From 293c480d4cf0bed442e02395dfd012c686d2ca59 Mon Sep 17 00:00:00 2001
From: yiyixu <yixu310@gmail.com>
Date: Mon, 27 Nov 2023 09:18:47 +0000
Subject: [PATCH 01/20]  remove kandinsky specific attention and attention
 processor

---
 src/diffusers/models/attention_processor.py | 49 ++-------------
 src/diffusers/models/unet_kandi3.py         | 68 ++++-----------------
 2 files changed, 18 insertions(+), 99 deletions(-)

diff --git a/src/diffusers/models/attention_processor.py b/src/diffusers/models/attention_processor.py
index 21eb3a32dc09..40a335527ace 100644
--- a/src/diffusers/models/attention_processor.py
+++ b/src/diffusers/models/attention_processor.py
@@ -16,7 +16,7 @@
 
 import torch
 import torch.nn.functional as F
-from torch import einsum, nn
+from torch import nn
 
 from ..utils import USE_PEFT_BACKEND, deprecate, logging
 from ..utils.import_utils import is_xformers_available
@@ -109,15 +109,17 @@ def __init__(
         residual_connection: bool = False,
         _from_deprecated_attn_block: bool = False,
         processor: Optional["AttnProcessor"] = None,
+        out_dim: int = None,
     ):
         super().__init__()
-        self.inner_dim = dim_head * heads
+        self.inner_dim = out_dim if out_dim is not None else dim_head * heads
         self.cross_attention_dim = cross_attention_dim if cross_attention_dim is not None else query_dim
         self.upcast_attention = upcast_attention
         self.upcast_softmax = upcast_softmax
         self.rescale_output_factor = rescale_output_factor
         self.residual_connection = residual_connection
         self.dropout = dropout
+        self.out_dim = out_dim if out_dim is not None else query_dim
 
         # we make use of this private variable to know whether this class is loaded
         # with an deprecated state dict so that we can convert it on the fly
@@ -126,7 +128,7 @@ def __init__(
         self.scale_qk = scale_qk
         self.scale = dim_head**-0.5 if self.scale_qk else 1.0
 
-        self.heads = heads
+        self.heads = out_dim // dim_head if out_dim is not None else heads
         # for slice_size > 0 the attention score computation
         # is split across the batch axis to save memory
         # You can set slice_size with `set_attention_slice`
@@ -193,7 +195,7 @@ def __init__(
             self.add_v_proj = linear_cls(added_kv_proj_dim, self.inner_dim)
 
         self.to_out = nn.ModuleList([])
-        self.to_out.append(linear_cls(self.inner_dim, query_dim, bias=out_bias))
+        self.to_out.append(linear_cls(self.inner_dim, self.out_dim, bias=out_bias))
         self.to_out.append(nn.Dropout(dropout))
 
         # set attention processor
@@ -2219,44 +2221,6 @@ def __call__(
         return hidden_states
 
 
-# TODO(Yiyi): This class should not exist, we can replace it with a normal attention processor I believe
-# this way torch.compile and co. will work as well
-class Kandi3AttnProcessor:
-    r"""
-    Default kandinsky3 proccesor for performing attention-related computations.
-    """
-
-    @staticmethod
-    def _reshape(hid_states, h):
-        b, n, f = hid_states.shape
-        d = f // h
-        return hid_states.unsqueeze(-1).reshape(b, n, h, d).permute(0, 2, 1, 3)
-
-    def __call__(
-        self,
-        attn,
-        x,
-        context,
-        context_mask=None,
-    ):
-        query = self._reshape(attn.to_q(x), h=attn.num_heads)
-        key = self._reshape(attn.to_k(context), h=attn.num_heads)
-        value = self._reshape(attn.to_v(context), h=attn.num_heads)
-
-        attention_matrix = einsum("b h i d, b h j d -> b h i j", query, key)
-
-        if context_mask is not None:
-            max_neg_value = -torch.finfo(attention_matrix.dtype).max
-            context_mask = context_mask.unsqueeze(1).unsqueeze(1)
-            attention_matrix = attention_matrix.masked_fill(~(context_mask != 0), max_neg_value)
-        attention_matrix = (attention_matrix * attn.scale).softmax(dim=-1)
-
-        out = einsum("b h i j, b h j d -> b h i d", attention_matrix, value)
-        out = out.permute(0, 2, 1, 3).reshape(out.shape[0], out.shape[2], -1)
-        out = attn.to_out[0](out)
-        return out
-
-
 LORA_ATTENTION_PROCESSORS = (
     LoRAAttnProcessor,
     LoRAAttnProcessor2_0,
@@ -2282,7 +2246,6 @@ def __call__(
     LoRAXFormersAttnProcessor,
     IPAdapterAttnProcessor,
     IPAdapterAttnProcessor2_0,
-    Kandi3AttnProcessor,
 )
 
 AttentionProcessor = Union[
diff --git a/src/diffusers/models/unet_kandi3.py b/src/diffusers/models/unet_kandi3.py
index 42e25a942f7d..036d2b2195e1 100644
--- a/src/diffusers/models/unet_kandi3.py
+++ b/src/diffusers/models/unet_kandi3.py
@@ -3,13 +3,12 @@
 from typing import Dict, Tuple, Union
 
 import torch
-import torch.nn.functional as F
 import torch.utils.checkpoint
 from torch import nn
 
 from ..configuration_utils import ConfigMixin, register_to_config
 from ..utils import BaseOutput, logging
-from .attention_processor import AttentionProcessor, Kandi3AttnProcessor
+from .attention_processor import Attention, AttentionProcessor, AttnProcessor
 from .embeddings import TimestepEmbedding
 from .modeling_utils import ModelMixin
 
@@ -30,7 +29,7 @@ def set_default_item(condition, item_1, item_2=None):
         return item_2
 
 
-# TODO(Yiyi): This class needs to be removed
+# TODO(Yiyi): This class needs to be removed: either layer_1 or nn.identity
 def set_default_layer(condition, layer_1, args_1=[], kwargs_1={}, layer_2=torch.nn.Identity, args_2=[], kwargs_2={}):
     if condition:
         return layer_1(*args_1, **kwargs_1)
@@ -223,7 +222,7 @@ def set_default_attn_processor(self):
         """
         Disables custom attention processors and sets the default attention implementation.
         """
-        self.set_attn_processor(Kandi3AttnProcessor())
+        self.set_attn_processor(AttnProcessor())
 
     def _set_gradient_checkpointing(self, module, value=False):
         if hasattr(module, "gradient_checkpointing"):
@@ -412,6 +411,7 @@ def forward(self, x, time_embed, context=None, context_mask=None, image_mask=Non
         return x
 
 
+# yiyi notes: should not have a seperate class here either
 class Kandinsky3ConditionalGroupNorm(nn.Module):
     def __init__(self, groups, normalized_shape, context_dim):
         super().__init__()
@@ -431,49 +431,6 @@ def forward(self, x, context):
         return x
 
 
-# TODO(Yiyi): This class should ideally not even exist, it slows everything needlessly down. I'm pretty
-# sure we can delete it and instead just pass an attention_mask
-class Attention(nn.Module):
-    def __init__(self, in_channels, out_channels, context_dim, head_dim=64):
-        super().__init__()
-        assert out_channels % head_dim == 0
-        self.num_heads = out_channels // head_dim
-        self.scale = head_dim**-0.5
-
-        # to_q
-        self.to_q = nn.Linear(in_channels, out_channels, bias=False)
-        # to_k
-        self.to_k = nn.Linear(context_dim, out_channels, bias=False)
-        # to_v
-        self.to_v = nn.Linear(context_dim, out_channels, bias=False)
-        processor = Kandi3AttnProcessor()
-        self.set_processor(processor)
-        # to_out
-        self.to_out = nn.ModuleList([])
-        self.to_out.append(nn.Linear(out_channels, out_channels, bias=False))
-
-    def set_processor(self, processor: "AttnProcessor"):  # noqa: F821
-        # if current processor is in `self._modules` and if passed `processor` is not, we need to
-        # pop `processor` from `self._modules`
-        if (
-            hasattr(self, "processor")
-            and isinstance(self.processor, torch.nn.Module)
-            and not isinstance(processor, torch.nn.Module)
-        ):
-            logger.info(f"You are removing possibly trained weights of {self.processor} with {processor}")
-            self._modules.pop("processor")
-
-        self.processor = processor
-
-    def forward(self, x, context, context_mask=None, image_mask=None):
-        return self.processor(
-            self,
-            x,
-            context=context,
-            context_mask=context_mask,
-        )
-
-
 class Kandinsky3Block(nn.Module):
     def __init__(self, in_channels, out_channels, time_embed_dim, kernel_size=3, norm_groups=32, up_resolution=None):
         super().__init__()
@@ -546,9 +503,10 @@ def forward(self, x, time_embed):
 class Kandinsky3AttentionPooling(nn.Module):
     def __init__(self, num_channels, context_dim, head_dim=64):
         super().__init__()
-        self.attention = Attention(context_dim, num_channels, context_dim, head_dim)
+        self.attention = Attention(context_dim, context_dim, dim_head=head_dim, out_dim=num_channels, out_bias=False)
 
     def forward(self, x, context, context_mask=None):
+        context_mask = context_mask.unsqueeze(1).to(dtype=context.dtype)
         context = self.attention(context.mean(dim=1, keepdim=True), context, context_mask)
         return x + context.squeeze(1)
 
@@ -557,7 +515,9 @@ class Kandinsky3AttentionBlock(nn.Module):
     def __init__(self, num_channels, time_embed_dim, context_dim=None, norm_groups=32, head_dim=64, expansion_ratio=4):
         super().__init__()
         self.in_norm = Kandinsky3ConditionalGroupNorm(norm_groups, num_channels, time_embed_dim)
-        self.attention = Attention(num_channels, num_channels, context_dim or num_channels, head_dim)
+        self.attention = Attention(
+            num_channels, context_dim or num_channels, dim_head=head_dim, out_dim=num_channels, out_bias=False
+        )
 
         hidden_channels = expansion_ratio * num_channels
         self.out_norm = Kandinsky3ConditionalGroupNorm(norm_groups, num_channels, time_embed_dim)
@@ -572,14 +532,10 @@ def forward(self, x, time_embed, context=None, context_mask=None, image_mask=Non
         out = self.in_norm(x, time_embed)
         out = out.reshape(x.shape[0], -1, height * width).permute(0, 2, 1)
         context = context if context is not None else out
+        if context_mask is not None:
+            context_mask = context_mask.unsqueeze(1).to(dtype=context.dtype)
 
-        if image_mask is not None:
-            mask_height, mask_width = image_mask.shape[-2:]
-            kernel_size = (mask_height // height, mask_width // width)
-            image_mask = F.max_pool2d(image_mask, kernel_size, kernel_size)
-            image_mask = image_mask.reshape(image_mask.shape[0], -1)
-
-        out = self.attention(out, context, context_mask, image_mask)
+        out = self.attention(out, context, context_mask)
         out = out.permute(0, 2, 1).unsqueeze(-1).reshape(out.shape[0], -1, height, width)
         x = x + out
 

From eca5c18542d721619e404f6461b553feeb379adc Mon Sep 17 00:00:00 2001
From: yiyixu <yixu310@gmail.com>
Date: Tue, 28 Nov 2023 01:10:26 +0000
Subject: [PATCH 02/20] remove set_default_layer and set_default_item

---
 src/diffusers/models/unet_kandi3.py | 118 ++++++++++++----------------
 1 file changed, 52 insertions(+), 66 deletions(-)

diff --git a/src/diffusers/models/unet_kandi3.py b/src/diffusers/models/unet_kandi3.py
index 036d2b2195e1..355ce47d624c 100644
--- a/src/diffusers/models/unet_kandi3.py
+++ b/src/diffusers/models/unet_kandi3.py
@@ -21,22 +21,6 @@ class Kandinsky3UNetOutput(BaseOutput):
     sample: torch.FloatTensor = None
 
