Hugging Face's logo

WetLabRoboData
/
diffusion-bimanual_pick-scratch

Robotics
LeRobot
Safetensors
diffusion-policy
imitation-learning
ur3e
bimanual
Model card Files
xet
 Community
RamaE commited on
Commit
3ec8f7d
·
verified ·
1 Parent(s): c0aa16d

Phase B.3 populate: from old/train/1_success_bimanual_pick_20260624_ds5/deploy/checkpoints/1-suuces-040000/pretrained_model

Browse files
Files changed (7) hide show
  1. config.json +103 -0
  2. model.safetensors +3 -0
  3. policy_postprocessor.json +32 -0
  4. policy_postprocessor_step_0_unnormalizer_processor.safetensors +3 -0
  5. policy_preprocessor.json +72 -0
  6. policy_preprocessor_step_3_normalizer_processor.safetensors +3 -0
  7. train_config.json +239 -0
config.json ADDED
@@ -0,0 +1,103 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "diffusion",
3
+ "n_obs_steps": 1,
4
+ "input_features": {
5
+ "observation.images.left_wrist_top": {
6
+ "type": "VISUAL",
7
+ "shape": [
8
+ 3,
9
+ 144,
10
+ 256
11
+ ]
12
+ },
13
+ "observation.images.right_wrist_top": {
14
+ "type": "VISUAL",
15
+ "shape": [
16
+ 3,
17
+ 144,
18
+ 256
19
+ ]
20
+ },
21
+ "observation.images.front": {
22
+ "type": "VISUAL",
23
+ "shape": [
24
+ 3,
25
+ 144,
26
+ 256
27
+ ]
28
+ },
29
+ "observation.state": {
30
+ "type": "STATE",
31
+ "shape": [
32
+ 14
33
+ ]
34
+ }
35
+ },
36
+ "output_features": {
37
+ "action": {
38
+ "type": "ACTION",
39
+ "shape": [
40
+ 14
41
+ ]
42
+ }
43
+ },
44
+ "device": "cuda",
45
+ "use_amp": false,
46
+ "use_peft": false,
47
+ "push_to_hub": false,
48
+ "repo_id": null,
49
+ "private": null,
50
+ "tags": null,
51
+ "license": null,
52
+ "pretrained_path": null,
53
+ "horizon": 64,
54
+ "n_action_steps": 16,
55
+ "normalization_mapping": {
56
+ "VISUAL": "MEAN_STD",
57
+ "STATE": "MIN_MAX",
58
+ "ACTION": "MIN_MAX"
59
+ },
60
+ "drop_n_last_frames": 7,
61
+ "vision_backbone": "resnet18",
62
+ "resize_shape": null,
63
+ "crop_ratio": 1.0,
64
+ "crop_shape": [
65
+ 134,
66
+ 246
67
+ ],
68
+ "crop_is_random": true,
69
+ "pretrained_backbone_weights": null,
70
+ "use_group_norm": true,
71
+ "spatial_softmax_num_keypoints": 32,
72
+ "use_separate_rgb_encoder_per_camera": false,
73
+ "down_dims": [
74
+ 512,
75
+ 1024,
76
+ 2048
77
+ ],
78
+ "kernel_size": 5,
79
+ "n_groups": 8,
80
+ "diffusion_step_embed_dim": 128,
81
+ "use_film_scale_modulation": true,
82
+ "noise_scheduler_type": "DDPM",
83
+ "num_train_timesteps": 100,
84
+ "beta_schedule": "squaredcos_cap_v2",
85
+ "beta_start": 0.0001,
86
+ "beta_end": 0.02,
87
+ "prediction_type": "epsilon",
88
+ "clip_sample": true,
89
+ "clip_sample_range": 1.0,
90
+ "num_inference_steps": 10,
91
+ "compile_model": false,
92
+ "compile_mode": "reduce-overhead",
93
+ "do_mask_loss_for_padding": false,
94
+ "optimizer_lr": 0.0001,
95
+ "optimizer_betas": [
96
+ 0.95,
97
+ 0.999
98
+ ],
99
+ "optimizer_eps": 1e-08,
100
+ "optimizer_weight_decay": 1e-06,
101
+ "scheduler_name": "cosine",
102
+ "scheduler_warmup_steps": 500
103
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c36092ad457f884ebd1f32b9f32f5c3afa09f609b69d03cf95eb318f83d0a708
3
+ size 1059520696
policy_postprocessor.json ADDED
@@ -0,0 +1,32 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "name": "policy_postprocessor",
3
+ "steps": [
4
+ {
5
+ "registry_name": "unnormalizer_processor",
6
+ "config": {
7
+ "eps": 1e-08,
8
+ "features": {
9
+ "action": {
10
+ "type": "ACTION",
11
+ "shape": [
12
+ 14
13
+ ]
14
+ }
15
+ },
16
+ "norm_map": {
17
+ "VISUAL": "MEAN_STD",
18
+ "STATE": "MIN_MAX",
19
