Commit
·
ee085ff
1
Parent(s):
17a4f1e
Add model files
Browse files- .gitignore +8 -0
- clone-wars-diffusion-v1.ckpt +3 -0
- convert_diffusers_to_sd.py +235 -0
- feature_extractor/preprocessor_config.json +20 -0
- model_index.json +32 -0
- safety_checker/config.json +175 -0
- safety_checker/pytorch_model.bin +3 -0
- scheduler/scheduler_config.json +12 -0
- text_encoder/config.json +25 -0
- text_encoder/pytorch_model.bin +3 -0
- tokenizer/merges.txt +0 -0
- tokenizer/special_tokens_map.json +24 -0
- tokenizer/tokenizer_config.json +34 -0
- tokenizer/vocab.json +0 -0
- unet/config.json +37 -0
- unet/diffusion_pytorch_model.bin +3 -0
- vae/README.md +84 -0
- vae/config.json +30 -0
- vae/diffusion_pytorch_model.bin +3 -0
.gitignore
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
/.idea
|
2 |
+
.DS_Store
|
3 |
+
__pycache__
|
4 |
+
*/**/__pycache__
|
5 |
+
logs
|
6 |
+
/tmp
|
7 |
+
/train
|
8 |
+
/data
|
clone-wars-diffusion-v1.ckpt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2e7f6f95a061a62caa9c3bc0715cd77ae668ba4b0e33ec8cc9c581a5e0a1a2c2
|
3 |
+
size 2132856622
|
convert_diffusers_to_sd.py
ADDED
@@ -0,0 +1,235 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# Script for converting a HF Diffusers saved pipeline to a Stable Diffusion checkpoint.
|
2 |
+
# *Only* converts the UNet, VAE, and Text Encoder.
|
3 |
+
# Does not convert optimizer state or any other thing.
|
4 |
+
# Written by jachiam
|
5 |
+
|
6 |
+
import argparse
|
7 |
+
import os.path as osp
|
8 |
+
|
9 |
+
import torch
|
10 |
+
|
11 |
+
|
12 |
+
# =================#
|
13 |
+
# UNet Conversion #
|
14 |
+
# =================#
|
15 |
+
|
16 |
+
unet_conversion_map = [
|
17 |
+
# (stable-diffusion, HF Diffusers)
|
18 |
+
("time_embed.0.weight", "time_embedding.linear_1.weight"),
|
19 |
+
("time_embed.0.bias", "time_embedding.linear_1.bias"),
|
20 |
+
("time_embed.2.weight", "time_embedding.linear_2.weight"),
|
21 |
+
("time_embed.2.bias", "time_embedding.linear_2.bias"),
|
22 |
+
("input_blocks.0.0.weight", "conv_in.weight"),
|
23 |
+
("input_blocks.0.0.bias", "conv_in.bias"),
|
24 |
+
("out.0.weight", "conv_norm_out.weight"),
|
25 |
+
("out.0.bias", "conv_norm_out.bias"),
|
26 |
+
("out.2.weight", "conv_out.weight"),
|
27 |
+
("out.2.bias", "conv_out.bias"),
|
28 |
+
]
|
29 |
+
|
30 |
+
unet_conversion_map_resnet = [
|
31 |
+
# (stable-diffusion, HF Diffusers)
|
32 |
+
("in_layers.0", "norm1"),
|
33 |
+
("in_layers.2", "conv1"),
|
34 |
+
("out_layers.0", "norm2"),
|
35 |
+
("out_layers.3", "conv2"),
|
36 |
+
("emb_layers.1", "time_emb_proj"),
|
37 |
+
("skip_connection", "conv_shortcut"),
|
38 |
+
]
|
39 |
+
|
40 |
+
unet_conversion_map_layer = []
|
41 |
+
# hardcoded number of downblocks and resnets/attentions...
|
42 |
+
# would need smarter logic for other networks.
|
43 |
+
for i in range(4):
|
44 |
+
# loop over downblocks/upblocks
|
45 |
+
|
46 |
+
for j in range(2):
|
47 |
+
# loop over resnets/attentions for downblocks
|
48 |
+
hf_down_res_prefix = f"down_blocks.{i}.resnets.{j}."
|
49 |
+
sd_down_res_prefix = f"input_blocks.{3*i + j + 1}.0."
|
50 |
+
unet_conversion_map_layer.append((sd_down_res_prefix, hf_down_res_prefix))
|
51 |
+
|
52 |
+
if i < 3:
|
53 |
+
# no attention layers in down_blocks.3
|
54 |
+
hf_down_atn_prefix = f"down_blocks.{i}.attentions.{j}."
|
55 |
+
sd_down_atn_prefix = f"input_blocks.{3*i + j + 1}.1."
|
56 |
+
unet_conversion_map_layer.append((sd_down_atn_prefix, hf_down_atn_prefix))
|
57 |
+
|
58 |
+
for j in range(3):
|
59 |
+
# loop over resnets/attentions for upblocks
|
60 |
+
hf_up_res_prefix = f"up_blocks.{i}.resnets.{j}."
|
61 |
+
sd_up_res_prefix = f"output_blocks.{3*i + j}.0."
|
62 |
+
unet_conversion_map_layer.append((sd_up_res_prefix, hf_up_res_prefix))
|
63 |
+
|
64 |
+
if i > 0:
|
65 |
+
# no attention layers in up_blocks.0
|
66 |
+
hf_up_atn_prefix = f"up_blocks.{i}.attentions.{j}."