 
-# TODO(Yiyi): This class needs to be removed
-def set_default_item(condition, item_1, item_2=None):
-    if condition:
-        return item_1
-    else:
-        return item_2
-
-
-# TODO(Yiyi): This class needs to be removed: either layer_1 or nn.identity
-def set_default_layer(condition, layer_1, args_1=[], kwargs_1={}, layer_2=torch.nn.Identity, args_2=[], kwargs_2={}):
-    if condition:
-        return layer_1(*args_1, **kwargs_1)
-    else:
-        return layer_2(*args_2, **kwargs_2)
-
-
 # TODO(Yiyi): This class should be removed and be replaced by Timesteps
 class SinusoidalPosEmb(nn.Module):
     def __init__(self, dim):
@@ -105,7 +89,7 @@ def __init__(
 
         hidden_dims = [init_channels] + list(block_out_channels)
         in_out_dims = list(zip(hidden_dims[:-1], hidden_dims[1:]))
-        text_dims = [set_default_item(is_exist, cross_attention_dim) for is_exist in add_cross_attention]
+        text_dims = [cross_attention_dim if is_exist else None for is_exist in add_cross_attention]
         num_blocks = len(block_out_channels) * [layers_per_block]
         layer_params = [num_blocks, text_dims, add_self_attention]
         rev_layer_params = map(reversed, layer_params)
@@ -117,7 +101,7 @@ def __init__(
             zip(in_out_dims, *layer_params)
         ):
             down_sample = level != (self.num_levels - 1)
-            cat_dims.append(set_default_item(level != (self.num_levels - 1), out_dim, 0))
+            cat_dims.append(out_dim if level != (self.num_levels - 1) else 0)
             self.down_blocks.append(
                 Kandinsky3DownSampleBlock(
                     in_dim,
@@ -289,7 +273,7 @@ def __init__(
         self_attention=True,
     ):
         super().__init__()
-        up_resolutions = [[None, set_default_item(up_sample, True), None, None]] + [[None] * 4] * (num_blocks - 1)
+        up_resolutions = [[None, True if up_sample else None, None, None]] + [[None] * 4] * (num_blocks - 1)
         hidden_channels = (
             [(in_channels + cat_dim, in_channels)]
             + [(in_channels, in_channels)] * (num_blocks - 2)
@@ -302,27 +286,27 @@ def __init__(
         self.self_attention = self_attention
         self.context_dim = context_dim
 
-        attentions.append(
-            set_default_layer(
-                self_attention,
-                Kandinsky3AttentionBlock,
-                (out_channels, time_embed_dim, None, groups, head_dim, expansion_ratio),
-                layer_2=nn.Identity,
+        if self_attention:
+            attentions.append(
+                Kandinsky3AttentionBlock(out_channels, time_embed_dim, None, groups, head_dim, expansion_ratio)
             )
-        )
+        else:
+            attentions.append(nn.Identity())
 
         for (in_channel, out_channel), up_resolution in zip(hidden_channels, up_resolutions):
             resnets_in.append(
                 Kandinsky3ResNetBlock(in_channel, in_channel, time_embed_dim, groups, compression_ratio, up_resolution)
             )
-            attentions.append(
-                set_default_layer(
-                    context_dim is not None,
-                    Kandinsky3AttentionBlock,
-                    (in_channel, time_embed_dim, context_dim, groups, head_dim, expansion_ratio),
-                    layer_2=nn.Identity,
+
+            if context_dim is not None:
+                attentions.append(
+                    Kandinsky3AttentionBlock(
+                        in_channel, time_embed_dim, context_dim, groups, head_dim, expansion_ratio
+                    )
                 )
-            )
+            else:
+                attentions.append(nn.Identity())
+
             resnets_out.append(
                 Kandinsky3ResNetBlock(in_channel, out_channel, time_embed_dim, groups, compression_ratio)
             )
@@ -366,29 +350,29 @@ def __init__(
         self.self_attention = self_attention
         self.context_dim = context_dim
 
-        attentions.append(
-            set_default_layer(
-                self_attention,
-                Kandinsky3AttentionBlock,
-                (in_channels, time_embed_dim, None, groups, head_dim, expansion_ratio),
-                layer_2=nn.Identity,
+        if self_attention:
+            attentions.append(
+                Kandinsky3AttentionBlock(in_channels, time_embed_dim, None, groups, head_dim, expansion_ratio)
             )
-        )
+        else:
+            attentions.append(nn.Identity())
 
-        up_resolutions = [[None] * 4] * (num_blocks - 1) + [[None, None, set_default_item(down_sample, False), None]]
+        up_resolutions = [[None] * 4] * (num_blocks - 1) + [[None, None, False if down_sample else None, None]]
         hidden_channels = [(in_channels, out_channels)] + [(out_channels, out_channels)] * (num_blocks - 1)
         for (in_channel, out_channel), up_resolution in zip(hidden_channels, up_resolutions):
             resnets_in.append(
                 Kandinsky3ResNetBlock(in_channel, out_channel, time_embed_dim, groups, compression_ratio)
             )
-            attentions.append(
-                set_default_layer(
-                    context_dim is not None,
-                    Kandinsky3AttentionBlock,
-                    (out_channel, time_embed_dim, context_dim, groups, head_dim, expansion_ratio),
-                    layer_2=nn.Identity,
+
+            if context_dim is not None:
+                attentions.append(
+                    Kandinsky3AttentionBlock(
+                        out_channel, time_embed_dim, context_dim, groups, head_dim, expansion_ratio
+                    )
                 )
-            )
+            else:
+                attentions.append(nn.Identity())
+
             resnets_out.append(
                 Kandinsky3ResNetBlock(
                     out_channel, out_channel, time_embed_dim, groups, compression_ratio, up_resolution
@@ -436,20 +420,18 @@ def __init__(self, in_channels, out_channels, time_embed_dim, kernel_size=3, nor
         super().__init__()
         self.group_norm = Kandinsky3ConditionalGroupNorm(norm_groups, in_channels, time_embed_dim)
         self.activation = nn.SiLU()
-        self.up_sample = set_default_layer(
-            up_resolution is not None and up_resolution,
-            nn.ConvTranspose2d,
-            (in_channels, in_channels),
-            {"kernel_size": 2, "stride": 2},
-        )
+        if up_resolution is not None and up_resolution:
+            self.up_sample = nn.ConvTranspose2d(in_channels, in_channels, kernel_size=2, stride=2)
+        else:
+            self.up_sample = nn.Identity()
+
         padding = int(kernel_size > 1)
         self.projection = nn.Conv2d(in_channels, out_channels, kernel_size=kernel_size, padding=padding)
-        self.down_sample = set_default_layer(
-            up_resolution is not None and not up_resolution,
-            nn.Conv2d,
-            (out_channels, out_channels),
-            {"kernel_size": 2, "stride": 2},
-        )
+
+        if up_resolution is not None and not up_resolution:
+            self.down_sample = nn.Conv2d(out_channels, out_channels, kernel_size=2, stride=2)
+        else:
+            self.down_sample = nn.Identity()
 
     def forward(self, x, time_embed):
         x = self.group_norm(x, time_embed)
@@ -478,14 +460,18 @@ def __init__(
                 )
             ]
         )
-        self.shortcut_up_sample = set_default_layer(
-            True in up_resolutions, nn.ConvTranspose2d, (in_channels, in_channels), {"kernel_size": 2, "stride": 2}
+        self.shortcut_up_sample = (
+            nn.ConvTranspose2d(in_channels, in_channels, kernel_size=2, stride=2)
+            if True in up_resolutions
+            else nn.Identity()
         )
-        self.shortcut_projection = set_default_layer(
-            in_channels != out_channels, nn.Conv2d, (in_channels, out_channels), {"kernel_size": 1}
+        self.shortcut_projection = (
+            nn.Conv2d(in_channels, out_channels, kernel_size=1) if in_channels != out_channels else nn.Identity()
         )
-        self.shortcut_down_sample = set_default_layer(
-            False in up_resolutions, nn.Conv2d, (out_channels, out_channels), {"kernel_size": 2, "stride": 2}
+        self.shortcut_down_sample = (
+            nn.Conv2d(out_channels, out_channels, kernel_size=2, stride=2)
+            if False in up_resolutions
+            else nn.Identity()
         )
 
     def forward(self, x, time_embed):

From b8bb288001b14c08f7d90b4dccf614ef8c8e0245 Mon Sep 17 00:00:00 2001
From: yiyixu <yixu310@gmail.com>
Date: Tue, 28 Nov 2023 01:55:23 +0000
Subject: [PATCH 03/20] remove SinusoidalPosEmb

---
 src/diffusers/models/unet_kandi3.py | 21 ++-------------------
 1 file changed, 2 insertions(+), 19 deletions(-)

diff --git a/src/diffusers/models/unet_kandi3.py b/src/diffusers/models/unet_kandi3.py
index 355ce47d624c..257e9b2dd7a3 100644
--- a/src/diffusers/models/unet_kandi3.py
+++ b/src/diffusers/models/unet_kandi3.py
@@ -1,4 +1,3 @@
-import math
 from dataclasses import dataclass
 from typing import Dict, Tuple, Union
 
@@ -9,7 +8,7 @@
 from ..configuration_utils import ConfigMixin, register_to_config
 from ..utils import BaseOutput, logging
 from .attention_processor import Attention, AttentionProcessor, AttnProcessor
-from .embeddings import TimestepEmbedding
+from .embeddings import TimestepEmbedding, Timesteps
 from .modeling_utils import ModelMixin
 
 
@@ -21,20 +20,6 @@ class Kandinsky3UNetOutput(BaseOutput):
     sample: torch.FloatTensor = None
 
 
-# TODO(Yiyi): This class should be removed and be replaced by Timesteps
-class SinusoidalPosEmb(nn.Module):
-    def __init__(self, dim):
-        super().__init__()
-        self.dim = dim
-
-    def forward(self, x, type_tensor=None):
-        half_dim = self.dim // 2
-        emb = math.log(10000) / (half_dim - 1)
-        emb = torch.exp(torch.arange(half_dim, device=x.device) * -emb)
-        emb = x[:, None] * emb[None, :]
-        return torch.cat((emb.sin(), emb.cos()), dim=-1)
-
-
 class Kandinsky3EncoderProj(nn.Module):
     def __init__(self, encoder_hid_dim, cross_attention_dim):
         super().__init__()
@@ -70,9 +55,7 @@ def __init__(
 
         out_channels = in_channels
         init_channels = block_out_channels[0] // 2
-        # TODO(Yiyi): Should be replaced with Timesteps class -> make sure that results are the same
-        # self.time_proj = Timesteps(init_channels, flip_sin_to_cos=False, downscale_freq_shift=1)
-        self.time_proj = SinusoidalPosEmb(init_channels)
+        self.time_proj = Timesteps(init_channels, flip_sin_to_cos=False, downscale_freq_shift=1)
 
         self.time_embedding = TimestepEmbedding(
             init_channels,

From 84ce3d67d154db2a7454ae3265d83520f8e53f90 Mon Sep 17 00:00:00 2001
From: yiyixu <yixu310@gmail.com>
Date: Tue, 28 Nov 2023 04:32:56 +0000
Subject: [PATCH 04/20] more

---
 src/diffusers/models/unet_kandi3.py | 35 ++++++++++++-----------------
 1 file changed, 14 insertions(+), 21 deletions(-)

diff --git a/src/diffusers/models/unet_kandi3.py b/src/diffusers/models/unet_kandi3.py
index 257e9b2dd7a3..3eea77ccb841 100644
--- a/src/diffusers/models/unet_kandi3.py
+++ b/src/diffusers/models/unet_kandi3.py
@@ -196,12 +196,6 @@ def _set_gradient_checkpointing(self, module, value=False):
             module.gradient_checkpointing = value
 
     def forward(self, sample, timestep, encoder_hidden_states=None, encoder_attention_mask=None, return_dict=True):
-        # TODO(Yiyi): Clean up the following variables - these names should not be used
-        # but instead only the ones that we pass to forward
-        x = sample
-        context_mask = encoder_attention_mask
-        context = encoder_hidden_states
-
         if not torch.is_tensor(timestep):
             dtype = torch.float32 if isinstance(timestep, float) else torch.int32
             timestep = torch.tensor([timestep], dtype=dtype, device=sample.device)
@@ -210,33 +204,33 @@ def forward(self, sample, timestep, encoder_hidden_states=None, encoder_attentio
 
         # broadcast to batch dimension in a way that's compatible with ONNX/Core ML
         timestep = timestep.expand(sample.shape[0])
-        time_embed_input = self.time_proj(timestep).to(x.dtype)
+        time_embed_input = self.time_proj(timestep).to(sample.dtype)
         time_embed = self.time_embedding(time_embed_input)
 
-        context = self.encoder_hid_proj(context)
+        encoder_hidden_states = self.encoder_hid_proj(encoder_hidden_states)
 
-        if context is not None:
-            time_embed = self.add_time_condition(time_embed, context, context_mask)
+        if encoder_hidden_states is not None:
+            time_embed = self.add_time_condition(time_embed, encoder_hidden_states, encoder_attention_mask)
 
         hidden_states = []
-        x = self.conv_in(x)
+        sample = self.conv_in(sample)
         for level, down_sample in enumerate(self.down_blocks):
-            x = down_sample(x, time_embed, context, context_mask)
+            sample = down_sample(sample, time_embed, encoder_hidden_states, encoder_attention_mask)
             if level != self.num_levels - 1:
-                hidden_states.append(x)
+                hidden_states.append(sample)
 
         for level, up_sample in enumerate(self.up_blocks):
             if level != 0:
-                x = torch.cat([x, hidden_states.pop()], dim=1)
-            x = up_sample(x, time_embed, context, context_mask)
+                sample = torch.cat([sample, hidden_states.pop()], dim=1)
+            sample = up_sample(sample, time_embed, encoder_hidden_states, encoder_attention_mask)
 
