Hugging Face's logo

AbstractFramework
/
seedvr2-7b-4bit

Image-to-Image
Safetensors
MLX
mlx-gen
mflux
apple-silicon
4-bit precision
seedvr2
image-upscaling
super-resolution
Model card Files
xet
 Community
lpalbou commited on
Commit
22f491b
·
verified ·
1 Parent(s): 0674755

Upload folder using huggingface_hub

Browse files
Files changed (7) hide show
  1. README.md +82 -0
  2. transformer/0.safetensors +3 -0
  3. transformer/1.safetensors +3 -0
  4. transformer/2.safetensors +3 -0
  5. transformer/model.safetensors.index.json +0 -0
  6. vae/0.safetensors +3 -0
  7. vae/model.safetensors.index.json +274 -0
README.md ADDED
@@ -0,0 +1,82 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ base_model: ByteDance-Seed/SeedVR2-7B
4
+ pipeline_tag: image-to-image
5
+ library_name: mlx-gen
6
+ tags:
7
+ - mlx
8
+ - mlx-gen
9
+ - mflux
10
+ - apple-silicon
11
+ - 4-bit
12
+ - seedvr2
13
+ - image-upscaling
14
+ - super-resolution
15
+ ---
16
+ # seedvr2-7b-4bit
17
+
18
+ This repository contains MLX-Gen saved weights for `ByteDance-Seed/SeedVR2-7B`. The checkpoint is designed for local Apple Silicon inference with [`mlx-gen`](https://github.com/lpalbou/mlx-gen).
19
+
20
+ It uses the mflux/MLX saved-weight layout. Quantized checkpoints include MLX quantization tensors. It is not a Diffusers or Transformers `from_pretrained()` checkpoint.
21
+
22
+ ## Source Model
23
+
24
+ Original model: [`ByteDance-Seed/SeedVR2-7B`](https://huggingface.co/ByteDance-Seed/SeedVR2-7B).
25
+
26
+ ## License and Access
27
+
28
+ This quantized derivative follows the Apache 2.0 license of the source model.
29
+
30
+ ## Quantization
31
+
32
+ This is an MLX 4-bit checkpoint for SeedVR2 7B image super-resolution.
33
+
34
+ - q4 for quantizable SeedVR2 transformer linears and VAE attention linears.
35
+ - BF16 for convolutions, normalization layers, and other non-quantizable parameters.
36
+ - The static SeedVR2 positive embedding is provided by the MLX-Gen runtime.
37
+
38
+ See the [MLX-Gen quantization docs](https://github.com/lpalbou/mlx-gen/blob/main/docs/quantization.md) for compatibility notes and measured behavior.
39
+
40
+ ## Validation
41
+
42
+ Measured on an Apple M5 Max with 128 GB unified memory using `mlx-gen 0.18.13`.
43
+ The profile upscales `docs/assets/upscaling/seedvr2-5x-source.jpg` from `133x113`
44
+ to `658x560` with `--resolution 5x --seed 42 --metadata`.
45
+
46
+ | Package | Storage | Generation time | Wall time | Max RSS |
47
+ | --- | ---: | ---: | ---: | ---: |
48
+ | `ByteDance-Seed/SeedVR2-7B` source generation files | 31.63 GiB | 2.64 s | 8.69 s | 61.62 GiB |
49
+ | `AbstractFramework/seedvr2-7b-4bit` | 4.79 GiB | 2.21 s | 3.24 s | 5.10 GiB |
50
+
51
+ See the [SeedVR2 upscaling guide](https://github.com/lpalbou/mlx-gen/blob/main/docs/upscaling.md)
52
+ for the 3B/7B source, q8, and q4 comparison sheet generated from this profile.
53
+
54
+ ## Compatibility
55
+
56
+ Requires `mlx-gen >= 0.18.13`.
57
+
58
+ Generated with `mlx-gen 0.18.13`.
