{
"cells": [
{
"cell_type": "code",
"execution_count": 3,
"metadata": {},
"outputs": [],
"source": [
"from templates import *\n",
"from templates_cls import *\n",
"from experiment_classifier import ClsModel\n",
"\n",
"import matplotlib.pyplot as plt\n"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [],
"source": [
"def show_images(images, cols = 1, titles = None, apply_convert=False):\n",
" if apply_convert: images = [convert(img) for img in images]\n",
" assert((titles is None)or (len(images) == len(titles)))\n",
" n_images = len(images)\n",
" if titles is None: titles = ['Image (%d)' % i for i in range(1,n_images + 1)]\n",
" fig = plt.figure()\n",
" for n, (image, title) in enumerate(zip(images, titles)):\n",
" a = fig.add_subplot(cols, int(np.ceil(n_images/float(cols))), n + 1)\n",
" if image.ndim == 2:\n",
" plt.gray()\n",
" plt.imshow(image)\n",
" a.set_title(title)\n",
" fig.set_size_inches(np.array(fig.get_size_inches()) * n_images*2)\n",
" plt.show()\n",
" \n"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
" Model configuaration "
]
},
{
"cell_type": "code",
"execution_count": 4,
"metadata": {},
"outputs": [],
"source": [
"device = 'cuda:3'\n",
"img_resolution = 256 # img output's resolution; we have provided 128 and 256 pretrained models; you may train our model with your own preference, the training script is also provided.\n",
"T_step = 100 # number of steps for DDPM [0,1000]\n",
"\n",
"model_config = ffhq256_autoenc() if img_resolution == 256 else ffhq128_autoenc()\n",
"model_config.T_eval = T_step"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
" Model initialization "
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"Global seed set to 0\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Model params: 160.69 M\n"
]
},
{
"data": {
"text/plain": [
"BeatGANsAutoencModel(\n",
" (time_embed): TimeStyleSeperateEmbed(\n",
" (time_embed): Sequential(\n",
" (0): Linear(in_features=128, out_features=512, bias=True)\n",
" (1): SiLU()\n",
" (2): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (style): Identity()\n",
" )\n",
" (input_blocks): ModuleList(\n",
" (0): TimestepEmbedSequential(\n",
" (0): Conv2d(3, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (1): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (2): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (3): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (4): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (5): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (6): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (7): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(128, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (8): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (9): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (10): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (11): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (12): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (13): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(256, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" )\n",
" (14): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" )\n",
" (15): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (16): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (17): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" )\n",
" (middle_block): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" (2): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (output_blocks): ModuleList(\n",
" (0): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 1024, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(1024, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (1): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 1024, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(1024, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (2): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 1024, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(1024, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Upsample()\n",
" (x_upd): Upsample()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (3): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 1024, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(1024, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" )\n",
" (4): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 1024, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(1024, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" )\n",
" (5): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 768, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(768, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(768, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" (2): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Upsample()\n",
" (x_upd): Upsample()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (6): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 768, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(768, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(768, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (7): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (8): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Upsample()\n",
" (x_upd): Upsample()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (9): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (10): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (11): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 384, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(384, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(384, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Upsample()\n",
" (x_upd): Upsample()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (12): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 384, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(384, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(384, 128, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (13): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (14): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Upsample()\n",
" (x_upd): Upsample()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (15): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (16): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (17): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" )\n",
" (out): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 3, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (encoder): BeatGANsEncoderModel(\n",
" (input_blocks): ModuleList(\n",
" (0): TimestepEmbedSequential(\n",
" (0): Conv2d(3, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (1): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (2): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (3): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (4): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (5): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (6): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (7): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Conv2d(128, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (8): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (9): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (10): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (11): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (12): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (13): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Conv2d(256, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" )\n",
