Hugging Face's logo

ywu67
/
smolvla-task0000

Safetensors
Model card Files
xet
 Community
ywu67 commited on
Commit
46f9f1b
·
verified ·
1 Parent(s): c995da7

Add checkpoint at step 020000

Browse files
Files changed (7) hide show
  1. checkpoints/step_020000/config.json +157 -0
  2. checkpoints/step_020000/model.safetensors +3 -0
  3. checkpoints/step_020000/policy_postprocessor.json +32 -0
  4. checkpoints/step_020000/policy_postprocessor_step_0_unnormalizer_processor.safetensors +3 -0
  5. checkpoints/step_020000/policy_preprocessor.json +147 -0
  6. checkpoints/step_020000/policy_preprocessor_step_5_normalizer_processor.safetensors +3 -0
  7. checkpoints/step_020000/train_config.json +286 -0
checkpoints/step_020000/config.json ADDED
@@ -0,0 +1,157 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "smolvla",
3
+ "n_obs_steps": 1,
4
+ "input_features": {
5
+ "observation.images.rgb.left_wrist": {
6
+ "type": "VISUAL",
7
+ "shape": [
8
+ 480,
9
+ 480,
10
+ 3
11
+ ]
12
+ },
13
+ "observation.images.rgb.right_wrist": {
14
+ "type": "VISUAL",
15
+ "shape": [
16
+ 480,
17
+ 480,
18
+ 3
19
+ ]
20
+ },
21
+ "observation.images.rgb.head": {
22
+ "type": "VISUAL",
23
+ "shape": [
24
+ 720,
25
+ 720,
26
+ 3
27
+ ]
28
+ },
29
+ "observation.images.depth.left_wrist": {
30
+ "type": "VISUAL",
31
+ "shape": [
32
+ 480,
33
+ 480,
34
+ 3
35
+ ]
36
+ },
37
+ "observation.images.depth.right_wrist": {
38
+ "type": "VISUAL",
39
+ "shape": [
40
+ 480,
41
+ 480,
42
+ 3
43
+ ]
44
+ },
45
+ "observation.images.depth.head": {
46
+ "type": "VISUAL",
47
+ "shape": [
48
+ 720,
49
+ 720,
50
+ 3
51
+ ]
52
+ },
53
+ "observation.images.seg_instance_id.left_wrist": {
54
+ "type": "VISUAL",
55
+ "shape": [
56
+ 480,
57
+ 480,
58
+ 3
59
+ ]
60
+ },
61
+ "observation.images.seg_instance_id.right_wrist": {
62
+ "type": "VISUAL",
63
+ "shape": [
64
+ 480,
65
+ 480,
66
+ 3
67
+ ]
68
+ },
69
+ "observation.images.seg_instance_id.head": {
70
+ "type": "VISUAL",
71
+ "shape": [
72
+ 720,
73
+ 720,
74
+ 3
75
+ ]
76
+ },
77
+ "observation.cam_rel_poses": {
78
+ "type": "STATE",
79
+ "shape": [
80
+ 21
81
+ ]
82
+ },
83
+ "observation.state": {
84
+ "type": "STATE",
85
+ "shape": [
86
+ 256
87
+ ]
88
+ },
89
+ "observation.task_info": {
90
+ "type": "STATE",
91
+ "shape": [
92
+ 46
93
+ ]
94
+ }
95
+ },
96
+ "output_features": {
97
+ "action": {
98
+ "type": "ACTION",
99
+ "shape": [
100
+ 23
101
+ ]
102
+ }
103
+ },
104
+ "device": "cuda",
105
+ "use_amp": false,
106
+ "push_to_hub": true,
107
+ "repo_id": "ywu67/smolvla-task0000",
108
+ "private": null,
109
+ "tags": null,
110
+ "license": null,
111
+ "pretrained_path": null,
112
+ "chunk_size": 50,
113
+ "n_action_steps": 50,
114
+ "normalization_mapping": {
115
+ "VISUAL": "IDENTITY",
116
+ "STATE": "MEAN_STD",
117
+ "ACTION": "MEAN_STD"
118
+ },
119
+ "max_state_dim": 256,
120
+ "max_action_dim": 32,
121
+ "resize_imgs_with_padding": [
122
+ 512,
123
+ 512
124
+ ],
125
+ "empty_cameras": 0,
126
+ "adapt_to_pi_aloha": false,
127
+ "use_delta_joint_actions_aloha": false,
128
+ "tokenizer_max_length": 48,
129
+ "num_steps": 10,
130
+ "use_cache": true,
131
+ "freeze_vision_encoder": true,
132
+ "train_expert_only": true,
133
+ "train_state_proj": true,
134
+ "optimizer_lr": 0.0001,
135
+ "optimizer_betas": [
136
+ 0.9,
137
+ 0.95
138
+ ],
139
+ "optimizer_eps": 1e-08,
140
+ "optimizer_weight_decay": 1e-10,
141
+ "optimizer_grad_clip_norm": 10,
142
+ "scheduler_warmup_steps": 1000,
143
+ "scheduler_decay_steps": 30000,
144
+ "scheduler_decay_lr": 2.5e-06,
145
+ "vlm_model_name": "HuggingFaceTB/SmolVLM2-500M-Video-Instruct",
146
+ "load_vlm_weights": false,
