Hugging Face's logo

woozziam
/
251204_diffusion

Safetensors
Model card Files
xet
Community
woozziam commited on
Commit
ecdc898
·
verified ·
1 Parent(s): 3866f97

Upload folder using huggingface_hub

Browse files
Files changed (3) hide show
  1. config.json +89 -0
  2. model.safetensors +3 -0
  3. train_config.json +200 -0
config.json ADDED
@@ -0,0 +1,89 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "diffusion",
3
+ "n_obs_steps": 2,
4
+ "normalization_mapping": {
5
+ "VISUAL": "MEAN_STD",
6
+ "STATE": "MIN_MAX",
7
+ "ACTION": "MIN_MAX"
8
+ },
9
+ "input_features": {
10
+ "observation.state": {
11
+ "type": "STATE",
12
+ "shape": [
13
+ 6
14
+ ]
15
+ },
16
+ "observation.images.top": {
17
+ "type": "VISUAL",
18
+ "shape": [
19
+ 3,
20
+ 480,
21
+ 640
22
+ ]
23
+ },
24
+ "observation.images.top2": {
25
+ "type": "VISUAL",
26
+ "shape": [
27
+ 3,
28
+ 480,
29
+ 640
30
+ ]
31
+ }
32
+ },
33
+ "output_features": {
34
+ "action": {
35
+ "type": "ACTION",
36
+ "shape": [
37
+ 6
38
+ ]
39
+ }
40
+ },
41
+ "device": "cuda",
42
+ "use_amp": false,
43
+ "push_to_hub": false,
44
+ "repo_id": "outputs/train/251203_picknplace_diffusion",
45
+ "private": null,
46
+ "tags": null,
47
+ "license": null,
48
+ "horizon": 16,
49
+ "n_action_steps": 8,
50
+ "drop_n_last_frames": 7,
51
+ "vision_backbone": "resnet18",
52
+ "crop_shape": [
53
+ 84,
54
+ 84
55
+ ],
56
+ "crop_is_random": true,
57
+ "pretrained_backbone_weights": null,
58
+ "use_group_norm": true,
59
+ "spatial_softmax_num_keypoints": 32,
60
+ "use_separate_rgb_encoder_per_camera": false,
61
+ "down_dims": [
62
+ 512,
63
+ 1024,
64
+ 2048
65
+ ],
66
+ "kernel_size": 5,
67
+ "n_groups": 8,
68
+ "diffusion_step_embed_dim": 128,
69
+ "use_film_scale_modulation": true,
70
+ "noise_scheduler_type": "DDPM",
71
+ "num_train_timesteps": 50,
72
+ "beta_schedule": "squaredcos_cap_v2",
73
+ "beta_start": 0.0001,
74
+ "beta_end": 0.02,
75
+ "prediction_type": "epsilon",
76
+ "clip_sample": true,
77
+ "clip_sample_range": 1.0,
78
+ "num_inference_steps": 20,
79
+ "do_mask_loss_for_padding": false,
80
+ "optimizer_lr": 0.0001,
81
+ "optimizer_betas": [
82
+ 0.95,
83
+ 0.999
84
+ ],
85
+ "optimizer_eps": 1e-08,
86
+ "optimizer_weight_decay": 1e-06,
87
+ "scheduler_name": "cosine",
88
+ "scheduler_warmup_steps": 500
89
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:583f8c6d1fe1ab00489958c0377e1f0601f04a4efc3bd5c0ba990b75823fba09
3
+ size 1066517728
train_config.json ADDED
@@ -0,0 +1,200 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "./251203_picknplace_expand",
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": true,
65
+ "video_backend": "pyav",
66
+ "streaming": false
67
+ },
68
+ "env": null,
69
+ "policy": {
70
+ "type": "diffusion",
71
+ "n_obs_steps": 2,
72
+ "normalization_mapping": {
73
+ "VISUAL": "MEAN_STD",
74
