save intermediate (#87)

* save intermediate

* up

* up

conversion.py

+# coding=utf-8
+# Copyright 2022 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 inspect
+import tempfile
+import unittest
+
+import numpy as np
+import torch
+
+from diffusers import (
+    AutoencoderKL,
+    DDIMPipeline,
+    DDIMScheduler,
+    DDPMPipeline,
+    DDPMScheduler,
+    GlidePipeline,
+    GlideSuperResUNetModel,
+    GlideTextToImageUNetModel,
+    LatentDiffusionPipeline,
+    LatentDiffusionUncondPipeline,
+    NCSNpp,
+    PNDMPipeline,
+    PNDMScheduler,
+    ScoreSdeVePipeline,
+    ScoreSdeVeScheduler,
+    ScoreSdeVpPipeline,
+    ScoreSdeVpScheduler,
+    UNetLDMModel,
+    UNetModel,
+    UNetUnconditionalModel,
+    VQModel,
+)
+from diffusers.configuration_utils import ConfigMixin
+from diffusers.pipeline_utils import DiffusionPipeline
+from diffusers.testing_utils import floats_tensor, slow, torch_device
+from diffusers.training_utils import EMAModel
+
+
+def test_output_pretrained_ldm_dummy():
+    model = UNetUnconditionalModel.from_pretrained("fusing/unet-ldm-dummy", ldm=True)
+    model.eval()
+
+    torch.manual_seed(0)
+    if torch.cuda.is_available():
+        torch.cuda.manual_seed_all(0)
+
+    noise = torch.randn(1, model.config.in_channels, model.config.image_size, model.config.image_size)
+    time_step = torch.tensor([10] * noise.shape[0])
+
+    with torch.no_grad():
+        output = model(noise, time_step)
+        
+    print(model)
+    import ipdb; ipdb.set_trace()
+
+
+def test_output_pretrained_ldm():
+    model = UNetUnconditionalModel.from_pretrained("fusing/latent-diffusion-celeba-256", subfolder="unet", ldm=True)
+    model.eval()
+
+    torch.manual_seed(0)
+    if torch.cuda.is_available():
+        torch.cuda.manual_seed_all(0)
+
+    noise = torch.randn(1, model.config.in_channels, model.config.image_size, model.config.image_size)
+    time_step = torch.tensor([10] * noise.shape[0])
+
+    with torch.no_grad():
+        output = model(noise, time_step)
+            
+    print(model)
+    import ipdb; ipdb.set_trace()
+
+# To see the how the final model should look like
+test_output_pretrained_ldm_dummy()
+test_output_pretrained_ldm()
+# => this is the architecture in which the model should be saved in the new format
+# -> verify new repo with the following tests (in `test_modeling_utils.py`)
+# - test_ldm_uncond (in PipelineTesterMixin)
+# - test_output_pretrained  ( in UNetLDMModelTests)

docs/source/examples/diffusers_for_vision.mdx

@@ -111,7 +111,7 @@ prompt = "A painting of a squirrel eating a burger"
 image = ldm([prompt], generator=generator, eta=0.3, guidance_scale=6.0, num_inference_steps=50)
 
 image_processed = image.cpu().permute(0, 2, 3, 1)
-image_processed = image_processed  * 255.
+image_processed = image_processed * 255.0
 image_processed = image_processed.numpy().astype(np.uint8)
 image_pil = PIL.Image.fromarray(image_processed[0])
 
@@ -143,6 +143,7 @@ audio = bddm(mel_spec, generator, torch_device=torch_device)
 
 # save generated audio
 from scipy.io.wavfile import write as wavwrite
+
 sampling_rate = 22050
 wavwrite("generated_audio.wav", sampling_rate, audio.squeeze().cpu().numpy())
 ```

src/diffusers/configuration_utils.py

@@ -116,6 +116,7 @@ class ConfigMixin:
         use_auth_token = kwargs.pop("use_auth_token", None)
         local_files_only = kwargs.pop("local_files_only", False)
         revision = kwargs.pop("revision", None)
+        subfolder = kwargs.pop("subfolder", None)
 
         user_agent = {"file_type": "config"}
 
@@ -150,6 +151,7 @@ class ConfigMixin:
                     local_files_only=local_files_only,
                     use_auth_token=use_auth_token,
                     user_agent=user_agent,
+                    subfolder=subfolder,
                 )
 
             except RepositoryNotFoundError:

src/diffusers/modeling_utils.py

@@ -321,6 +321,7 @@ class ModelMixin(torch.nn.Module):
         use_auth_token = kwargs.pop("use_auth_token", None)
         revision = kwargs.pop("revision", None)
         from_auto_class = kwargs.pop("_from_auto", False)
+        subfolder = kwargs.pop("subfolder", None)
 
         user_agent = {"file_type": "model", "framework": "pytorch", "from_auto_class": from_auto_class}
 
@@ -336,6 +337,7 @@ class ModelMixin(torch.nn.Module):
             local_files_only=local_files_only,
             use_auth_token=use_auth_token,
             revision=revision,
+            subfolder=subfolder,
             **kwargs,
         )
         model.register_to_config(name_or_path=pretrained_model_name_or_path)
@@ -363,6 +365,7 @@ class ModelMixin(torch.nn.Module):
                     local_files_only=local_files_only,
                     use_auth_token=use_auth_token,
                     user_agent=user_agent,
+                    subfolder=subfolder,
                 )
 
             except RepositoryNotFoundError:

src/diffusers/models/attention.py

@@ -51,6 +51,7 @@ class AttentionBlock(nn.Module):
         overwrite_qkv=False,
         overwrite_linear=False,
         rescale_output_factor=1.0,
+        eps=1e-5,
     ):
         super().__init__()
         self.channels = channels
@@ -62,7 +63,7 @@ class AttentionBlock(nn.Module):
             ), f"q,k,v channels {channels} is not divisible by num_head_channels {num_head_channels}"
             self.num_heads = channels // num_head_channels
 
-        self.norm = nn.GroupNorm(num_channels=channels, num_groups=num_groups, eps=1e-5, affine=True)
+        self.norm = nn.GroupNorm(num_channels=channels, num_groups=num_groups, eps=eps, affine=True)
         self.qkv = nn.Conv1d(channels, channels * 3, 1)
         self.n_heads = self.num_heads
         self.rescale_output_factor = rescale_output_factor
@@ -165,7 +166,7 @@ class AttentionBlock(nn.Module):
         return result
 
 
-class AttentionBlockNew(nn.Module):
+class AttentionBlockNew_2(nn.Module):
     """
     An attention block that allows spatial positions to attend to each other.
