Hugging Face's logo

DanqingZ
/
diffusion-pusht-model

Safetensors
Model card Files
xet
Community
DanqingZ commited on
Commit
353ae68
·
verified ·
1 Parent(s): b831bcb

Upload folder using huggingface_hub

Browse files
Files changed (3) hide show
  1. config.json +76 -0
  2. model.safetensors +3 -0
  3. train_config.json +223 -0
config.json ADDED
@@ -0,0 +1,76 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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.image": {
11
+ "type": "VISUAL",
12
+ "shape": [
13
+ 3,
14
+ 96,
15
+ 96
16
+ ]
17
+ },
18
+ "observation.state": {
19
+ "type": "STATE",
20
+ "shape": [
21
+ 2
22
+ ]
23
+ }
24
+ },
25
+ "output_features": {
26
+ "action": {
27
+ "type": "ACTION",
28
+ "shape": [
29
+ 2
30
+ ]
31
+ }
32
+ },
33
+ "device": "cuda",
34
+ "use_amp": false,
35
+ "horizon": 16,
36
+ "n_action_steps": 8,
37
+ "drop_n_last_frames": 7,
38
+ "vision_backbone": "resnet18",
39
+ "crop_shape": [
40
+ 84,
41
+ 84
42
+ ],
43
+ "crop_is_random": true,
44
+ "pretrained_backbone_weights": null,
45
+ "use_group_norm": true,
46
+ "spatial_softmax_num_keypoints": 32,
47
+ "use_separate_rgb_encoder_per_camera": false,
48
+ "down_dims": [
49
+ 512,
50
+ 1024,
51
+ 2048
52
+ ],
53
+ "kernel_size": 5,
54
+ "n_groups": 8,
55
+ "diffusion_step_embed_dim": 128,
56
+ "use_film_scale_modulation": true,
57
+ "noise_scheduler_type": "DDPM",
58
+ "num_train_timesteps": 100,
59
+ "beta_schedule": "squaredcos_cap_v2",
60
+ "beta_start": 0.0001,
61
+ "beta_end": 0.02,
62
+ "prediction_type": "epsilon",
63
+ "clip_sample": true,
64
+ "clip_sample_range": 1.0,
65
+ "num_inference_steps": null,
66
+ "do_mask_loss_for_padding": false,
67
+ "optimizer_lr": 0.0001,
68
+ "optimizer_betas": [
69
+ 0.95,
70
+ 0.999
71
+ ],
72
+ "optimizer_eps": 1e-08,
73
+ "optimizer_weight_decay": 1e-06,
74
+ "scheduler_name": "cosine",
75
+ "scheduler_warmup_steps": 500
76
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:abb83b01c5b398a550904121839bc714f6832f874b6b7beb745f1e3a38a4253b
3
+ size 1050862408
train_config.json ADDED
@@ -0,0 +1,223 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "lerobot/pusht",
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": "torchcodec"
66
+ },
67
+ "env": {
68
+ "type": "pusht",
69
+ "task": "PushT-v0",
70
+ "fps": 10,
71
+ "features": {
72
+ "action": {
73
+ "type": "ACTION",
74
+ "shape": [
75
+ 2
76
+ ]
77
+ },
78
+ "agent_pos": {
79
+ "type": "STATE",
80
+ "shape": [
81
+ 2
82
+ ]
83
+ },
84
+ "pixels": {
85
+ "type": "VISUAL",
86
+ "shape": [
87
+ 384,
88
+ 384,
89
+ 3
90
+ ]
91
+ }
92
+ },
93
+ "features_map": {
94
+ "action": "action",
95
+ "agent_pos": "observation.state",
96
+ "environment_state": "observation.environment_state",
97
+ "pixels": "observation.image"
98
+ },
99
+ "episode_length": 300,
100
+ "obs_type": "pixels_agent_pos",