-        x = self.conv_norm_out(x)
-        x = self.conv_act_out(x)
-        x = self.conv_out(x)
+        sample = self.conv_norm_out(sample)
+        sample = self.conv_act_out(sample)
+        sample = self.conv_out(sample)
 
         if not return_dict:
-            return (x,)
-        return Kandinsky3UNetOutput(sample=x)
+            return (sample,)
+        return Kandinsky3UNetOutput(sample=sample)
 
 
 class Kandinsky3UpSampleBlock(nn.Module):
@@ -378,7 +372,6 @@ def forward(self, x, time_embed, context=None, context_mask=None, image_mask=Non
         return x
 
 
-# yiyi notes: should not have a seperate class here either
 class Kandinsky3ConditionalGroupNorm(nn.Module):
     def __init__(self, groups, normalized_shape, context_dim):
         super().__init__()

From 123dafcdea49c154a2dda7f7086d18bd981c6fb8 Mon Sep 17 00:00:00 2001
From: yiyixu <yixu310@gmail.com>
Date: Wed, 29 Nov 2023 03:28:04 +0000
Subject: [PATCH 05/20] disable batch test

---
 tests/pipelines/kandinsky3/test_kandinsky3.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tests/pipelines/kandinsky3/test_kandinsky3.py b/tests/pipelines/kandinsky3/test_kandinsky3.py
index 65297a36b157..82eed67f39d1 100644
--- a/tests/pipelines/kandinsky3/test_kandinsky3.py
+++ b/tests/pipelines/kandinsky3/test_kandinsky3.py
@@ -163,7 +163,7 @@ def test_float16_inference(self):
         super().test_float16_inference(expected_max_diff=1e-1)
 
     def test_inference_batch_single_identical(self):
-        super().test_inference_batch_single_identical(expected_max_diff=1e-2)
+        pass
 
     def test_model_cpu_offload_forward_pass(self):
         # TODO(Yiyi) - this test should work, skipped for time reasons for now

From 145ddad6889f835e68368df88f01296811104e46 Mon Sep 17 00:00:00 2001
From: yiyixu <yixu310@gmail.com>
Date: Wed, 29 Nov 2023 04:36:56 +0000
Subject: [PATCH 06/20] fix cpu model offload

---
 src/diffusers/pipelines/kandinsky3/kandinsky3_pipeline.py | 4 +++-
 tests/pipelines/kandinsky3/test_kandinsky3.py             | 4 ----
 2 files changed, 3 insertions(+), 5 deletions(-)

diff --git a/src/diffusers/pipelines/kandinsky3/kandinsky3_pipeline.py b/src/diffusers/pipelines/kandinsky3/kandinsky3_pipeline.py
index f116fb7894f0..a6a0e01084de 100644
--- a/src/diffusers/pipelines/kandinsky3/kandinsky3_pipeline.py
+++ b/src/diffusers/pipelines/kandinsky3/kandinsky3_pipeline.py
@@ -50,7 +50,7 @@ def remove_all_hooks(self):
         else:
             raise ImportError("Please install accelerate via `pip install accelerate`")
 
-        for model in [self.text_encoder, self.unet]:
+        for model in [self.text_encoder, self.unet, self.movq]:
             if model is not None:
                 remove_hook_from_module(model, recurse=True)
 
@@ -446,6 +446,8 @@ def __call__(
             if output_type == "pil":
                 image = self.numpy_to_pil(image)
 
+            self.maybe_free_model_hooks()
+
             if not return_dict:
                 return (image,)
 
diff --git a/tests/pipelines/kandinsky3/test_kandinsky3.py b/tests/pipelines/kandinsky3/test_kandinsky3.py
index 82eed67f39d1..7b5691409383 100644
--- a/tests/pipelines/kandinsky3/test_kandinsky3.py
+++ b/tests/pipelines/kandinsky3/test_kandinsky3.py
@@ -165,10 +165,6 @@ def test_float16_inference(self):
     def test_inference_batch_single_identical(self):
         pass
 
-    def test_model_cpu_offload_forward_pass(self):
-        # TODO(Yiyi) - this test should work, skipped for time reasons for now
-        pass
-
 
 @slow
 @require_torch_gpu

From 5a2dd24f589357516bd5bb38e09f02749c0b93e3 Mon Sep 17 00:00:00 2001
From: yiyixu <yixu310@gmail.com>
Date: Wed, 29 Nov 2023 04:40:43 +0000
Subject: [PATCH 07/20] take off last to-do

---
 src/diffusers/pipelines/kandinsky3/kandinsky3_pipeline.py | 2 --
 1 file changed, 2 deletions(-)

diff --git a/src/diffusers/pipelines/kandinsky3/kandinsky3_pipeline.py b/src/diffusers/pipelines/kandinsky3/kandinsky3_pipeline.py
index a6a0e01084de..3494d3873ab4 100644
--- a/src/diffusers/pipelines/kandinsky3/kandinsky3_pipeline.py
+++ b/src/diffusers/pipelines/kandinsky3/kandinsky3_pipeline.py
@@ -397,8 +397,6 @@ def __call__(
             self.text_encoder_offload_hook.offload()
 
         # 7. Denoising loop
-        # TODO(Yiyi): Correct the following line and use correctly
-        # num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
         with self.progress_bar(total=num_inference_steps) as progress_bar:
             for i, t in enumerate(timesteps):
                 latent_model_input = torch.cat([latents] * 2) if do_classifier_free_guidance else latents

From 89fdee4910e3d29e19bdaadf431b6255ff482815 Mon Sep 17 00:00:00 2001
From: yiyixu <yixu310@gmail.com>
Date: Wed, 29 Nov 2023 04:44:02 +0000
Subject: [PATCH 08/20] another to-do

---
 .../pipelines/kandinsky3/kandinsky3img2img_pipeline.py          | 2 --
 1 file changed, 2 deletions(-)

diff --git a/src/diffusers/pipelines/kandinsky3/kandinsky3img2img_pipeline.py b/src/diffusers/pipelines/kandinsky3/kandinsky3img2img_pipeline.py
index b043110cf1d7..a4ea60291606 100644
--- a/src/diffusers/pipelines/kandinsky3/kandinsky3img2img_pipeline.py
+++ b/src/diffusers/pipelines/kandinsky3/kandinsky3img2img_pipeline.py
@@ -409,8 +409,6 @@ def __call__(
             self.text_encoder_offload_hook.offload()
 
         # 7. Denoising loop
-        # TODO(Yiyi): Correct the following line and use correctly
-        # num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
         with self.progress_bar(total=num_inference_steps) as progress_bar:
             for i, t in enumerate(timesteps):
                 latent_model_input = torch.cat([latents] * 2) if do_classifier_free_guidance else latents

From e408cdf98816f503439306d61fcc2a98b5df41f9 Mon Sep 17 00:00:00 2001
From: yiyixu <yixu310@gmail.com>
Date: Wed, 29 Nov 2023 09:25:00 +0000
Subject: [PATCH 09/20] refactor

---
 src/diffusers/models/attention_processor.py   |   4 +-
 src/diffusers/models/unet_kandi3.py           |  16 +-
 .../kandinsky3/kandinsky3img2img_pipeline.py  |   2 +
 tests/pipelines/kandinsky3/test_kandinsky3.py |   2 +-
 .../kandinsky3/test_kandinsky3_img2img.py     | 212 ++++++++++++++++++
 tests/pipelines/test_pipelines_common.py      |   4 +
 6 files changed, 236 insertions(+), 4 deletions(-)
 create mode 100644 tests/pipelines/kandinsky3/test_kandinsky3_img2img.py

diff --git a/src/diffusers/models/attention_processor.py b/src/diffusers/models/attention_processor.py
index 40a335527ace..12939f6eabf2 100644
--- a/src/diffusers/models/attention_processor.py
+++ b/src/diffusers/models/attention_processor.py
@@ -109,6 +109,7 @@ def __init__(
         residual_connection: bool = False,
         _from_deprecated_attn_block: bool = False,
         processor: Optional["AttnProcessor"] = None,
+        scale_mask_factor: float = 1.0,
         out_dim: int = None,
     ):
         super().__init__()
@@ -120,6 +121,7 @@ def __init__(
         self.residual_connection = residual_connection
         self.dropout = dropout
         self.out_dim = out_dim if out_dim is not None else query_dim
+        self.scale_mask_factor = scale_mask_factor
 
         # we make use of this private variable to know whether this class is loaded
         # with an deprecated state dict so that we can convert it on the fly
@@ -595,7 +597,7 @@ def get_attention_scores(
             beta = 0
         else:
             baddbmm_input = attention_mask
-            beta = 1
+            beta = self.scale_mask_factor
 
         attention_scores = torch.baddbmm(
             baddbmm_input,
diff --git a/src/diffusers/models/unet_kandi3.py b/src/diffusers/models/unet_kandi3.py
index 3eea77ccb841..5dbbd095d069 100644
--- a/src/diffusers/models/unet_kandi3.py
+++ b/src/diffusers/models/unet_kandi3.py
@@ -465,7 +465,14 @@ def forward(self, x, time_embed):
 class Kandinsky3AttentionPooling(nn.Module):
     def __init__(self, num_channels, context_dim, head_dim=64):
         super().__init__()
-        self.attention = Attention(context_dim, context_dim, dim_head=head_dim, out_dim=num_channels, out_bias=False)
+        self.attention = Attention(
+            context_dim,
+            context_dim,
+            dim_head=head_dim,
+            out_dim=num_channels,
+            out_bias=False,
+            scale_mask_factor=-60000.0,
+        )
 
     def forward(self, x, context, context_mask=None):
         context_mask = context_mask.unsqueeze(1).to(dtype=context.dtype)
@@ -478,7 +485,12 @@ def __init__(self, num_channels, time_embed_dim, context_dim=None, norm_groups=3
         super().__init__()
         self.in_norm = Kandinsky3ConditionalGroupNorm(norm_groups, num_channels, time_embed_dim)
         self.attention = Attention(
-            num_channels, context_dim or num_channels, dim_head=head_dim, out_dim=num_channels, out_bias=False
+            num_channels,
+            context_dim or num_channels,
+            dim_head=head_dim,
+            out_dim=num_channels,
+            out_bias=False,
+            scale_mask_factor=-60000.0,
         )
 
         hidden_channels = expansion_ratio * num_channels
diff --git a/src/diffusers/pipelines/kandinsky3/kandinsky3img2img_pipeline.py b/src/diffusers/pipelines/kandinsky3/kandinsky3img2img_pipeline.py
index a4ea60291606..d698c15d68fa 100644
--- a/src/diffusers/pipelines/kandinsky3/kandinsky3img2img_pipeline.py
+++ b/src/diffusers/pipelines/kandinsky3/kandinsky3img2img_pipeline.py
@@ -452,6 +452,8 @@ def __call__(
             if output_type == "pil":
                 image = self.numpy_to_pil(image)
 
+            self.maybe_free_model_hooks()
+
             if not return_dict:
                 return (image,)
 
diff --git a/tests/pipelines/kandinsky3/test_kandinsky3.py b/tests/pipelines/kandinsky3/test_kandinsky3.py
index 7b5691409383..c163fe3102c4 100644
--- a/tests/pipelines/kandinsky3/test_kandinsky3.py
+++ b/tests/pipelines/kandinsky3/test_kandinsky3.py
@@ -163,7 +163,7 @@ def test_float16_inference(self):
         super().test_float16_inference(expected_max_diff=1e-1)
 
     def test_inference_batch_single_identical(self):
-        pass
+        super().test_inference_batch_single_identical(expected_max_diff=1e-2)
 