 
@@ -180,11 +181,14 @@ class AttentionBlockNew(nn.Module):
         num_groups=32,
         encoder_channels=None,
         rescale_output_factor=1.0,
+        eps=1e-5,
     ):
         super().__init__()
-        self.norm = nn.GroupNorm(num_channels=channels, num_groups=num_groups, eps=1e-5, affine=True)
+        self.channels = channels
+        self.norm = nn.GroupNorm(num_channels=channels, num_groups=num_groups, eps=eps, affine=True)
         self.qkv = nn.Conv1d(channels, channels * 3, 1)
         self.n_heads = channels // num_head_channels
+        self.num_head_size = num_head_channels
         self.rescale_output_factor = rescale_output_factor
 
         if encoder_channels is not None:
@@ -192,6 +196,28 @@ class AttentionBlockNew(nn.Module):
 
         self.proj = zero_module(nn.Conv1d(channels, channels, 1))
 
+        # ------------------------- new -----------------------
+        num_heads = self.n_heads
+        self.channels = channels
+        if num_head_channels is None:
+            self.num_heads = num_heads
+        else:
+            assert (
+                channels % num_head_channels == 0
+            ), f"q,k,v channels {channels} is not divisible by num_head_channels {num_head_channels}"
+            self.num_heads = channels // num_head_channels
+
+        self.group_norm = nn.GroupNorm(num_channels=channels, num_groups=num_groups, eps=eps, affine=True)
+
+        # define q,k,v as linear layers
+        self.query = nn.Linear(channels, channels)
+        self.key = nn.Linear(channels, channels)
+        self.value = nn.Linear(channels, channels)
+
+        self.rescale_output_factor = rescale_output_factor
+        self.proj_attn = zero_module(nn.Linear(channels, channels, 1))
+        # ------------------------- new -----------------------
+
     def set_weight(self, attn_layer):
         self.norm.weight.data = attn_layer.norm.weight.data
         self.norm.bias.data = attn_layer.norm.bias.data
@@ -202,6 +228,89 @@ class AttentionBlockNew(nn.Module):
         self.proj.weight.data = attn_layer.proj.weight.data
         self.proj.bias.data = attn_layer.proj.bias.data
 
+        if hasattr(attn_layer, "q"):
+            module = attn_layer
+            qkv_weight = torch.cat([module.q.weight.data, module.k.weight.data, module.v.weight.data], dim=0)[
+                :, :, :, 0
+            ]
+            qkv_bias = torch.cat([module.q.bias.data, module.k.bias.data, module.v.bias.data], dim=0)
+
+            self.qkv.weight.data = qkv_weight
+            self.qkv.bias.data = qkv_bias
+
+            proj_out = zero_module(nn.Conv1d(self.channels, self.channels, 1))
+            proj_out.weight.data = module.proj_out.weight.data[:, :, :, 0]
+            proj_out.bias.data = module.proj_out.bias.data
+
+            self.proj = proj_out
+
+        self.set_weights_2(attn_layer)
+
+    def transpose_for_scores(self, projection: torch.Tensor) -> torch.Tensor:
+        new_projection_shape = projection.size()[:-1] + (self.n_heads, self.num_head_size)
+        # move heads to 2nd position (B, T, H * D) -> (B, T, H, D) -> (B, H, T, D)
+        new_projection = projection.view(new_projection_shape).permute(0, 2, 1, 3)
+        return new_projection
+
+    def set_weights_2(self, attn_layer):
+        self.group_norm.weight.data = attn_layer.norm.weight.data
+        self.group_norm.bias.data = attn_layer.norm.bias.data
+
+        qkv_weight = attn_layer.qkv.weight.data.reshape(self.n_heads, 3 * self.channels // self.n_heads, self.channels)
+        qkv_bias = attn_layer.qkv.bias.data.reshape(self.n_heads, 3 * self.channels // self.n_heads)
+
+        q_w, k_w, v_w = qkv_weight.split(self.channels // self.n_heads, dim=1)
+        q_b, k_b, v_b = qkv_bias.split(self.channels // self.n_heads, dim=1)
+
+        self.query.weight.data = q_w.reshape(-1, self.channels)
+        self.key.weight.data = k_w.reshape(-1, self.channels)
+        self.value.weight.data = v_w.reshape(-1, self.channels)
+
+        self.query.bias.data = q_b.reshape(-1)
+        self.key.bias.data = k_b.reshape(-1)
+        self.value.bias.data = v_b.reshape(-1)
+
+        self.proj_attn.weight.data = attn_layer.proj.weight.data[:, :, 0]
+        self.proj_attn.bias.data = attn_layer.proj.bias.data
+
+    def forward_2(self, hidden_states):
+        residual = hidden_states
+        batch, channel, height, width = hidden_states.shape
+
+        # norm
+        hidden_states = self.group_norm(hidden_states)
+        hidden_states = hidden_states.view(batch, channel, height * width).transpose(1, 2)
+
+        # proj to q, k, v
+        query_proj = self.query(hidden_states)
+        key_proj = self.key(hidden_states)
+        value_proj = self.value(hidden_states)
+
+        # transpose
+        query_states = self.transpose_for_scores(query_proj)
+        key_states = self.transpose_for_scores(key_proj)
+        value_states = self.transpose_for_scores(value_proj)
+
+        # get scores
+        attention_scores = torch.matmul(query_states, key_states.transpose(-1, -2))
+        attention_scores = attention_scores / math.sqrt(self.channels // self.n_heads)
+        attention_probs = nn.functional.softmax(attention_scores, dim=-1)
+
+        # compute attention output
+        context_states = torch.matmul(attention_probs, value_states)
+
+        context_states = context_states.permute(0, 2, 1, 3).contiguous()
+        new_context_states_shape = context_states.size()[:-2] + (self.channels,)
+        context_states = context_states.view(new_context_states_shape)
+
+        # compute next hidden_states
+        hidden_states = self.proj_attn(context_states)
+        hidden_states = hidden_states.transpose(-1, -2).reshape(batch, channel, height, width)
+
+        # res connect and rescale
+        hidden_states = (hidden_states + residual) / self.rescale_output_factor
+        return hidden_states
+
     def forward(self, x, encoder_out=None):
         b, c, *spatial = x.shape
         hid_states = self.norm(x).view(b, c, -1)
@@ -230,10 +339,119 @@ class AttentionBlockNew(nn.Module):
         h = h.reshape(b, c, *spatial)
 
         result = x + h
-
         result = result / self.rescale_output_factor
 
-        return result
+        result_2 = self.forward_2(x)
+
+        print((result - result_2).abs().sum())
+
+        return result_2
+
+
+class AttentionBlockNew(nn.Module):
+    """
+    An attention block that allows spatial positions to attend to each other. Originally ported from here, but adapted
+    to the N-d case.
+    https://github.com/hojonathanho/diffusion/blob/1e0dceb3b3495bbe19116a5e1b3596cd0706c543/diffusion_tf/models/unet.py#L66.
