn.ModuleList()
            attn = nn.ModuleList()
            block_in = ch*in_ch_mult[i_level]
            block_out = ch*ch_mult[i_level]
            for i_block in range(self.num_res_blocks):
                block.append(ResnetBlock(in_channels=block_in,
                                         out_channels=block_out,
                                         temb_channels=self.temb_ch,
                                         dropout=dropout))
                block_in = block_out
                if curr_res in attn_resolutions:
                    attn.append(make_attn(block_in, attn_type=attn_type))
            down = nn.Module()
            down.block = block
            down.attn = attn
            if i_level != self.num_resolutions-1:
                down.downsample = Downsample(block_in, resamp_with_conv)
                curr_res = curr_res // 2
            self.down.append(down)

        # middle
        self.mid = nn.Module()
        self.mid.block_1 = ResnetBlock(in_channels=block_in,
                                       out_channels=block_in,
                                       temb_channels=self.temb_ch,
                                       dropout=dropout)
        self.mid.attn_1 = make_attn(block_in, attn_type=attn_type)
        self.mid.block_2 = ResnetBlock(in_channels=block_in,
                                       out_channels=block_in,
                                       temb_channels=self.temb_ch,
                                       dropout=dropout)

        # upsampling
        self.up = nn.ModuleList()
        for i_level in reversed(range(self.num_resolutions)):
            block = nn.ModuleList()
            attn = nn.ModuleList()
            block_out = ch*ch_mult[i_level]
            skip_in = ch*ch_mult[i_level]
            for i_block in range(self.num_res_blocks+1):
                if i_block == self.num_res_blocks:
                    skip_in = ch*in_ch_mult[i_level]
                block.append(ResnetBlock(in_channels=block_in+skip_in,
                                         out_channels=block_out,
                                         temb_channels=self.temb_ch,
                                         dropout=dropout))
                block_in = block_out
                if curr_res in attn_resolutions:
                    attn.append(make_attn(block_in, attn_type=attn_type))
            up = nn.Module()
            up.block = block
            up.attn = attn
            if i_level != 0:
                up.upsample = Upsample(block_in, resamp_with_conv)
                curr_res = curr_res * 2
            self.up.insert(0, up) # prepend to get consistent order

        # end
        self.norm_out = Normalize(block_in)
        self.conv_out = comfy.ops.Conv2d(block_in,
                                        out_ch,
                                        kernel_size=3,
                                        stride=1,
                                        padding=1)

    def forward(self, x, t=None, context=None):
        #assert x.shape[2] == x.shape[3] == self.resolution
        if context is not None:
            # assume aligned context, cat along channel axis
            x = torch.cat((x, context), dim=1)
        if self.use_timestep:
            # timestep embedding
            assert t is not None
            temb = get_timestep_embedding(t, self.ch)
            temb = self.temb.dense[0](temb)
            temb = nonlinearity(temb)
            temb = self.temb.dense[1](temb)
        else:
            temb = None

        # downsampling
        hs = [self.conv_in(x)]
        for i_level in range(self.num_resolutions):
            for i_block in range(self.num_res_blocks):
                h = self.down[i_level].block[i_block](hs[-1], temb)
                if len(self.down[i_level].attn) > 0:
                    h = self.down[i_level].attn[i_block](h)
                hs.append(h)
            if i_level != self.num_resolutions-1:
                hs.append(self.down[i_level].downsample(hs[-1]))

        # middle
        h = hs[-1]
        h = self.mid.block_1(h, temb)
        h = self.mid.attn_1(h)
        h = self.mid.block_2(h, temb)

        # upsampling
        for i_level in reversed(range(self.num_resolutions)):
            for i_block in range(self.num_res_blocks+1):
                h = self.up[i_level].block[i_block](
                    torch.cat([h, hs.pop()], dim=1), temb)
                if len(self.up[i_level].attn) > 0:
                    h = self.up[i_level].attn[i_block](h)
            if i_level != 0:
                h = self.up[i_level].upsample(h)

        # end
        h = self.norm_out(h)
        h = nonlinearity(h)
        h = self.conv_out(h)
        return h

    def get_last_layer(self):
        return self.conv_out.weight


class Encoder(nn.Module):
    def __init__(self, *, ch, out_ch, ch_mult=(1,2,4,8), num_res_blocks,
                 attn_resolutions, dropout=0.0, resamp_with_conv=True, in_channels,
                 resolution, z_channels, double_z=True, use_linear_attn=False, attn_type="vanilla",
                 **ignore_kwargs):
        super().__init__()
        if use_linear_attn: attn_type = "linear"
        self.ch = ch
        self.temb_ch = 0
        self.num_resolutions = len(ch_mult)
        self.num_res_blocks = num_res_blocks
        self.resolution = resolution
        self.in_channels = in_channels

        # downsampling
        self.conv_in = comfy.ops.Conv2d(in_channels,
                                       self.ch,
                                       kernel_size=3,
                                       stride=1,
                                       padding=1)

        curr_res = resolution
        in_ch_mult = (1,)+tuple(ch_mult)
        self.in_ch_mult = in_ch_mult
        self.down = nn.ModuleList()
        for i_level in range(self.num_resolutions):
            block = nn.ModuleList()
            attn = nn.ModuleList()
            block_in = ch*in_ch_mult[i_level]
            block_out = ch*ch_mult[i_level]
            for i_block in range(self.num_res_blocks):
                block.append(ResnetBlock(in_channels=block_in,
                                         out_channels=block_out,
                                         temb_channels=self.temb_ch,
                                         dropout=dropout))
                block_in = block_out
                if curr_res in attn_resolutions:
                    attn.append(make_attn(block_in, attn_type=attn_type))
            down = nn.Module()
            down.block = block
            down.attn = attn
            if i_level != self.num_resolutions-1:
                down.downsample = Downsample(block_in, resamp_with_conv)
                curr_res = curr_res // 2
            self.down.append(down)

