File size: 8,937 Bytes
69e1a8d
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
# Copyright 2025 The HuggingFace Team. All rights reserved.
#
# 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.

from dataclasses import dataclass
from typing import Any

import torch
from torch import nn

from ...configuration_utils import ConfigMixin, register_to_config
from ...loaders import PeftAdapterMixin
from ...utils import BaseOutput, apply_lora_scale, logging
from ..attention import AttentionMixin
from ..embeddings import PatchEmbed, PixArtAlphaTextProjection
from ..modeling_outputs import Transformer2DModelOutput
from ..modeling_utils import ModelMixin
from ..normalization import AdaLayerNormSingle, RMSNorm
from ..transformers.sana_transformer import SanaTransformerBlock
from .controlnet import zero_module


logger = logging.get_logger(__name__)  # pylint: disable=invalid-name


@dataclass
class SanaControlNetOutput(BaseOutput):
    controlnet_block_samples: tuple[torch.Tensor]


class SanaControlNetModel(ModelMixin, AttentionMixin, ConfigMixin, PeftAdapterMixin):
    _supports_gradient_checkpointing = True
    _no_split_modules = ["SanaTransformerBlock", "PatchEmbed"]
    _skip_layerwise_casting_patterns = ["patch_embed", "norm"]

    @register_to_config
    def __init__(
        self,
        in_channels: int = 32,
        out_channels: int | None = 32,
        num_attention_heads: int = 70,
        attention_head_dim: int = 32,
        num_layers: int = 7,
        num_cross_attention_heads: int | None = 20,
        cross_attention_head_dim: int | None = 112,
        cross_attention_dim: int | None = 2240,
        caption_channels: int = 2304,
        mlp_ratio: float = 2.5,
        dropout: float = 0.0,
        attention_bias: bool = False,
        sample_size: int = 32,
        patch_size: int = 1,
        norm_elementwise_affine: bool = False,
        norm_eps: float = 1e-6,
        interpolation_scale: int | None = None,
    ) -> None:
        super().__init__()

        out_channels = out_channels or in_channels
        inner_dim = num_attention_heads * attention_head_dim

        # 1. Patch Embedding
        self.patch_embed = PatchEmbed(
            height=sample_size,
            width=sample_size,
            patch_size=patch_size,
            in_channels=in_channels,
            embed_dim=inner_dim,
            interpolation_scale=interpolation_scale,
            pos_embed_type="sincos" if interpolation_scale is not None else None,
        )

        # 2. Additional condition embeddings
        self.time_embed = AdaLayerNormSingle(inner_dim)

        self.caption_projection = PixArtAlphaTextProjection(in_features=caption_channels, hidden_size=inner_dim)
        self.caption_norm = RMSNorm(inner_dim, eps=1e-5, elementwise_affine=True)

        # 3. Transformer blocks
        self.transformer_blocks = nn.ModuleList(
            [
                SanaTransformerBlock(
                    inner_dim,
                    num_attention_heads,
                    attention_head_dim,
                    dropout=dropout,
                    num_cross_attention_heads=num_cross_attention_heads,
                    cross_attention_head_dim=cross_attention_head_dim,
                    cross_attention_dim=cross_attention_dim,
                    attention_bias=attention_bias,
                    norm_elementwise_affine=norm_elementwise_affine,
                    norm_eps=norm_eps,
                    mlp_ratio=mlp_ratio,
                )
                for _ in range(num_layers)
            ]
        )