+    Uses three q, k, v linear layers to compute attention
+    """
+
+    def __init__(
+        self,
+        channels,
+        num_heads=1,
+        num_head_channels=None,
+        num_groups=32,
+        rescale_output_factor=1.0,
+        eps=1e-5,
+    ):
+        super().__init__()
+        self.channels = channels
+        if num_head_channels is None:
+            self.num_heads = num_heads
+        else:
+            assert (
+                channels % num_head_channels == 0
+            ), f"q,k,v channels {channels} is not divisible by num_head_channels {num_head_channels}"
+            self.num_heads = channels // num_head_channels
+
+        self.num_head_size = num_head_channels
+        self.group_norm = nn.GroupNorm(num_channels=channels, num_groups=num_groups, eps=eps, affine=True)
+
+        # define q,k,v as linear layers
+        self.query = nn.Linear(channels, channels)
+        self.key = nn.Linear(channels, channels)
+        self.value = nn.Linear(channels, channels)
+
+        self.rescale_output_factor = rescale_output_factor
+        self.proj_attn = zero_module(nn.Linear(channels, channels, 1))
+
+    def transpose_for_scores(self, projection: torch.Tensor) -> torch.Tensor:
+        new_projection_shape = projection.size()[:-1] + (self.num_heads, self.num_head_size)
+        # move heads to 2nd position (B, T, H * D) -> (B, T, H, D) -> (B, H, T, D)
+        new_projection = projection.view(new_projection_shape).permute(0, 2, 1, 3)
+        return new_projection
+
+    def forward(self, hidden_states):
+        residual = hidden_states
+        batch, channel, height, width = hidden_states.shape
+
+        # norm
+        hidden_states = self.group_norm(hidden_states)
+        hidden_states = hidden_states.view(batch, channel, height * width).transpose(1, 2)
+
+        # proj to q, k, v
+        query_proj = self.query(hidden_states)
+        key_proj = self.key(hidden_states)
+        value_proj = self.value(hidden_states)
+
+        # transpose
+        query_states = self.transpose_for_scores(query_proj)
+        key_states = self.transpose_for_scores(key_proj)
+        value_states = self.transpose_for_scores(value_proj)
+
+        # get scores
+        attention_scores = torch.matmul(query_states, key_states.transpose(-1, -2))
+        attention_scores = attention_scores / math.sqrt(self.channels // self.num_heads)
+        attention_probs = nn.functional.softmax(attention_scores, dim=-1)
+
+        # compute attention output
+        context_states = torch.matmul(attention_probs, value_states)
+
+        context_states = context_states.permute(0, 2, 1, 3).contiguous()
+        new_context_states_shape = context_states.size()[:-2] + (self.channels,)
+        context_states = context_states.view(new_context_states_shape)
+
+        # compute next hidden_states
+        hidden_states = self.proj_attn(context_states)
+        hidden_states = hidden_states.transpose(-1, -2).reshape(batch, channel, height, width)
+
+        # res connect and rescale
+        hidden_states = (hidden_states + residual) / self.rescale_output_factor
+        return hidden_states
+
+    def set_weight(self, attn_layer):
+        self.group_norm.weight.data = attn_layer.norm.weight.data
+        self.group_norm.bias.data = attn_layer.norm.bias.data
+
+        qkv_weight = attn_layer.qkv.weight.data.reshape(
+            self.num_heads, 3 * self.channels // self.num_heads, self.channels
+        )
+        qkv_bias = attn_layer.qkv.bias.data.reshape(self.num_heads, 3 * self.channels // self.num_heads)
+
+        q_w, k_w, v_w = qkv_weight.split(self.channels // self.num_heads, dim=1)
+        q_b, k_b, v_b = qkv_bias.split(self.channels // self.num_heads, dim=1)
+
+        self.query.weight.data = q_w.reshape(-1, self.channels)
+        self.key.weight.data = k_w.reshape(-1, self.channels)
+        self.value.weight.data = v_w.reshape(-1, self.channels)
+
+        self.query.bias.data = q_b.reshape(-1)
+        self.key.bias.data = k_b.reshape(-1)
+        self.value.bias.data = v_b.reshape(-1)
+
+        self.proj_attn.weight.data = attn_layer.proj.weight.data[:, :, 0]
+        self.proj_attn.bias.data = attn_layer.proj.bias.data
 
 
 class SpatialTransformer(nn.Module):

src/diffusers/models/resnet.py

@@ -81,8 +81,10 @@ class Downsample2D(nn.Module):
             self.conv = conv
         elif name == "Conv2d_0":
             self.Conv2d_0 = conv
+            self.conv = conv
         else:
             self.op = conv
+            self.conv = conv
 
     def forward(self, x):
         assert x.shape[1] == self.channels
@@ -90,13 +92,16 @@ class Downsample2D(nn.Module):
             pad = (0, 1, 0, 1)
             x = F.pad(x, pad, mode="constant", value=0)
 
-        # TODO(Suraj, Patrick) - clean up after weight dicts are correctly renamed
-        if self.name == "conv":
         return self.conv(x)
-        elif self.name == "Conv2d_0":
-            return self.Conv2d_0(x)
-        else:
-            return self.op(x)
+        # TODO(Suraj, Patrick) - clean up after weight dicts are correctly renamed
+
+
+#        if self.name == "conv":
+#            return self.conv(x)
+#        elif self.name == "Conv2d_0":
+#            return self.Conv2d_0(x)
+#        else:
+#            return self.op(x)
 
 
 class Upsample1D(nn.Module):
@@ -656,9 +661,9 @@ class ResnetBlock(nn.Module):
         self.conv1 = torch.nn.Conv2d(in_channels, out_channels, kernel_size=3, stride=1, padding=1)
 
         if time_embedding_norm == "default" and temb_channels > 0:
-            self.temb_proj = torch.nn.Linear(temb_channels, out_channels)
+            self.time_emb_proj = torch.nn.Linear(temb_channels, out_channels)
         elif time_embedding_norm == "scale_shift" and temb_channels > 0:
-            self.temb_proj = torch.nn.Linear(temb_channels, 2 * out_channels)
+            self.time_emb_proj = torch.nn.Linear(temb_channels, 2 * out_channels)
 
         self.norm2 = torch.nn.GroupNorm(num_groups=groups_out, num_channels=out_channels, eps=eps, affine=True)
         self.dropout = torch.nn.Dropout(dropout)
@@ -691,9 +696,9 @@ class ResnetBlock(nn.Module):
 
         self.use_nin_shortcut = self.in_channels != self.out_channels if use_nin_shortcut is None else use_nin_shortcut
 
-        self.nin_shortcut = None
+        self.conv_shortcut = None
         if self.use_nin_shortcut:
-            self.nin_shortcut = torch.nn.Conv2d(in_channels, out_channels, kernel_size=1, stride=1, padding=0)
+            self.conv_shortcut = torch.nn.Conv2d(in_channels, out_channels, kernel_size=1, stride=1, padding=0)
 
     def forward(self, x, temb):
         h = x
@@ -715,7 +720,7 @@ class ResnetBlock(nn.Module):
             h = self.nonlinearity(h)
 
         if temb is not None:
-            temb = self.temb_proj(self.nonlinearity(temb))[:, :, None, None]
+            temb = self.time_emb_proj(self.nonlinearity(temb))[:, :, None, None]
         else:
             temb = 0
 
@@ -738,8 +743,8 @@ class ResnetBlock(nn.Module):
             h = self.norm2(h)
             h = self.nonlinearity(h)
 
-        if self.nin_shortcut is not None:
-            x = self.nin_shortcut(x)
+        if self.conv_shortcut is not None:
+            x = self.conv_shortcut(x)
 
         return (x + h) / self.output_scale_factor
 
@@ -750,8 +755,8 @@ class ResnetBlock(nn.Module):
         self.conv1.weight.data = resnet.conv1.weight.data
         self.conv1.bias.data = resnet.conv1.bias.data
 
-        self.temb_proj.weight.data = resnet.temb_proj.weight.data
-        self.temb_proj.bias.data = resnet.temb_proj.bias.data
+        self.time_emb_proj.weight.data = resnet.temb_proj.weight.data
+        self.time_emb_proj.bias.data = resnet.temb_proj.bias.data
 
         self.norm2.weight.data = resnet.norm2.weight.data
         self.norm2.bias.data = resnet.norm2.bias.data