59
+
60
+ Use `mlxgen download` and `mlxgen upscale` for this SeedVR2 package.
61
+
62
+ ## Usage
63
+
64
+ ```bash
65
+ python -m pip install -U mlx-gen
66
+
67
+ mlxgen download --model AbstractFramework/seedvr2-7b-4bit
68
+
69
+ mlxgen upscale \
70
+ --model AbstractFramework/seedvr2-7b-4bit \
71
+ --image-path input.png \
72
+ --resolution 2x \
73
+ --seed 42 \
74
+ --metadata \
75
+ --output upscaled.png
76
+ ```
77
+
78
+ ## Attribution
79
+
80
+ MLX-Gen is based on [mflux](https://github.com/filipstrand/mflux) by Filip Strand and the original mflux contributors. This model card is generated by MLX-Gen so derived checkpoints keep that attribution visible.
81
+
82
+ Quantized and contributed by [@lpalbou](https://huggingface.co/lpalbou).
transformer/0.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ac398fa0983edca076627fd4dc8bb13cd9ea4094c60d56ca3b852b31f83a31ff
3
+ size 2130692977
transformer/1.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2773096ae8d324c69f8816789221e9344c06c8951bf6a3db049347b1d22c8759
3
+ size 2147202171
transformer/2.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a520462eac3e3a3711db5aac0340093e501ebaaaddb6afe284f7d8ecdad6e9df
3
+ size 361536118
transformer/model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
vae/0.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:df03120c5c02d862dd0b78c42cc58b9dce3c7b28252ac9604e7544ddcd14e37c
3
+ size 498312006
vae/model.safetensors.index.json ADDED
@@ -0,0 +1,274 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "quantization_level": "4",
4
+ "mflux_version": "0.18.13"
5
+ },
6
+ "weight_map": {
7
+ "encoder.conv_in.weight": "0.safetensors",
8
+ "encoder.conv_in.bias": "0.safetensors",
9
+ "encoder.down_blocks.0.resnets.0.norm1.bias": "0.safetensors",
10
+ "encoder.down_blocks.0.resnets.0.norm1.weight": "0.safetensors",
11
+ "encoder.down_blocks.0.resnets.0.norm2.bias": "0.safetensors",
12
+ "encoder.down_blocks.0.resnets.0.norm2.weight": "0.safetensors",
13
+ "encoder.down_blocks.0.resnets.0.conv1.weight": "0.safetensors",
14
+ "encoder.down_blocks.0.resnets.0.conv1.bias": "0.safetensors",
15
+ "encoder.down_blocks.0.resnets.0.conv2.weight": "0.safetensors",
16
+ "encoder.down_blocks.0.resnets.0.conv2.bias": "0.safetensors",
17
+ "encoder.down_blocks.0.resnets.1.norm1.bias": "0.safetensors",
18
+ "encoder.down_blocks.0.resnets.1.norm1.weight": "0.safetensors",
19
+ "encoder.down_blocks.0.resnets.1.norm2.bias": "0.safetensors",
20
+ "encoder.down_blocks.0.resnets.1.norm2.weight": "0.safetensors",
21
+ "encoder.down_blocks.0.resnets.1.conv1.weight": "0.safetensors",
22
+ "encoder.down_blocks.0.resnets.1.conv1.bias": "0.safetensors",
23
+ "encoder.down_blocks.0.resnets.1.conv2.weight": "0.safetensors",
24
+ "encoder.down_blocks.0.resnets.1.conv2.bias": "0.safetensors",
25
