Hugging Face's logo

jjang318
/
digit_diffusion_door_opening

Safetensors
Model card Files
xet
Community
jjang318 commited on
Commit
fcbf1c8
·
verified ·
1 Parent(s): 576746c

Upload 3 files

Browse files
Files changed (3) hide show
  1. config.json +108 -0
  2. model.safetensors +3 -0
  3. train_config.json +218 -0
config.json ADDED
@@ -0,0 +1,108 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "diffusion",
3
+ "n_obs_steps": 1,
4
+ "normalization_mapping": {
5
+ "VISUAL": "IDENTITY",
6
+ "STATE": "MIN_MAX",
7
+ "ACTION": "MIN_MAX"
8
+ },
9
+ "input_features": {
10
+ "observation.state": {
11
+ "type": "STATE",
12
+ "shape": [
13
+ 26
14
+ ]
15
+ },
16
+ "observation.images.forward_chest_realsense_d435.rgb": {
17
+ "type": "VISUAL",
18
+ "shape": [
19
+ 3,
20
+ 240,
21
+ 424
22
+ ]
23
+ },
24
+ "observation.images.forward_chest_realsense_d435.depth": {
25
+ "type": "VISUAL",
26
+ "shape": [
27
+ 1,
28
+ 240,
29
+ 424
30
+ ]
31
+ },
32
+ "observation.images.zed.rgb": {
33
+ "type": "VISUAL",
34
+ "shape": [
35
+ 3,
36
+ 240,
37
+ 424
38
+ ]
39
+ },
40
+ "observation.images.zed.depth": {
41
+ "type": "VISUAL",
42
+ "shape": [
43
+ 1,
44
+ 240,
45
+ 424
46
+ ]
47
+ }
48
+ },
49
+ "output_features": {
50
+ "action": {
51
+ "type": "ACTION",
52
+ "shape": [
53
+ 23
54
+ ]
55
+ }
56
+ },
57
+ "device": "cuda",
58
+ "use_amp": false,
59
+ "push_to_hub": true,
60
+ "repo_id": "kelvinzhaozg/digit3a_hardware_door_open",
61
+ "private": null,
62
+ "tags": null,
63
+ "license": null,
64
+ "horizon": 64,
65
+ "n_action_steps": 32,
66
+ "per_feature_normalization_overrides": null,
67
+ "drop_n_last_frames": 7,
68
+ "vision_backbone": "resnet18",
69
+ "crop_shape": null,
70
+ "crop_is_random": true,
71
+ "pretrained_backbone_weights": null,
72
+ "use_group_norm": true,
73
+ "spatial_softmax_num_keypoints": 32,
74
+ "use_separate_rgb_encoder_per_camera": false,
75
+ "use_depth_as_channel": true,
76
+ "use_depth_only": false,
77
+ "rgbd_pretrained_backbone_weights": "r3m",
78
+ "norm_rgb": true,
79
+ "norm_depth": false,
80
+ "down_dims": [
81
+ 512,
82
+ 1024,
83
+ 2048
84
+ ],
85
+ "kernel_size": 5,
86
+ "n_groups": 8,
87
+ "diffusion_step_embed_dim": 128,
88
+ "use_film_scale_modulation": true,
89
+ "noise_scheduler_type": "DDIM",
90
+ "num_train_timesteps": 100,
91
+ "beta_schedule": "squaredcos_cap_v2",
92
+ "beta_start": 0.0001,
93
+ "beta_end": 0.02,
94
+ "prediction_type": "sample",
95
+ "clip_sample": true,
96
+ "clip_sample_range": 1.0,
97
+ "num_inference_steps": null,
98
+ "do_mask_loss_for_padding": false,
99
+ "optimizer_lr": 0.0001,
100
+ "optimizer_betas": [
101
+ 0.95,
102
+ 0.999
103
+ ],
104
+ "optimizer_eps": 1e-08,
105
+ "optimizer_weight_decay": 1e-06,
106
+ "scheduler_name": "cosine",
107
+ "scheduler_warmup_steps": 500
108
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e496454c8962c90a9d38cd9826b68b3ed02b88a665335b32b308ed4db9b045ce
3
+ size 1053700868
train_config.json ADDED
@@ -0,0 +1,218 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "kelvinzhaozg/digit3a_hardware_door_open_dataset",
4
+ "root": null,
5
+ "episodes": null,
6
+ "image_transforms": {
7
+ "enable": false,
8
+ "max_num_transforms": 3,
9
+ "random_order": false,
10
+ "tfs": {
11
+ "brightness": {
12
+ "weight": 1.0,
13
+ "type": "ColorJitter",
14
+ "kwargs": {
15
+ "brightness": [
16
+ 0.8,
17
+ 1.2
18
+ ]
19
+ }
20
+ },
21
+ "contrast": {
22
+ "weight": 1.0,
23
+ "type": "ColorJitter",
24
+ "kwargs": {
25
+ "contrast": [
26
+ 0.8,
27
+ 1.2
28
+ ]
29
+ }
30
+ },
31
+ "saturation": {
32
+ "weight": 1.0,
33
+ "type": "ColorJitter",
34
+ "kwargs": {
35
+ "saturation": [
36
+ 0.5,
37
+ 1.5
38
+ ]
39
+ }
40
+ },
41
+ "hue": {
42
+ "weight": 1.0,
43
+ "type": "ColorJitter",
44
+ "kwargs": {
45
+ "hue": [
46
+ -0.05,
47
+ 0.05
48