@@ -760,8 +765,8 @@ class ResnetBlock(nn.Module):
         self.conv2.bias.data = resnet.conv2.bias.data
 
         if self.use_nin_shortcut:
-            self.nin_shortcut.weight.data = resnet.nin_shortcut.weight.data
-            self.nin_shortcut.bias.data = resnet.nin_shortcut.bias.data
+            self.conv_shortcut.weight.data = resnet.nin_shortcut.weight.data
+            self.conv_shortcut.bias.data = resnet.nin_shortcut.bias.data
 
 
 # TODO(Patrick) - just there to convert the weights; can delete afterward

src/diffusers/models/unet.py

@@ -177,7 +177,9 @@ class UNetModel(ModelMixin, ConfigMixin):
                 hs.append(self.down[i_level].downsample(hs[-1]))
 
         # middle
+        print("hs", hs[-1].abs().sum())
         h = self.mid_new(hs[-1], temb)
+        print("h", h.abs().sum())
 
         # upsampling
         for i_level in reversed(range(self.num_resolutions)):

src/diffusers/models/unet_new.py

@@ -29,9 +29,10 @@ def get_down_block(
     resnet_eps,
     resnet_act_fn,
     attn_num_head_channels,
+    downsample_padding=None,
 ):
     if down_block_type == "UNetResDownBlock2D":
-        return UNetResAttnDownBlock2D(
+        return UNetResDownBlock2D(
             num_layers=num_layers,
             in_channels=in_channels,
             out_channels=out_channels,
@@ -39,6 +40,7 @@ def get_down_block(
             add_downsample=add_downsample,
             resnet_eps=resnet_eps,
             resnet_act_fn=resnet_act_fn,
+            downsample_padding=downsample_padding,
         )
     elif down_block_type == "UNetResAttnDownBlock2D":
         return UNetResAttnDownBlock2D(
@@ -57,7 +59,8 @@ def get_up_block(
     up_block_type,
     num_layers,
     in_channels,
-    next_channels,
+    out_channels,
+    prev_output_channel,
     temb_channels,
     add_upsample,
     resnet_eps,
@@ -68,7 +71,8 @@ def get_up_block(
         return UNetResUpBlock2D(
             num_layers=num_layers,
             in_channels=in_channels,
-            next_channels=next_channels,
+            out_channels=out_channels,
+            prev_output_channel=prev_output_channel,
             temb_channels=temb_channels,
             add_upsample=add_upsample,
             resnet_eps=resnet_eps,
@@ -78,7 +82,8 @@ def get_up_block(
         return UNetResAttnUpBlock2D(
             num_layers=num_layers,
             in_channels=in_channels,
-            next_channels=next_channels,
+            out_channels=out_channels,
+            prev_output_channel=prev_output_channel,
             temb_channels=temb_channels,
             add_upsample=add_upsample,
             resnet_eps=resnet_eps,
@@ -100,11 +105,14 @@ class UNetMidBlock2D(nn.Module):
         resnet_groups: int = 32,
         resnet_pre_norm: bool = True,
         attn_num_head_channels=1,
+        attention_type="default",
         output_scale_factor=1.0,
         **kwargs,
     ):
         super().__init__()
 
+        self.attention_type = attention_type
+
         # there is always at least one resnet
         resnets = [
             ResnetBlock(
@@ -128,6 +136,7 @@ class UNetMidBlock2D(nn.Module):
                     in_channels,
                     num_head_channels=attn_num_head_channels,
                     rescale_output_factor=output_scale_factor,
+                    eps=resnet_eps,
                 )
             )
             resnets.append(
@@ -148,17 +157,14 @@ class UNetMidBlock2D(nn.Module):
         self.attentions = nn.ModuleList(attentions)
         self.resnets = nn.ModuleList(resnets)
 
-    def forward(self, hidden_states, temb=None, encoder_states=None, mask=None):
-        if mask is not None:
-            hidden_states = self.resnets[0](hidden_states, temb, mask=mask)
-        else:
+    def forward(self, hidden_states, temb=None, encoder_states=None):
         hidden_states = self.resnets[0](hidden_states, temb)
 
         for attn, resnet in zip(self.attentions, self.resnets[1:]):
-            hidden_states = attn(hidden_states, encoder_states)
-            if mask is not None:
-                hidden_states = resnet(hidden_states, temb, mask=mask)
+            if self.attention_type == "default":
+                hidden_states = attn(hidden_states)
             else:
+                hidden_states = attn(hidden_states, encoder_states)
             hidden_states = resnet(hidden_states, temb)
 
         return hidden_states
@@ -178,6 +184,7 @@ class UNetResAttnDownBlock2D(nn.Module):
         resnet_groups: int = 32,
         resnet_pre_norm: bool = True,
         attn_num_head_channels=1,
+        attention_type="default",
         output_scale_factor=1.0,
         add_downsample=True,
     ):
@@ -185,6 +192,8 @@ class UNetResAttnDownBlock2D(nn.Module):
         resnets = []
         attentions = []
 
+        self.attention_type = attention_type
+
         for i in range(num_layers):
             in_channels = in_channels if i == 0 else out_channels
             resnets.append(
@@ -206,6 +215,7 @@ class UNetResAttnDownBlock2D(nn.Module):
                     out_channels,
                     num_head_channels=attn_num_head_channels,
                     rescale_output_factor=output_scale_factor,
+                    eps=resnet_eps,
                 )
             )
 
@@ -251,6 +261,7 @@ class UNetResDownBlock2D(nn.Module):
         resnet_pre_norm: bool = True,
         output_scale_factor=1.0,
         add_downsample=True,
+        downsample_padding=1,
     ):
         super().__init__()
         resnets = []
@@ -276,7 +287,11 @@ class UNetResDownBlock2D(nn.Module):
 
         if add_downsample:
             self.downsamplers = nn.ModuleList(
-                [Downsample2D(in_channels, use_conv=True, out_channels=out_channels, padding=1, name="op")]
+                [
+                    Downsample2D(
+                        in_channels, use_conv=True, out_channels=out_channels, padding=downsample_padding, name="op"
+                    )
+                ]
             )
         else:
             self.downsamplers = None
@@ -301,7 +316,8 @@ class UNetResAttnUpBlock2D(nn.Module):
     def __init__(
         self,
         in_channels: int,
-        next_channels: int,
+        prev_output_channel: int,
+        out_channels: int,
         temb_channels: int,
         dropout: float = 0.0,
         num_layers: int = 1,
@@ -310,7 +326,7 @@ class UNetResAttnUpBlock2D(nn.Module):
         resnet_act_fn: str = "swish",
         resnet_groups: int = 32,
         resnet_pre_norm: bool = True,
-        attention_layer_type: str = "self",
+        attention_type="default",
         attn_num_head_channels=1,
         output_scale_factor=1.0,
         add_upsample=True,
@@ -319,12 +335,16 @@ class UNetResAttnUpBlock2D(nn.Module):
         resnets = []
         attentions = []
 
+        self.attention_type = attention_type
+
         for i in range(num_layers):
-            resnet_channels = in_channels if i < num_layers - 1 else next_channels
+            res_skip_channels = in_channels if (i == num_layers - 1) else out_channels
+            resnet_in_channels = prev_output_channel if i == 0 else out_channels
+
             resnets.append(
                 ResnetBlock(
-                    in_channels=in_channels + resnet_channels,
-                    out_channels=in_channels,
+                    in_channels=resnet_in_channels + res_skip_channels,
+                    out_channels=out_channels,
                     temb_channels=temb_channels,
                     eps=resnet_eps,
                     groups=resnet_groups,
@@ -337,9 +357,10 @@ class UNetResAttnUpBlock2D(nn.Module):
             )
             attentions.append(
                 AttentionBlockNew(
-                    in_channels,
+                    out_channels,
                     num_head_channels=attn_num_head_channels,