+ "encoder.down_blocks.0.downsamplers.0.conv.weight": "0.safetensors",
26
+ "encoder.down_blocks.0.downsamplers.0.conv.bias": "0.safetensors",
27
+ "encoder.down_blocks.1.resnets.0.norm1.bias": "0.safetensors",
28
+ "encoder.down_blocks.1.resnets.0.norm1.weight": "0.safetensors",
29
+ "encoder.down_blocks.1.resnets.0.norm2.bias": "0.safetensors",
30
+ "encoder.down_blocks.1.resnets.0.norm2.weight": "0.safetensors",
31
+ "encoder.down_blocks.1.resnets.0.conv1.weight": "0.safetensors",
32
+ "encoder.down_blocks.1.resnets.0.conv1.bias": "0.safetensors",
33
+ "encoder.down_blocks.1.resnets.0.conv2.weight": "0.safetensors",
34
+ "encoder.down_blocks.1.resnets.0.conv2.bias": "0.safetensors",
35
+ "encoder.down_blocks.1.resnets.0.conv_shortcut.weight": "0.safetensors",
36
+ "encoder.down_blocks.1.resnets.0.conv_shortcut.bias": "0.safetensors",
37
+ "encoder.down_blocks.1.resnets.1.norm1.bias": "0.safetensors",
38
+ "encoder.down_blocks.1.resnets.1.norm1.weight": "0.safetensors",
39
+ "encoder.down_blocks.1.resnets.1.norm2.bias": "0.safetensors",
40
+ "encoder.down_blocks.1.resnets.1.norm2.weight": "0.safetensors",
41
+ "encoder.down_blocks.1.resnets.1.conv1.weight": "0.safetensors",
42
+ "encoder.down_blocks.1.resnets.1.conv1.bias": "0.safetensors",
43
+ "encoder.down_blocks.1.resnets.1.conv2.weight": "0.safetensors",
44
+ "encoder.down_blocks.1.resnets.1.conv2.bias": "0.safetensors",
45
+ "encoder.down_blocks.1.downsamplers.0.conv.weight": "0.safetensors",
46
+ "encoder.down_blocks.1.downsamplers.0.conv.bias": "0.safetensors",
47
+ "encoder.down_blocks.2.resnets.0.norm1.bias": "0.safetensors",
48
+ "encoder.down_blocks.2.resnets.0.norm1.weight": "0.safetensors",
49
+ "encoder.down_blocks.2.resnets.0.norm2.bias": "0.safetensors",
50
+ "encoder.down_blocks.2.resnets.0.norm2.weight": "0.safetensors",
51
+ "encoder.down_blocks.2.resnets.0.conv1.weight": "0.safetensors",
52
+ "encoder.down_blocks.2.resnets.0.conv1.bias": "0.safetensors",
53
+ "encoder.down_blocks.2.resnets.0.conv2.weight": "0.safetensors",
54
+ "encoder.down_blocks.2.resnets.0.conv2.bias": "0.safetensors",
55
+ "encoder.down_blocks.2.resnets.0.conv_shortcut.weight": "0.safetensors",
56
+ "encoder.down_blocks.2.resnets.0.conv_shortcut.bias": "0.safetensors",
57
+ "encoder.down_blocks.2.resnets.1.norm1.bias": "0.safetensors",
58
+ "encoder.down_blocks.2.resnets.1.norm1.weight": "0.safetensors",
59
+ "encoder.down_blocks.2.resnets.1.norm2.bias": "0.safetensors",
60
+ "encoder.down_blocks.2.resnets.1.norm2.weight": "0.safetensors",
61
+ "encoder.down_blocks.2.resnets.1.conv1.weight": "0.safetensors",
62
+ "encoder.down_blocks.2.resnets.1.conv1.bias": "0.safetensors",
63
+ "encoder.down_blocks.2.resnets.1.conv2.weight": "0.safetensors",
64
+ "encoder.down_blocks.2.resnets.1.conv2.bias": "0.safetensors",
65