|
67 |
+
sd_up_atn_prefix = f"output_blocks.{3*i + j}.1."
|
68 |
+
unet_conversion_map_layer.append((sd_up_atn_prefix, hf_up_atn_prefix))
|
69 |
+
|
70 |
+
if i < 3:
|
71 |
+
# no downsample in down_blocks.3
|
72 |
+
hf_downsample_prefix = f"down_blocks.{i}.downsamplers.0.conv."
|
73 |
+
sd_downsample_prefix = f"input_blocks.{3*(i+1)}.0.op."
|
74 |
+
unet_conversion_map_layer.append((sd_downsample_prefix, hf_downsample_prefix))
|
75 |
+
|
76 |
+
# no upsample in up_blocks.3
|
77 |
+
hf_upsample_prefix = f"up_blocks.{i}.upsamplers.0."
|
78 |
+
sd_upsample_prefix = f"output_blocks.{3*i + 2}.{1 if i == 0 else 2}."
|
79 |
+
unet_conversion_map_layer.append((sd_upsample_prefix, hf_upsample_prefix))
|
80 |
+
|
81 |
+
hf_mid_atn_prefix = "mid_block.attentions.0."
|
82 |
+
sd_mid_atn_prefix = "middle_block.1."
|
83 |
+
unet_conversion_map_layer.append((sd_mid_atn_prefix, hf_mid_atn_prefix))
|
84 |
+
|
85 |
+
for j in range(2):
|
86 |
+
hf_mid_res_prefix = f"mid_block.resnets.{j}."
|
87 |
+
sd_mid_res_prefix = f"middle_block.{2*j}."
|
88 |
+
unet_conversion_map_layer.append((sd_mid_res_prefix, hf_mid_res_prefix))
|
89 |
+
|
90 |
+
|
91 |
+
def convert_unet_state_dict(unet_state_dict):
|
92 |
+
# buyer beware: this is a *brittle* function,
|
93 |
+
# and correct output requires that all of these pieces interact in
|
94 |
+
# the exact order in which I have arranged them.
|
95 |
+
mapping = {k: k for k in unet_state_dict.keys()}
|
96 |
+
for sd_name, hf_name in unet_conversion_map:
|
97 |
+
mapping[hf_name] = sd_name
|
98 |
+
for k, v in mapping.items():
|
99 |
+
if "resnets" in k:
|
100 |
+
for sd_part, hf_part in unet_conversion_map_resnet:
|
101 |
+
v = v.replace(hf_part, sd_part)
|
102 |
+
mapping[k] = v
|
103 |
+
for k, v in mapping.items():
|
104 |
+
for sd_part, hf_part in unet_conversion_map_layer:
|
105 |
+
v = v.replace(hf_part, sd_part)
|
106 |
+
mapping[k] = v
|
107 |
+
new_state_dict = {v: unet_state_dict[k] for k, v in mapping.items()}
|
108 |
+
return new_state_dict
|
109 |
+
|
110 |
+
|
111 |
+
# ================#
|
112 |
+
# VAE Conversion #
|
113 |
+
# ================#
|
114 |
+
|
115 |
+
vae_conversion_map = [
|
116 |
+
# (stable-diffusion, HF Diffusers)
|
117 |
+
("nin_shortcut", "conv_shortcut"),
|
118 |
+
("norm_out", "conv_norm_out"),
|
119 |
+
("mid.attn_1.", "mid_block.attentions.0."),
|
120 |
+
]
|
121 |
+
|
122 |
+
for i in range(4):
|
123 |
+
# down_blocks have two resnets
|
124 |
+
for j in range(2):
|
125 |
+
hf_down_prefix = f"encoder.down_blocks.{i}.resnets.{j}."
|
126 |
+
sd_down_prefix = f"encoder.down.{i}.block.{j}."
|
127 |
+
vae_conversion_map.append((sd_down_prefix, hf_down_prefix))
|
128 |
+
|
129 |
+
if i < 3:
|
130 |
+
hf_downsample_prefix = f"down_blocks.{i}.downsamplers.0."
|
131 |
+
sd_downsample_prefix = f"down.{i}.downsample."
|
132 |
+
vae_conversion_map.append((sd_downsample_prefix, hf_downsample_prefix))
|
133 |
+
|
134 |
+
hf_upsample_prefix = f"up_blocks.{i}.upsamplers.0."
|
135 |
+
sd_upsample_prefix = f"up.{3-i}.upsample."
|
136 |
+
vae_conversion_map.append((sd_upsample_prefix, hf_upsample_prefix))
|
137 |
+
|
138 |
+
# up_blocks have three resnets
|
139 |
+
# also, up blocks in hf are numbered in reverse from sd
|
140 |
+
for j in range(3):
|
141 |
+
hf_up_prefix = f"decoder.up_blocks.{i}.resnets.{j}."
|
142 |
+
sd_up_prefix = f"decoder.up.{3-i}.block.{j}."
|
143 |
+
vae_conversion_map.append((sd_up_prefix, hf_up_prefix))
|
144 |
+
|
145 |
+
# this part accounts for mid blocks in both the encoder and the decoder
|
146 |
+
for i in range(2):
|
147 |
+
hf_mid_res_prefix = f"mid_block.resnets.{i}."