        # middle
        self.mid = nn.Module()
        self.mid.block_1 = ResnetBlock(in_channels=block_in,
                                       out_channels=block_in,
                                       temb_channels=self.temb_ch,
                                       dropout=dropout)
        self.mid.attn_1 = make_attn(block_in, attn_type=attn_type)
        self.mid.block_2 = ResnetBlock(in_channels=block_in,
                                       out_channels=block_in,
                                       temb_channels=self.temb_ch,
                                       dropout=dropout)

        # end
        self.norm_out = Normalize(block_in)
        self.conv_out = comfy.ops.Conv2d(block_in,
                                        2*z_channels if double_z else z_channels,
                                        kernel_size=3,
                                        stride=1,
                                        padding=1)

    def forward(self, x):
        # timestep embedding
        temb = None
        # downsampling
        h = self.conv_in(x)
        for i_level in range(self.num_resolutions):
            for i_block in range(self.num_res_blocks):
                h = self.down[i_level].block[i_block](h, temb)
                if len(self.down[i_level].attn) > 0:
                    h = self.down[i_level].attn[i_block](h)
            if i_level != self.num_resolutions-1:
                h = self.down[i_level].downsample(h)

        # middle
        h = self.mid.block_1(h, temb)
        h = self.mid.attn_1(h)
        h = self.mid.block_2(h, temb)

        # end
        h = self.norm_out(h)
        h = nonlinearity(h)
        h = self.conv_out(h)
        return h


class Decoder(nn.Module):
    def __init__(self, *, ch, out_ch, ch_mult=(1,2,4,8), num_res_blocks,
                 attn_resolutions, dropout=0.0, resamp_with_conv=True, in_channels,
                 resolution, z_channels, give_pre_end=False, tanh_out=False, use_linear_attn=False,
                 attn_type="vanilla", **ignorekwargs):
        super().__init__()
        if use_linear_attn: attn_type = "linear"
        self.ch = ch
        self.temb_ch = 0
        self.num_resolutions = len(ch_mult)
        self.num_res_blocks = num_res_blocks
        self.resolution = resolution
        self.in_channels = in_channels
        self.give_pre_end = give_pre_end
        self.tanh_out = tanh_out

        # compute in_ch_mult, block_in and curr_res at lowest res
        in_ch_mult = (1,)+tuple(ch_mult)
        block_in = ch*ch_mult[self.num_resolutions-1]
        curr_res = resolution // 2**(self.num_resolutions-1)
        self.z_shape = (1,z_channels,curr_res,curr_res)
        print("Working with z of shape {} = {} dimensions.".format(
            self.z_shape, np.prod(self.z_shape)))

        # z to block_in
        self.conv_in = comfy.ops.Conv2d(z_channels,
                                       block_in,
                                       kernel_size=3,
                                       stride=1,
                                       padding=1)

        # middle
        self.mid = nn.Module()
        self.mid.block_1 = ResnetBlock(in_channels=block_in,
                                       out_channels=block_in,
                                       temb_channels=self.temb_ch,
                                       dropout=dropout)
        self.mid.attn_1 = make_attn(block_in, attn_type=attn_type)
        self.mid.block_2 = ResnetBlock(in_channels=block_in,
                                       out_channels=block_in,
                                       temb_channels=self.temb_ch,
                                       dropout=dropout)

        # upsampling
        self.up = nn.ModuleList()
        for i_level in reversed(range(self.num_resolutions)):
            block = nn.ModuleList()
            attn = nn.ModuleList()
            block_out = ch*ch_mult[i_level]
            for i_block in range(self.num_res_blocks+1):
                block.append(ResnetBlock(in_channels=block_in,
                                         out_channels=block_out,
                                         temb_channels=self.temb_ch,
                                         dropout=dropout))
                block_in = block_out
                if curr_res in attn_resolutions:
                    attn.append(make_attn(block_in, attn_type=attn_type))
            up = nn.Module()
            up.block = block
            up.attn = attn
            if i_level != 0:
                up.upsample = Upsample(block_in, resamp_with_conv)
                curr_res = curr_res * 2
            self.up.insert(0, up) # prepend to get consistent order

        # end
        self.norm_out = Normalize(block_in)
        self.conv_out = comfy.ops.Conv2d(block_in,
                                        out_ch,
                                        kernel_size=3,
                                        stride=1,
                                        padding=1)

    def forward(self, z):
        #assert z.shape[1:] == self.z_shape[1:]
        self.last_z_shape = z.shape

        # timestep embedding
        temb = None

        # z to block_in
        h = self.conv_in(z)

        # middle
        h = self.mid.block_1(h, temb)
        h = self.mid.attn_1(h)
        h = self.mid.block_2(h, temb)

        # upsampling
        for i_level in reversed(range(self.num_resolutions)):
            for i_block in range(self.num_res_blocks+1):
                h = self.up[i_level].block[i_block](h, temb)
                if len(self.up[i_level].attn) > 0:
                    h = self.up[i_level].attn[i_block](h)
            if i_level != 0:
                h = self.up[i_level].upsample(h)

        # end
        if self.give_pre_end:
            return h

        h = self.norm_out(h)
        h = nonlinearity(h)
        h = self.conv_out(h)
        if self.tanh_out:
            h = torch.tanh(h)
        return h