+ "encoder.down_blocks.2.downsamplers.0.conv.weight": "0.safetensors",
66
+ "encoder.down_blocks.2.downsamplers.0.conv.bias": "0.safetensors",
67
+ "encoder.down_blocks.3.resnets.0.norm1.bias": "0.safetensors",
68
+ "encoder.down_blocks.3.resnets.0.norm1.weight": "0.safetensors",
69
+ "encoder.down_blocks.3.resnets.0.norm2.bias": "0.safetensors",
70
+ "encoder.down_blocks.3.resnets.0.norm2.weight": "0.safetensors",
71
+ "encoder.down_blocks.3.resnets.0.conv1.weight": "0.safetensors",
72
+ "encoder.down_blocks.3.resnets.0.conv1.bias": "0.safetensors",
73
+ "encoder.down_blocks.3.resnets.0.conv2.weight": "0.safetensors",
74
+ "encoder.down_blocks.3.resnets.0.conv2.bias": "0.safetensors",
75
+ "encoder.down_blocks.3.resnets.1.norm1.bias": "0.safetensors",
76
+ "encoder.down_blocks.3.resnets.1.norm1.weight": "0.safetensors",
77
+ "encoder.down_blocks.3.resnets.1.norm2.bias": "0.safetensors",
78
+ "encoder.down_blocks.3.resnets.1.norm2.weight": "0.safetensors",
79
+ "encoder.down_blocks.3.resnets.1.conv1.weight": "0.safetensors",
80
+ "encoder.down_blocks.3.resnets.1.conv1.bias": "0.safetensors",
81
+ "encoder.down_blocks.3.resnets.1.conv2.weight": "0.safetensors",
82
+ "encoder.down_blocks.3.resnets.1.conv2.bias": "0.safetensors",
83
+ "encoder.mid_block.attentions.0.group_norm.bias": "0.safetensors",
84
+ "encoder.mid_block.attentions.0.group_norm.weight": "0.safetensors",
85
+ "encoder.mid_block.attentions.0.to_q.weight": "0.safetensors",
86
+ "encoder.mid_block.attentions.0.to_q.scales": "0.safetensors",
87
+ "encoder.mid_block.attentions.0.to_q.biases": "0.safetensors",
88
+ "encoder.mid_block.attentions.0.to_q.bias": "0.safetensors",
89
+ "encoder.mid_block.attentions.0.to_k.weight": "0.safetensors",
90
+ "encoder.mid_block.attentions.0.to_k.scales": "0.safetensors",
91
+ "encoder.mid_block.attentions.0.to_k.biases": "0.safetensors",
92
+ "encoder.mid_block.attentions.0.to_k.bias": "0.safetensors",
93
+ "encoder.mid_block.attentions.0.to_v.weight": "0.safetensors",
94
+ "encoder.mid_block.attentions.0.to_v.scales": "0.safetensors",
95
+ "encoder.mid_block.attentions.0.to_v.biases": "0.safetensors",
96
+ "encoder.mid_block.attentions.0.to_v.bias": "0.safetensors",
97
+ "encoder.mid_block.attentions.0.to_out.0.weight": "0.safetensors",
98
+ "encoder.mid_block.attentions.0.to_out.0.scales": "0.safetensors",
99
+ "encoder.mid_block.attentions.0.to_out.0.biases": "0.safetensors",
100
+ "encoder.mid_block.attentions.0.to_out.0.bias": "0.safetensors",
101
+ "encoder.mid_block.resnets.0.norm1.bias": "0.safetensors",
102
+ "encoder.mid_block.resnets.0.norm1.weight": "0.safetensors",
103
+ "encoder.mid_block.resnets.0.norm2.bias": "0.safetensors",
104
+ "encoder.mid_block.resnets.0.norm2.weight": "0.safetensors",
105
+ "encoder.mid_block.resnets.0.conv1.weight": "0.safetensors",
106