+ ]
49
+ }
50
+ },
51
+ "sharpness": {
52
+ "weight": 1.0,
53
+ "type": "SharpnessJitter",
54
+ "kwargs": {
55
+ "sharpness": [
56
+ 0.5,
57
+ 1.5
58
+ ]
59
+ }
60
+ }
61
+ }
62
+ },
63
+ "revision": null,
64
+ "use_imagenet_stats": false,
65
+ "video_backend": "torchcodec"
66
+ },
67
+ "env": null,
68
+ "policy": {
69
+ "type": "diffusion",
70
+ "n_obs_steps": 1,
71
+ "normalization_mapping": {
72
+ "VISUAL": "IDENTITY",
73
+ "STATE": "MIN_MAX",
74
+ "ACTION": "MIN_MAX"
75
+ },
76
+ "input_features": {
77
+ "observation.state": {
78
+ "type": "STATE",
79
+ "shape": [
80
+ 26
81
+ ]
82
+ },
83
+ "observation.images.forward_chest_realsense_d435.rgb": {
84
+ "type": "VISUAL",
85
+ "shape": [
86
+ 3,
87
+ 240,
88
+ 424
89
+ ]
90
+ },
91
+ "observation.images.forward_chest_realsense_d435.depth": {
92
+ "type": "VISUAL",
93
+ "shape": [
94
+ 1,
95
+ 240,
96
+ 424
97
+ ]
98
+ },
99
+ "observation.images.zed.rgb": {
100
+ "type": "VISUAL",
101
+ "shape": [
102
+ 3,
103
+ 240,
104
+ 424
105
+ ]
106
+ },
107
+ "observation.images.zed.depth": {
108
+ "type": "VISUAL",
109
+ "shape": [
110
+ 1,
111
+ 240,
112
+ 424
113
+ ]
114
+ }
115
+ },
116
+ "output_features": {
117
+ "action": {
118
+ "type": "ACTION",
119
+ "shape": [
120
+ 23
121
+ ]
122
+ }
123
+ },
124
+ "device": "cuda",
125
+ "use_amp": false,
126
+ "push_to_hub": true,
127
+ "repo_id": "kelvinzhaozg/digit3a_hardware_door_open",
128
+ "private": null,
129
+ "tags": null,
130
+ "license": null,
131
+ "horizon": 64,
132
+ "n_action_steps": 32,
133
+ "per_feature_normalization_overrides": null,
134
+ "drop_n_last_frames": 7,
135
+ "vision_backbone": "resnet18",
136
+ "crop_shape": null,
137
+ "crop_is_random": true,
138
+ "pretrained_backbone_weights": null,
139
+ "use_group_norm": true,
140
+ "spatial_softmax_num_keypoints": 32,
141
+ "use_separate_rgb_encoder_per_camera": false,
142
+ "use_depth_as_channel": true,
143
+ "use_depth_only": false,
144
+ "rgbd_pretrained_backbone_weights": "r3m",
145
+ "norm_rgb": true,
146
+ "norm_depth": false,
147
+ "down_dims": [
148
+ 512,
149
+ 1024,
150
+ 2048
151
+ ],
152
+ "kernel_size": 5,
153
+ "n_groups": 8,
154
+ "diffusion_step_embed_dim": 128,
155
+ "use_film_scale_modulation": true,
156
+ "noise_scheduler_type": "DDIM",
157
+ "num_train_timesteps": 100,
158
+ "beta_schedule": "squaredcos_cap_v2",
159
+ "beta_start": 0.0001,
160
+ "beta_end": 0.02,
161
+ "prediction_type": "sample",
162
+ "clip_sample": true,
163
+ "clip_sample_range": 1.0,
164
+ "num_inference_steps": null,
165
+ "do_mask_loss_for_padding": false,
166
+ "optimizer_lr": 0.0001,
167
+ "optimizer_betas": [
168
+ 0.95,
169
+ 0.999
170
+ ],
171
+ "optimizer_eps": 1e-08,
172
+ "optimizer_weight_decay": 1e-06,
173
+ "scheduler_name": "cosine",
174
+ "scheduler_warmup_steps": 500
175
+ },
176
+ "output_dir": "outputs/train/2025-11-03/12-35-58_diffusion",
177
+ "job_name": "diffusion",
178
+ "resume": false,
179
+ "seed": 1000,
180
+ "num_workers": 4,
181
+ "batch_size": 8,
182
+ "steps": 200000,
183
+ "eval_freq": 20000,
184
+ "log_freq": 200,
185
+ "save_checkpoint": true,
186
+ "save_freq": 20000,
187
+ "use_policy_training_preset": true,
188
+ "optimizer": {
189
+ "type": "adam",
190
+ "lr": 0.0001,
191
+ "weight_decay": 1e-06,
192
+ "grad_clip_norm": 10.0,
193
+ "betas": [
194
+ 0.95,
195
+ 0.999
196
+ ],
197
+ "eps": 1e-08
198
+ },
199
+ "scheduler": {
200
+ "type": "diffuser",
201
+ "num_warmup_steps": 500,
202
+ "name": "cosine"
203
+ },
204
+ "eval": {
205
+ "n_episodes": 50,
206
+ "batch_size": 50,
207
+ "use_async_envs": false
208
+ },
209
+ "wandb": {
210
+ "enable": true,
211
+ "disable_artifact": true,
212
+ "project": "digit3a_door_open_rgbd",
213
+ "entity": null,
214
+ "notes": null,
215
+ "run_id": "ziobsrpl",
216
+ "mode": null
217
+ }
218
+ }