        # controlnet_blocks
        self.controlnet_blocks = nn.ModuleList([])

        self.input_block = zero_module(nn.Linear(inner_dim, inner_dim))
        for _ in range(len(self.transformer_blocks)):
            controlnet_block = nn.Linear(inner_dim, inner_dim)
            controlnet_block = zero_module(controlnet_block)
            self.controlnet_blocks.append(controlnet_block)

        self.gradient_checkpointing = False

    @apply_lora_scale("attention_kwargs")
    def forward(
        self,
        hidden_states: torch.Tensor,
        encoder_hidden_states: torch.Tensor,
        timestep: torch.LongTensor,
        controlnet_cond: torch.Tensor,
        conditioning_scale: float = 1.0,
        encoder_attention_mask: torch.Tensor | None = None,
        attention_mask: torch.Tensor | None = None,
        attention_kwargs: dict[str, Any] | None = None,
        return_dict: bool = True,
    ) -> tuple[torch.Tensor, ...] | Transformer2DModelOutput:
        # ensure attention_mask is a bias, and give it a singleton query_tokens dimension.
        #   we may have done this conversion already, e.g. if we came here via UNet2DConditionModel#forward.
        #   we can tell by counting dims; if ndim == 2: it's a mask rather than a bias.
        # expects mask of shape:
        #   [batch, key_tokens]
        # adds singleton query_tokens dimension:
        #   [batch,                    1, key_tokens]
        # this helps to broadcast it as a bias over attention scores, which will be in one of the following shapes:
        #   [batch,  heads, query_tokens, key_tokens] (e.g. torch sdp attn)
        #   [batch * heads, query_tokens, key_tokens] (e.g. xformers or classic attn)
        if attention_mask is not None and attention_mask.ndim == 2:
            # assume that mask is expressed as:
            #   (1 = keep,      0 = discard)
            # convert mask into a bias that can be added to attention scores:
            #       (keep = +0,     discard = -10000.0)
            attention_mask = (1 - attention_mask.to(hidden_states.dtype)) * -10000.0
            attention_mask = attention_mask.unsqueeze(1)

        # convert encoder_attention_mask to a bias the same way we do for attention_mask
        if encoder_attention_mask is not None and encoder_attention_mask.ndim == 2:
            encoder_attention_mask = (1 - encoder_attention_mask.to(hidden_states.dtype)) * -10000.0
            encoder_attention_mask = encoder_attention_mask.unsqueeze(1)

        # 1. Input
        batch_size, num_channels, height, width = hidden_states.shape
        p = self.config.patch_size
        post_patch_height, post_patch_width = height // p, width // p

        hidden_states = self.patch_embed(hidden_states)
        hidden_states = hidden_states + self.input_block(self.patch_embed(controlnet_cond.to(hidden_states.dtype)))

        timestep, embedded_timestep = self.time_embed(
            timestep, batch_size=batch_size, hidden_dtype=hidden_states.dtype
        )

        encoder_hidden_states = self.caption_projection(encoder_hidden_states)
        encoder_hidden_states = encoder_hidden_states.view(batch_size, -1, hidden_states.shape[-1])

        encoder_hidden_states = self.caption_norm(encoder_hidden_states)

        # 2. Transformer blocks
        block_res_samples = ()
        if torch.is_grad_enabled() and self.gradient_checkpointing:
            for block in self.transformer_blocks:
                hidden_states = self._gradient_checkpointing_func(
                    block,
                    hidden_states,
                    attention_mask,
                    encoder_hidden_states,
                    encoder_attention_mask,
                    timestep,
                    post_patch_height,
                    post_patch_width,
                )
                block_res_samples = block_res_samples + (hidden_states,)
        else:
            for block in self.transformer_blocks:
                hidden_states = block(
                    hidden_states,
                    attention_mask,
                    encoder_hidden_states,
                    encoder_attention_mask,
                    timestep,
                    post_patch_height,
                    post_patch_width,
                )
                block_res_samples = block_res_samples + (hidden_states,)

        # 3. ControlNet blocks
        controlnet_block_res_samples = ()
        for block_res_sample, controlnet_block in zip(block_res_samples, self.controlnet_blocks):
            block_res_sample = controlnet_block(block_res_sample)
            controlnet_block_res_samples = controlnet_block_res_samples + (block_res_sample,)

        controlnet_block_res_samples = [sample * conditioning_scale for sample in controlnet_block_res_samples]

        if not return_dict:
            return (controlnet_block_res_samples,)

        return SanaControlNetOutput(controlnet_block_samples=controlnet_block_res_samples)