+ "encoder.mid_block.resnets.0.conv1.bias": "0.safetensors",
107
+ "encoder.mid_block.resnets.0.conv2.weight": "0.safetensors",
108
+ "encoder.mid_block.resnets.0.conv2.bias": "0.safetensors",
109
+ "encoder.mid_block.resnets.1.norm1.bias": "0.safetensors",
110
+ "encoder.mid_block.resnets.1.norm1.weight": "0.safetensors",
111
+ "encoder.mid_block.resnets.1.norm2.bias": "0.safetensors",
112
+ "encoder.mid_block.resnets.1.norm2.weight": "0.safetensors",
113
+ "encoder.mid_block.resnets.1.conv1.weight": "0.safetensors",
114
+ "encoder.mid_block.resnets.1.conv1.bias": "0.safetensors",
115
+ "encoder.mid_block.resnets.1.conv2.weight": "0.safetensors",
116
+ "encoder.mid_block.resnets.1.conv2.bias": "0.safetensors",
117
+ "encoder.conv_norm_out.bias": "0.safetensors",
118
+ "encoder.conv_norm_out.weight": "0.safetensors",
119
+ "encoder.conv_out.weight": "0.safetensors",
120
+ "encoder.conv_out.bias": "0.safetensors",
121
+ "decoder.conv_in.weight": "0.safetensors",
122
+ "decoder.conv_in.bias": "0.safetensors",
123
+ "decoder.mid_block.attentions.0.group_norm.bias": "0.safetensors",
124
+ "decoder.mid_block.attentions.0.group_norm.weight": "0.safetensors",
125
+ "decoder.mid_block.attentions.0.to_q.weight": "0.safetensors",
126
+ "decoder.mid_block.attentions.0.to_q.scales": "0.safetensors",
127
+ "decoder.mid_block.attentions.0.to_q.biases": "0.safetensors",
128
+ "decoder.mid_block.attentions.0.to_q.bias": "0.safetensors",
129
+ "decoder.mid_block.attentions.0.to_k.weight": "0.safetensors",
130
+ "decoder.mid_block.attentions.0.to_k.scales": "0.safetensors",
131
+ "decoder.mid_block.attentions.0.to_k.biases": "0.safetensors",
132
+ "decoder.mid_block.attentions.0.to_k.bias": "0.safetensors",
133
+ "decoder.mid_block.attentions.0.to_v.weight": "0.safetensors",
134
+ "decoder.mid_block.attentions.0.to_v.scales": "0.safetensors",
135
+ "decoder.mid_block.attentions.0.to_v.biases": "0.safetensors",
136
+ "decoder.mid_block.attentions.0.to_v.bias": "0.safetensors",
137
+ "decoder.mid_block.attentions.0.to_out.0.weight": "0.safetensors",
138
+ "decoder.mid_block.attentions.0.to_out.0.scales": "0.safetensors",
139
+ "decoder.mid_block.attentions.0.to_out.0.biases": "0.safetensors",
140
+ "decoder.mid_block.attentions.0.to_out.0.bias": "0.safetensors",
141
+ "decoder.mid_block.resnets.0.norm1.bias": "0.safetensors",
142
+ "decoder.mid_block.resnets.0.norm1.weight": "0.safetensors",
143
+ "decoder.mid_block.resnets.0.norm2.bias": "0.safetensors",
144
+ "decoder.mid_block.resnets.0.norm2.weight": "0.safetensors",
145
+ "decoder.mid_block.resnets.0.conv1.weight": "0.safetensors",
146
+ "decoder.mid_block.resnets.0.conv1.bias": "0.safetensors",
147
+ "decoder.mid_block.resnets.0.conv2.weight": "0.safetensors",
148
+ "decoder.mid_block.resnets.0.conv2.bias": "0.safetensors",
149