 
 @slow
diff --git a/tests/pipelines/kandinsky3/test_kandinsky3_img2img.py b/tests/pipelines/kandinsky3/test_kandinsky3_img2img.py
new file mode 100644
index 000000000000..a3c2697d0fbc
--- /dev/null
+++ b/tests/pipelines/kandinsky3/test_kandinsky3_img2img.py
@@ -0,0 +1,212 @@
+# coding=utf-8
+# Copyright 2023 HuggingFace Inc.
+#
+# 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.
+
+import gc
+import random
+import unittest
+
+import numpy as np
+import torch
+from PIL import Image
+from transformers import AutoTokenizer, T5EncoderModel
+
+from diffusers import (
+    AutoPipelineForImage2Image,
+    Kandinsky3Img2ImgPipeline,
+    Kandinsky3UNet,
+    VQModel,
+)
+from diffusers.image_processor import VaeImageProcessor
+from diffusers.schedulers.scheduling_ddpm import DDPMScheduler
+from diffusers.utils.testing_utils import (
+    enable_full_determinism,
+    floats_tensor,
+    load_image,
+    require_torch_gpu,
+    slow,
+)
+
+from ..pipeline_params import (
+    IMAGE_TO_IMAGE_IMAGE_PARAMS,
+    TEXT_GUIDED_IMAGE_VARIATION_BATCH_PARAMS,
+    TEXT_GUIDED_IMAGE_VARIATION_PARAMS,
+)
+from ..test_pipelines_common import PipelineTesterMixin
+
+
+enable_full_determinism()
+
+
+class Kandinsky3Img2ImgPipelineFastTests(PipelineTesterMixin, unittest.TestCase):
+    pipeline_class = Kandinsky3Img2ImgPipeline
+    params = TEXT_GUIDED_IMAGE_VARIATION_PARAMS - {"height", "width"}
+    batch_params = TEXT_GUIDED_IMAGE_VARIATION_BATCH_PARAMS
+    image_params = IMAGE_TO_IMAGE_IMAGE_PARAMS
+    test_xformers_attention = False
+
+    @property
+    def dummy_movq_kwargs(self):
+        return {
+            "block_out_channels": [32, 64],
+            "down_block_types": ["DownEncoderBlock2D", "AttnDownEncoderBlock2D"],
+            "in_channels": 3,
+            "latent_channels": 4,
+            "layers_per_block": 1,
+            "norm_num_groups": 8,
+            "norm_type": "spatial",
+            "num_vq_embeddings": 12,
+            "out_channels": 3,
+            "up_block_types": [
+                "AttnUpDecoderBlock2D",
+                "UpDecoderBlock2D",
+            ],
+            "vq_embed_dim": 4,
+        }
+
+    @property
+    def dummy_movq(self):
+        torch.manual_seed(0)
+        model = VQModel(**self.dummy_movq_kwargs)
+        return model
+
+    def get_dummy_components(self, time_cond_proj_dim=None):
+        torch.manual_seed(0)
+        unet = Kandinsky3UNet(
+            in_channels=4,
+            time_embedding_dim=4,
+            groups=2,
+            attention_head_dim=4,
+            layers_per_block=3,
+            block_out_channels=(32, 64),
+            cross_attention_dim=4,
+            encoder_hid_dim=32,
+        )
+        scheduler = DDPMScheduler(
+            beta_start=0.00085,
+            beta_end=0.012,
+            steps_offset=1,
+            beta_schedule="squaredcos_cap_v2",
+            clip_sample=True,
+            thresholding=False,
+        )
+        torch.manual_seed(0)
+        movq = self.dummy_movq
+        torch.manual_seed(0)
+        text_encoder = T5EncoderModel.from_pretrained("hf-internal-testing/tiny-random-t5")
+
+        torch.manual_seed(0)
+        tokenizer = AutoTokenizer.from_pretrained("hf-internal-testing/tiny-random-t5")
+
+        components = {
+            "unet": unet,
+            "scheduler": scheduler,
+            "movq": movq,
+            "text_encoder": text_encoder,
+            "tokenizer": tokenizer,
+        }
+        return components
+
+    def get_dummy_inputs(self, device, seed=0):
+        # create init_image
+        image = floats_tensor((1, 3, 64, 64), rng=random.Random(seed)).to(device)
+        image = image.cpu().permute(0, 2, 3, 1)[0]
+        init_image = Image.fromarray(np.uint8(image)).convert("RGB")
+
+        if str(device).startswith("mps"):
+            generator = torch.manual_seed(seed)
+        else:
+            generator = torch.Generator(device=device).manual_seed(seed)
+        inputs = {
+            "prompt": "A painting of a squirrel eating a burger",
+            "image": init_image,
+            "generator": generator,
+            "strength": 0.75,
+            "num_inference_steps": 10,
+            "guidance_scale": 6.0,
+            "output_type": "np",
+        }
+        return inputs
+
+    def test_kandinsky3_img2img(self):
+        device = "cpu"
+
+        components = self.get_dummy_components()
+
+        pipe = self.pipeline_class(**components)
+        pipe = pipe.to(device)
+
+        pipe.set_progress_bar_config(disable=None)
+
+        output = pipe(**self.get_dummy_inputs(device))
+        image = output.images
+
+        image_slice = image[0, -3:, -3:, -1]
+
+        assert image.shape == (1, 64, 64, 3)
+
+        expected_slice = np.array(
+            [0.576259, 0.6132097, 0.41703486, 0.603196, 0.62062526, 0.4655338, 0.5434324, 0.5660727, 0.65433365]
+        )
+
+        assert (
+            np.abs(image_slice.flatten() - expected_slice).max() < 1e-2
+        ), f" expected_slice {expected_slice}, but got {image_slice.flatten()}"
+
+    def test_float16_inference(self):
+        super().test_float16_inference(expected_max_diff=1e-1)
+
+    def test_inference_batch_single_identical(self):
+        super().test_inference_batch_single_identical(expected_max_diff=1e-2)
+
+
+@slow
+@require_torch_gpu
+class Kandinsky3Img2ImgPipelineIntegrationTests(unittest.TestCase):
+    def tearDown(self):
+        # clean up the VRAM after each test
+        super().tearDown()
+        gc.collect()
+        torch.cuda.empty_cache()
+
+    def test_kandinskyV3_img2img(self):
+        pipe = AutoPipelineForImage2Image.from_pretrained(
+            "kandinsky-community/kandinsky-3", variant="fp16", torch_dtype=torch.float16
+        )
+        pipe.enable_model_cpu_offload()
+        pipe.set_progress_bar_config(disable=None)
+
+        generator = torch.Generator(device="cpu").manual_seed(0)
+
+        image = load_image(
+            "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/kandinsky3/t2i.png"
+        )
+        w, h = 512, 512
+        image = image.resize((w, h), resample=Image.BICUBIC, reducing_gap=1)
+        prompt = "A painting of the inside of a subway train with tiny raccoons."
+
+        image = pipe(prompt, image=image, strength=0.75, num_inference_steps=25, generator=generator).images[0]
+
+        assert image.size == (512, 512)
+
+        expected_image = load_image(
+            "https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/kandinsky3/i2i.png"
+        )
+
+        image_processor = VaeImageProcessor()
+
+        image_np = image_processor.pil_to_numpy(image)
+        expected_image_np = image_processor.pil_to_numpy(expected_image)
+
+        self.assertTrue(np.allclose(image_np, expected_image_np, atol=5e-2))
diff --git a/tests/pipelines/test_pipelines_common.py b/tests/pipelines/test_pipelines_common.py
index e11175921184..cac5ee442ae6 100644
--- a/tests/pipelines/test_pipelines_common.py
+++ b/tests/pipelines/test_pipelines_common.py
@@ -377,6 +377,10 @@ def test_save_load_local(self, expected_max_difference=5e-4):
             with CaptureLogger(logger) as cap_logger:
                 pipe_loaded = self.pipeline_class.from_pretrained(tmpdir)
 
+            for component in pipe_loaded.components.values():
+                if hasattr(component, "set_default_attn_processor"):
+                    component.set_default_attn_processor()
+
             for name in pipe_loaded.components.keys():
                 if name not in pipe_loaded._optional_components:
                     assert name in str(cap_logger)

From 7dced94116d007bf49ace508811275783feed3c9 Mon Sep 17 00:00:00 2001
From: yiyixu <yixu310@gmail.com>
Date: Wed, 29 Nov 2023 10:08:06 +0000
Subject: [PATCH 10/20] add callback and latent output for text2img

---
 .../kandinsky3/kandinsky3_pipeline.py         | 133 ++++++++++++++----
 1 file changed, 103 insertions(+), 30 deletions(-)

diff --git a/src/diffusers/pipelines/kandinsky3/kandinsky3_pipeline.py b/src/diffusers/pipelines/kandinsky3/kandinsky3_pipeline.py
index 3494d3873ab4..8af4716da1c8 100644
--- a/src/diffusers/pipelines/kandinsky3/kandinsky3_pipeline.py
+++ b/src/diffusers/pipelines/kandinsky3/kandinsky3_pipeline.py
@@ -1,4 +1,4 @@
-from typing import Callable, List, Optional, Union
+from typing import Callable, Dict, List, Optional, Union
 
 import torch
 from transformers import T5EncoderModel, T5Tokenizer
@@ -7,6 +7,7 @@
 from ...models import Kandinsky3UNet, VQModel
 from ...schedulers import DDPMScheduler
 from ...utils import (
+    deprecate,
     is_accelerate_available,
     logging,
 )
@@ -29,6 +30,13 @@ def downscale_height_and_width(height, width, scale_factor=8):
 
 class Kandinsky3Pipeline(DiffusionPipeline, LoraLoaderMixin):
     model_cpu_offload_seq = "text_encoder->unet->movq"
+    _callback_tensor_inputs = [
+        "latents",
+        "prompt_embeds",
+        "negative_prompt_embeds",
+        "negative_attention_mask",
+        "attention_mask",
+    ]
 
     def __init__(
         self,
@@ -228,14 +236,19 @@ def check_inputs(
         negative_prompt=None,
         prompt_embeds=None,
         negative_prompt_embeds=None,
+        callback_on_step_end_tensor_inputs=None,
     ):
-        if (callback_steps is None) or (
-            callback_steps is not None and (not isinstance(callback_steps, int) or callback_steps <= 0)
-        ):
+        if callback_steps is not None and (not isinstance(callback_steps, int) or callback_steps <= 0):
             raise ValueError(
                 f"`callback_steps` has to be a positive integer but is {callback_steps} of type"
                 f" {type(callback_steps)}."
             )
+        if callback_on_step_end_tensor_inputs is not None and not all(
+            k in self._callback_tensor_inputs for k in callback_on_step_end_tensor_inputs
+        ):
+            raise ValueError(
+                f"`callback_on_step_end_tensor_inputs` has to be in {self._callback_tensor_inputs}, but found {[k for k in callback_on_step_end_tensor_inputs if k not in self._callback_tensor_inputs]}"
+            )
 
         if prompt is not None and prompt_embeds is not None:
             raise ValueError(
@@ -263,6 +276,18 @@ def check_inputs(
                     f" {negative_prompt_embeds.shape}."
                 )
 
+    @property
+    def guidance_scale(self):
+        return self._guidance_scale
+
+    @property
+    def do_classifier_free_guidance(self):
+        return self._guidance_scale > 1
+
+    @property
+    def num_timesteps(self):
+        return self._num_timesteps
+
     @torch.no_grad()
     def __call__(
         self,
@@ -278,9 +303,10 @@ def __call__(
         negative_prompt_embeds: Optional[torch.FloatTensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
-        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
-        callback_steps: int = 1,
         latents=None,
+        callback_on_step_end: Optional[Callable[[int, int, Dict], None]] = None,
+        callback_on_step_end_tensor_inputs: List[str] = ["latents"],
+        **kwargs,
     ):
         """
         Function invoked when calling the pipeline for generation.
@@ -344,11 +370,44 @@ def __call__(
                 `self.processor` in
                 [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
         """
+
+        callback = kwargs.pop("callback", None)
+        callback_steps = kwargs.pop("callback_steps", None)
+
+        if callback is not None:
+            deprecate(
+                "callback",
+                "1.0.0",
+                "Passing `callback` as an input argument to `__call__` is deprecated, consider use `callback_on_step_end`",
+            )
+        if callback_steps is not None:
+            deprecate(
+                "callback_steps",
+                "1.0.0",
+                "Passing `callback_steps` as an input argument to `__call__` is deprecated, consider use `callback_on_step_end`",
+            )
+
+        if callback_on_step_end_tensor_inputs is not None and not all(
+            k in self._callback_tensor_inputs for k in callback_on_step_end_tensor_inputs
+        ):
+            raise ValueError(
+                f"`callback_on_step_end_tensor_inputs` has to be in {self._callback_tensor_inputs}, but found {[k for k in callback_on_step_end_tensor_inputs if k not in self._callback_tensor_inputs]}"
+            )
+
         cut_context = True
         device = self._execution_device
 
         # 1. Check inputs. Raise error if not correct
-        self.check_inputs(prompt, callback_steps, negative_prompt, prompt_embeds, negative_prompt_embeds)
+        self.check_inputs(
+            prompt,
+            callback_steps,
+            negative_prompt,
+            prompt_embeds,
+            negative_prompt_embeds,
+            callback_on_step_end_tensor_inputs,
+        )
+
+        self._guidance_scale = guidance_scale
 
         if prompt is not None and isinstance(prompt, str):
             batch_size = 1
@@ -357,15 +416,10 @@ def __call__(
         else:
             batch_size = prompt_embeds.shape[0]
 
-        # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
-        # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
-        # corresponds to doing no classifier free guidance.
-        do_classifier_free_guidance = guidance_scale > 1.0
-
         # 3. Encode input prompt
         prompt_embeds, negative_prompt_embeds, attention_mask, negative_attention_mask = self.encode_prompt(
             prompt,
-            do_classifier_free_guidance,
+            self.do_classifier_free_guidance,
             num_images_per_prompt=num_images_per_prompt,
             device=device,
             negative_prompt=negative_prompt,
@@ -374,7 +428,7 @@ def __call__(
             _cut_context=cut_context,
         )
 