|
148 |
+
sd_mid_res_prefix = f"mid.block_{i+1}."
|
149 |
+
vae_conversion_map.append((sd_mid_res_prefix, hf_mid_res_prefix))
|
150 |
+
|
151 |
+
|
152 |
+
vae_conversion_map_attn = [
|
153 |
+
# (stable-diffusion, HF Diffusers)
|
154 |
+
("norm.", "group_norm."),
|
155 |
+
("q.", "query."),
|
156 |
+
("k.", "key."),
|
157 |
+
("v.", "value."),
|
158 |
+
("proj_out.", "proj_attn."),
|
159 |
+
]
|
160 |
+
|
161 |
+
|
162 |
+
def reshape_weight_for_sd(w):
|
163 |
+
# convert HF linear weights to SD conv2d weights
|
164 |
+
return w.reshape(*w.shape, 1, 1)
|
165 |
+
|
166 |
+
|
167 |
+
def convert_vae_state_dict(vae_state_dict):
|
168 |
+
mapping = {k: k for k in vae_state_dict.keys()}
|
169 |
+
for k, v in mapping.items():
|
170 |
+
for sd_part, hf_part in vae_conversion_map:
|
171 |
+
v = v.replace(hf_part, sd_part)
|
172 |
+
mapping[k] = v
|
173 |
+
for k, v in mapping.items():
|
174 |
+
if "attentions" in k:
|
175 |
+
for sd_part, hf_part in vae_conversion_map_attn:
|
176 |
+
v = v.replace(hf_part, sd_part)
|
177 |
+
mapping[k] = v
|
178 |
+
new_state_dict = {v: vae_state_dict[k] for k, v in mapping.items()}
|
179 |
+
weights_to_convert = ["q", "k", "v", "proj_out"]
|
180 |
+
for k, v in new_state_dict.items():
|
181 |
+
for weight_name in weights_to_convert:
|
182 |
+
if f"mid.attn_1.{weight_name}.weight" in k:
|
183 |
+
print(f"Reshaping {k} for SD format")
|
184 |
+
new_state_dict[k] = reshape_weight_for_sd(v)
|
185 |
+
return new_state_dict
|
186 |
+
|
187 |
+
|
188 |
+
# =========================#
|
189 |
+
# Text Encoder Conversion #
|
190 |
+
# =========================#
|
191 |
+
# pretty much a no-op
|
192 |
+
|
193 |
+
|
194 |
+
def convert_text_enc_state_dict(text_enc_dict):
|
195 |
+
return text_enc_dict
|
196 |
+
|
197 |
+
|
198 |
+
if __name__ == "__main__":
|
199 |
+
parser = argparse.ArgumentParser()
|
200 |
+
|
201 |
+
parser.add_argument("--model_path", default=None, type=str, required=True, help="Path to the model to convert.")
|
202 |
+
parser.add_argument("--checkpoint_path", default=None, type=str, required=True, help="Path to the output model.")
|
203 |
+
parser.add_argument("--half", action="store_true", help="Save weights in half precision.")
|
204 |
+
|
205 |
+
args = parser.parse_args()
|
206 |
+
|
207 |
+
assert args.model_path is not None, "Must provide a model path!"
|
208 |
+
|
209 |
+
assert args.checkpoint_path is not None, "Must provide a checkpoint path!"
|
210 |
+
|
211 |
+
unet_path = osp.join(args.model_path, "unet", "diffusion_pytorch_model.bin")
|
212 |
+
vae_path = osp.join(args.model_path, "vae", "diffusion_pytorch_model.bin")
|
213 |
+
text_enc_path = osp.join(args.model_path, "text_encoder", "pytorch_model.bin")
|
214 |
+
|
215 |
+
# Convert the UNet model
|
216 |
+
unet_state_dict = torch.load(unet_path, map_location='cpu')
|
217 |
+
unet_state_dict = convert_unet_state_dict(unet_state_dict)
|
218 |
+
unet_state_dict = {"model.diffusion_model." + k: v for k, v in unet_state_dict.items()}
|
219 |
+
|
220 |
+
# Convert the VAE model
|
221 |
+
vae_state_dict = torch.load(vae_path, map_location='cpu')
|
222 |
+
vae_state_dict = convert_vae_state_dict(vae_state_dict)
|
223 |
+
vae_state_dict = {"first_stage_model." + k: v for k, v in vae_state_dict.items()}
|
224 |
+
|
225 |
+
# Convert the text encoder model
|
226 |
+
text_enc_dict = torch.load(text_enc_path, map_location='cpu')
|
227 |
+
text_enc_dict = convert_text_enc_state_dict(text_enc_dict)
|
228 |
+
text_enc_dict = {"cond_stage_model.transformer." + k: v for k, v in text_enc_dict.items()}
|
229 |
+
|
230 |
+
# Put together new checkpoint
|
231 |
+