+ "STATE": "MIN_MAX",
75
+ "ACTION": "MIN_MAX"
76
+ },
77
+ "input_features": {
78
+ "observation.state": {
79
+ "type": "STATE",
80
+ "shape": [
81
+ 6
82
+ ]
83
+ },
84
+ "observation.images.top": {
85
+ "type": "VISUAL",
86
+ "shape": [
87
+ 3,
88
+ 480,
89
+ 640
90
+ ]
91
+ },
92
+ "observation.images.top2": {
93
+ "type": "VISUAL",
94
+ "shape": [
95
+ 3,
96
+ 480,
97
+ 640
98
+ ]
99
+ }
100
+ },
101
+ "output_features": {
102
+ "action": {
103
+ "type": "ACTION",
104
+ "shape": [
105
+ 6
106
+ ]
107
+ }
108
+ },
109
+ "device": "cuda",
110
+ "use_amp": false,
111
+ "push_to_hub": false,
112
+ "repo_id": "outputs/train/251203_picknplace_diffusion",
113
+ "private": null,
114
+ "tags": null,
115
+ "license": null,
116
+ "horizon": 16,
117
+ "n_action_steps": 8,
118
+ "drop_n_last_frames": 7,
119
+ "vision_backbone": "resnet18",
120
+ "crop_shape": [
121
+ 84,
122
+ 84
123
+ ],
124
+ "crop_is_random": true,
125
+ "pretrained_backbone_weights": null,
126
+ "use_group_norm": true,
127
+ "spatial_softmax_num_keypoints": 32,
128
+ "use_separate_rgb_encoder_per_camera": false,
129
+ "down_dims": [
130
+ 512,
131
+ 1024,
132
+ 2048
133
+ ],
134
+ "kernel_size": 5,
135
+ "n_groups": 8,
136
+ "diffusion_step_embed_dim": 128,
137
+ "use_film_scale_modulation": true,
138
+ "noise_scheduler_type": "DDPM",
139
+ "num_train_timesteps": 50,
140
+ "beta_schedule": "squaredcos_cap_v2",
141
+ "beta_start": 0.0001,
142
+ "beta_end": 0.02,
143
+ "prediction_type": "epsilon",
144
+ "clip_sample": true,
145
+ "clip_sample_range": 1.0,
146
+ "num_inference_steps": 20,
147
+ "do_mask_loss_for_padding": false,
148
+ "optimizer_lr": 0.0001,
149
+ "optimizer_betas": [
150
+ 0.95,
151
+ 0.999
152
+ ],
153
+ "optimizer_eps": 1e-08,
154
+ "optimizer_weight_decay": 1e-06,
155
+ "scheduler_name": "cosine",
156
+ "scheduler_warmup_steps": 500
157
+ },
158
+ "output_dir": "outputs/train/251203_picknplace_diffusion",
159
+ "job_name": "251203_picknplace_diffusion",
160
+ "resume": false,
161
+ "seed": 1000,
162
+ "num_workers": 4,
163
+ "batch_size": 8,
164
+ "steps": 1000000,
165
+ "eval_freq": 50000,
166
+ "log_freq": 500,
167
+ "save_checkpoint": true,
168
+ "save_freq": 50000,
169
+ "use_policy_training_preset": true,
170
+ "optimizer": {
171
+ "type": "adam",
172
+ "lr": 0.0001,
173
+ "weight_decay": 1e-06,
174
+ "grad_clip_norm": 10.0,
175
+ "betas": [
176
+ 0.95,
177
+ 0.999
178
+ ],
179
+ "eps": 1e-08
180
+ },
181
+ "scheduler": {
182
+ "type": "diffuser",
183
+ "num_warmup_steps": 500,
184
+ "name": "cosine"
185
+ },
186
+ "eval": {
187
+ "n_episodes": 50,
188
+ "batch_size": 50,
189
+ "use_async_envs": false
190
+ },
191
+ "wandb": {
192
+ "enable": false,
193
+ "disable_artifact": false,
194
+ "project": "lerobot",
195
+ "entity": null,
196
+ "notes": null,
197
+ "run_id": null,
198
+ "mode": null
199
+ }
200
+ }