-        if do_classifier_free_guidance:
+        if self.do_classifier_free_guidance:
             prompt_embeds = torch.cat([negative_prompt_embeds, prompt_embeds])
             attention_mask = torch.cat([negative_attention_mask, attention_mask]).bool()
         # 4. Prepare timesteps
@@ -397,9 +451,11 @@ def __call__(
             self.text_encoder_offload_hook.offload()
 
         # 7. Denoising loop
+        num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
+        self._num_timesteps = len(timesteps)
         with self.progress_bar(total=num_inference_steps) as progress_bar:
             for i, t in enumerate(timesteps):
-                latent_model_input = torch.cat([latents] * 2) if do_classifier_free_guidance else latents
+                latent_model_input = torch.cat([latents] * 2) if self.do_classifier_free_guidance else latents
 
                 # predict the noise residual
                 noise_pred = self.unet(
@@ -410,7 +466,7 @@ def __call__(
                     return_dict=False,
                 )[0]
 
-                if do_classifier_free_guidance:
+                if self.do_classifier_free_guidance:
                     noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
 
                     noise_pred = (guidance_scale + 1.0) * noise_pred_text - guidance_scale * noise_pred_uncond
@@ -423,26 +479,43 @@ def __call__(
                     latents,
                     generator=generator,
                 ).prev_sample
-                progress_bar.update()
-                if callback is not None and i % callback_steps == 0:
-                    step_idx = i // getattr(self.scheduler, "order", 1)
-                    callback(step_idx, t, latents)
 
-            # post-processing
-            image = self.movq.decode(latents, force_not_quantize=True)["sample"]
+                if callback_on_step_end is not None:
+                    callback_kwargs = {}
+                    for k in callback_on_step_end_tensor_inputs:
+                        callback_kwargs[k] = locals()[k]
+                    callback_outputs = callback_on_step_end(self, i, t, callback_kwargs)
+
+                    latents = callback_outputs.pop("latents", latents)
+                    prompt_embeds = callback_outputs.pop("prompt_embeds", prompt_embeds)
+                    negative_prompt_embeds = callback_outputs.pop("negative_prompt_embeds", negative_prompt_embeds)
+                    attention_mask = callback_outputs.pop("attention_mask", attention_mask)
+                    negative_attention_mask = callback_outputs.pop("negative_attention_mask", negative_attention_mask)
+
+                if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
+                    progress_bar.update()
+                    if callback is not None and i % callback_steps == 0:
+                        step_idx = i // getattr(self.scheduler, "order", 1)
+                        callback(step_idx, t, latents)
 
-            if output_type not in ["pt", "np", "pil"]:
+            # post-processing
+            if output_type not in ["pt", "np", "pil", "latent"]:
                 raise ValueError(
-                    f"Only the output types `pt`, `pil` and `np` are supported not output_type={output_type}"
+                    f"Only the output types `pt`, `pil`, `np` and `latent` are supported not output_type={output_type}"
                 )
 
-            if output_type in ["np", "pil"]:
-                image = image * 0.5 + 0.5
-                image = image.clamp(0, 1)
-                image = image.cpu().permute(0, 2, 3, 1).float().numpy()
+            if not output_type == "latent":
+                image = self.movq.decode(latents, force_not_quantize=True)["sample"]
+
+                if output_type in ["np", "pil"]:
+                    image = image * 0.5 + 0.5
+                    image = image.clamp(0, 1)
+                    image = image.cpu().permute(0, 2, 3, 1).float().numpy()
 
-            if output_type == "pil":
-                image = self.numpy_to_pil(image)
+                if output_type == "pil":
+                    image = self.numpy_to_pil(image)
+            else:
+                image = latents
 
             self.maybe_free_model_hooks()
 

From 51fe17bcd8d76b2b16403f4bf15efcfa848eb494 Mon Sep 17 00:00:00 2001
From: yiyixu <yixu310@gmail.com>
Date: Wed, 29 Nov 2023 10:29:14 +0000
Subject: [PATCH 11/20] refactor img2img

---
 .../kandinsky3/kandinsky3img2img_pipeline.py  | 133 ++++++++++++++----
 .../kandinsky3/test_kandinsky3_img2img.py     |   4 +
 2 files changed, 107 insertions(+), 30 deletions(-)

diff --git a/src/diffusers/pipelines/kandinsky3/kandinsky3img2img_pipeline.py b/src/diffusers/pipelines/kandinsky3/kandinsky3img2img_pipeline.py
index d698c15d68fa..5a27baeefb19 100644
--- a/src/diffusers/pipelines/kandinsky3/kandinsky3img2img_pipeline.py
+++ b/src/diffusers/pipelines/kandinsky3/kandinsky3img2img_pipeline.py
@@ -1,5 +1,5 @@
 import inspect
-from typing import Callable, List, Optional, Union
+from typing import Callable, Dict, List, Optional, Union
 
 import numpy as np
 import PIL
@@ -11,6 +11,7 @@
 from ...models import Kandinsky3UNet, VQModel
 from ...schedulers import DDPMScheduler
 from ...utils import (
+    deprecate,
     is_accelerate_available,
     logging,
 )
@@ -41,6 +42,13 @@ def prepare_image(pil_image):
 
 class Kandinsky3Img2ImgPipeline(DiffusionPipeline, LoraLoaderMixin):
     model_cpu_offload_seq = "text_encoder->unet->movq"
+    _callback_tensor_inputs = [
+        "latents",
+        "prompt_embeds",
+        "negative_prompt_embeds",
+        "negative_attention_mask",
+        "attention_mask",
+    ]
 
     def __init__(
         self,
@@ -299,15 +307,21 @@ def check_inputs(
         negative_prompt=None,
         prompt_embeds=None,
         negative_prompt_embeds=None,
+        callback_on_step_end_tensor_inputs=None,
     ):
-        if (callback_steps is None) or (
-            callback_steps is not None and (not isinstance(callback_steps, int) or callback_steps <= 0)
-        ):
+        if callback_steps is not None and (not isinstance(callback_steps, int) or callback_steps <= 0):
             raise ValueError(
                 f"`callback_steps` has to be a positive integer but is {callback_steps} of type"
                 f" {type(callback_steps)}."
             )
 
+        if callback_on_step_end_tensor_inputs is not None and not all(
+            k in self._callback_tensor_inputs for k in callback_on_step_end_tensor_inputs
+        ):
+            raise ValueError(
+                f"`callback_on_step_end_tensor_inputs` has to be in {self._callback_tensor_inputs}, but found {[k for k in callback_on_step_end_tensor_inputs if k not in self._callback_tensor_inputs]}"
+            )
+
         if prompt is not None and prompt_embeds is not None:
             raise ValueError(
                 f"Cannot forward both `prompt`: {prompt} and `prompt_embeds`: {prompt_embeds}. Please make sure to"
@@ -334,6 +348,18 @@ def check_inputs(
                     f" {negative_prompt_embeds.shape}."
                 )
 
+    @property
+    def guidance_scale(self):
+        return self._guidance_scale
+
+    @property
+    def do_classifier_free_guidance(self):
+        return self._guidance_scale > 1
+
+    @property
+    def num_timesteps(self):
+        return self._num_timesteps
+
     @torch.no_grad()
     def __call__(
         self,
@@ -349,13 +375,46 @@ def __call__(
         negative_prompt_embeds: Optional[torch.FloatTensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
-        callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
-        callback_steps: int = 1,
         latents=None,
+        callback_on_step_end: Optional[Callable[[int, int, Dict], None]] = None,
+        callback_on_step_end_tensor_inputs: List[str] = ["latents"],
+        **kwargs,
     ):
+        callback = kwargs.pop("callback", None)
+        callback_steps = kwargs.pop("callback_steps", None)
+
+        if callback is not None:
+            deprecate(
+                "callback",
+                "1.0.0",
+                "Passing `callback` as an input argument to `__call__` is deprecated, consider use `callback_on_step_end`",
+            )
+        if callback_steps is not None:
+            deprecate(
+                "callback_steps",
+                "1.0.0",
+                "Passing `callback_steps` as an input argument to `__call__` is deprecated, consider use `callback_on_step_end`",
+            )
+
+        if callback_on_step_end_tensor_inputs is not None and not all(
+            k in self._callback_tensor_inputs for k in callback_on_step_end_tensor_inputs
+        ):
+            raise ValueError(
+                f"`callback_on_step_end_tensor_inputs` has to be in {self._callback_tensor_inputs}, but found {[k for k in callback_on_step_end_tensor_inputs if k not in self._callback_tensor_inputs]}"
+            )
+
         cut_context = True
         # 1. Check inputs. Raise error if not correct
-        self.check_inputs(prompt, callback_steps, negative_prompt, prompt_embeds, negative_prompt_embeds)
+        self.check_inputs(
+            prompt,
+            callback_steps,
+            negative_prompt,
+            prompt_embeds,
+            negative_prompt_embeds,
+            callback_on_step_end_tensor_inputs,
+        )
+
+        self._guidance_scale = guidance_scale
 
         if prompt is not None and isinstance(prompt, str):
             batch_size = 1
@@ -366,15 +425,10 @@ def __call__(
 
         device = self._execution_device
 
-        # here `guidance_scale` is defined analog to the guidance weight `w` of equation (2)
-        # of the Imagen paper: https://arxiv.org/pdf/2205.11487.pdf . `guidance_scale = 1`
-        # corresponds to doing no classifier free guidance.
-        do_classifier_free_guidance = guidance_scale > 1.0
-
         # 3. Encode input prompt
         prompt_embeds, negative_prompt_embeds, attention_mask, negative_attention_mask = self.encode_prompt(
             prompt,
-            do_classifier_free_guidance,
+            self.do_classifier_free_guidance,
             num_images_per_prompt=num_images_per_prompt,
             device=device,
             negative_prompt=negative_prompt,
@@ -383,7 +437,7 @@ def __call__(
             _cut_context=cut_context,
         )
 
-        if do_classifier_free_guidance:
+        if self.do_classifier_free_guidance:
             prompt_embeds = torch.cat([negative_prompt_embeds, prompt_embeds])
             attention_mask = torch.cat([negative_attention_mask, attention_mask]).bool()
         if not isinstance(image, list):
@@ -409,9 +463,11 @@ def __call__(
             self.text_encoder_offload_hook.offload()
 
         # 7. Denoising loop
+        num_warmup_steps = len(timesteps) - num_inference_steps * self.scheduler.order
+        self._num_timesteps = len(timesteps)
         with self.progress_bar(total=num_inference_steps) as progress_bar:
             for i, t in enumerate(timesteps):
-                latent_model_input = torch.cat([latents] * 2) if do_classifier_free_guidance else latents
+                latent_model_input = torch.cat([latents] * 2) if self.do_classifier_free_guidance else latents
 
                 # predict the noise residual
                 noise_pred = self.unet(
@@ -420,7 +476,7 @@ def __call__(
                     encoder_hidden_states=prompt_embeds,
                     encoder_attention_mask=attention_mask,
                 )[0]
-                if do_classifier_free_guidance:
+                if self.do_classifier_free_guidance:
                     noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
 
                     noise_pred = (guidance_scale + 1.0) * noise_pred_text - guidance_scale * noise_pred_uncond
@@ -432,25 +488,42 @@ def __call__(
                     latents,
                     generator=generator,
                 ).prev_sample
-                progress_bar.update()
-                if callback is not None and i % callback_steps == 0:
-                    step_idx = i // getattr(self.scheduler, "order", 1)
-                    callback(step_idx, t, latents)
-            # post-processing
-            image = self.movq.decode(latents, force_not_quantize=True)["sample"]
 
-            if output_type not in ["pt", "np", "pil"]:
+                if callback_on_step_end is not None:
+                    callback_kwargs = {}
+                    for k in callback_on_step_end_tensor_inputs:
+                        callback_kwargs[k] = locals()[k]
+                    callback_outputs = callback_on_step_end(self, i, t, callback_kwargs)
+
+                    latents = callback_outputs.pop("latents", latents)
+                    prompt_embeds = callback_outputs.pop("prompt_embeds", prompt_embeds)
+                    negative_prompt_embeds = callback_outputs.pop("negative_prompt_embeds", negative_prompt_embeds)
+                    attention_mask = callback_outputs.pop("attention_mask", attention_mask)
+                    negative_attention_mask = callback_outputs.pop("negative_attention_mask", negative_attention_mask)
+
+                if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
+                    progress_bar.update()
+                    if callback is not None and i % callback_steps == 0:
+                        step_idx = i // getattr(self.scheduler, "order", 1)
+                        callback(step_idx, t, latents)
+
+            # post-processing
+            if output_type not in ["pt", "np", "pil", "latent"]:
                 raise ValueError(
-                    f"Only the output types `pt`, `pil` and `np` are supported not output_type={output_type}"
+                    f"Only the output types `pt`, `pil`, `np` and `latent` are supported not output_type={output_type}"
                 )
+            if not output_type == "latent":
+                image = self.movq.decode(latents, force_not_quantize=True)["sample"]
 