                     rescale_output_factor=output_scale_factor,
+                    eps=resnet_eps,
                 )
             )
 
@@ -347,7 +368,7 @@ class UNetResAttnUpBlock2D(nn.Module):
         self.resnets = nn.ModuleList(resnets)
 
         if add_upsample:
-            self.upsamplers = nn.ModuleList([Upsample2D(in_channels, use_conv=True, out_channels=in_channels)])
+            self.upsamplers = nn.ModuleList([Upsample2D(out_channels, use_conv=True, out_channels=out_channels)])
         else:
             self.upsamplers = None
 
@@ -373,7 +394,8 @@ class UNetResUpBlock2D(nn.Module):
     def __init__(
         self,
         in_channels: int,
-        next_channels: int,
+        prev_output_channel: int,
+        out_channels: int,
         temb_channels: int,
         dropout: float = 0.0,
         num_layers: int = 1,
@@ -382,7 +404,6 @@ class UNetResUpBlock2D(nn.Module):
         resnet_act_fn: str = "swish",
         resnet_groups: int = 32,
         resnet_pre_norm: bool = True,
-        attention_layer_type: str = "self",
         output_scale_factor=1.0,
         add_upsample=True,
     ):
@@ -390,11 +411,13 @@ class UNetResUpBlock2D(nn.Module):
         resnets = []
 
         for i in range(num_layers):
-            resnet_channels = in_channels if i < num_layers - 1 else next_channels
+            res_skip_channels = in_channels if (i == num_layers - 1) else out_channels
+            resnet_in_channels = prev_output_channel if i == 0 else out_channels
+
             resnets.append(
                 ResnetBlock(
-                    in_channels=in_channels + resnet_channels,
-                    out_channels=in_channels,
+                    in_channels=resnet_in_channels + res_skip_channels,
+                    out_channels=out_channels,
                     temb_channels=temb_channels,
                     eps=resnet_eps,
                     groups=resnet_groups,
@@ -409,7 +432,7 @@ class UNetResUpBlock2D(nn.Module):
         self.resnets = nn.ModuleList(resnets)
 
         if add_upsample:
-            self.upsamplers = nn.ModuleList([Upsample2D(in_channels, use_conv=True, out_channels=in_channels)])
+            self.upsamplers = nn.ModuleList([Upsample2D(out_channels, use_conv=True, out_channels=out_channels)])
         else:
             self.upsamplers = None
 

src/diffusers/models/unet_unconditional.py

@@ -9,6 +9,30 @@ from .resnet import Downsample2D, ResnetBlock2D, Upsample2D
 from .unet_new import UNetMidBlock2D, get_down_block, get_up_block
 
 
+def nonlinearity(x):
+    # swish
+    return x * torch.sigmoid(x)
+
+
+def Normalize(in_channels):
+    return torch.nn.GroupNorm(num_groups=32, num_channels=in_channels, eps=1e-6, affine=True)
+
+
+class TimestepEmbedding(nn.Module):
+    def __init__(self, channel, time_embed_dim):
+        super().__init__()
+
+        self.linear_1 = nn.Linear(channel, time_embed_dim)
+        self.act = nn.SiLU()
+        self.linear_2 = nn.Linear(time_embed_dim, time_embed_dim)
+
+    def forward(self, sample):
+        sample = self.linear_1(sample)
+        sample = self.act(sample)
+        sample = self.linear_2(sample)
+        return sample
+
+
 class UNetUnconditionalModel(ModelMixin, ConfigMixin):
     """
     The full UNet model with attention and timestep embedding. :param in_channels: channels in the input Tensor. :param
@@ -35,35 +59,66 @@ class UNetUnconditionalModel(ModelMixin, ConfigMixin):
 
     def __init__(
         self,
-        image_size,
-        in_channels,
-        out_channels,
-        num_res_blocks,
+        image_size=None,
+        in_channels=None,
+        out_channels=None,
+        num_res_blocks=None,
         dropout=0,
-        block_input_channels=(224, 224, 448, 672),
-        block_output_channels=(224, 448, 672, 896),
+        block_channels=(224, 448, 672, 896),
         down_blocks=(
             "UNetResDownBlock2D",
             "UNetResAttnDownBlock2D",
             "UNetResAttnDownBlock2D",
             "UNetResAttnDownBlock2D",
         ),
+        downsample_padding=1,
         up_blocks=("UNetResAttnUpBlock2D", "UNetResAttnUpBlock2D", "UNetResAttnUpBlock2D", "UNetResUpBlock2D"),
         resnet_act_fn="silu",
         resnet_eps=1e-5,
         conv_resample=True,
         num_head_channels=32,
+        flip_sin_to_cos=True,
+        downscale_freq_shift=0,
         # To delete once weights are converted
+        # LDM
         attention_resolutions=(8, 4, 2),
+        ldm=False,
+        # DDPM
+        out_ch=None,
+        resolution=None,
+        attn_resolutions=None,
+        resamp_with_conv=None,
+        ch_mult=None,
+        ch=None,
+        ddpm=False,
     ):
         super().__init__()
 
+        # DELETE if statements if not necessary anymore
+        # DDPM
+        if ddpm:
+            out_channels = out_ch
+            image_size = resolution
+            block_channels = [x * ch for x in ch_mult]
+            conv_resample = resamp_with_conv
+            flip_sin_to_cos = False
+            downscale_freq_shift = 1
+            resnet_eps = 1e-6
+            block_channels = (32, 64)
+            down_blocks = (
+                "UNetResDownBlock2D",
+                "UNetResAttnDownBlock2D",
+            )
+            up_blocks = ("UNetResUpBlock2D", "UNetResAttnUpBlock2D")
+            downsample_padding = 0
+            num_head_channels = 64
+
         # register all __init__ params with self.register
         self.register_to_config(
             image_size=image_size,
             in_channels=in_channels,
-            block_input_channels=block_input_channels,
-            block_output_channels=block_output_channels,
+            block_channels=block_channels,
+            downsample_padding=downsample_padding,
             out_channels=out_channels,
             num_res_blocks=num_res_blocks,
             down_blocks=down_blocks,
@@ -71,37 +126,34 @@ class UNetUnconditionalModel(ModelMixin, ConfigMixin):
             dropout=dropout,
             conv_resample=conv_resample,
             num_head_channels=num_head_channels,
+            flip_sin_to_cos=flip_sin_to_cos,
+            downscale_freq_shift=downscale_freq_shift,
             # (TODO(PVP) - To delete once weights are converted
             attention_resolutions=attention_resolutions,
+            ldm=ldm,
+            ddpm=ddpm,
         )
 
         # To delete - replace with config values
         self.image_size = image_size
-        self.in_channels = in_channels
-        self.out_channels = out_channels
-        self.num_res_blocks = num_res_blocks
-        self.dropout = dropout
+        time_embed_dim = block_channels[0] * 4
 
-        time_embed_dim = block_input_channels[0] * 4
+        # # input
+        self.conv_in = nn.Conv2d(in_channels, block_channels[0], kernel_size=3, padding=(1, 1))
 
-        # ======================== Input ===================
-        self.conv_in = nn.Conv2d(in_channels, block_input_channels[0], kernel_size=3, padding=(1, 1))
-
-        # ======================== Time ====================
-        self.time_embed = nn.Sequential(
-            nn.Linear(block_input_channels[0], time_embed_dim),
-            nn.SiLU(),
-            nn.Linear(time_embed_dim, time_embed_dim),
-        )
-
-        # ======================== Down ====================
-        input_channels = list(block_input_channels)
-        output_channels = list(block_output_channels)
+        # # time
+        self.time_embedding = TimestepEmbedding(block_channels[0], time_embed_dim)
 
         self.downsample_blocks = nn.ModuleList([])
-        for i, (input_channel, output_channel) in enumerate(zip(input_channels, output_channels)):
-            down_block_type = down_blocks[i]
-            is_final_block = i == len(input_channels) - 1
+        self.mid = None
+        self.upsample_blocks = nn.ModuleList([])