" (14): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" )\n",
" (15): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (16): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (17): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (18): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (19): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (20): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" )\n",
" (middle_block): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" (2): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (out): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): AdaptiveAvgPool2d(output_size=(1, 1))\n",
" (3): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" (4): Flatten(start_dim=1, end_dim=-1)\n",
" )\n",
" )\n",
")"
]
},
"execution_count": 5,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"model = LitModel(model_config)\n",
"state = torch.load(f'checkpoints/{model_config.name}/last.ckpt', map_location='cpu')\n",
"model.load_state_dict(state['state_dict'], strict=False)\n",
"model.ema_model.eval()\n",
"model.ema_model.to(device)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
" Classifer initialization \n",
"We'll use a trained classifer weight to provide a class direction in our image manipulation process"
]
},
{
"cell_type": "code",
"execution_count": 6,
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"Global seed set to 0\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"loading pretrain ... 130M\n",
"step: 1563562\n",
"_IncompatibleKeys(missing_keys=[], unexpected_keys=['x_T'])\n",
"loading latent stats ...\n",
"latent step: 9375\n"
]
},
{
"data": {
"text/plain": [
"ClsModel(\n",
" (model): BeatGANsAutoencModel(\n",
" (time_embed): TimeStyleSeperateEmbed(\n",
" (time_embed): Sequential(\n",
" (0): Linear(in_features=128, out_features=512, bias=True)\n",
" (1): SiLU()\n",
" (2): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (style): Identity()\n",
" )\n",
" (input_blocks): ModuleList(\n",
" (0): TimestepEmbedSequential(\n",
" (0): Conv2d(3, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (1): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (2): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (3): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (4): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (5): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (6): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (7): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(128, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (8): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (9): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (10): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (11): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (12): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (13): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(256, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" )\n",
" (14): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" )\n",
" (15): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (16): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (17): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" )\n",
" (middle_block): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" (2): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (output_blocks): ModuleList(\n",
" (0): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 1024, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(1024, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (1): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 1024, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(1024, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (2): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 1024, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(1024, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Upsample()\n",
" (x_upd): Upsample()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (3): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 1024, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(1024, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" )\n",
" (4): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 1024, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(1024, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" )\n",
" (5): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 768, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(768, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(768, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" (2): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Upsample()\n",
" (x_upd): Upsample()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (6): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 768, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(768, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(768, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (7): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (8): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Upsample()\n",
" (x_upd): Upsample()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (9): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (10): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (11): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 384, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(384, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(384, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Upsample()\n",
" (x_upd): Upsample()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (12): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 384, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(384, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(384, 128, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (13): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (14): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Upsample()\n",
" (x_upd): Upsample()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (15): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (16): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (17): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" )\n",
" (out): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 3, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (encoder): BeatGANsEncoderModel(\n",
" (input_blocks): ModuleList(\n",
" (0): TimestepEmbedSequential(\n",
" (0): Conv2d(3, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (1): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (2): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (3): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (4): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (5): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (6): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (7): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Conv2d(128, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (8): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (9): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (10): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (11): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (12): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (13): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Conv2d(256, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" )\n",