+ "ACTION": "MIN_MAX"
20
+ }
21
+ },
22
+ "state_file": "policy_postprocessor_step_0_unnormalizer_processor.safetensors"
23
+ },
24
+ {
25
+ "registry_name": "device_processor",
26
+ "config": {
27
+ "device": "cpu",
28
+ "float_dtype": null
29
+ }
30
+ }
31
+ ]
32
+ }
policy_postprocessor_step_0_unnormalizer_processor.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:97a7ced2965a02cbf57e7f1e301491e69b68e40d2793497989cce1a7f1ff7bf7
3
+ size 9392
policy_preprocessor.json ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "name": "policy_preprocessor",
3
+ "steps": [
4
+ {
5
+ "registry_name": "rename_observations_processor",
6
+ "config": {
7
+ "rename_map": {}
8
+ }
9
+ },
10
+ {
11
+ "registry_name": "to_batch_processor",
12
+ "config": {}
13
+ },
14
+ {
15
+ "registry_name": "device_processor",
16
+ "config": {
17
+ "device": "cuda",
18
+ "float_dtype": null
19
+ }
20
+ },
21
+ {
22
+ "registry_name": "normalizer_processor",
23
+ "config": {
24
+ "eps": 1e-08,
25
+ "features": {
26
+ "observation.images.left_wrist_top": {
27
+ "type": "VISUAL",
28
+ "shape": [
29
+ 3,
30
+ 144,
31
+ 256
32
+ ]
33
+ },
34
+ "observation.images.right_wrist_top": {
35
+ "type": "VISUAL",
36
+ "shape": [
37
+ 3,
38
+ 144,
39
+ 256
40
+ ]
41
+ },
42
+ "observation.images.front": {
43
+ "type": "VISUAL",
44
+ "shape": [
45
+ 3,
46
+ 144,
47
+ 256
48
+ ]
49
+ },
50
+ "observation.state": {
51
+ "type": "STATE",
52
+ "shape": [
53
+ 14
54
+ ]
55
+ },
56
+ "action": {
57
+ "type": "ACTION",
58
+ "shape": [
59
+ 14
60
+ ]
61
+ }
62
+ },
63
+ "norm_map": {
64
+ "VISUAL": "MEAN_STD",
65
+ "STATE": "MIN_MAX",
66
+ "ACTION": "MIN_MAX"
67
+ }
68
+ },
69
+ "state_file": "policy_preprocessor_step_3_normalizer_processor.safetensors"
70
+ }
71
+ ]
72
+ }
policy_preprocessor_step_3_normalizer_processor.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:97a7ced2965a02cbf57e7f1e301491e69b68e40d2793497989cce1a7f1ff7bf7
3
+ size 9392
train_config.json ADDED
@@ -0,0 +1,239 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "WetLabRoboData/lerobot-data-bimanual_pick",
4
+ "episodes": null,
5
+ "image_transforms": {
6
+ "enable": true,
7
+ "max_num_transforms": 3,
8
+ "random_order": true,
9
+ "tfs": {
10
+ "brightness": {
11
+ "weight": 1.0,
12
+ "type": "ColorJitter",
13
+ "kwargs": {
14
+ "brightness": [
15
+ 0.8,
16
+ 1.2
17
+ ]
18
+ }
19
+ },
20
+ "contrast": {
21
+ "weight": 1.0,
22
+ "type": "ColorJitter",
23
+ "kwargs": {
24
+ "contrast": [
25
+ 0.8,
26
+ 1.2
27
+ ]
28
+ }
29
+ },
30
+ "saturation": {
31
+ "weight": 1.0,
32
+ "type": "ColorJitter",
33
+ "kwargs": {
34
+ "saturation": [
35
+ 0.5,
36
+ 1.5
37
+ ]
38
+ }
39
+ },
40
+ "hue": {
41
+ "weight": 1.0,
42
+ "type": "ColorJitter",
43
+ "kwargs": {
44
+ "hue": [
45
+ -0.05,
46
+ 0.05
47
+ ]
48
+ }
49
+ },
50
+ "sharpness": {
51
+ "weight": 1.0,
52
+ "type": "SharpnessJitter",
53
+ "kwargs": {
54
+ "sharpness": [
55
+ 0.5,
56
+ 1.5
57
+ ]
58
+ }
59
+ },
60
+ "affine": {
61
+ "weight": 1.0,
62
+ "type": "RandomAffine",
63
+ "kwargs": {
64
+ "degrees": [
65
+ -5.0,
66
+ 5.0
67
+ ],
68
+ "translate": [
69
+ 0.05,
70
+ 0.05
71
+ ]
72
+ }
73
+ }
74
+ },
75
+ "per_camera_crops": {}
76
+ },
77
+ "revision": null,
78
+ "use_imagenet_stats": true,
79
+ "video_backend": "torchcodec",
80
+ "streaming": false
81
+ },
82
+ "env": null,
83
+ "policy": {
84
+ "type": "diffusion",
85
+ "n_obs_steps": 1,
86
+ "input_features": {
87
+ "observation.images.left_wrist_top": {
88
+ "type": "VISUAL",
89
+ "shape": [
90
+ 3,
91
+ 144,
92
+ 256
93
+ ]
94
+ },
