Hugging Face's logo

yzembodied
/
libero10_task1_diffusion

Safetensors
Model card Files
xet
Community
Yi Zhang commited on
Commit
7ce6bc5
·
1 Parent(s): 3db9c93

Initial commit of pretrained policy

Browse files
Files changed (3) hide show
  1. config.json +87 -0
  2. model.safetensors +3 -0
  3. train_config.json +246 -0
config.json ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "diffusion",
3
+ "n_obs_steps": 5,
4
+ "normalization_mapping": {
5
+ "VISUAL": "MEAN_STD",
6
+ "STATE": "MIN_MAX",
7
+ "ACTION": "MIN_MAX"
8
+ },
9
+ "input_features": {
10
+ "observation.images.image": {
11
+ "type": "VISUAL",
12
+ "shape": [
13
+ 3,
14
+ 256,
15
+ 256
16
+ ]
17
+ },
18
+ "observation.images.wrist_image": {
19
+ "type": "VISUAL",
20
+ "shape": [
21
+ 3,
22
+ 256,
23
+ 256
24
+ ]
25
+ },
26
+ "observation.state": {
27
+ "type": "STATE",
28
+ "shape": [
29
+ 8
30
+ ]
31
+ },
32
+ "observation.state.joint": {
33
+ "type": "STATE",
34
+ "shape": [
35
+ 7
36
+ ]
37
+ }
38
+ },
39
+ "output_features": {
40
+ "action": {
41
+ "type": "ACTION",
42
+ "shape": [
43
+ 7
44
+ ]
45
+ }
46
+ },
47
+ "device": "cuda",
48
+ "use_amp": false,
49
+ "horizon": 16,
50
+ "n_action_steps": 8,
51
+ "drop_n_last_frames": 7,
52
+ "vision_backbone": "resnet18",
53
+ "crop_shape": null,
54
+ "crop_is_random": false,
55
+ "pretrained_backbone_weights": null,
56
+ "use_group_norm": true,
57
+ "spatial_softmax_num_keypoints": 32,
58
+ "use_separate_rgb_encoder_per_camera": true,
59
+ "down_dims": [
60
+ 512,
61
+ 1024,
62
+ 2048
63
+ ],
64
+ "kernel_size": 5,
65
+ "n_groups": 8,
66
+ "diffusion_step_embed_dim": 128,
67
+ "use_film_scale_modulation": true,
68
+ "noise_scheduler_type": "DDPM",
69
+ "num_train_timesteps": 100,
70
+ "beta_schedule": "squaredcos_cap_v2",
71
+ "beta_start": 0.0001,
72
+ "beta_end": 0.02,
73
+ "prediction_type": "epsilon",
74
+ "clip_sample": true,
75
+ "clip_sample_range": 1.0,
76
+ "num_inference_steps": null,
77
+ "do_mask_loss_for_padding": false,
78
+ "optimizer_lr": 0.0001,
79
+ "optimizer_betas": [
80
+ 0.95,
81
+ 0.999
82
+ ],
83
+ "optimizer_eps": 1e-08,
84
+ "optimizer_weight_decay": 1e-06,
85
+ "scheduler_name": "cosine",
86
+ "scheduler_warmup_steps": 500
87
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fc10534f97c04fb5198641275fe2e36f6daa7797fe344dfd7dd70b89006acec2
3
+ size 1158580996
train_config.json ADDED
@@ -0,0 +1,246 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "yzembodied/libero_10_image_task_1",
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
+ },
67
+ "env": {
68
+ "type": "libero",
69
+ "task": "Libero_10_Task_1",
70
+ "fps": 20,
71
+ "features": {
72
+ "action": {
73
+ "type": "ACTION",
74
+ "shape": [
75
+ 7
76
+ ]
77
+ },
78
+ "agent_pos": {
79
+ "type": "STATE",
80
+ "shape": [
81
+ 8
82
+ ]
83
+ },
84
+ "joint_state": {
85
+ "type": "STATE",
86
+ "shape": [
87
+ 7
88
+ ]
89
+ },
90
+ "pixels/image": {
91
+ "type": "VISUAL",
92
+ "shape": [
93
+ 256,
94
+ 256,
95
+ 3
96
+ ]
97
+ },
98
+ "pixels/wrist_image": {
99
+ "type": "VISUAL",
100
+ "shape": [
101
+ 256,
102
+ 256,
103
+ 3
104
+ ]
105
+ }
106
+ },
107
+ "features_map": {
108
+ "action": "action",
109
+ "agent_pos": "observation.state",
110
+ "joint_state": "observation.state.joint",