" (14): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" )\n",
" (15): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (16): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (17): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (18): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (19): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (20): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" )\n",
" (middle_block): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" (2): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (out): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): AdaptiveAvgPool2d(output_size=(1, 1))\n",
" (3): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" (4): Flatten(start_dim=1, end_dim=-1)\n",
" )\n",
" )\n",
" )\n",
" (ema_model): BeatGANsAutoencModel(\n",
" (time_embed): TimeStyleSeperateEmbed(\n",
" (time_embed): Sequential(\n",
" (0): Linear(in_features=128, out_features=512, bias=True)\n",
" (1): SiLU()\n",
" (2): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (style): Identity()\n",
" )\n",
" (input_blocks): ModuleList(\n",
" (0): TimestepEmbedSequential(\n",
" (0): Conv2d(3, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (1): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (2): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (3): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (4): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (5): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (6): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (7): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(128, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (8): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (9): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (10): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (11): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (12): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (13): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(256, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" )\n",
" (14): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" )\n",
" (15): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (16): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (17): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" )\n",
" (middle_block): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" (2): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (output_blocks): ModuleList(\n",
" (0): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 1024, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(1024, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (1): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 1024, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(1024, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (2): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 1024, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(1024, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Upsample()\n",
" (x_upd): Upsample()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (3): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 1024, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(1024, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" )\n",
" (4): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 1024, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(1024, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(1024, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" )\n",
" (5): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 768, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(768, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(768, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" (2): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Upsample()\n",
" (x_upd): Upsample()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=1024, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (6): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 768, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(768, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(768, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (7): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (8): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Upsample()\n",
" (x_upd): Upsample()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (9): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (10): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (11): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 384, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(384, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(384, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Upsample()\n",
" (x_upd): Upsample()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=512, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (12): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 384, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(384, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(384, 128, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (13): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (14): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Upsample()\n",
" (x_upd): Upsample()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (15): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (16): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (17): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=256, bias=True)\n",
" )\n",
" (cond_emb_layers): Sequential(\n",
" (0): SiLU()\n",
" (1): Linear(in_features=512, out_features=128, bias=True)\n",
" )\n",
" (out_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Dropout(p=0.1, inplace=False)\n",
" (3): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (skip_connection): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" )\n",
" (out): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 3, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (encoder): BeatGANsEncoderModel(\n",
" (input_blocks): ModuleList(\n",
" (0): TimestepEmbedSequential(\n",
" (0): Conv2d(3, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (1): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (2): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (3): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (4): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (5): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (6): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (7): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 128, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Conv2d(128, 256, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" )\n",
" (8): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (9): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (10): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (11): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (12): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (13): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 256, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Conv2d(256, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" )\n",
