diffusion/model/nets/sana_U_shape.py

# Copyright 2024 NVIDIA CORPORATION & AFFILIATES
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#      http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
# SPDX-License-Identifier: Apache-2.0

# This file is modified from https://github.com/PixArt-alpha/PixArt-sigma
import os

import torch
import torch.nn as nn
from timm.models.layers import DropPath

from diffusion.model.builder import MODELS
from diffusion.model.nets.basic_modules import DWMlp, GLUMBConv, MBConvPreGLU, Mlp
from diffusion.model.nets.fastlinear.modules import TritonLiteMLA
from diffusion.model.nets.sana import Sana, get_2d_sincos_pos_embed
from diffusion.model.nets.sana_blocks import (
    Attention,
    CaptionEmbedder,
    FlashAttention,
    LiteLA,
    MultiHeadCrossAttention,
    PatchEmbed,
    T2IFinalLayer,
    TimestepEmbedder,
    t2i_modulate,
)
from diffusion.model.norms import RMSNorm
from diffusion.model.utils import auto_grad_checkpoint, to_2tuple
from diffusion.utils.logger import get_root_logger


class SanaUBlock(nn.Module):
    """
    A SanaU block with global shared adaptive layer norm (adaLN-single) conditioning and U-shaped model.
    """

    def __init__(
        self,
        hidden_size,
        num_heads,
        mlp_ratio=4.0,
        drop_path=0,
        input_size=None,
        qk_norm=False,
        attn_type="flash",
        ffn_type="mlp",
        mlp_acts=("silu", "silu", None),
        skip_linear=False,
        **block_kwargs,
    ):
        super().__init__()
        self.norm1 = nn.LayerNorm(hidden_size, elementwise_affine=False, eps=1e-6)
        if attn_type == "flash":
            # flash self attention
            self.attn = FlashAttention(
                hidden_size,
                num_heads=num_heads,
                qkv_bias=True,
                qk_norm=qk_norm,
                **block_kwargs,
            )
        elif attn_type == "linear":
            # linear self attention
            # TODO: Here the num_heads set to 36 for tmp used
            self_num_heads = hidden_size // 32
            self.attn = LiteLA(hidden_size, hidden_size, heads=self_num_heads, eps=1e-8, qk_norm=qk_norm)
        elif attn_type == "triton_linear":
            # linear self attention with triton kernel fusion
            # TODO: Here the num_heads set to 36 for tmp used
            self_num_heads = hidden_size // 32
            self.attn = TritonLiteMLA(hidden_size, num_heads=self_num_heads, eps=1e-8)
        elif attn_type == "vanilla":
            # vanilla self attention
            self.attn = Attention(hidden_size, num_heads=num_heads, qkv_bias=True)
        else:
            raise ValueError(f"{attn_type} type is not defined.")

        self.cross_attn = MultiHeadCrossAttention(hidden_size, num_heads, **block_kwargs)
        self.norm2 = nn.LayerNorm(hidden_size, elementwise_affine=False, eps=1e-6)
        # to be compatible with lower version pytorch
        if ffn_type == "dwmlp":
            approx_gelu = lambda: nn.GELU(approximate="tanh")
            self.mlp = DWMlp(
                in_features=hidden_size, hidden_features=int(hidden_size * mlp_ratio), act_layer=approx_gelu, drop=0
            )
        elif ffn_type == "glumbconv":
            self.mlp = GLUMBConv(
                in_features=hidden_size,
                hidden_features=int(hidden_size * mlp_ratio),
                use_bias=(True, True, False),
                norm=(None, None, None),
                act=mlp_acts,
            )
        elif ffn_type == "mbconvpreglu":
            self.mlp = MBConvPreGLU(
                in_dim=hidden_size,
                out_dim=hidden_size,
                mid_dim=int(hidden_size * mlp_ratio),
                use_bias=(True, True, False),
                norm=None,
                act=("silu", "silu", None),
            )
        elif ffn_type == "mlp":
            approx_gelu = lambda: nn.GELU(approximate="tanh")
            self.mlp = Mlp(
                in_features=hidden_size, hidden_features=int(hidden_size * mlp_ratio), act_layer=approx_gelu, drop=0
            )
        else:
            raise ValueError(f"{ffn_type} type is not defined.")
        self.drop_path = DropPath(drop_path) if drop_path > 0.0 else nn.Identity()
        self.scale_shift_table = nn.Parameter(torch.randn(6, hidden_size) / hidden_size**0.5)