state_dict = {**unet_state_dict, **vae_state_dict, **text_enc_dict}
|
232 |
+
if args.half:
|
233 |
+
state_dict = {k:v.half() for k,v in state_dict.items()}
|
234 |
+
state_dict = {"state_dict": state_dict}
|
235 |
+
torch.save(state_dict, args.checkpoint_path)
|
feature_extractor/preprocessor_config.json
ADDED
@@ -0,0 +1,20 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"crop_size": 224,
|
3 |
+
"do_center_crop": true,
|
4 |
+
"do_convert_rgb": true,
|
5 |
+
"do_normalize": true,
|
6 |
+
"do_resize": true,
|
7 |
+
"feature_extractor_type": "CLIPFeatureExtractor",
|
8 |
+
"image_mean": [
|
9 |
+
0.48145466,
|
10 |
+
0.4578275,
|
11 |
+
0.40821073
|
12 |
+
],
|
13 |
+
"image_std": [
|
14 |
+
0.26862954,
|
15 |
+
0.26130258,
|
16 |
+
0.27577711
|
17 |
+
],
|
18 |
+
"resample": 3,
|
19 |
+
"size": 224
|
20 |
+
}
|
model_index.json
ADDED
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_class_name": "StableDiffusionPipeline",
|
3 |
+
"_diffusers_version": "0.6.0",
|
4 |
+
"feature_extractor": [
|
5 |
+
"transformers",
|
6 |
+
"CLIPFeatureExtractor"
|
7 |
+
],
|
8 |
+
"safety_checker": [
|
9 |
+
"stable_diffusion",
|
10 |
+
"StableDiffusionSafetyChecker"
|
11 |
+
],
|
12 |
+
"scheduler": [
|
13 |
+
"diffusers",
|
14 |
+
"PNDMScheduler"
|
15 |
+
],
|
16 |
+
"text_encoder": [
|
17 |
+
"transformers",
|
18 |
+
"CLIPTextModel"
|
19 |
+
],
|
20 |
+
"tokenizer": [
|
21 |
+
"transformers",
|
22 |
+
"CLIPTokenizer"
|
23 |
+
],
|
24 |
+
"unet": [
|
25 |
+
"diffusers",
|
26 |
+
"UNet2DConditionModel"
|
27 |
+
],
|
28 |
+
"vae": [
|
29 |
+
"diffusers",
|
30 |
+
"AutoencoderKL"
|
31 |
+
]
|
32 |
+
}
|
safety_checker/config.json
ADDED
@@ -0,0 +1,175 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_commit_hash": "4bb648a606ef040e7685bde262611766a5fdd67b",
|
3 |
+
"_name_or_path": "CompVis/stable-diffusion-safety-checker",
|
4 |
+
"architectures": [
|
5 |
+
"StableDiffusionSafetyChecker"
|
6 |
+
],
|
7 |
+
"initializer_factor": 1.0,
|
8 |
+
"logit_scale_init_value": 2.6592,
|
9 |
+
"model_type": "clip",
|
10 |
+
"projection_dim": 768,
|
11 |
+
"text_config": {
|
12 |
+
"_name_or_path": "",
|
13 |
+
"add_cross_attention": false,
|
14 |
+
"architectures": null,
|
15 |
+
"attention_dropout": 0.0,
|
16 |
+
"bad_words_ids": null,
|
17 |
+
"bos_token_id": 0,
|
18 |
+
"chunk_size_feed_forward": 0,
|
19 |
+
"cross_attention_hidden_size": null,
|
20 |
+
"decoder_start_token_id": null,
|
21 |
+
"diversity_penalty": 0.0,
|
22 |
+
"do_sample": false,
|
23 |
+
"dropout": 0.0,
|
24 |
+
"early_stopping": false,
|
25 |
+
"encoder_no_repeat_ngram_size": 0,
|
26 |
+
"eos_token_id": 2,
|
27 |
+
"exponential_decay_length_penalty": null,
|
28 |
+
"finetuning_task": null,
|
29 |
+
"forced_bos_token_id": null,
|
30 |
+
"forced_eos_token_id": null,
|
31 |
+
"hidden_act": "quick_gelu",
|
32 |
+
"hidden_size": 768,
|
33 |
+
"id2label": {
|
34 |
+
"0": "LABEL_0",
|
35 |
+
"1": "LABEL_1"
|
36 |
+
},
|
37 |
+
"initializer_factor": 1.0,
|
38 |
+
"initializer_range": 0.02,
|
39 |
+
"intermediate_size": 3072,
|
40 |
+
"is_decoder": false,
|
41 |
+
"is_encoder_decoder": false,
|
42 |
+
"label2id": {
|
43 |
+
"LABEL_0": 0,
|
44 |
+
"LABEL_1": 1
|
45 |
+
},
|
46 |
+
"layer_norm_eps": 1e-05,
|
47 |
+
"length_penalty": 1.0,
|
48 |
+
"max_length": 20,
|
49 |
+
"max_position_embeddings": 77,
|
50 |
+
"min_length": 0,
|
51 |
+
"model_type": "clip_text_model",
|
52 |
+
"no_repeat_ngram_size": 0,
|
53 |
+
"num_attention_heads": 12,
|
54 |
+
"num_beam_groups": 1,
|
55 |
+
"num_beams": 1,
|
56 |
+
"num_hidden_layers": 12,
|
57 |
+
"num_return_sequences": 1,
|
58 |
+
"output_attentions": false,
|
59 |
+
"output_hidden_states": false,