" (14): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" )\n",
" (15): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (16): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (17): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (18): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (x_upd): Downsample(\n",
" (op): AvgPool2d(kernel_size=2, stride=2, padding=0)\n",
" )\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (19): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (20): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" )\n",
" (middle_block): TimestepEmbedSequential(\n",
" (0): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" (1): AttentionBlock(\n",
" (norm): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (qkv): Conv1d(512, 1536, kernel_size=(1,), stride=(1,))\n",
" (attention): QKVAttentionLegacy()\n",
" (proj_out): Conv1d(512, 512, kernel_size=(1,), stride=(1,))\n",
" )\n",
" (2): ResBlock(\n",
" (in_layers): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
" )\n",
" (h_upd): Identity()\n",
" (x_upd): Identity()\n",
" (skip_connection): Identity()\n",
" )\n",
" )\n",
" (out): Sequential(\n",
" (0): GroupNorm32(32, 512, eps=1e-05, affine=True)\n",
" (1): SiLU()\n",
" (2): AdaptiveAvgPool2d(output_size=(1, 1))\n",
" (3): Conv2d(512, 512, kernel_size=(1, 1), stride=(1, 1))\n",
" (4): Flatten(start_dim=1, end_dim=-1)\n",
" )\n",
" )\n",
" )\n",
" (classifier): Linear(in_features=512, out_features=40, bias=True)\n",
" (ema_classifier): Linear(in_features=512, out_features=40, bias=True)\n",
")"
]
},
"execution_count": 6,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"classifer_config = ffhq256_autoenc_cls() if img_resolution == 256 else ffhq128_autoenc_cls()\n",
"classifier = ClsModel(classifer_config)\n",
"state = torch.load(f'checkpoints/{classifer_config.name}/last.ckpt', map_location='cpu')\n",
"print('latent step:', state['global_step'])\n",
"classifier.load_state_dict(state['state_dict'], strict=False)\n",
"classifier.to(device)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
" Data initialization "
]
},
{
"cell_type": "code",
"execution_count": 7,
"metadata": {},
"outputs": [],
"source": [
"def convert2rgb(img,adjust_scale=True):\n",
" convert_img = torch.tensor(img)\n",
" if adjust_scale: convert_img = (convert_img+1)/2\n",
" return (convert_img).permute(1, 2, 0).cpu()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
" Custom dataset \n",
"In addition, you may feed any well-aligned portrait face image into our model. The alignment and image preparation script is also provided; check out align.py"
]
},
{
"cell_type": "code",
"execution_count": 8,
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"/home/nessessence/anaconda3/envs/auto_ddpm/lib/python3.7/site-packages/ipykernel_launcher.py:2: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).\n",
" \n"
]
},
{
"data": {
"text/plain": [
""
]
},
"execution_count": 8,
"metadata": {},
"output_type": "execute_result"
},
{
"data": {
"image/png": "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",
"text/plain": [
""
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"imgdataset_path = 'imgs_align' # path to your own image dataset \n",
"data = ImageDataset(imgdataset_path, image_size=model_config.img_size, exts=['jpg', 'JPG', 'png'], do_augment=False)\n",
"plt.imshow(convert2rgb(data[0]['img']))"
]
},
{
"cell_type": "code",
"execution_count": 9,
"metadata": {},
"outputs": [],
"source": [
"batch = data[0]['img'][None]\n",
"\n",
"semantic_latent = model.encode(batch.to(device))\n",
"stochastic_latent = model.encode_stochastic(batch.to(device), semantic_latent)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
" Manipulation "
]
},
{
"cell_type": "code",
"execution_count": 10,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"['5_o_Clock_Shadow',\n",
" 'Arched_Eyebrows',\n",
" 'Attractive',\n",
" 'Bags_Under_Eyes',\n",
" 'Bald',\n",
" 'Bangs',\n",
" 'Big_Lips',\n",
" 'Big_Nose',\n",
" 'Black_Hair',\n",
" 'Blond_Hair',\n",
" 'Blurry',\n",
" 'Brown_Hair',\n",
" 'Bushy_Eyebrows',\n",
" 'Chubby',\n",
" 'Double_Chin',\n",
" 'Eyeglasses',\n",
" 'Goatee',\n",
" 'Gray_Hair',\n",
" 'Heavy_Makeup',\n",
" 'High_Cheekbones',\n",
" 'Male',\n",
" 'Mouth_Slightly_Open',\n",
" 'Mustache',\n",
" 'Narrow_Eyes',\n",
" 'No_Beard',\n",
" 'Oval_Face',\n",
" 'Pale_Skin',\n",
" 'Pointy_Nose',\n",
" 'Receding_Hairline',\n",
" 'Rosy_Cheeks',\n",
" 'Sideburns',\n",
" 'Smiling',\n",
" 'Straight_Hair',\n",
" 'Wavy_Hair',\n",
" 'Wearing_Earrings',\n",
" 'Wearing_Hat',\n",
" 'Wearing_Lipstick',\n",
" 'Wearing_Necklace',\n",
" 'Wearing_Necktie',\n",
" 'Young']"
]
},
"execution_count": 10,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"CelebAttrDataset.id_to_cls"
]
},
{
"cell_type": "code",
"execution_count": 24,
"metadata": {},
"outputs": [],
"source": [
"target_class = 'Bangs' \n",
"manipulation_amp = 0.4 # how strong of the class direction; too strong will be resulted in artifact as it could dominate the original image information\n",
"\n",
"cls_id = CelebAttrDataset.cls_to_id[target_class]\n",
"class_direction = classifier.classifier.weight[cls_id]\n",
"normalized_class_direction = F.normalize(class_direction[None, :], dim=1)\n",
"\n",
"normalized_semantic_latent = classifier.normalize(semantic_latent)\n",
"normalized_manipulation_amp = manipulation_amp * math.sqrt(512)\n",
"normalized_manipulated_semantic_latent = normalized_semantic_latent + normalized_manipulation_amp*normalized_class_direction\n",
"\n",
"manipulated_semantic_latent = classifier.denormalize(normalized_manipulated_semantic_latent)"
]
},
{
"cell_type": "code",
"execution_count": 25,
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"/home/nessessence/anaconda3/envs/auto_ddpm/lib/python3.7/site-packages/ipykernel_launcher.py:2: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).\n",
" \n"
]
},
{
"data": {
"text/plain": [
""
]
},
"execution_count": 25,
"metadata": {},
"output_type": "execute_result"
},
{
"data": {
"image/png": "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",
"text/plain": [
""
]
},
"metadata": {
"needs_background": "light"
},
"output_type": "display_data"
}
],
"source": [
"manipulated_img = model.render(stochastic_latent, manipulated_semantic_latent)[0]\n",
"original_img = data[0]['img']\n",
"\n",
"fig, ax = plt.subplots(1, 2, figsize=(10, 5))\n",
"ax[0].imshow(convert2rgb(original_img))\n",
"ax[1].imshow(convert2rgb(manipulated_img,adjust_scale=False))\n",
"# plt.savefig('imgs_manipulated/compare.png')"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
<<<<<<< HEAD
"source": [
"To do:\n",
"- colab (optional)\n",
" - check if the model can run on a free gpu\n",
" (if the colab works)\n",
" - upload weight to gdrive\n",
" - ddpm\n",
" - classifier\n",
" - ddpm-latent (for unconditional sample)\n",
" - upload image button\n",
" - a scroll for manipuation amplitude\n",
"- friendly explaination \n",
"\n",
"nex"
]
=======
"source": []
>>>>>>> 3fccf65c747dd578943b8429da7fe2d5a6f31df0
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"interpreter": {
"hash": "c34c6593d99c9985a1b30927262ec4c88246550b24b160c694e84311d56c55f2"
},
"kernelspec": {
"display_name": "Python 3 (ipykernel)",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.7.12"
}
},
"nbformat": 4,
"nbformat_minor": 2
}