95
+ "observation.images.right_wrist_top": {
96
+ "type": "VISUAL",
97
+ "shape": [
98
+ 3,
99
+ 144,
100
+ 256
101
+ ]
102
+ },
103
+ "observation.images.front": {
104
+ "type": "VISUAL",
105
+ "shape": [
106
+ 3,
107
+ 144,
108
+ 256
109
+ ]
110
+ },
111
+ "observation.state": {
112
+ "type": "STATE",
113
+ "shape": [
114
+ 14
115
+ ]
116
+ }
117
+ },
118
+ "output_features": {
119
+ "action": {
120
+ "type": "ACTION",
121
+ "shape": [
122
+ 14
123
+ ]
124
+ }
125
+ },
126
+ "device": "cuda",
127
+ "use_amp": false,
128
+ "use_peft": false,
129
+ "push_to_hub": false,
130
+ "repo_id": null,
131
+ "private": null,
132
+ "tags": null,
133
+ "license": null,
134
+ "pretrained_path": null,
135
+ "horizon": 64,
136
+ "n_action_steps": 8,
137
+ "normalization_mapping": {
138
+ "VISUAL": "MEAN_STD",
139
+ "STATE": "MIN_MAX",
140
+ "ACTION": "MIN_MAX"
141
+ },
142
+ "drop_n_last_frames": 7,
143
+ "vision_backbone": "resnet18",
144
+ "resize_shape": null,
145
+ "crop_ratio": 1.0,
146
+ "crop_shape": [
147
+ 134,
148
+ 246
149
+ ],
150
+ "crop_is_random": true,
151
+ "pretrained_backbone_weights": null,
152
+ "use_group_norm": true,
153
+ "spatial_softmax_num_keypoints": 32,
154
+ "use_separate_rgb_encoder_per_camera": false,
155
+ "down_dims": [
156
+ 512,
157
+ 1024,
158
+ 2048
159
+ ],
160
+ "kernel_size": 5,
161
+ "n_groups": 8,
162
+ "diffusion_step_embed_dim": 128,
163
+ "use_film_scale_modulation": true,
164
+ "noise_scheduler_type": "DDPM",
165
+ "num_train_timesteps": 100,
166
+ "beta_schedule": "squaredcos_cap_v2",
167
+ "beta_start": 0.0001,
168
+ "beta_end": 0.02,
169
+ "prediction_type": "epsilon",
170
+ "clip_sample": true,
171
+ "clip_sample_range": 1.0,
172
+ "num_inference_steps": 10,
173
+ "compile_model": false,
174
+ "compile_mode": "reduce-overhead",
175
+ "do_mask_loss_for_padding": false,
176
+ "optimizer_lr": 0.0001,
177
+ "optimizer_betas": [
178
+ 0.95,
179
+ 0.999
180
+ ],
181
+ "optimizer_eps": 1e-08,
182
+ "optimizer_weight_decay": 1e-06,
183
+ "scheduler_name": "cosine",
184
+ "scheduler_warmup_steps": 500
185
+ },
186
+ "output_dir": "/lerobot-ros/data/smrithi/bimanual_pick_20260624/train/bimanual_pick_20260624_ds5/deploy",
187
+ "job_name": "dp64_3cam_smrithi_bimanual_pick_20260624_train_bimanual_pick_20260624_ds5",
188
+ "resume": false,
189
+ "seed": 42,
190
+ "cudnn_deterministic": false,
191
+ "num_workers": 4,
192
+ "batch_size": 128,
193
+ "steps": 40000,
194
+ "eval_freq": 2000,
195
+ "log_freq": 200,
196
+ "tolerance_s": 0.0001,
197
+ "save_checkpoint": true,
198
+ "save_freq": 5000,
199
+ "use_policy_training_preset": true,
200
+ "optimizer": {
201
+ "type": "adam",
202
+ "lr": 0.0001,
203
+ "weight_decay": 1e-06,
204
+ "grad_clip_norm": 10.0,
205
+ "betas": [
206
+ 0.95,
207
+ 0.999
208
+ ],
209
+ "eps": 1e-08
210
+ },
211
+ "scheduler": {
212
+ "type": "diffuser",
213
+ "num_warmup_steps": 500,
214
+ "name": "cosine"
215
+ },
216
+ "eval": {
217
+ "n_episodes": 50,
218
+ "batch_size": 50,
219
+ "use_async_envs": false
220
+ },
221
+ "wandb": {
222
+ "enable": true,
223
+ "disable_artifact": true,
224
+ "project": "smrithi_bimanual_pick_20260624_train_bimanual_pick_20260624_ds5",
225
+ "entity": null,
226
+ "notes": null,
227
+ "run_id": "ap31usu9",
228
+ "mode": null,
229
+ "add_tags": true
230
+ },
231
+ "peft": null,
232
+ "use_rabc": false,
233
+ "rabc_progress_path": null,
234
+ "rabc_kappa": 0.01,
235
+ "rabc_epsilon": 1e-06,
236
+ "rabc_head_mode": "sparse",
237
+ "rename_map": {},
238
+ "checkpoint_path": null
239
+ }