|
60 |
+
"output_scores": false,
|
61 |
+
"pad_token_id": 1,
|
62 |
+
"prefix": null,
|
63 |
+
"problem_type": null,
|
64 |
+
"pruned_heads": {},
|
65 |
+
"remove_invalid_values": false,
|
66 |
+
"repetition_penalty": 1.0,
|
67 |
+
"return_dict": true,
|
68 |
+
"return_dict_in_generate": false,
|
69 |
+
"sep_token_id": null,
|
70 |
+
"task_specific_params": null,
|
71 |
+
"temperature": 1.0,
|
72 |
+
"tf_legacy_loss": false,
|
73 |
+
"tie_encoder_decoder": false,
|
74 |
+
"tie_word_embeddings": true,
|
75 |
+
"tokenizer_class": null,
|
76 |
+
"top_k": 50,
|
77 |
+
"top_p": 1.0,
|
78 |
+
"torch_dtype": null,
|
79 |
+
"torchscript": false,
|
80 |
+
"transformers_version": "4.22.0.dev0",
|
81 |
+
"typical_p": 1.0,
|
82 |
+
"use_bfloat16": false,
|
83 |
+
"vocab_size": 49408
|
84 |
+
},
|
85 |
+
"text_config_dict": {
|
86 |
+
"hidden_size": 768,
|
87 |
+
"intermediate_size": 3072,
|
88 |
+
"num_attention_heads": 12,
|
89 |
+
"num_hidden_layers": 12
|
90 |
+
},
|
91 |
+
"torch_dtype": "float32",
|
92 |
+
"transformers_version": null,
|
93 |
+
"vision_config": {
|
94 |
+
"_name_or_path": "",
|
95 |
+
"add_cross_attention": false,
|
96 |
+
"architectures": null,
|
97 |
+
"attention_dropout": 0.0,
|
98 |
+
"bad_words_ids": null,
|
99 |
+
"bos_token_id": null,
|
100 |
+
"chunk_size_feed_forward": 0,
|
101 |
+
"cross_attention_hidden_size": null,
|
102 |
+
"decoder_start_token_id": null,
|
103 |
+
"diversity_penalty": 0.0,
|
104 |
+
"do_sample": false,
|
105 |
+
"dropout": 0.0,
|
106 |
+
"early_stopping": false,
|
107 |
+
"encoder_no_repeat_ngram_size": 0,
|
108 |
+
"eos_token_id": null,
|
109 |
+
"exponential_decay_length_penalty": null,
|
110 |
+
"finetuning_task": null,
|
111 |
+
"forced_bos_token_id": null,
|
112 |
+
"forced_eos_token_id": null,
|
113 |
+
"hidden_act": "quick_gelu",
|
114 |
+
"hidden_size": 1024,
|
115 |
+
"id2label": {
|
116 |
+
"0": "LABEL_0",
|
117 |
+
"1": "LABEL_1"
|
118 |
+
},
|
119 |
+
"image_size": 224,
|
120 |
+
"initializer_factor": 1.0,
|
121 |
+
"initializer_range": 0.02,
|
122 |
+
"intermediate_size": 4096,
|
123 |
+
"is_decoder": false,
|
124 |
+
"is_encoder_decoder": false,
|
125 |
+
"label2id": {
|
126 |
+
"LABEL_0": 0,
|
127 |
+
"LABEL_1": 1
|
128 |
+
},
|
129 |
+
"layer_norm_eps": 1e-05,
|
130 |
+
"length_penalty": 1.0,
|
131 |
+
"max_length": 20,
|
132 |
+
"min_length": 0,
|
133 |
+
"model_type": "clip_vision_model",
|
134 |
+
"no_repeat_ngram_size": 0,
|
135 |
+
"num_attention_heads": 16,
|
136 |
+
"num_beam_groups": 1,
|
137 |
+
"num_beams": 1,
|
138 |
+
"num_channels": 3,
|
139 |
+
"num_hidden_layers": 24,
|
140 |
+
"num_return_sequences": 1,
|
141 |
+
"output_attentions": false,
|
142 |
+
"output_hidden_states": false,
|
143 |
+
"output_scores": false,
|
144 |
+
"pad_token_id": null,
|
145 |
+
"patch_size": 14,
|
146 |
+
"prefix": null,
|
147 |
+
"problem_type": null,
|
148 |
+
"pruned_heads": {},
|
149 |
+
"remove_invalid_values": false,
|
150 |
+
"repetition_penalty": 1.0,
|
151 |
+
"return_dict": true,
|
152 |
+
"return_dict_in_generate": false,
|
153 |
+
"sep_token_id": null,
|
154 |
+
"task_specific_params": null,
|
155 |
+
"temperature": 1.0,
|
156 |
+
"tf_legacy_loss": false,
|
157 |
+
"tie_encoder_decoder": false,
|
158 |
+
"tie_word_embeddings": true,
|
159 |
+
"tokenizer_class": null,
|
160 |
+
"top_k": 50,
|
161 |
+
"top_p": 1.0,
|
162 |
+
"torch_dtype": null,
|
163 |
+
"torchscript": false,
|
164 |
+
"transformers_version": "4.22.0.dev0",
|
165 |
+
"typical_p": 1.0,
|
166 |
+
"use_bfloat16": false
|
167 |
+
},
|
168 |
+
"vision_config_dict": {
|
169 |
+
"hidden_size": 1024,
|
170 |
+
"intermediate_size": 4096,
|