+
+        # down
+        output_channel = block_channels[0]
+        for i, down_block_type in enumerate(down_blocks):
+            input_channel = output_channel
+            output_channel = block_channels[i]
+            is_final_block = i == len(block_channels) - 1
 
             down_block = get_down_block(
                 down_block_type,
@@ -113,12 +165,24 @@ class UNetUnconditionalModel(ModelMixin, ConfigMixin):
                 resnet_eps=resnet_eps,
                 resnet_act_fn=resnet_act_fn,
                 attn_num_head_channels=num_head_channels,
+                downsample_padding=downsample_padding,
             )
             self.downsample_blocks.append(down_block)
 
-        # ======================== Mid ====================
+        # mid
+        if self.config.ddpm:
+            self.mid_new_2 = UNetMidBlock2D(
+                in_channels=block_channels[-1],
+                dropout=dropout,
+                temb_channels=time_embed_dim,
+                resnet_eps=resnet_eps,
+                resnet_act_fn=resnet_act_fn,
+                resnet_time_scale_shift="default",
+                attn_num_head_channels=num_head_channels,
+            )
+        else:
             self.mid = UNetMidBlock2D(
-            in_channels=output_channels[-1],
+                in_channels=block_channels[-1],
                 dropout=dropout,
                 temb_channels=time_embed_dim,
                 resnet_eps=resnet_eps,
@@ -127,16 +191,22 @@ class UNetUnconditionalModel(ModelMixin, ConfigMixin):
                 attn_num_head_channels=num_head_channels,
             )
 
-        self.upsample_blocks = nn.ModuleList([])
-        for i, (input_channel, output_channel) in enumerate(zip(reversed(input_channels), reversed(output_channels))):
-            up_block_type = up_blocks[i]
-            is_final_block = i == len(input_channels) - 1
+        # up
+        reversed_block_channels = list(reversed(block_channels))
+        output_channel = reversed_block_channels[0]
+        for i, up_block_type in enumerate(up_blocks):
+            prev_output_channel = output_channel
+            output_channel = reversed_block_channels[i]
+            input_channel = reversed_block_channels[min(i + 1, len(block_channels) - 1)]
+
+            is_final_block = i == len(block_channels) - 1
 
             up_block = get_up_block(
                 up_block_type,
                 num_layers=num_res_blocks + 1,
-                in_channels=output_channel,
-                next_channels=input_channel,
+                in_channels=input_channel,
+                out_channels=output_channel,
+                prev_output_channel=prev_output_channel,
                 temb_channels=time_embed_dim,
                 add_upsample=not is_final_block,
                 resnet_eps=resnet_eps,
@@ -144,21 +214,24 @@ class UNetUnconditionalModel(ModelMixin, ConfigMixin):
                 attn_num_head_channels=num_head_channels,
             )
             self.upsample_blocks.append(up_block)
+            prev_output_channel = output_channel
+
+        # out
+        self.conv_norm_out = nn.GroupNorm(num_channels=block_channels[0], num_groups=32, eps=1e-5)
+        self.conv_act = nn.SiLU()
+        self.conv_out = nn.Conv2d(block_channels[0], out_channels, 3, padding=1)
 
         # ======================== Out ====================
-        self.out = nn.Sequential(
-            nn.GroupNorm(num_channels=output_channels[0], num_groups=32, eps=1e-5),
-            nn.SiLU(),
-            nn.Conv2d(block_input_channels[0], out_channels, 3, padding=1),
-        )
 
+        self.is_overwritten = False
+        if ldm:
             # =========== TO DELETE AFTER CONVERSION ==========
             transformer_depth = 1
             context_dim = None
             legacy = True
             num_heads = -1
-        model_channels = block_input_channels[0]
-        channel_mult = tuple([x // model_channels for x in block_output_channels])
+            model_channels = block_channels[0]
+            channel_mult = tuple([x // model_channels for x in block_channels])
             self.init_for_ldm(
                 in_channels,
                 model_channels,
@@ -176,18 +249,37 @@ class UNetUnconditionalModel(ModelMixin, ConfigMixin):
                 conv_resample,
                 out_channels,
             )
+        if ddpm:
+            self.init_for_ddpm(
+                ch_mult,
+                ch,
+                num_res_blocks,
+                resolution,
+                in_channels,
+                resamp_with_conv,
+                attn_resolutions,
+                out_ch,
+                dropout=0.1,
+            )
 
     def forward(self, sample, timesteps=None):
+        # TODO(PVP) - to delete later
+        if not self.is_overwritten:
+            self.set_weights()
+
         # 1. time step embeddings
         if not torch.is_tensor(timesteps):
             timesteps = torch.tensor([timesteps], dtype=torch.long, device=sample.device)
+
         t_emb = get_timestep_embedding(
-            timesteps, self.config.block_input_channels[0], flip_sin_to_cos=True, downscale_freq_shift=0
+            timesteps,
+            self.config.block_channels[0],
+            flip_sin_to_cos=self.config.flip_sin_to_cos,
+            downscale_freq_shift=self.config.downscale_freq_shift,
         )
-        emb = self.time_embed(t_emb)
+        emb = self.time_embedding(t_emb)
 
         # 2. pre-process sample
-        #        sample = sample.type(self.dtype_)
         sample = self.conv_in(sample)
 
         # 3. down blocks
@@ -198,8 +290,13 @@ class UNetUnconditionalModel(ModelMixin, ConfigMixin):
             # append to tuple
             down_block_res_samples += res_samples
 
+        print("sample", sample.abs().sum())
         # 4. mid block
+        if self.config.ddpm:
+            sample = self.mid_new_2(sample, emb)
+        else:
             sample = self.mid(sample, emb)
+        print("sample", sample.abs().sum())
 
         # 5. up blocks
         for upsample_block in self.upsample_blocks:
@@ -211,10 +308,192 @@ class UNetUnconditionalModel(ModelMixin, ConfigMixin):
             sample = upsample_block(sample, res_samples, emb)
 
         # 6. post-process sample
-        sample = self.out(sample)
+        sample = self.conv_norm_out(sample)
+        sample = self.conv_act(sample)
+        sample = self.conv_out(sample)
 
         return sample
 
+    def set_weights(self):
+        self.is_overwritten = True
+        if self.config.ldm:
+
+            self.time_embedding.linear_1.weight.data = self.time_embed[0].weight.data
+            self.time_embedding.linear_1.bias.data = self.time_embed[0].bias.data
+            self.time_embedding.linear_2.weight.data = self.time_embed[2].weight.data
+            self.time_embedding.linear_2.bias.data = self.time_embed[2].bias.data
+
+            # ================ SET WEIGHTS OF ALL WEIGHTS ==================
+            for i, input_layer in enumerate(self.input_blocks[1:]):
+                block_id = i // (self.config.num_res_blocks + 1)
+                layer_in_block_id = i % (self.config.num_res_blocks + 1)
+
+                if layer_in_block_id == 2:
+                    self.downsample_blocks[block_id].downsamplers[0].conv.weight.data = input_layer[0].op.weight.data
+                    self.downsample_blocks[block_id].downsamplers[0].conv.bias.data = input_layer[0].op.bias.data
+                elif len(input_layer) > 1:
+                    self.downsample_blocks[block_id].resnets[layer_in_block_id].set_weight(input_layer[0])
+                    self.downsample_blocks[block_id].attentions[layer_in_block_id].set_weight(input_layer[1])
+                else:
+                    self.downsample_blocks[block_id].resnets[layer_in_block_id].set_weight(input_layer[0])