        # skip connection
        if skip_linear:
            self.skip_linear = nn.Linear(hidden_size * 2, hidden_size, bias=True)

    def forward(self, x, y, t, mask=None, skip_x=None, **kwargs):
        B, N, C = x.shape
        if skip_x is not None:
            x = self.skip_linear(torch.cat([x, skip_x], dim=-1))

        shift_msa, scale_msa, gate_msa, shift_mlp, scale_mlp, gate_mlp = (
            self.scale_shift_table[None] + t.reshape(B, 6, -1)
        ).chunk(6, dim=1)
        x = x + self.drop_path(gate_msa * self.attn(t2i_modulate(self.norm1(x), shift_msa, scale_msa)).reshape(B, N, C))
        x = x + self.cross_attn(x, y, mask)
        x = x + self.drop_path(gate_mlp * self.mlp(t2i_modulate(self.norm2(x), shift_mlp, scale_mlp)))

        return x


#############################################################################
#                                 Core SanaU Model                                #
#################################################################################
@MODELS.register_module()
class SanaU(Sana):
    """
    Diffusion model with a Transformer backbone.
    """

    def __init__(
        self,
        input_size=32,
        patch_size=2,
        in_channels=4,
        hidden_size=1152,
        depth=29,
        num_heads=16,
        mlp_ratio=4.0,
        class_dropout_prob=0.1,
        learn_sigma=True,
        pred_sigma=True,
        drop_path: float = 0.0,
        caption_channels=2304,
        pe_interpolation=1.0,
        config=None,
        model_max_length=300,
        micro_condition=False,
        qk_norm=False,
        y_norm=False,
        norm_eps=1e-5,
        attn_type="flash",
        ffn_type="mlp",
        use_pe=True,
        y_norm_scale_factor=1.0,
        patch_embed_kernel=None,
        mlp_acts=("silu", "silu", None),
        **kwargs,
    ):
        super().__init__(
            input_size=input_size,
            patch_size=patch_size,
            in_channels=in_channels,
            hidden_size=hidden_size,
            depth=depth,
            num_heads=num_heads,
            mlp_ratio=mlp_ratio,
            class_dropout_prob=class_dropout_prob,
            learn_sigma=learn_sigma,
            pred_sigma=pred_sigma,
            drop_path=drop_path,
            caption_channels=caption_channels,
            pe_interpolation=pe_interpolation,
            config=config,
            model_max_length=model_max_length,
            micro_condition=micro_condition,
            qk_norm=qk_norm,
            y_norm=y_norm,
            norm_eps=norm_eps,
            attn_type=attn_type,
            ffn_type=ffn_type,
            use_pe=use_pe,
            y_norm_scale_factor=y_norm_scale_factor,
            patch_embed_kernel=patch_embed_kernel,
            mlp_acts=mlp_acts,
            **kwargs,
        )

        kernel_size = patch_embed_kernel or patch_size
        self.x_embedder = PatchEmbed(
            input_size, patch_size, in_channels, hidden_size, kernel_size=kernel_size, bias=True
        )
        self.t_embedder = TimestepEmbedder(hidden_size)
        num_patches = self.x_embedder.num_patches
        self.base_size = input_size // self.patch_size
        # Will use fixed sin-cos embedding:
        self.register_buffer("pos_embed", torch.zeros(1, num_patches, hidden_size))

        approx_gelu = lambda: nn.GELU(approximate="tanh")
        self.t_block = nn.Sequential(nn.SiLU(), nn.Linear(hidden_size, 6 * hidden_size, bias=True))

        self.y_embedder = CaptionEmbedder(
            in_channels=caption_channels,
            hidden_size=hidden_size,
            uncond_prob=class_dropout_prob,
            act_layer=approx_gelu,
            token_num=model_max_length,
        )
        if self.y_norm:
            self.attention_y_norm = RMSNorm(hidden_size, scale_factor=y_norm_scale_factor, eps=norm_eps)
        drop_path = [x.item() for x in torch.linspace(0, drop_path, depth)]  # stochastic depth decay rule
        self.blocks = nn.ModuleList(
            [
                SanaUBlock(
                    hidden_size,
                    num_heads,
                    mlp_ratio=mlp_ratio,
                    drop_path=drop_path[i],
                    input_size=(input_size // patch_size, input_size // patch_size),
                    qk_norm=qk_norm,
                    attn_type=attn_type,
                    ffn_type=ffn_type,
                    mlp_acts=mlp_acts,
                    skip_linear=i > depth // 2,
                )
                for i in range(depth)
            ]
        )
        self.final_layer = T2IFinalLayer(hidden_size, patch_size, self.out_channels)

        self.initialize_weights()

        if config:
            logger = get_root_logger(os.path.join(config.work_dir, "train_log.log"))
            logger = logger.warning
        else:
            logger = print
        logger(f"use pe: {use_pe}, position embed interpolation: {self.pe_interpolation}, base size: {self.base_size}")
        logger(
            f"attention type: {attn_type}; ffn type: {ffn_type}; "
            f"autocast linear attn: {os.environ.get('AUTOCAST_LINEAR_ATTN', False)}"
        )