-            if output_type in ["np", "pil"]:
-                image = image * 0.5 + 0.5
-                image = image.clamp(0, 1)
-                image = image.cpu().permute(0, 2, 3, 1).float().numpy()
+                if output_type in ["np", "pil"]:
+                    image = image * 0.5 + 0.5
+                    image = image.clamp(0, 1)
+                    image = image.cpu().permute(0, 2, 3, 1).float().numpy()
 
-            if output_type == "pil":
-                image = self.numpy_to_pil(image)
+                if output_type == "pil":
+                    image = self.numpy_to_pil(image)
+            else:
+                image = latents
 
             self.maybe_free_model_hooks()
 
diff --git a/tests/pipelines/kandinsky3/test_kandinsky3_img2img.py b/tests/pipelines/kandinsky3/test_kandinsky3_img2img.py
index a3c2697d0fbc..fcd1ed0dc94d 100644
--- a/tests/pipelines/kandinsky3/test_kandinsky3_img2img.py
+++ b/tests/pipelines/kandinsky3/test_kandinsky3_img2img.py
@@ -42,6 +42,8 @@
     IMAGE_TO_IMAGE_IMAGE_PARAMS,
     TEXT_GUIDED_IMAGE_VARIATION_BATCH_PARAMS,
     TEXT_GUIDED_IMAGE_VARIATION_PARAMS,
+    TEXT_TO_IMAGE_CALLBACK_CFG_PARAMS,
+    TEXT_TO_IMAGE_IMAGE_PARAMS,
 )
 from ..test_pipelines_common import PipelineTesterMixin
 
@@ -54,6 +56,8 @@ class Kandinsky3Img2ImgPipelineFastTests(PipelineTesterMixin, unittest.TestCase)
     params = TEXT_GUIDED_IMAGE_VARIATION_PARAMS - {"height", "width"}
     batch_params = TEXT_GUIDED_IMAGE_VARIATION_BATCH_PARAMS
     image_params = IMAGE_TO_IMAGE_IMAGE_PARAMS
+    image_latents_params = TEXT_TO_IMAGE_IMAGE_PARAMS
+    callback_cfg_params = TEXT_TO_IMAGE_CALLBACK_CFG_PARAMS
     test_xformers_attention = False
 
     @property

From f5cfa5ad9223e2b7e633f951289a26c5893f3fc8 Mon Sep 17 00:00:00 2001
From: yiyixu <yixu310@gmail.com>
Date: Wed, 29 Nov 2023 10:42:35 +0000
Subject: [PATCH 12/20] change unet file name

---
 src/diffusers/models/__init__.py                   |  4 ++--
 ...{unet_kandi3.py => unet_2d_condition_kandi3.py} | 14 ++++++++++++++
 2 files changed, 16 insertions(+), 2 deletions(-)
 rename src/diffusers/models/{unet_kandi3.py => unet_2d_condition_kandi3.py} (97%)

diff --git a/src/diffusers/models/__init__.py b/src/diffusers/models/__init__.py
index de2e2848b848..13f05f78b769 100644
--- a/src/diffusers/models/__init__.py
+++ b/src/diffusers/models/__init__.py
@@ -36,7 +36,7 @@
     _import_structure["unet_2d"] = ["UNet2DModel"]
     _import_structure["unet_2d_condition"] = ["UNet2DConditionModel"]
     _import_structure["unet_3d_condition"] = ["UNet3DConditionModel"]
-    _import_structure["unet_kandi3"] = ["Kandinsky3UNet"]
+    _import_structure["unet_2d_condition_kandi3"] = ["Kandinsky3UNet"]
     _import_structure["unet_motion_model"] = ["MotionAdapter", "UNetMotionModel"]
     _import_structure["vq_model"] = ["VQModel"]
 
@@ -64,7 +64,7 @@
         from .unet_2d import UNet2DModel
         from .unet_2d_condition import UNet2DConditionModel
         from .unet_3d_condition import UNet3DConditionModel
-        from .unet_kandi3 import Kandinsky3UNet
+        from .unet_2d_condition_kandi3 import Kandinsky3UNet
         from .unet_motion_model import MotionAdapter, UNetMotionModel
         from .vq_model import VQModel
 
diff --git a/src/diffusers/models/unet_kandi3.py b/src/diffusers/models/unet_2d_condition_kandi3.py
similarity index 97%
rename from src/diffusers/models/unet_kandi3.py
rename to src/diffusers/models/unet_2d_condition_kandi3.py
index 5dbbd095d069..1ce831393c37 100644
--- a/src/diffusers/models/unet_kandi3.py
+++ b/src/diffusers/models/unet_2d_condition_kandi3.py
@@ -1,3 +1,17 @@
+# Copyright 2023 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 Dict, Tuple, Union
 

From 334cd2eb64c97d57f02b9df8de0147c9fc22a7cd Mon Sep 17 00:00:00 2001
From: yiyixu <yixu310@gmail.com>
Date: Wed, 29 Nov 2023 11:09:21 +0000
Subject: [PATCH 13/20] add doc string

---
 src/diffusers/models/__init__.py              |  4 +-
 .../kandinsky3/kandinsky3_pipeline.py         | 25 ++++++
 .../kandinsky3/kandinsky3img2img_pipeline.py  | 83 ++++++++++++++++++-
 3 files changed, 109 insertions(+), 3 deletions(-)

diff --git a/src/diffusers/models/__init__.py b/src/diffusers/models/__init__.py
index 13f05f78b769..6346523a7004 100644
--- a/src/diffusers/models/__init__.py
+++ b/src/diffusers/models/__init__.py
@@ -35,8 +35,8 @@
     _import_structure["unet_1d"] = ["UNet1DModel"]
     _import_structure["unet_2d"] = ["UNet2DModel"]
     _import_structure["unet_2d_condition"] = ["UNet2DConditionModel"]
-    _import_structure["unet_3d_condition"] = ["UNet3DConditionModel"]
     _import_structure["unet_2d_condition_kandi3"] = ["Kandinsky3UNet"]
+    _import_structure["unet_3d_condition"] = ["UNet3DConditionModel"]
     _import_structure["unet_motion_model"] = ["MotionAdapter", "UNetMotionModel"]
     _import_structure["vq_model"] = ["VQModel"]
 
@@ -63,8 +63,8 @@
         from .unet_1d import UNet1DModel
         from .unet_2d import UNet2DModel
         from .unet_2d_condition import UNet2DConditionModel
-        from .unet_3d_condition import UNet3DConditionModel
         from .unet_2d_condition_kandi3 import Kandinsky3UNet
+        from .unet_3d_condition import UNet3DConditionModel
         from .unet_motion_model import MotionAdapter, UNetMotionModel
         from .vq_model import VQModel
 
diff --git a/src/diffusers/pipelines/kandinsky3/kandinsky3_pipeline.py b/src/diffusers/pipelines/kandinsky3/kandinsky3_pipeline.py
index 8af4716da1c8..dab0d15918b7 100644
--- a/src/diffusers/pipelines/kandinsky3/kandinsky3_pipeline.py
+++ b/src/diffusers/pipelines/kandinsky3/kandinsky3_pipeline.py
@@ -10,6 +10,7 @@
     deprecate,
     is_accelerate_available,
     logging,
+    replace_example_docstring,
 )
 from ...utils.torch_utils import randn_tensor
 from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
@@ -17,6 +18,23 @@
 
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
 
+EXAMPLE_DOC_STRING = """
+    Examples:
+        ```py
+        >>> from diffusers import AutoPipelineForText2Image
+        >>> import torch
+
+        >>> pipe = AutoPipelineForText2Image.from_pretrained("kandinsky-community/kandinsky-3", variant="fp16", torch_dtype=torch.float16)
+        >>> pipe.enable_model_cpu_offload()
+
+        >>> prompt = "A photograph of the inside of a subway train. There are raccoons sitting on the seats. One of them is reading a newspaper. The window shows the city in the background."
+
+        >>> generator = torch.Generator(device="cpu").manual_seed(0)
+        >>> image = pipe(prompt, num_inference_steps=25, generator=generator).images[0]
+        ```
+
+"""
+
 
 def downscale_height_and_width(height, width, scale_factor=8):
     new_height = height // scale_factor**2
@@ -289,6 +307,7 @@ def num_timesteps(self):
         return self._num_timesteps
 
     @torch.no_grad()
+    @replace_example_docstring(EXAMPLE_DOC_STRING)
     def __call__(
         self,
         prompt: Union[str, List[str]] = None,
@@ -369,6 +388,12 @@ def __call__(
                 A kwargs dictionary that if specified is passed along to the `AttentionProcessor` as defined under
                 `self.processor` in
                 [diffusers.models.attention_processor](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention_processor.py).
+
+        Examples:
+
+        Returns:
+            [`~pipelines.ImagePipelineOutput`] or `tuple`
+
         """
 
         callback = kwargs.pop("callback", None)
diff --git a/src/diffusers/pipelines/kandinsky3/kandinsky3img2img_pipeline.py b/src/diffusers/pipelines/kandinsky3/kandinsky3img2img_pipeline.py
index 5a27baeefb19..b830179196a5 100644
--- a/src/diffusers/pipelines/kandinsky3/kandinsky3img2img_pipeline.py
+++ b/src/diffusers/pipelines/kandinsky3/kandinsky3img2img_pipeline.py
@@ -14,6 +14,7 @@
     deprecate,
     is_accelerate_available,
     logging,
+    replace_example_docstring,
 )
 from ...utils.torch_utils import randn_tensor
 from ..pipeline_utils import DiffusionPipeline, ImagePipelineOutput
@@ -21,6 +22,24 @@
 
 logger = logging.get_logger(__name__)  # pylint: disable=invalid-name
 
+EXAMPLE_DOC_STRING = """
+    Examples:
+        ```py
+        >>> from diffusers import AutoPipelineForImage2Image
+        >>> from diffusers.utils import load_image
+        >>> import torch
+
+        >>> pipe = AutoPipelineForImage2Image.from_pretrained("kandinsky-community/kandinsky-3", variant="fp16", torch_dtype=torch.float16)
+        >>> pipe.enable_model_cpu_offload()
+
+        >>> prompt = "A painting of the inside of a subway train with tiny raccoons."
+        >>> image = load_image("https://huggingface.co/datasets/hf-internal-testing/diffusers-images/resolve/main/kandinsky3/t2i.png")
+
+        >>> generator = torch.Generator(device="cpu").manual_seed(0)
+        >>> image = pipe(prompt, image=image, strength=0.75, num_inference_steps=25, generator=generator).images[0]
+        ```
+"""
+
 
 def downscale_height_and_width(height, width, scale_factor=8):
     new_height = height // scale_factor**2
@@ -361,6 +380,7 @@ def num_timesteps(self):
         return self._num_timesteps
 