171 |
+
"num_attention_heads": 16,
|
172 |
+
"num_hidden_layers": 24,
|
173 |
+
"patch_size": 14
|
174 |
+
}
|
175 |
+
}
|
safety_checker/pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:193490b58ef62739077262e833bf091c66c29488058681ac25cf7df3d8190974
|
3 |
+
size 1216061799
|
scheduler/scheduler_config.json
ADDED
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_class_name": "DDIMScheduler",
|
3 |
+
"_diffusers_version": "0.7.0.dev0",
|
4 |
+
"beta_end": 0.012,
|
5 |
+
"beta_schedule": "scaled_linear",
|
6 |
+
"beta_start": 0.00085,
|
7 |
+
"clip_sample": false,
|
8 |
+
"num_train_timesteps": 1000,
|
9 |
+
"set_alpha_to_one": false,
|
10 |
+
"steps_offset": 1,
|
11 |
+
"trained_betas": null
|
12 |
+
}
|
text_encoder/config.json
ADDED
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "models/b_step1/20000",
|
3 |
+
"architectures": [
|
4 |
+
"CLIPTextModel"
|
5 |
+
],
|
6 |
+
"attention_dropout": 0.0,
|
7 |
+
"bos_token_id": 0,
|
8 |
+
"dropout": 0.0,
|
9 |
+
"eos_token_id": 2,
|
10 |
+
"hidden_act": "quick_gelu",
|
11 |
+
"hidden_size": 768,
|
12 |
+
"initializer_factor": 1.0,
|
13 |
+
"initializer_range": 0.02,
|
14 |
+
"intermediate_size": 3072,
|
15 |
+
"layer_norm_eps": 1e-05,
|
16 |
+
"max_position_embeddings": 77,
|
17 |
+
"model_type": "clip_text_model",
|
18 |
+
"num_attention_heads": 12,
|
19 |
+
"num_hidden_layers": 12,
|
20 |
+
"pad_token_id": 1,
|
21 |
+
"projection_dim": 768,
|
22 |
+
"torch_dtype": "float32",
|
23 |
+
"transformers_version": "4.24.0",
|
24 |
+
"vocab_size": 49408
|
25 |
+
}
|
text_encoder/pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d9763dce458c00c60d8c3a8206310410fac434b8c7094d6b9993542991cb048f
|
3 |
+
size 492309793
|
tokenizer/merges.txt
ADDED
The diff for this file is too large to render.
See raw diff
|
|
tokenizer/special_tokens_map.json
ADDED
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"bos_token": {
|
3 |
+
"content": "<|startoftext|>",
|
4 |
+
"lstrip": false,
|
5 |
+
"normalized": true,
|
6 |
+
"rstrip": false,
|
7 |
+
"single_word": false
|
8 |
+
},
|
9 |
+
"eos_token": {
|
10 |
+
"content": "<|endoftext|>",
|
11 |
+
"lstrip": false,
|
12 |
+
"normalized": true,
|
13 |
+
"rstrip": false,
|
14 |
+
"single_word": false
|
15 |
+
},
|
16 |
+
"pad_token": "<|endoftext|>",
|
17 |
+
"unk_token": {
|
18 |
+
"content": "<|endoftext|>",
|
19 |
+
"lstrip": false,
|
20 |
+
"normalized": true,
|
21 |
+
"rstrip": false,
|
22 |
+
"single_word": false
|
23 |
+
}
|
24 |
+
}
|
tokenizer/tokenizer_config.json
ADDED
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"add_prefix_space": false,
|
3 |
+
"bos_token": {
|
4 |
+
"__type": "AddedToken",
|
5 |
+
"content": "<|startoftext|>",
|
6 |
+
"lstrip": false,
|
7 |
+
"normalized": true,
|
8 |
+
"rstrip": false,
|
9 |
+
"single_word": false
|
10 |
+
},
|
11 |
+
"do_lower_case": true,
|
12 |
+
"eos_token": {
|
13 |
+
"__type": "AddedToken",
|
14 |
+
"content": "<|endoftext|>",
|
15 |
+
"lstrip": false,
|
16 |
+
"normalized": true,
|
17 |
+
"rstrip": false,
|
18 |
+
"single_word": false
|
19 |
+
},
|
20 |
+
"errors": "replace",
|
21 |
+
"model_max_length": 77,
|
22 |
+
"name_or_path": "models/b_step1/20000\\tokenizer",
|
23 |
+
"pad_token": "<|endoftext|>",
|
24 |
+
"special_tokens_map_file": "./special_tokens_map.json",
|
25 |
+
"tokenizer_class": "CLIPTokenizer",
|
26 |
+
"unk_token": {
|
27 |
+
"__type": "AddedToken",
|
28 |
+
"content": "<|endoftext|>",
|
29 |
+
"lstrip": false,
|
30 |
+
"normalized": true,
|
31 |
+
"rstrip": false,
|
32 |
+
"single_word": false
|
33 |
+
}
|
34 |
+
}
|
tokenizer/vocab.json
ADDED
The diff for this file is too large to render.