101
+ "render_mode": "rgb_array",
102
+ "visualization_width": 384,
103
+ "visualization_height": 384
104
+ },
105
+ "policy": {
106
+ "type": "diffusion",
107
+ "n_obs_steps": 2,
108
+ "normalization_mapping": {
109
+ "VISUAL": "MEAN_STD",
110
+ "STATE": "MIN_MAX",
111
+ "ACTION": "MIN_MAX"
112
+ },
113
+ "input_features": {
114
+ "observation.image": {
115
+ "type": "VISUAL",
116
+ "shape": [
117
+ 3,
118
+ 96,
119
+ 96
120
+ ]
121
+ },
122
+ "observation.state": {
123
+ "type": "STATE",
124
+ "shape": [
125
+ 2
126
+ ]
127
+ }
128
+ },
129
+ "output_features": {
130
+ "action": {
131
+ "type": "ACTION",
132
+ "shape": [
133
+ 2
134
+ ]
135
+ }
136
+ },
137
+ "device": "cuda",
138
+ "use_amp": false,
139
+ "horizon": 16,
140
+ "n_action_steps": 8,
141
+ "drop_n_last_frames": 7,
142
+ "vision_backbone": "resnet18",
143
+ "crop_shape": [
144
+ 84,
145
+ 84
146
+ ],
147
+ "crop_is_random": true,
148
+ "pretrained_backbone_weights": null,
149
+ "use_group_norm": true,
150
+ "spatial_softmax_num_keypoints": 32,
151
+ "use_separate_rgb_encoder_per_camera": false,
152
+ "down_dims": [
153
+ 512,
154
+ 1024,
155
+ 2048
156
+ ],
157
+ "kernel_size": 5,
158
+ "n_groups": 8,
159
+ "diffusion_step_embed_dim": 128,
160
+ "use_film_scale_modulation": true,
161
+ "noise_scheduler_type": "DDPM",
162
+ "num_train_timesteps": 100,
163
+ "beta_schedule": "squaredcos_cap_v2",
164
+ "beta_start": 0.0001,
165
+ "beta_end": 0.02,
166
+ "prediction_type": "epsilon",
167
+ "clip_sample": true,
168
+ "clip_sample_range": 1.0,
169
+ "num_inference_steps": null,
170
+ "do_mask_loss_for_padding": false,
171
+ "optimizer_lr": 0.0001,
172
+ "optimizer_betas": [
173
+ 0.95,
174
+ 0.999
175
+ ],
176
+ "optimizer_eps": 1e-08,
177
+ "optimizer_weight_decay": 1e-06,
178
+ "scheduler_name": "cosine",
179
+ "scheduler_warmup_steps": 500
180
+ },
181
+ "output_dir": "outputs/train/diffusion_pusht",
182
+ "job_name": "pusht_diffusion",
183
+ "resume": false,
184
+ "seed": 100000,
185
+ "num_workers": 4,
186
+ "batch_size": 64,
187
+ "steps": 200000,
188
+ "eval_freq": 25000,
189
+ "log_freq": 200,
190
+ "save_checkpoint": true,
191
+ "save_freq": 25000,
192
+ "use_policy_training_preset": true,
193
+ "optimizer": {
194
+ "type": "adam",
195
+ "lr": 0.0001,
196
+ "weight_decay": 1e-06,
197
+ "grad_clip_norm": 10.0,
198
+ "betas": [
199
+ 0.95,
200
+ 0.999
201
+ ],
202
+ "eps": 1e-08
203
+ },
204
+ "scheduler": {
205
+ "type": "diffuser",
206
+ "num_warmup_steps": 500,
207
+ "name": "cosine"
208
+ },
209
+ "eval": {
210
+ "n_episodes": 50,
211
+ "batch_size": 50,
212
+ "use_async_envs": false
213
+ },
214
+ "wandb": {
215
+ "enable": true,
216
+ "disable_artifact": false,
217
+ "project": "lerobot",
218
+ "entity": null,
219
+ "notes": null,
220
+ "run_id": null,
221
+ "mode": null
222
+ }
223
+ }