+ "decoder.mid_block.resnets.1.norm1.bias": "0.safetensors",
150
+ "decoder.mid_block.resnets.1.norm1.weight": "0.safetensors",
151
+ "decoder.mid_block.resnets.1.norm2.bias": "0.safetensors",
152
+ "decoder.mid_block.resnets.1.norm2.weight": "0.safetensors",
153
+ "decoder.mid_block.resnets.1.conv1.weight": "0.safetensors",
154
+ "decoder.mid_block.resnets.1.conv1.bias": "0.safetensors",
155
+ "decoder.mid_block.resnets.1.conv2.weight": "0.safetensors",
156
+ "decoder.mid_block.resnets.1.conv2.bias": "0.safetensors",
157
+ "decoder.up_blocks.0.resnets.0.norm1.bias": "0.safetensors",
158
+ "decoder.up_blocks.0.resnets.0.norm1.weight": "0.safetensors",
159
+ "decoder.up_blocks.0.resnets.0.norm2.bias": "0.safetensors",
160
+ "decoder.up_blocks.0.resnets.0.norm2.weight": "0.safetensors",
161
+ "decoder.up_blocks.0.resnets.0.conv1.weight": "0.safetensors",
162
+ "decoder.up_blocks.0.resnets.0.conv1.bias": "0.safetensors",
163
+ "decoder.up_blocks.0.resnets.0.conv2.weight": "0.safetensors",
164
+ "decoder.up_blocks.0.resnets.0.conv2.bias": "0.safetensors",
165
+ "decoder.up_blocks.0.resnets.1.norm1.bias": "0.safetensors",
166
+ "decoder.up_blocks.0.resnets.1.norm1.weight": "0.safetensors",
167
+ "decoder.up_blocks.0.resnets.1.norm2.bias": "0.safetensors",
168
+ "decoder.up_blocks.0.resnets.1.norm2.weight": "0.safetensors",
169
+ "decoder.up_blocks.0.resnets.1.conv1.weight": "0.safetensors",
170
+ "decoder.up_blocks.0.resnets.1.conv1.bias": "0.safetensors",
171
+ "decoder.up_blocks.0.resnets.1.conv2.weight": "0.safetensors",
172
+ "decoder.up_blocks.0.resnets.1.conv2.bias": "0.safetensors",
173
+ "decoder.up_blocks.0.resnets.2.norm1.bias": "0.safetensors",
174
+ "decoder.up_blocks.0.resnets.2.norm1.weight": "0.safetensors",
175
+ "decoder.up_blocks.0.resnets.2.norm2.bias": "0.safetensors",
176
+ "decoder.up_blocks.0.resnets.2.norm2.weight": "0.safetensors",
177
+ "decoder.up_blocks.0.resnets.2.conv1.weight": "0.safetensors",
178
+ "decoder.up_blocks.0.resnets.2.conv1.bias": "0.safetensors",
179
+ "decoder.up_blocks.0.resnets.2.conv2.weight": "0.safetensors",
180
+ "decoder.up_blocks.0.resnets.2.conv2.bias": "0.safetensors",
181
+ "decoder.up_blocks.0.upsamplers.0.conv.weight": "0.safetensors",
182
+ "decoder.up_blocks.0.upsamplers.0.conv.bias": "0.safetensors",
183
+ "decoder.up_blocks.0.upsamplers.0.upscale_conv.weight": "0.safetensors",
184
+ "decoder.up_blocks.0.upsamplers.0.upscale_conv.bias": "0.safetensors",
185
+ "decoder.up_blocks.1.resnets.0.norm1.bias": "0.safetensors",
186
+ "decoder.up_blocks.1.resnets.0.norm1.weight": "0.safetensors",
187
+ "decoder.up_blocks.1.resnets.0.norm2.bias": "0.safetensors",
188
+ "decoder.up_blocks.1.resnets.0.norm2.weight": "0.safetensors",
189
+ "decoder.up_blocks.1.resnets.0.conv1.weight": "0.safetensors",
190