     @torch.no_grad()
+    @replace_example_docstring(EXAMPLE_DOC_STRING)
     def __call__(
         self,
         prompt: Union[str, List[str]] = None,
@@ -375,11 +395,72 @@ def __call__(
         negative_prompt_embeds: Optional[torch.FloatTensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
-        latents=None,
         callback_on_step_end: Optional[Callable[[int, int, Dict], None]] = None,
         callback_on_step_end_tensor_inputs: List[str] = ["latents"],
         **kwargs,
     ):
+        """
+        Function invoked when calling the pipeline for generation.
+
+        Args:
+            prompt (`str` or `List[str]`, *optional*):
+                The prompt or prompts to guide the image generation. If not defined, one has to pass `prompt_embeds`.
+                instead.
+            image (`torch.FloatTensor`, `PIL.Image.Image`, `np.ndarray`, `List[torch.FloatTensor]`, `List[PIL.Image.Image]`, or `List[np.ndarray]`):
+                `Image`, or tensor representing an image batch, that will be used as the starting point for the
+                process.
+            strength (`float`, *optional*, defaults to 0.8):
+                Indicates extent to transform the reference `image`. Must be between 0 and 1. `image` is used as a
+                starting point and more noise is added the higher the `strength`. The number of denoising steps depends
+                on the amount of noise initially added. When `strength` is 1, added noise is maximum and the denoising
+                process runs for the full number of iterations specified in `num_inference_steps`. A value of 1
+                essentially ignores `image`.
+            num_inference_steps (`int`, *optional*, defaults to 50):
+                The number of denoising steps. More denoising steps usually lead to a higher quality image at the
+                expense of slower inference.
+            guidance_scale (`float`, *optional*, defaults to 3.0):
+                Guidance scale as defined in [Classifier-Free Diffusion Guidance](https://arxiv.org/abs/2207.12598).
+                `guidance_scale` is defined as `w` of equation 2. of [Imagen
+                Paper](https://arxiv.org/pdf/2205.11487.pdf). Guidance scale is enabled by setting `guidance_scale >
+                1`. Higher guidance scale encourages to generate images that are closely linked to the text `prompt`,
+                usually at the expense of lower image quality.
+            negative_prompt (`str` or `List[str]`, *optional*):
+                The prompt or prompts not to guide the image generation. If not defined, one has to pass
+                `negative_prompt_embeds` instead. Ignored when not using guidance (i.e., ignored if `guidance_scale` is
+                less than `1`).
+            num_images_per_prompt (`int`, *optional*, defaults to 1):
+                The number of images to generate per prompt.
+            generator (`torch.Generator` or `List[torch.Generator]`, *optional*):
+                One or a list of [torch generator(s)](https://pytorch.org/docs/stable/generated/torch.Generator.html)
+                to make generation deterministic.
+            prompt_embeds (`torch.FloatTensor`, *optional*):
+                Pre-generated text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt weighting. If not
+                provided, text embeddings will be generated from `prompt` input argument.
+            negative_prompt_embeds (`torch.FloatTensor`, *optional*):
+                Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
+                weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
+                argument.
+            output_type (`str`, *optional*, defaults to `"pil"`):
+                The output format of the generate image. Choose between
+                [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
+            return_dict (`bool`, *optional*, defaults to `True`):
+                Whether or not to return a [`~pipelines.stable_diffusion.IFPipelineOutput`] instead of a plain tuple.
+            callback_on_step_end (`Callable`, *optional*):
+                A function that calls at the end of each denoising steps during the inference. The function is called
+                with the following arguments: `callback_on_step_end(self: DiffusionPipeline, step: int, timestep: int,
+                callback_kwargs: Dict)`. `callback_kwargs` will include a list of all tensors as specified by
+                `callback_on_step_end_tensor_inputs`.
+            callback_on_step_end_tensor_inputs (`List`, *optional*):
+                The list of tensor inputs for the `callback_on_step_end` function. The tensors specified in the list
+                will be passed as `callback_kwargs` argument. You will only be able to include variables listed in the
+                `._callback_tensor_inputs` attribute of your pipeline class.
+
+        Examples:
+
+        Returns:
+            [`~pipelines.ImagePipelineOutput`] or `tuple`
+
+        """
         callback = kwargs.pop("callback", None)
         callback_steps = kwargs.pop("callback_steps", None)
 

From dbf5135de4c7244bbd4baace1e33ba2d64f456df Mon Sep 17 00:00:00 2001
From: yiyixu <yixu310@gmail.com>
Date: Wed, 29 Nov 2023 11:15:16 +0000
Subject: [PATCH 14/20] change pipeline file name

---
 src/diffusers/pipelines/kandinsky3/__init__.py            | 8 ++++----
 .../{kandinsky3_pipeline.py => pipeline_kandinsky3.py}    | 0
 ...img2img_pipeline.py => pipeline_kandinsky3_img2img.py} | 0
 3 files changed, 4 insertions(+), 4 deletions(-)
 rename src/diffusers/pipelines/kandinsky3/{kandinsky3_pipeline.py => pipeline_kandinsky3.py} (100%)
 rename src/diffusers/pipelines/kandinsky3/{kandinsky3img2img_pipeline.py => pipeline_kandinsky3_img2img.py} (100%)

diff --git a/src/diffusers/pipelines/kandinsky3/__init__.py b/src/diffusers/pipelines/kandinsky3/__init__.py
index 4da3a83c0448..e8a3063141b5 100644
--- a/src/diffusers/pipelines/kandinsky3/__init__.py
+++ b/src/diffusers/pipelines/kandinsky3/__init__.py
@@ -21,8 +21,8 @@
 
     _dummy_objects.update(get_objects_from_module(dummy_torch_and_transformers_objects))
 else:
-    _import_structure["kandinsky3_pipeline"] = ["Kandinsky3Pipeline"]
-    _import_structure["kandinsky3img2img_pipeline"] = ["Kandinsky3Img2ImgPipeline"]
+    _import_structure["pipeline_kandinsky3"] = ["Kandinsky3Pipeline"]
+    _import_structure["pipeline_kandinsky3_img2img"] = ["Kandinsky3Img2ImgPipeline"]
 
 
 if TYPE_CHECKING or DIFFUSERS_SLOW_IMPORT:
@@ -33,8 +33,8 @@
     except OptionalDependencyNotAvailable:
         from ...utils.dummy_torch_and_transformers_objects import *
     else:
-        from .kandinsky3_pipeline import Kandinsky3Pipeline
-        from .kandinsky3img2img_pipeline import Kandinsky3Img2ImgPipeline
+        from .pipeline_kandinsky3 import Kandinsky3Pipeline
+        from .pipeline_kandinsky3_img2img import Kandinsky3Img2ImgPipeline
 else:
     import sys
 
diff --git a/src/diffusers/pipelines/kandinsky3/kandinsky3_pipeline.py b/src/diffusers/pipelines/kandinsky3/pipeline_kandinsky3.py
similarity index 100%
rename from src/diffusers/pipelines/kandinsky3/kandinsky3_pipeline.py
rename to src/diffusers/pipelines/kandinsky3/pipeline_kandinsky3.py
diff --git a/src/diffusers/pipelines/kandinsky3/kandinsky3img2img_pipeline.py b/src/diffusers/pipelines/kandinsky3/pipeline_kandinsky3_img2img.py
similarity index 100%
rename from src/diffusers/pipelines/kandinsky3/kandinsky3img2img_pipeline.py
rename to src/diffusers/pipelines/kandinsky3/pipeline_kandinsky3_img2img.py

From 25c4e07b42ad77b4ca0edfcad15745438bf29e1e Mon Sep 17 00:00:00 2001
From: yiyixu <yixu310@gmail.com>
Date: Wed, 29 Nov 2023 12:04:14 +0000
Subject: [PATCH 15/20] fix failing test

---
 tests/pipelines/kandinsky3/test_kandinsky3_img2img.py | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/tests/pipelines/kandinsky3/test_kandinsky3_img2img.py b/tests/pipelines/kandinsky3/test_kandinsky3_img2img.py
index fcd1ed0dc94d..581251a81639 100644
--- a/tests/pipelines/kandinsky3/test_kandinsky3_img2img.py
+++ b/tests/pipelines/kandinsky3/test_kandinsky3_img2img.py
@@ -59,6 +59,15 @@ class Kandinsky3Img2ImgPipelineFastTests(PipelineTesterMixin, unittest.TestCase)
     image_latents_params = TEXT_TO_IMAGE_IMAGE_PARAMS
     callback_cfg_params = TEXT_TO_IMAGE_CALLBACK_CFG_PARAMS
     test_xformers_attention = False
+    required_optional_params = frozenset(
+        [
+            "num_inference_steps",
+            "num_images_per_prompt",
+            "generator",
+            "output_type",
+            "return_dict",
+        ]
+    )
 
     @property
     def dummy_movq_kwargs(self):

From dd198cb729d450890a5e36eab636c2188010f2e6 Mon Sep 17 00:00:00 2001
From: yiyixu <yixu310@gmail.com>
Date: Wed, 29 Nov 2023 17:34:58 +0000
Subject: [PATCH 16/20]  rename unet file

---
 src/diffusers/models/__init__.py                              | 4 ++--
 .../{unet_2d_condition_kandi3.py => unet_kandinsky3.py}       | 0
 2 files changed, 2 insertions(+), 2 deletions(-)
 rename src/diffusers/models/{unet_2d_condition_kandi3.py => unet_kandinsky3.py} (100%)

diff --git a/src/diffusers/models/__init__.py b/src/diffusers/models/__init__.py
index 6346523a7004..103f014a2b89 100644
--- a/src/diffusers/models/__init__.py
+++ b/src/diffusers/models/__init__.py
@@ -35,8 +35,8 @@
     _import_structure["unet_1d"] = ["UNet1DModel"]
     _import_structure["unet_2d"] = ["UNet2DModel"]
     _import_structure["unet_2d_condition"] = ["UNet2DConditionModel"]
-    _import_structure["unet_2d_condition_kandi3"] = ["Kandinsky3UNet"]
     _import_structure["unet_3d_condition"] = ["UNet3DConditionModel"]
+    _import_structure["unet_kandinsky3"] = ["Kandinsky3UNet"]
     _import_structure["unet_motion_model"] = ["MotionAdapter", "UNetMotionModel"]
     _import_structure["vq_model"] = ["VQModel"]
 
@@ -63,8 +63,8 @@
         from .unet_1d import UNet1DModel
         from .unet_2d import UNet2DModel
         from .unet_2d_condition import UNet2DConditionModel
-        from .unet_2d_condition_kandi3 import Kandinsky3UNet
         from .unet_3d_condition import UNet3DConditionModel
+        from .unet_kandinsky3 import Kandinsky3UNet
         from .unet_motion_model import MotionAdapter, UNetMotionModel
         from .vq_model import VQModel
 
diff --git a/src/diffusers/models/unet_2d_condition_kandi3.py b/src/diffusers/models/unet_kandinsky3.py
similarity index 100%
rename from src/diffusers/models/unet_2d_condition_kandi3.py
rename to src/diffusers/models/unet_kandinsky3.py

From d60bc4e4c7d6f21cbe8d561319ccf788de553367 Mon Sep 17 00:00:00 2001
From: yiyixu <yixu310@gmail.com>
Date: Wed, 29 Nov 2023 18:41:53 +0000
Subject: [PATCH 17/20] testing prints

---
 src/diffusers/models/attention_processor.py |  4 +---
 src/diffusers/models/unet_kandinsky3.py     | 11 +++++++----
 2 files changed, 8 insertions(+), 7 deletions(-)

diff --git a/src/diffusers/models/attention_processor.py b/src/diffusers/models/attention_processor.py
index 12939f6eabf2..40a335527ace 100644
--- a/src/diffusers/models/attention_processor.py
+++ b/src/diffusers/models/attention_processor.py
@@ -109,7 +109,6 @@ def __init__(
         residual_connection: bool = False,
         _from_deprecated_attn_block: bool = False,
         processor: Optional["AttnProcessor"] = None,
-        scale_mask_factor: float = 1.0,
         out_dim: int = None,
     ):
         super().__init__()
@@ -121,7 +120,6 @@ def __init__(
         self.residual_connection = residual_connection
         self.dropout = dropout
         self.out_dim = out_dim if out_dim is not None else query_dim
-        self.scale_mask_factor = scale_mask_factor
 
         # we make use of this private variable to know whether this class is loaded
         # with an deprecated state dict so that we can convert it on the fly
@@ -597,7 +595,7 @@ def get_attention_scores(
             beta = 0
         else:
             baddbmm_input = attention_mask
-            beta = self.scale_mask_factor
+            beta = 1
 
         attention_scores = torch.baddbmm(
             baddbmm_input,
diff --git a/src/diffusers/models/unet_kandinsky3.py b/src/diffusers/models/unet_kandinsky3.py
index 1ce831393c37..620220b9c9ed 100644
--- a/src/diffusers/models/unet_kandinsky3.py
+++ b/src/diffusers/models/unet_kandinsky3.py
@@ -210,6 +210,11 @@ def _set_gradient_checkpointing(self, module, value=False):
             module.gradient_checkpointing = value
 
     def forward(self, sample, timestep, encoder_hidden_states=None, encoder_attention_mask=None, return_dict=True):
+        
+        if encoder_attention_mask is not None:
+            encoder_attention_mask = (1 - encoder_attention_mask.to(sample.dtype)) * -10000.0
+            encoder_attention_mask = encoder_attention_mask.unsqueeze(1)
+        
         if not torch.is_tensor(timestep):
             dtype = torch.float32 if isinstance(timestep, float) else torch.int32
             timestep = torch.tensor([timestep], dtype=dtype, device=sample.device)
@@ -485,11 +490,10 @@ def __init__(self, num_channels, context_dim, head_dim=64):
             dim_head=head_dim,
             out_dim=num_channels,
             out_bias=False,
-            scale_mask_factor=-60000.0,
         )
 
     def forward(self, x, context, context_mask=None):
-        context_mask = context_mask.unsqueeze(1).to(dtype=context.dtype)
+        context_mask = context_mask.to(dtype=context.dtype)
         context = self.attention(context.mean(dim=1, keepdim=True), context, context_mask)
         return x + context.squeeze(1)
 
@@ -504,7 +508,6 @@ def __init__(self, num_channels, time_embed_dim, context_dim=None, norm_groups=3
             dim_head=head_dim,
             out_dim=num_channels,
             out_bias=False,
-            scale_mask_factor=-60000.0,
         )
 
         hidden_channels = expansion_ratio * num_channels
@@ -521,7 +524,7 @@ def forward(self, x, time_embed, context=None, context_mask=None, image_mask=Non
         out = out.reshape(x.shape[0], -1, height * width).permute(0, 2, 1)
         context = context if context is not None else out
         if context_mask is not None:
-            context_mask = context_mask.unsqueeze(1).to(dtype=context.dtype)
+            context_mask = context_mask.to(dtype=context.dtype)
 
         out = self.attention(out, context, context_mask)
         out = out.permute(0, 2, 1).unsqueeze(-1).reshape(out.shape[0], -1, height, width)