+
+            self.mid.resnets[0].set_weight(self.middle_block[0])
+            self.mid.resnets[1].set_weight(self.middle_block[2])
+            self.mid.attentions[0].set_weight(self.middle_block[1])
+
+            for i, input_layer in enumerate(self.output_blocks):
+                block_id = i // (self.config.num_res_blocks + 1)
+                layer_in_block_id = i % (self.config.num_res_blocks + 1)
+
+                if len(input_layer) > 2:
+                    self.upsample_blocks[block_id].resnets[layer_in_block_id].set_weight(input_layer[0])
+                    self.upsample_blocks[block_id].attentions[layer_in_block_id].set_weight(input_layer[1])
+                    self.upsample_blocks[block_id].upsamplers[0].conv.weight.data = input_layer[2].conv.weight.data
+                    self.upsample_blocks[block_id].upsamplers[0].conv.bias.data = input_layer[2].conv.bias.data
+                elif len(input_layer) > 1 and "Upsample2D" in input_layer[1].__class__.__name__:
+                    self.upsample_blocks[block_id].resnets[layer_in_block_id].set_weight(input_layer[0])
+                    self.upsample_blocks[block_id].upsamplers[0].conv.weight.data = input_layer[1].conv.weight.data
+                    self.upsample_blocks[block_id].upsamplers[0].conv.bias.data = input_layer[1].conv.bias.data
+                elif len(input_layer) > 1:
+                    self.upsample_blocks[block_id].resnets[layer_in_block_id].set_weight(input_layer[0])
+                    self.upsample_blocks[block_id].attentions[layer_in_block_id].set_weight(input_layer[1])
+                else:
+                    self.upsample_blocks[block_id].resnets[layer_in_block_id].set_weight(input_layer[0])
+
+            self.conv_in.weight.data = self.input_blocks[0][0].weight.data
+            self.conv_in.bias.data = self.input_blocks[0][0].bias.data
+
+            self.conv_norm_out.weight.data = self.out[0].weight.data
+            self.conv_norm_out.bias.data = self.out[0].bias.data
+            self.conv_out.weight.data = self.out[2].weight.data
+            self.conv_out.bias.data = self.out[2].bias.data
+
+            self.remove_ldm()
+
+        elif self.config.ddpm:
+            # =============== SET WEIGHTS ===============
+            # =============== TIME ======================
+            self.time_embed[0] = self.temb.dense[0]
+            self.time_embed[2] = self.temb.dense[1]
+
+            for i, block in enumerate(self.down):
+                if hasattr(block, "downsample"):
+                    self.downsample_blocks[i].downsamplers[0].conv.weight.data = block.downsample.conv.weight.data
+                    self.downsample_blocks[i].downsamplers[0].conv.bias.data = block.downsample.conv.bias.data
+                if hasattr(block, "block") and len(block.block) > 0:
+                    for j in range(self.num_res_blocks):
+                        self.downsample_blocks[i].resnets[j].set_weight(block.block[j])
+                if hasattr(block, "attn") and len(block.attn) > 0:
+                    for j in range(self.num_res_blocks):
+                        self.downsample_blocks[i].attentions[j].set_weight(block.attn[j])
+
+            self.mid_new_2.resnets[0].set_weight(self.mid.block_1)
+            self.mid_new_2.resnets[1].set_weight(self.mid.block_2)
+            self.mid_new_2.attentions[0].set_weight(self.mid.attn_1)
+
+    def init_for_ddpm(
+        self,
+        ch_mult,
+        ch,
+        num_res_blocks,
+        resolution,
+        in_channels,
+        resamp_with_conv,
+        attn_resolutions,
+        out_ch,
+        dropout=0.1,
+    ):
+        ch_mult = tuple(ch_mult)
+        self.ch = ch
+        self.temb_ch = self.ch * 4
+        self.num_resolutions = len(ch_mult)
+        self.num_res_blocks = num_res_blocks
+        self.resolution = resolution
+
+        # timestep embedding
+        self.temb = nn.Module()
+        self.temb.dense = nn.ModuleList(
+            [
+                torch.nn.Linear(self.ch, self.temb_ch),
+                torch.nn.Linear(self.temb_ch, self.temb_ch),
+            ]
+        )
+
+        # downsampling
+        self.conv_in = torch.nn.Conv2d(in_channels, self.ch, kernel_size=3, stride=1, padding=1)
+
+        curr_res = resolution
+        in_ch_mult = (1,) + ch_mult
+        self.down = nn.ModuleList()
+        for i_level in range(self.num_resolutions):
+            block = nn.ModuleList()
+            attn = nn.ModuleList()
+            block_in = ch * in_ch_mult[i_level]
+            block_out = ch * ch_mult[i_level]
+            for i_block in range(self.num_res_blocks):
+                block.append(
+                    ResnetBlock2D(
+                        in_channels=block_in, out_channels=block_out, temb_channels=self.temb_ch, dropout=dropout
+                    )
+                )
+                block_in = block_out
+                if curr_res in attn_resolutions:
+                    attn.append(AttentionBlock(block_in, overwrite_qkv=True))
+            down = nn.Module()
+            down.block = block
+            down.attn = attn
+            if i_level != self.num_resolutions - 1:
+                down.downsample = Downsample2D(block_in, use_conv=resamp_with_conv, padding=0)
+                curr_res = curr_res // 2
+            self.down.append(down)
+
+        # middle
+        self.mid = nn.Module()
+        self.mid.block_1 = ResnetBlock2D(
+            in_channels=block_in, out_channels=block_in, temb_channels=self.temb_ch, dropout=dropout
+        )
+        self.mid.attn_1 = AttentionBlock(block_in, overwrite_qkv=True)
+        self.mid.block_2 = ResnetBlock2D(
+            in_channels=block_in, out_channels=block_in, temb_channels=self.temb_ch, dropout=dropout
+        )
+        self.mid_new = UNetMidBlock2D(in_channels=block_in, temb_channels=self.temb_ch, dropout=dropout)
+        self.mid_new.resnets[0] = self.mid.block_1
+        self.mid_new.attentions[0] = self.mid.attn_1
+        self.mid_new.resnets[1] = self.mid.block_2
+
+        # upsampling
+        self.up = nn.ModuleList()
+        for i_level in reversed(range(self.num_resolutions)):
+            block = nn.ModuleList()
+            attn = nn.ModuleList()
+            block_out = ch * ch_mult[i_level]
+            skip_in = ch * ch_mult[i_level]
+            for i_block in range(self.num_res_blocks + 1):
+                if i_block == self.num_res_blocks:
+                    skip_in = ch * in_ch_mult[i_level]
+                block.append(
+                    ResnetBlock2D(
+                        in_channels=block_in + skip_in,
+                        out_channels=block_out,
+                        temb_channels=self.temb_ch,
+                        dropout=dropout,
+                    )
+                )
+                block_in = block_out
+                if curr_res in attn_resolutions:
+                    attn.append(AttentionBlock(block_in, overwrite_qkv=True))
+            up = nn.Module()
+            up.block = block
+            up.attn = attn
+            if i_level != 0:
+                up.upsample = Upsample2D(block_in, use_conv=resamp_with_conv)
+                curr_res = curr_res * 2
+            self.up.insert(0, up)  # prepend to get consistent order
+
+        # end
+        self.norm_out = Normalize(block_in)
+        self.conv_out = torch.nn.Conv2d(block_in, out_ch, kernel_size=3, stride=1, padding=1)
+
     def init_for_ldm(
         self,
         in_channels,
@@ -234,7 +513,6 @@ class UNetUnconditionalModel(ModelMixin, ConfigMixin):
         out_channels,
     ):
         # TODO(PVP) - delete after weight conversion
-
         class TimestepEmbedSequential(nn.Sequential):
             """
             A sequential module that passes timestep embeddings to the children that support it as an extra input.