See raw diff
|
|
unet/config.json
ADDED
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_class_name": "UNet2DConditionModel",
|
3 |
+
"_diffusers_version": "0.7.0.dev0",
|
4 |
+
"_name_or_path": "models/b_step1/20000",
|
5 |
+
"act_fn": "silu",
|
6 |
+
"attention_head_dim": 8,
|
7 |
+
"block_out_channels": [
|
8 |
+
320,
|
9 |
+
640,
|
10 |
+
1280,
|
11 |
+
1280
|
12 |
+
],
|
13 |
+
"center_input_sample": false,
|
14 |
+
"cross_attention_dim": 768,
|
15 |
+
"down_block_types": [
|
16 |
+
"CrossAttnDownBlock2D",
|
17 |
+
"CrossAttnDownBlock2D",
|
18 |
+
"CrossAttnDownBlock2D",
|
19 |
+
"DownBlock2D"
|
20 |
+
],
|
21 |
+
"downsample_padding": 1,
|
22 |
+
"flip_sin_to_cos": true,
|
23 |
+
"freq_shift": 0,
|
24 |
+
"in_channels": 4,
|
25 |
+
"layers_per_block": 2,
|
26 |
+
"mid_block_scale_factor": 1,
|
27 |
+
"norm_eps": 1e-05,
|
28 |
+
"norm_num_groups": 32,
|
29 |
+
"out_channels": 4,
|
30 |
+
"sample_size": 64,
|
31 |
+
"up_block_types": [
|
32 |
+
"UpBlock2D",
|
33 |
+
"CrossAttnUpBlock2D",
|
34 |
+
"CrossAttnUpBlock2D",
|
35 |
+
"CrossAttnUpBlock2D"
|
36 |
+
]
|
37 |
+
}
|
unet/diffusion_pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ea0b39120d0adb2d9736d2ffada4e4b7d56487562456927e55b71cd27f180570
|
3 |
+
size 3438375973
|
vae/README.md
ADDED
@@ -0,0 +1,84 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: mit
|
3 |
+
tags:
|
4 |
+
- stable-diffusion
|
5 |
+
- stable-diffusion-diffusers
|
6 |
+
- text-to-image
|
7 |
+
inference: false
|
8 |
+
---
|
9 |
+
# Improved Autoencoders
|
10 |
+
|
11 |
+
## Utilizing
|
12 |
+
These weights are intended to be used with the [🧨 diffusers library](https://github.com/huggingface/diffusers). If you are looking for the model to use with the original [CompVis Stable Diffusion codebase](https://github.com/CompVis/stable-diffusion), [come here](https://huggingface.co/CompVis/stabilityai/sd-vae-ft-ema-original).
|
13 |
+
|
14 |
+
#### How to use with 🧨 diffusers
|
15 |
+
You can integrate this fine-tuned VAE decoder to your existing `diffusers` workflows, by including a `vae` argument to the `StableDiffusionPipeline`
|
16 |
+
```py
|
17 |
+
from diffusers.models import AutoencoderKL
|
18 |
+
from diffusers import StableDiffusionPipeline
|
19 |
+
|
20 |
+
model = "CompVis/stable-diffusion-v1-4"
|
21 |
+
vae = AutoencoderKL.from_pretrained("stabilityai/sd-vae-ft-mse")
|
22 |
+
pipe = StableDiffusionPipeline.from_pretrained(model, vae=vae)
|
23 |
+
```
|
24 |
+
|
25 |
+
## Decoder Finetuning
|
26 |
+
We publish two kl-f8 autoencoder versions, finetuned from the original [kl-f8 autoencoder](https://github.com/CompVis/latent-diffusion#pretrained-autoencoding-models).
|
27 |
+
The first, _ft-EMA_, was resumed from the original checkpoint, trained for 313198 steps and uses EMA weights.
|
28 |
+
The second, _ft-MSE_, was resumed from _ft-EMA_ and uses EMA weights and was trained for another 280k steps using a re-weighted loss, with more emphasis
|
29 |
+
on MSE reconstruction (producing somewhat ``smoother'' outputs).
|
30 |
+
To keep compatibility with existing models, only the decoder part was finetuned; the checkpoints can be used as a drop-in replacement for the existing autoencoder.