    def forward(self, x, timestep, y, mask=None, data_info=None, **kwargs):
        """
        Forward pass of SanaU.
        x: (N, C, H, W) tensor of spatial inputs (images or latent representations of images)
        t: (N,) tensor of diffusion timesteps
        y: (N, 1, 120, C) tensor of class labels
        """
        x = x.to(self.dtype)
        timestep = timestep.to(self.dtype)
        y = y.to(self.dtype)
        pos_embed = self.pos_embed.to(self.dtype)
        self.h, self.w = x.shape[-2] // self.patch_size, x.shape[-1] // self.patch_size
        x = self.x_embedder(x) + pos_embed  # (N, T, D), where T = H * W / patch_size ** 2
        t = self.t_embedder(timestep.to(x.dtype))  # (N, D)
        t0 = self.t_block(t)
        y = self.y_embedder(y, self.training)  # (N, 1, L, D)
        if self.y_norm:
            y = self.attention_y_norm(y)
        if mask is not None:
            if mask.shape[0] != y.shape[0]:
                mask = mask.repeat(y.shape[0] // mask.shape[0], 1)
            mask = mask.squeeze(1).squeeze(1)
            y = y.squeeze(1).masked_select(mask.unsqueeze(-1) != 0).view(1, -1, x.shape[-1])
            y_lens = mask.sum(dim=1).tolist()
        else:
            y_lens = [y.shape[2]] * y.shape[0]
            y = y.squeeze(1).view(1, -1, x.shape[-1])
        results_hooker = {}
        for i, block in enumerate(self.blocks):
            if i > len(self.blocks) // 2:
                x = auto_grad_checkpoint(block, x, y, t0, y_lens, skip_x=results_hooker[len(self.blocks) - i - 1])
            else:
                x = auto_grad_checkpoint(block, x, y, t0, y_lens)  # (N, T, D) #support grad checkpoint
            results_hooker[i] = x
        x = self.final_layer(x, t)  # (N, T, patch_size ** 2 * out_channels)
        x = self.unpatchify(x)  # (N, out_channels, H, W)
        return x

    def initialize_weights(self):
        # Initialize transformer layers:
        def _basic_init(module):
            if isinstance(module, nn.Linear):
                torch.nn.init.xavier_uniform_(module.weight)
                if module.bias is not None:
                    nn.init.constant_(module.bias, 0)

        self.apply(_basic_init)

        if self.use_pe:
            # Initialize (and freeze) pos_embed by sin-cos embedding:
            pos_embed = get_2d_sincos_pos_embed(
                self.pos_embed.shape[-1],
                int(self.x_embedder.num_patches**0.5),
                pe_interpolation=self.pe_interpolation,
                base_size=self.base_size,
            )
            self.pos_embed.data.copy_(torch.from_numpy(pos_embed).float().unsqueeze(0))

        # Initialize patch_embed like nn.Linear (instead of nn.Conv2d):
        w = self.x_embedder.proj.weight.data
        nn.init.xavier_uniform_(w.view([w.shape[0], -1]))

        # Initialize timestep embedding MLP:
        nn.init.normal_(self.t_embedder.mlp[0].weight, std=0.02)
        nn.init.normal_(self.t_embedder.mlp[2].weight, std=0.02)
        nn.init.normal_(self.t_block[1].weight, std=0.02)

        # Initialize caption embedding MLP:
        nn.init.normal_(self.y_embedder.y_proj.fc1.weight, std=0.02)
        nn.init.normal_(self.y_embedder.y_proj.fc2.weight, std=0.02)

    @property
    def dtype(self):
        return next(self.parameters()).dtype


#################################################################################
#                                 SanaU Configs                                 #
#################################################################################
@MODELS.register_module()
def SanaMSU_600M_P1_D28(**kwargs):
    return SanaU(depth=28, hidden_size=1152, patch_size=1, num_heads=16, **kwargs)


@MODELS.register_module()
def SanaMSU_600M_P2_D28(**kwargs):
    return SanaU(depth=28, hidden_size=1152, patch_size=2, num_heads=16, **kwargs)


@MODELS.register_module()
def SanaMSU_600M_P4_D28(**kwargs):
    return SanaU(depth=28, hidden_size=1152, patch_size=4, num_heads=16, **kwargs)


@MODELS.register_module()
def SanaMSU_P1_D20(**kwargs):
    # 20 layers, 1648.48M
    return SanaU(depth=20, hidden_size=2240, patch_size=1, num_heads=20, **kwargs)


@MODELS.register_module()
def SanaMSU_P2_D20(**kwargs):
    # 28 layers, 1648.48M
    return SanaU(depth=20, hidden_size=2240, patch_size=2, num_heads=20, **kwargs)