+ "decoder.up_blocks.1.resnets.0.conv1.bias": "0.safetensors",
191
+ "decoder.up_blocks.1.resnets.0.conv2.weight": "0.safetensors",
192
+ "decoder.up_blocks.1.resnets.0.conv2.bias": "0.safetensors",
193
+ "decoder.up_blocks.1.resnets.1.norm1.bias": "0.safetensors",
194
+ "decoder.up_blocks.1.resnets.1.norm1.weight": "0.safetensors",
195
+ "decoder.up_blocks.1.resnets.1.norm2.bias": "0.safetensors",
196
+ "decoder.up_blocks.1.resnets.1.norm2.weight": "0.safetensors",
197
+ "decoder.up_blocks.1.resnets.1.conv1.weight": "0.safetensors",
198
+ "decoder.up_blocks.1.resnets.1.conv1.bias": "0.safetensors",
199
+ "decoder.up_blocks.1.resnets.1.conv2.weight": "0.safetensors",
200
+ "decoder.up_blocks.1.resnets.1.conv2.bias": "0.safetensors",
201
+ "decoder.up_blocks.1.resnets.2.norm1.bias": "0.safetensors",
202
+ "decoder.up_blocks.1.resnets.2.norm1.weight": "0.safetensors",
203
+ "decoder.up_blocks.1.resnets.2.norm2.bias": "0.safetensors",
204
+ "decoder.up_blocks.1.resnets.2.norm2.weight": "0.safetensors",
205
+ "decoder.up_blocks.1.resnets.2.conv1.weight": "0.safetensors",
206
+ "decoder.up_blocks.1.resnets.2.conv1.bias": "0.safetensors",
207
+ "decoder.up_blocks.1.resnets.2.conv2.weight": "0.safetensors",
208
+ "decoder.up_blocks.1.resnets.2.conv2.bias": "0.safetensors",
209
+ "decoder.up_blocks.1.upsamplers.0.conv.weight": "0.safetensors",
210
+ "decoder.up_blocks.1.upsamplers.0.conv.bias": "0.safetensors",
211
+ "decoder.up_blocks.1.upsamplers.0.upscale_conv.weight": "0.safetensors",
212
+ "decoder.up_blocks.1.upsamplers.0.upscale_conv.bias": "0.safetensors",
213
+ "decoder.up_blocks.2.resnets.0.norm1.bias": "0.safetensors",
214
+ "decoder.up_blocks.2.resnets.0.norm1.weight": "0.safetensors",
215
+ "decoder.up_blocks.2.resnets.0.norm2.bias": "0.safetensors",
216
+ "decoder.up_blocks.2.resnets.0.norm2.weight": "0.safetensors",
217
+ "decoder.up_blocks.2.resnets.0.conv1.weight": "0.safetensors",
218
+ "decoder.up_blocks.2.resnets.0.conv1.bias": "0.safetensors",
219
+ "decoder.up_blocks.2.resnets.0.conv2.weight": "0.safetensors",
220
+ "decoder.up_blocks.2.resnets.0.conv2.bias": "0.safetensors",
221
+ "decoder.up_blocks.2.resnets.0.conv_shortcut.weight": "0.safetensors",
222
+ "decoder.up_blocks.2.resnets.0.conv_shortcut.bias": "0.safetensors",
223
+ "decoder.up_blocks.2.resnets.1.norm1.bias": "0.safetensors",
224
+ "decoder.up_blocks.2.resnets.1.norm1.weight": "0.safetensors",
225
+ "decoder.up_blocks.2.resnets.1.norm2.bias": "0.safetensors",
226
+ "decoder.up_blocks.2.resnets.1.norm2.weight": "0.safetensors",
227
+ "decoder.up_blocks.2.resnets.1.conv1.weight": "0.safetensors",
228
+ "decoder.up_blocks.2.resnets.1.conv1.bias": "0.safetensors",
229
+ "decoder.up_blocks.2.resnets.1.conv2.weight": "0.safetensors",
230
+ "decoder.up_blocks.2.resnets.1.conv2.bias": "0.safetensors",
231