111
+ "pixels/image": "observation.images.image",
112
+ "pixels/wrist_image": "observation.images.wrist_image"
113
+ },
114
+ "benchmark": "libero_10",
115
+ "task_id": "task_0",
116
+ "episode_length": 500
117
+ },
118
+ "policy": {
119
+ "type": "diffusion",
120
+ "n_obs_steps": 5,
121
+ "normalization_mapping": {
122
+ "VISUAL": "MEAN_STD",
123
+ "STATE": "MIN_MAX",
124
+ "ACTION": "MIN_MAX"
125
+ },
126
+ "input_features": {
127
+ "observation.images.image": {
128
+ "type": "VISUAL",
129
+ "shape": [
130
+ 3,
131
+ 256,
132
+ 256
133
+ ]
134
+ },
135
+ "observation.images.wrist_image": {
136
+ "type": "VISUAL",
137
+ "shape": [
138
+ 3,
139
+ 256,
140
+ 256
141
+ ]
142
+ },
143
+ "observation.state": {
144
+ "type": "STATE",
145
+ "shape": [
146
+ 8
147
+ ]
148
+ },
149
+ "observation.state.joint": {
150
+ "type": "STATE",
151
+ "shape": [
152
+ 7
153
+ ]
154
+ }
155
+ },
156
+ "output_features": {
157
+ "action": {
158
+ "type": "ACTION",
159
+ "shape": [
160
+ 7
161
+ ]
162
+ }
163
+ },
164
+ "device": "cuda",
165
+ "use_amp": false,
166
+ "horizon": 16,
167
+ "n_action_steps": 8,
168
+ "drop_n_last_frames": 7,
169
+ "vision_backbone": "resnet18",
170
+ "crop_shape": null,
171
+ "crop_is_random": false,
172
+ "pretrained_backbone_weights": null,
173
+ "use_group_norm": true,
174
+ "spatial_softmax_num_keypoints": 32,
175
+ "use_separate_rgb_encoder_per_camera": true,
176
+ "down_dims": [
177
+ 512,
178
+ 1024,
179
+ 2048
180
+ ],
181
+ "kernel_size": 5,
182
+ "n_groups": 8,
183
+ "diffusion_step_embed_dim": 128,
184
+ "use_film_scale_modulation": true,
185
+ "noise_scheduler_type": "DDPM",
186
+ "num_train_timesteps": 100,
187
+ "beta_schedule": "squaredcos_cap_v2",
188
+ "beta_start": 0.0001,
189
+ "beta_end": 0.02,
190
+ "prediction_type": "epsilon",
191
+ "clip_sample": true,
192
+ "clip_sample_range": 1.0,
193
+ "num_inference_steps": null,
194
+ "do_mask_loss_for_padding": false,
195
+ "optimizer_lr": 0.0001,
196
+ "optimizer_betas": [
197
+ 0.95,
198
+ 0.999
199
+ ],
200
+ "optimizer_eps": 1e-08,
201
+ "optimizer_weight_decay": 1e-06,
202
+ "scheduler_name": "cosine",
203
+ "scheduler_warmup_steps": 500
204
+ },
205
+ "output_dir": "outputs/train/2025-03-25/14-25-48_libero10_task1_diffusion",
206
+ "job_name": "libero10_task1_diffusion",
207
+ "resume": false,
208
+ "seed": 1000,
209
+ "num_workers": 16,
210
+ "batch_size": 32,
211
+ "steps": 80000,
212
+ "eval_freq": 20000,
213
+ "log_freq": 500,
214
+ "save_checkpoint": true,
215
+ "save_freq": 20000,
216
+ "use_policy_training_preset": true,
217
+ "optimizer": {
218
+ "type": "adam",
219
+ "lr": 0.0001,
220
+ "weight_decay": 1e-06,
221
+ "grad_clip_norm": 10.0,
222
+ "betas": [
223
+ 0.95,
224
+ 0.999
225
+ ],
226
+ "eps": 1e-08
227
+ },
228
+ "scheduler": {
229
+ "type": "diffuser",
230
+ "num_warmup_steps": 500,
231
+ "name": "cosine"
232
+ },
233
+ "eval": {
234
+ "n_episodes": 50,
235
+ "batch_size": 1,
236
+ "use_async_envs": false,
237
+ "max_episodes_rendered": 4
238
+ },
239
+ "wandb": {
240
+ "enable": true,
241
+ "disable_artifact": true,
242
+ "project": "lifelong_learning",
243
+ "entity": "470620104-technical-university-of-munich",
244
+ "notes": null
245
+ }
246
+ }