Hugging Face's logo

mlx-community
/
moge-3-vitl-mlx-fp32

Depth Estimation
MLX
Safetensors
moge3
monocular-geometry
moge
Model card Files
xet
 Community
jarjoura commited on
Commit
0a8acc1
·
verified ·
1 Parent(s): 2bdfc94

Upload folder using huggingface_hub

Browse files
Files changed (4) hide show
  1. README.md +0 -0
  2. config.json +194 -0
  3. model.safetensors +3 -0
  4. preprocessor_config.json +3 -0
README.md ADDED
File without changes
config.json ADDED
@@ -0,0 +1,194 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "moge3",
3
+ "encoder": {
4
+ "backbone": "dinov2_vitl14",
5
+ "intermediate_layers": [
6
+ 5,
7
+ 11,
8
+ 17,
9
+ 23
10
+ ],
11
+ "dim_out": 1024
12
+ },
13
+ "neck": {
14
+ "dim_in": [
15
+ 1026,
16
+ 2,
17
+ 2,
18
+ 2,
19
+ 2
20
+ ],
21
+ "dim_out": null,
22
+ "dim_res_blocks": [
23
+ 1024,
24
+ 256,
25
+ 128,
26
+ 64,
27
+ 32
28
+ ],
29
+ "num_res_blocks": [
30
+ 0,
31
+ 2,
32
+ 2,
33
+ 2,
34
+ 0
35
+ ],
36
+ "res_block_in_norm": "none",
37
+ "res_block_hidden_norm": "none",
38
+ "resamplers": [
39
+ "conv_transpose",
40
+ "conv_transpose",
41
+ "conv_transpose",
42
+ "bilinear"
43
+ ]
44
+ },
45
+ "points_head": {
46
+ "dim_in": [
47
+ 1024,
48
+ 256,
49
+ 128,
50
+ 64,
51
+ 32
52
+ ],
53
+ "dim_out": [
54
+ null,
55
+ null,
56
+ null,
57
+ null,
58
+ 3
59
+ ],
60
+ "dim_res_blocks": [
61
+ 1024,
62
+ 256,
63
+ 128,
64
+ 64,
65
+ 32
66
+ ],
67
+ "num_res_blocks": [
68
+ 0,
69
+ 1,
70
+ 1,
71
+ 1,
72
+ 0
73
+ ],
74
+ "res_block_in_norm": "none",
75
+ "res_block_hidden_norm": "none",
76
+ "resamplers": [
77
+ "conv_transpose",
78
+ "conv_transpose",
79
+ "conv_transpose",
80
+ "bilinear"
81
+ ]
82
+ },
83
+ "normal_head": {
84
+ "dim_in": [
85
+ 1024,
86
+ 256,
87
+ 128,
88
+ 64,
89
+ 32
90
+ ],
91
+ "dim_out": [
92
+ null,
93
+ null,
94
+ null,
95
+ null,
96
+ 3
97
+ ],
98
+ "dim_res_blocks": [
99
+ 1024,
100
+ 256,
101
+ 128,
102
+ 64,
103
+ 32
104
+ ],
105
+ "num_res_blocks": [
106
+ 0,
107
+ 1,
108
+ 1,
109
+ 1,
110
+ 0
111
+ ],
112
+ "res_block_in_norm": "none",
113
+ "res_block_hidden_norm": "none",
114
+ "resamplers": [
115
+ "conv_transpose",
116
+ "conv_transpose",
117
+ "conv_transpose",
118
+ "bilinear"
119
+ ]
120
+ },
121
+ "mask_head": {
122
+ "dim_in": [
123
+ 1024,
124
+ 256,
125
+ 128,
126
+ 64,
127
+ 32
128
+ ],
129
+ "dim_out": [
130
+ null,
131
+ null,
132
+ null,
133
+ null,
134
+ 1
135
+ ],
136
+ "dim_res_blocks": [
137
+ 1024,
138
+ 256,
139
+ 128,
140
+ 64,
141
+ 32
142
+ ],
143
+ "num_res_blocks": [
144
+ 0,
145
+ 1,
146
+ 1,
147
+ 1,
148
+ 0
149
+ ],
150
+ "res_block_in_norm": "none",
151
+ "res_block_hidden_norm": "none",
152
+ "resamplers": [
153
+ "conv_transpose",
154
+ "conv_transpose",
155
+ "conv_transpose",
156
+ "bilinear"
157
+ ]
158
+ },
159
+ "scale_head": {
160
+ "dims": [
161
+ 1024,
162
+ 1024,
163
+ 1024,
164
+ 1
165
+ ]
166
+ },
167
+ "num_tokens_range": [
168
+ 1200,
169
+ 3600
170
+ ],
171
+ "refiner": {
172
+ "encoder_channels": 1026,
173
+ "model_channels": [
174
+ 32,
175
+ 64,
176
+ 128,
177
+ 256,
178
+ 512
179
+ ],
180
+ "encoder_blocks_per_level": 1,
181
+ "decoder_blocks_per_level": 1,
182
+ "bottleneck_blocks": 1,
183
+ "downsample_factors": [
184
+ 2,
185
+ 2,
186
+ 2,
187
+ 2
188
+ ],
189
+ "encoder_downsample": 16,
190
+ "in_channels": 3,
191
+ "out_channels": 1
192
+ },
193
+ "refiner_depth_resolution": 256
194
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:46b886fb6fca6dccdc5338a67225bf6727efe480700d705f0c2ac6601ef12185
3
+ size 1481203814
preprocessor_config.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {
2
+ "resize_to": null
3
+ }