From 1d170cc08add986c5a61cacc67be73e8f37a52fd Mon Sep 17 00:00:00 2001
From: yiyixu <yixu310@gmail.com>
Date: Wed, 29 Nov 2023 23:31:02 +0000
Subject: [PATCH 18/20] style

---
 src/diffusers/models/unet_kandinsky3.py | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/src/diffusers/models/unet_kandinsky3.py b/src/diffusers/models/unet_kandinsky3.py
index 620220b9c9ed..eef3287e5d99 100644
--- a/src/diffusers/models/unet_kandinsky3.py
+++ b/src/diffusers/models/unet_kandinsky3.py
@@ -210,11 +210,10 @@ def _set_gradient_checkpointing(self, module, value=False):
             module.gradient_checkpointing = value
 
     def forward(self, sample, timestep, encoder_hidden_states=None, encoder_attention_mask=None, return_dict=True):
-        
         if encoder_attention_mask is not None:
             encoder_attention_mask = (1 - encoder_attention_mask.to(sample.dtype)) * -10000.0
             encoder_attention_mask = encoder_attention_mask.unsqueeze(1)
-        
+
         if not torch.is_tensor(timestep):
             dtype = torch.float32 if isinstance(timestep, float) else torch.int32
             timestep = torch.tensor([timestep], dtype=dtype, device=sample.device)

From c4eae7e6405bab60d66971de9c485216e3f92f89 Mon Sep 17 00:00:00 2001
From: yiyixu <yixu310@gmail.com>
Date: Thu, 30 Nov 2023 04:17:42 +0000
Subject: [PATCH 19/20] allow pass prompt_embeds

---
 .../kandinsky3/pipeline_kandinsky3.py         | 39 ++++++++++++++++++
 .../kandinsky3/pipeline_kandinsky3_img2img.py | 40 +++++++++++++++++++
 2 files changed, 79 insertions(+)

diff --git a/src/diffusers/pipelines/kandinsky3/pipeline_kandinsky3.py b/src/diffusers/pipelines/kandinsky3/pipeline_kandinsky3.py
index dab0d15918b7..4d14fc637b05 100644
--- a/src/diffusers/pipelines/kandinsky3/pipeline_kandinsky3.py
+++ b/src/diffusers/pipelines/kandinsky3/pipeline_kandinsky3.py
@@ -103,6 +103,8 @@ def encode_prompt(
         prompt_embeds: Optional[torch.FloatTensor] = None,
         negative_prompt_embeds: Optional[torch.FloatTensor] = None,
         _cut_context=False,
+        attention_mask: Optional[torch.FloatTensor] = None,
+        negative_attention_mask: Optional[torch.FloatTensor] = None,
     ):
         r"""
         Encodes the prompt into text encoder hidden states.
@@ -127,6 +129,10 @@ def encode_prompt(
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                 weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
                 argument.
+            attention_mask (`torch.FloatTensor`, *optional*):
+                Pre-generated attention mask. Must provide if passing `prompt_embeds` directly.
+            negative_attention_mask (`torch.FloatTensor`, *optional*):
+                Pre-generated negative attention mask. Must provide if passing `negative_prompt_embeds` directly.
         """
         if prompt is not None and negative_prompt is not None:
             if type(prompt) is not type(negative_prompt):
@@ -255,6 +261,8 @@ def check_inputs(
         prompt_embeds=None,
         negative_prompt_embeds=None,
         callback_on_step_end_tensor_inputs=None,
+        attention_mask=None,
+        negative_attention_mask=None,
     ):
         if callback_steps is not None and (not isinstance(callback_steps, int) or callback_steps <= 0):
             raise ValueError(
@@ -293,6 +301,27 @@ def check_inputs(
                     f" got: `prompt_embeds` {prompt_embeds.shape} != `negative_prompt_embeds`"
                     f" {negative_prompt_embeds.shape}."
                 )
+        if negative_prompt_embeds is not None and negative_attention_mask is None:
+            raise ValueError("Please provide `negative_attention_mask` along with `negative_prompt_embeds`")
+
+        if negative_prompt_embeds is not None and negative_attention_mask is not None:
+            if negative_prompt_embeds.shape[:2] != negative_attention_mask.shape:
+                raise ValueError(
+                    "`negative_prompt_embeds` and `negative_attention_mask` must have the same batch_size and token length when passed directly, but"
+                    f" got: `negative_prompt_embeds` {negative_prompt_embeds.shape[:2]} != `negative_attention_mask`"
+                    f" {negative_attention_mask.shape}."
+                )
+
+        if prompt_embeds is not None and attention_mask is None:
+            raise ValueError("Please provide `attention_mask` along with `prompt_embeds`")
+
+        if prompt_embeds is not None and attention_mask is not None:
+            if prompt_embeds.shape[:2] != attention_mask.shape:
+                raise ValueError(
+                    "`prompt_embeds` and `attention_mask` must have the same batch_size and token length when passed directly, but"
+                    f" got: `prompt_embeds` {prompt_embeds.shape[:2]} != `attention_mask`"
+                    f" {attention_mask.shape}."
+                )
 
     @property
     def guidance_scale(self):
@@ -320,6 +349,8 @@ def __call__(
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
         prompt_embeds: Optional[torch.FloatTensor] = None,
         negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        attention_mask: Optional[torch.FloatTensor] = None,
+        negative_attention_mask: Optional[torch.FloatTensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
         latents=None,
@@ -369,6 +400,10 @@ def __call__(
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                 weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
                 argument.
+            attention_mask (`torch.FloatTensor`, *optional*):
+                Pre-generated attention mask. Must provide if passing `prompt_embeds` directly.
+            negative_attention_mask (`torch.FloatTensor`, *optional*):
+                Pre-generated negative attention mask. Must provide if passing `negative_prompt_embeds` directly.
             output_type (`str`, *optional*, defaults to `"pil"`):
                 The output format of the generate image. Choose between
                 [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
@@ -430,6 +465,8 @@ def __call__(
             prompt_embeds,
             negative_prompt_embeds,
             callback_on_step_end_tensor_inputs,
+            attention_mask,
+            negative_attention_mask,
         )
 
         self._guidance_scale = guidance_scale
@@ -451,6 +488,8 @@ def __call__(
             prompt_embeds=prompt_embeds,
             negative_prompt_embeds=negative_prompt_embeds,
             _cut_context=cut_context,
+            attention_mask=attention_mask,
+            negative_attention_mask=negative_attention_mask,
         )
 
         if self.do_classifier_free_guidance:
diff --git a/src/diffusers/pipelines/kandinsky3/pipeline_kandinsky3_img2img.py b/src/diffusers/pipelines/kandinsky3/pipeline_kandinsky3_img2img.py
index b830179196a5..edeb3955ec76 100644
--- a/src/diffusers/pipelines/kandinsky3/pipeline_kandinsky3_img2img.py
+++ b/src/diffusers/pipelines/kandinsky3/pipeline_kandinsky3_img2img.py
@@ -126,6 +126,8 @@ def encode_prompt(
         prompt_embeds: Optional[torch.FloatTensor] = None,
         negative_prompt_embeds: Optional[torch.FloatTensor] = None,
         _cut_context=False,
+        attention_mask: Optional[torch.FloatTensor] = None,
+        negative_attention_mask: Optional[torch.FloatTensor] = None,
     ):
         r"""
         Encodes the prompt into text encoder hidden states.
@@ -150,6 +152,10 @@ def encode_prompt(
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                 weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
                 argument.
+            attention_mask (`torch.FloatTensor`, *optional*):
+                Pre-generated attention mask. Must provide if passing `prompt_embeds` directly.
+            negative_attention_mask (`torch.FloatTensor`, *optional*):
+                Pre-generated negative attention mask. Must provide if passing `negative_prompt_embeds` directly.
         """
         if prompt is not None and negative_prompt is not None:
             if type(prompt) is not type(negative_prompt):
@@ -327,6 +333,8 @@ def check_inputs(
         prompt_embeds=None,
         negative_prompt_embeds=None,
         callback_on_step_end_tensor_inputs=None,
+        attention_mask=None,
+        negative_attention_mask=None,
     ):
         if callback_steps is not None and (not isinstance(callback_steps, int) or callback_steps <= 0):
             raise ValueError(
@@ -367,6 +375,28 @@ def check_inputs(
                     f" {negative_prompt_embeds.shape}."
                 )
 
+        if negative_prompt_embeds is not None and negative_attention_mask is None:
+            raise ValueError("Please provide `negative_attention_mask` along with `negative_prompt_embeds`")
+
+        if negative_prompt_embeds is not None and negative_attention_mask is not None:
+            if negative_prompt_embeds.shape[:2] != negative_attention_mask.shape:
+                raise ValueError(
+                    "`negative_prompt_embeds` and `negative_attention_mask` must have the same batch_size and token length when passed directly, but"
+                    f" got: `negative_prompt_embeds` {negative_prompt_embeds.shape[:2]} != `negative_attention_mask`"
+                    f" {negative_attention_mask.shape}."
+                )
+
+        if prompt_embeds is not None and attention_mask is None:
+            raise ValueError("Please provide `attention_mask` along with `prompt_embeds`")
+
+        if prompt_embeds is not None and attention_mask is not None:
+            if prompt_embeds.shape[:2] != attention_mask.shape:
+                raise ValueError(
+                    "`prompt_embeds` and `attention_mask` must have the same batch_size and token length when passed directly, but"
+                    f" got: `prompt_embeds` {prompt_embeds.shape[:2]} != `attention_mask`"
+                    f" {attention_mask.shape}."
+                )
+
     @property
     def guidance_scale(self):
         return self._guidance_scale
@@ -393,6 +423,8 @@ def __call__(
         generator: Optional[Union[torch.Generator, List[torch.Generator]]] = None,
         prompt_embeds: Optional[torch.FloatTensor] = None,
         negative_prompt_embeds: Optional[torch.FloatTensor] = None,
+        attention_mask: Optional[torch.FloatTensor] = None,
+        negative_attention_mask: Optional[torch.FloatTensor] = None,
         output_type: Optional[str] = "pil",
         return_dict: bool = True,
         callback_on_step_end: Optional[Callable[[int, int, Dict], None]] = None,
@@ -440,6 +472,10 @@ def __call__(
                 Pre-generated negative text embeddings. Can be used to easily tweak text inputs, *e.g.* prompt
                 weighting. If not provided, negative_prompt_embeds will be generated from `negative_prompt` input
                 argument.
+            attention_mask (`torch.FloatTensor`, *optional*):
+                Pre-generated attention mask. Must provide if passing `prompt_embeds` directly.
+            negative_attention_mask (`torch.FloatTensor`, *optional*):
+                Pre-generated negative attention mask. Must provide if passing `negative_prompt_embeds` directly.
             output_type (`str`, *optional*, defaults to `"pil"`):
                 The output format of the generate image. Choose between
                 [PIL](https://pillow.readthedocs.io/en/stable/): `PIL.Image.Image` or `np.array`.
@@ -493,6 +529,8 @@ def __call__(
             prompt_embeds,
             negative_prompt_embeds,
             callback_on_step_end_tensor_inputs,
+            attention_mask,
+            negative_attention_mask,
         )
 
         self._guidance_scale = guidance_scale
@@ -516,6 +554,8 @@ def __call__(
             prompt_embeds=prompt_embeds,
             negative_prompt_embeds=negative_prompt_embeds,
             _cut_context=cut_context,
+            attention_mask=attention_mask,
+            negative_attention_mask=negative_attention_mask,
         )
 
         if self.do_classifier_free_guidance:

From 226f755091b70ba2ac78bad075aaa46a4cdba26a Mon Sep 17 00:00:00 2001
From: yiyixu <yixu310@gmail.com>
Date: Thu, 30 Nov 2023 06:14:50 +0000
Subject: [PATCH 20/20] offload

---
 .../pipelines/kandinsky3/pipeline_kandinsky3_img2img.py         | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/diffusers/pipelines/kandinsky3/pipeline_kandinsky3_img2img.py b/src/diffusers/pipelines/kandinsky3/pipeline_kandinsky3_img2img.py
index edeb3955ec76..7f4164a04d1e 100644
--- a/src/diffusers/pipelines/kandinsky3/pipeline_kandinsky3_img2img.py
+++ b/src/diffusers/pipelines/kandinsky3/pipeline_kandinsky3_img2img.py
@@ -60,7 +60,7 @@ def prepare_image(pil_image):
 
 
 class Kandinsky3Img2ImgPipeline(DiffusionPipeline, LoraLoaderMixin):
-    model_cpu_offload_seq = "text_encoder->unet->movq"
+    model_cpu_offload_seq = "text_encoder->movq->unet->movq"
     _callback_tensor_inputs = [
         "latents",
         "prompt_embeds",