|
31 |
+
|
32 |
+
_Original kl-f8 VAE vs f8-ft-EMA vs f8-ft-MSE_
|
33 |
+
|
34 |
+
## Evaluation
|
35 |
+
### COCO 2017 (256x256, val, 5000 images)
|
36 |
+
| Model | train steps | rFID | PSNR | SSIM | PSIM | Link | Comments
|
37 |
+
|----------|---------|------|--------------|---------------|---------------|-----------------------------------------------------------------------------------|-------------------------------------------------------------------------------------------------|
|
38 |
+
| | | | | | | | |
|
39 |
+
| original | 246803 | 4.99 | 23.4 +/- 3.8 | 0.69 +/- 0.14 | 1.01 +/- 0.28 | https://ommer-lab.com/files/latent-diffusion/kl-f8.zip | as used in SD |
|
40 |
+
| ft-EMA | 560001 | 4.42 | 23.8 +/- 3.9 | 0.69 +/- 0.13 | 0.96 +/- 0.27 | https://huggingface.co/stabilityai/sd-vae-ft-ema-original/resolve/main/vae-ft-ema-560000-ema-pruned.ckpt | slightly better overall, with EMA |
|
41 |
+
| ft-MSE | 840001 | 4.70 | 24.5 +/- 3.7 | 0.71 +/- 0.13 | 0.92 +/- 0.27 | https://huggingface.co/stabilityai/sd-vae-ft-mse-original/resolve/main/vae-ft-mse-840000-ema-pruned.ckpt | resumed with EMA from ft-EMA, emphasis on MSE (rec. loss = MSE + 0.1 * LPIPS), smoother outputs |
|
42 |
+
|
43 |
+
|
44 |
+
### LAION-Aesthetics 5+ (256x256, subset, 10000 images)
|
45 |
+
| Model | train steps | rFID | PSNR | SSIM | PSIM | Link | Comments
|
46 |
+
|----------|-----------|------|--------------|---------------|---------------|-----------------------------------------------------------------------------------|-------------------------------------------------------------------------------------------------|
|
47 |
+
| | | | | | | | |
|
48 |
+
| original | 246803 | 2.61 | 26.0 +/- 4.4 | 0.81 +/- 0.12 | 0.75 +/- 0.36 | https://ommer-lab.com/files/latent-diffusion/kl-f8.zip | as used in SD |
|
49 |
+
| ft-EMA | 560001 | 1.77 | 26.7 +/- 4.8 | 0.82 +/- 0.12 | 0.67 +/- 0.34 | https://huggingface.co/stabilityai/sd-vae-ft-ema-original/resolve/main/vae-ft-ema-560000-ema-pruned.ckpt | slightly better overall, with EMA |
|
50 |
+
| ft-MSE | 840001 | 1.88 | 27.3 +/- 4.7 | 0.83 +/- 0.11 | 0.65 +/- 0.34 | https://huggingface.co/stabilityai/sd-vae-ft-mse-original/resolve/main/vae-ft-mse-840000-ema-pruned.ckpt | resumed with EMA from ft-EMA, emphasis on MSE (rec. loss = MSE + 0.1 * LPIPS), smoother outputs |
|
51 |
+
|
52 |
+
|
53 |
+
### Visual
|
54 |
+
_Visualization of reconstructions on 256x256 images from the COCO2017 validation dataset._
|
55 |
+
|
56 |
+
<p align="center">
|
57 |
+
<br>
|
58 |
+
<b>
|
59 |
+
256x256: ft-EMA (left), ft-MSE (middle), original (right)</b>
|
60 |
+
</p>
|
61 |
+
|
62 |
+
<p align="center">
|
63 |
+
<img src=https://huggingface.co/stabilityai/stable-diffusion-decoder-finetune/resolve/main/eval/ae-decoder-tuning-reconstructions/merged/00025_merged.png />
|
64 |
+
</p>
|
65 |
+
|
66 |
+
<p align="center">
|
67 |
+
<img src=https://huggingface.co/stabilityai/stable-diffusion-decoder-finetune/resolve/main/eval/ae-decoder-tuning-reconstructions/merged/00011_merged.png />
|
68 |
+
</p>
|
69 |
+
|
70 |
+
<p align="center">
|
71 |
+
<img src=https://huggingface.co/stabilityai/stable-diffusion-decoder-finetune/resolve/main/eval/ae-decoder-tuning-reconstructions/merged/00037_merged.png />
|
72 |
+
</p>
|
73 |
+
|
74 |
+
<p align="center">
|
75 |
+
<img src=https://huggingface.co/stabilityai/stable-diffusion-decoder-finetune/resolve/main/eval/ae-decoder-tuning-reconstructions/merged/00043_merged.png />
|
76 |
+
</p>
|
77 |
+
|
78 |
+
<p align="center">
|
79 |
+
<img src=https://huggingface.co/stabilityai/stable-diffusion-decoder-finetune/resolve/main/eval/ae-decoder-tuning-reconstructions/merged/00053_merged.png />
|
80 |
+
</p>
|
81 |
+
|
82 |
+
<p align="center">
|
83 |
+
<img src=https://huggingface.co/stabilityai/stable-diffusion-decoder-finetune/resolve/main/eval/ae-decoder-tuning-reconstructions/merged/00029_merged.png />
|
84 |
+
</p>
|
vae/config.json
ADDED
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_class_name": "AutoencoderKL",
|
3 |
+
"_diffusers_version": "0.7.0.dev0",
|
4 |
+
"_name_or_path": "stabilityai/sd-vae-ft-mse",
|
5 |
+
"act_fn": "silu",
|
6 |
+
"block_out_channels": [
|
7 |
+
128,
|
8 |
+
256,
|
9 |
+
512,
|
10 |
+
512
|
11 |
+
],
|
12 |
+
"down_block_types": [
|
13 |
+
"DownEncoderBlock2D",
|
14 |
+
"DownEncoderBlock2D",
|
15 |
+
"DownEncoderBlock2D",
|
16 |
+
"DownEncoderBlock2D"
|
17 |
+
],
|
18 |
+
"in_channels": 3,
|
19 |
+
"latent_channels": 4,
|
20 |
+
"layers_per_block": 2,
|
21 |
+
"norm_num_groups": 32,
|
22 |
+
"out_channels": 3,
|
23 |
+
"sample_size": 256,
|
24 |
+
"up_block_types": [
|
25 |
+
"UpDecoderBlock2D",
|
26 |
+
"UpDecoderBlock2D",
|
27 |
+
"UpDecoderBlock2D",
|
28 |
+
"UpDecoderBlock2D"
|
29 |
+
]
|
30 |
+
}
|
vae/diffusion_pytorch_model.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:36bb8e1b54aba3a0914eb35fba13dcb107e9f18d379d1df2158732cd4bf56a94
|
3 |
+
size 334711857
|