+ "decoder.up_blocks.2.resnets.2.norm1.bias": "0.safetensors",
232
+ "decoder.up_blocks.2.resnets.2.norm1.weight": "0.safetensors",
233
+ "decoder.up_blocks.2.resnets.2.norm2.bias": "0.safetensors",
234
+ "decoder.up_blocks.2.resnets.2.norm2.weight": "0.safetensors",
235
+ "decoder.up_blocks.2.resnets.2.conv1.weight": "0.safetensors",
236
+ "decoder.up_blocks.2.resnets.2.conv1.bias": "0.safetensors",
237
+ "decoder.up_blocks.2.resnets.2.conv2.weight": "0.safetensors",
238
+ "decoder.up_blocks.2.resnets.2.conv2.bias": "0.safetensors",
239
+ "decoder.up_blocks.2.upsamplers.0.conv.weight": "0.safetensors",
240
+ "decoder.up_blocks.2.upsamplers.0.conv.bias": "0.safetensors",
241
+ "decoder.up_blocks.2.upsamplers.0.upscale_conv.weight": "0.safetensors",
242
+ "decoder.up_blocks.2.upsamplers.0.upscale_conv.bias": "0.safetensors",
243
+ "decoder.up_blocks.3.resnets.0.norm1.bias": "0.safetensors",
244
+ "decoder.up_blocks.3.resnets.0.norm1.weight": "0.safetensors",
245
+ "decoder.up_blocks.3.resnets.0.norm2.bias": "0.safetensors",
246
+ "decoder.up_blocks.3.resnets.0.norm2.weight": "0.safetensors",
247
+ "decoder.up_blocks.3.resnets.0.conv1.weight": "0.safetensors",
248
+ "decoder.up_blocks.3.resnets.0.conv1.bias": "0.safetensors",
249
+ "decoder.up_blocks.3.resnets.0.conv2.weight": "0.safetensors",
250
+ "decoder.up_blocks.3.resnets.0.conv2.bias": "0.safetensors",
251
+ "decoder.up_blocks.3.resnets.0.conv_shortcut.weight": "0.safetensors",
252
+ "decoder.up_blocks.3.resnets.0.conv_shortcut.bias": "0.safetensors",
253
+ "decoder.up_blocks.3.resnets.1.norm1.bias": "0.safetensors",
254
+ "decoder.up_blocks.3.resnets.1.norm1.weight": "0.safetensors",
255
+ "decoder.up_blocks.3.resnets.1.norm2.bias": "0.safetensors",
256
+ "decoder.up_blocks.3.resnets.1.norm2.weight": "0.safetensors",
257
+ "decoder.up_blocks.3.resnets.1.conv1.weight": "0.safetensors",
258
+ "decoder.up_blocks.3.resnets.1.conv1.bias": "0.safetensors",
259
+ "decoder.up_blocks.3.resnets.1.conv2.weight": "0.safetensors",
260
+ "decoder.up_blocks.3.resnets.1.conv2.bias": "0.safetensors",
261
+ "decoder.up_blocks.3.resnets.2.norm1.bias": "0.safetensors",
262
+ "decoder.up_blocks.3.resnets.2.norm1.weight": "0.safetensors",
263
+ "decoder.up_blocks.3.resnets.2.norm2.bias": "0.safetensors",
264
+ "decoder.up_blocks.3.resnets.2.norm2.weight": "0.safetensors",
265
+ "decoder.up_blocks.3.resnets.2.conv1.weight": "0.safetensors",
266
+ "decoder.up_blocks.3.resnets.2.conv1.bias": "0.safetensors",
267
+ "decoder.up_blocks.3.resnets.2.conv2.weight": "0.safetensors",
268
+ "decoder.up_blocks.3.resnets.2.conv2.bias": "0.safetensors",
269
+ "decoder.conv_norm_out.bias": "0.safetensors",
270
+ "decoder.conv_norm_out.weight": "0.safetensors",
271
+ "decoder.conv_out.weight": "0.safetensors",
272
+ "decoder.conv_out.bias": "0.safetensors"
273
+ }
274
+ }