@@ -255,6 +533,12 @@ class UNetUnconditionalModel(ModelMixin, ConfigMixin):
                 return nn.Conv3d(*args, **kwargs)
             raise ValueError(f"unsupported dimensions: {dims}")
 
+        self.time_embed = nn.Sequential(
+            nn.Linear(model_channels, time_embed_dim),
+            nn.SiLU(),
+            nn.Linear(time_embed_dim, time_embed_dim),
+        )
+
         dims = 2
         self.input_blocks = nn.ModuleList(
             [TimestepEmbedSequential(conv_nd(dims, in_channels, model_channels, 3, padding=1))]
@@ -389,42 +673,15 @@ class UNetUnconditionalModel(ModelMixin, ConfigMixin):
                 self.output_blocks.append(TimestepEmbedSequential(*layers))
                 self._feature_size += ch
 
-        # ================ SET WEIGHTS OF ALL WEIGHTS ==================
-        for i, input_layer in enumerate(self.input_blocks[1:]):
-            block_id = i // (num_res_blocks + 1)
-            layer_in_block_id = i % (num_res_blocks + 1)
-
-            if layer_in_block_id == 2:
-                self.downsample_blocks[block_id].downsamplers[0].op.weight.data = input_layer[0].op.weight.data
-                self.downsample_blocks[block_id].downsamplers[0].op.bias.data = input_layer[0].op.bias.data
-            elif len(input_layer) > 1:
-                self.downsample_blocks[block_id].resnets[layer_in_block_id].set_weight(input_layer[0])
-                self.downsample_blocks[block_id].attentions[layer_in_block_id].set_weight(input_layer[1])
-            else:
-                self.downsample_blocks[block_id].resnets[layer_in_block_id].set_weight(input_layer[0])
-
-        self.mid.resnets[0].set_weight(self.middle_block[0])
-        self.mid.resnets[1].set_weight(self.middle_block[2])
-        self.mid.attentions[0].set_weight(self.middle_block[1])
-
-        for i, input_layer in enumerate(self.output_blocks):
-            block_id = i // (num_res_blocks + 1)
-            layer_in_block_id = i % (num_res_blocks + 1)
-
-            if len(input_layer) > 2:
-                self.upsample_blocks[block_id].resnets[layer_in_block_id].set_weight(input_layer[0])
-                self.upsample_blocks[block_id].attentions[layer_in_block_id].set_weight(input_layer[1])
-                self.upsample_blocks[block_id].upsamplers[0].conv.weight.data = input_layer[2].conv.weight.data
-                self.upsample_blocks[block_id].upsamplers[0].conv.bias.data = input_layer[2].conv.bias.data
-            elif len(input_layer) > 1 and "Upsample2D" in input_layer[1].__class__.__name__:
-                self.upsample_blocks[block_id].resnets[layer_in_block_id].set_weight(input_layer[0])
-                self.upsample_blocks[block_id].upsamplers[0].conv.weight.data = input_layer[1].conv.weight.data
-                self.upsample_blocks[block_id].upsamplers[0].conv.bias.data = input_layer[1].conv.bias.data
-            elif len(input_layer) > 1:
-                self.upsample_blocks[block_id].resnets[layer_in_block_id].set_weight(input_layer[0])
-                self.upsample_blocks[block_id].attentions[layer_in_block_id].set_weight(input_layer[1])
-            else:
-                self.upsample_blocks[block_id].resnets[layer_in_block_id].set_weight(input_layer[0])
+        self.out = nn.Sequential(
+            nn.GroupNorm(num_channels=model_channels, num_groups=32, eps=1e-5),
+            nn.SiLU(),
+            nn.Conv2d(model_channels, out_channels, 3, padding=1),
+        )
 
-        self.conv_in.weight.data = self.input_blocks[0][0].weight.data
-        self.conv_in.bias.data = self.input_blocks[0][0].bias.data
+    def remove_ldm(self):
+        del self.time_embed
+        del self.input_blocks
+        del self.middle_block
+        del self.output_blocks
+        del self.out

tests/test_modeling_utils.py

@@ -271,6 +271,27 @@ class UnetModelTests(ModelTesterMixin, unittest.TestCase):
         # fmt: on
         self.assertTrue(torch.allclose(output_slice, expected_output_slice, rtol=1e-2))
 
+        print("Original success!!!")
+
+        model = UNetUnconditionalModel.from_pretrained("fusing/ddpm_dummy", ddpm=True)
+        model.eval()
+
+        torch.manual_seed(0)
+        if torch.cuda.is_available():
+            torch.cuda.manual_seed_all(0)
+
+        noise = torch.randn(1, model.config.in_channels, model.config.image_size, model.config.image_size)
+        time_step = torch.tensor([10])
+
+        with torch.no_grad():
+            output = model(noise, time_step)
+
+        output_slice = output[0, -1, -3:, -3:].flatten()
+        # fmt: off
+        expected_output_slice = torch.tensor([0.2891, -0.1899, 0.2595, -0.6214, 0.0968, -0.2622, 0.4688, 0.1311, 0.0053])
+        # fmt: on
+        self.assertTrue(torch.allclose(output_slice, expected_output_slice, rtol=1e-2))
+
 
 class GlideSuperResUNetTests(ModelTesterMixin, unittest.TestCase):
     model_class = GlideSuperResUNetModel
@@ -486,18 +507,20 @@ class UNetLDMModelTests(ModelTesterMixin, unittest.TestCase):
             "out_channels": 4,
             "num_res_blocks": 2,
             "attention_resolutions": (16,),
-            "block_input_channels": [32, 32],
-            "block_output_channels": [32, 64],
+            "block_channels": (32, 64),
             "num_head_channels": 32,
             "conv_resample": True,
             "down_blocks": ("UNetResDownBlock2D", "UNetResDownBlock2D"),
             "up_blocks": ("UNetResUpBlock2D", "UNetResUpBlock2D"),
+            "ldm": True,
         }
         inputs_dict = self.dummy_input
         return init_dict, inputs_dict
 
     def test_from_pretrained_hub(self):
-        model, loading_info = UNetUnconditionalModel.from_pretrained("fusing/unet-ldm-dummy", output_loading_info=True)
+        model, loading_info = UNetUnconditionalModel.from_pretrained(
+            "fusing/unet-ldm-dummy", output_loading_info=True, ldm=True
+        )
         self.assertIsNotNone(model)
         self.assertEqual(len(loading_info["missing_keys"]), 0)
 
@@ -507,7 +530,7 @@ class UNetLDMModelTests(ModelTesterMixin, unittest.TestCase):
         assert image is not None, "Make sure output is not None"
 
     def test_output_pretrained(self):
-        model = UNetUnconditionalModel.from_pretrained("fusing/unet-ldm-dummy")
+        model = UNetUnconditionalModel.from_pretrained("fusing/unet-ldm-dummy", ldm=True)
         model.eval()
 
         torch.manual_seed(0)