147
+ "add_image_special_tokens": false,
148
+ "attention_mode": "cross_attn",
149
+ "prefix_length": -1,
150
+ "pad_language_to": "longest",
151
+ "num_expert_layers": -1,
152
+ "num_vlm_layers": 16,
153
+ "self_attn_every_n_layers": 2,
154
+ "expert_width_multiplier": 0.75,
155
+ "min_period": 0.004,
156
+ "max_period": 4.0
157
+ }
checkpoints/step_020000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d7e3cf1992c82ce513b5e6e3216dd073da5904eb5d8364901922714c148070d7
3
+ size 1198649392
checkpoints/step_020000/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
+ 23
13
+ ]
14
+ }
15
+ },
16
+ "norm_map": {
17
+ "VISUAL": "IDENTITY",
18
+ "STATE": "MEAN_STD",
19
+ "ACTION": "MEAN_STD"
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
+ }
checkpoints/step_020000/policy_postprocessor_step_0_unnormalizer_processor.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:089d5ed6f26ba87471187b591e4b793af3c976b96dabb5912b5ee36591a2770f
3
+ size 10924
checkpoints/step_020000/policy_preprocessor.json ADDED
@@ -0,0 +1,147 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": "smolvla_new_line_processor",
16
+ "config": {}
17
+ },
18
+ {
19
+ "registry_name": "tokenizer_processor",
20
+ "config": {
21
+ "max_length": 48,
22
+ "task_key": "task",
23
+ "padding_side": "right",
24
+ "padding": "longest",
25
+ "truncation": true,
26
+ "tokenizer_name": "HuggingFaceTB/SmolVLM2-500M-Video-Instruct"
27
+ }
28
+ },
29
+ {
30
+ "registry_name": "device_processor",
31
+ "config": {
32
+ "device": "cuda",
33
+ "float_dtype": null
34
+ }
35
+ },
36
+ {
37
+ "registry_name": "normalizer_processor",
38
+ "config": {
39
+ "eps": 1e-08,
40
+ "features": {
41
+ "observation.images.rgb.left_wrist": {
42
+ "type": "VISUAL",
43
+ "shape": [
44
+ 480,
45
+ 480,
46
+ 3
47
+ ]
48
+ },
49
+ "observation.images.rgb.right_wrist": {
50
+ "type": "VISUAL",
51
+ "shape": [
52
+ 480,
53
+ 480,
54
+ 3
55
+ ]
56
+ },
57
+ "observation.images.rgb.head": {
58
+ "type": "VISUAL",
59
+ "shape": [
60
+ 720,
61
+ 720,
62
+ 3
63
+ ]
64
+ },
65
+ "observation.images.depth.left_wrist": {
66
+ "type": "VISUAL",
67
+ "shape": [
68
+ 480,
69
+ 480,
70
+ 3
71
+ ]
72
+ },
73
+ "observation.images.depth.right_wrist": {
74
+ "type": "VISUAL",
75
+ "shape": [
76
+ 480,
77
+ 480,
78
+ 3
79
+ ]
80
+ },
81
+ "observation.images.depth.head": {
82
+ "type": "VISUAL",
83
+ "shape": [
84
+ 720,
85
+ 720,
86
+ 3
87
+ ]
88
+ },
89
+ "observation.images.seg_instance_id.left_wrist": {
90
+ "type": "VISUAL",
91
+ "shape": [
92
+ 480,
93
+ 480,
94
+ 3
95
+ ]
96
+ },
97
+ "observation.images.seg_instance_id.right_wrist": {
98
+ "type": "VISUAL",
99
+ "shape": [
100
+ 480,
101
+ 480,
102
+ 3
103
+ ]
104
+ },
105
+ "observation.images.seg_instance_id.head": {
106
+ "type": "VISUAL",
107
+ "shape": [
108
+ 720,
109
+ 720,
110
+ 3
111
+ ]
112
+ },
113
+ "observation.cam_rel_poses": {
114
+ "type": "STATE",
115
+ "shape": [
116
+ 21
117
+ ]
118
+ },
119
+ "observation.state": {
120
+ "type": "STATE",
121
+ "shape": [
122
+ 256
123
+ ]
124
+ },
125
+ "observation.task_info": {
126
+ "type": "STATE",
127
+ "shape": [
128
+ 46
129
+ ]
130
+ },
131
+ "action": {
132
+ "type": "ACTION",
133
+ "shape": [
134
+ 23
135
+ ]
136
+ }
137
+ },
138
+ "norm_map": {
139
+ "VISUAL": "IDENTITY",
140
+ "STATE": "MEAN_STD",
141
+ "ACTION": "MEAN_STD"
142
+ }
143
+ },
144
+ "state_file": "policy_preprocessor_step_5_normalizer_processor.safetensors"
145
+ }
146
+ ]
147
+ }
checkpoints/step_020000/policy_preprocessor_step_5_normalizer_processor.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:089d5ed6f26ba87471187b591e4b793af3c976b96dabb5912b5ee36591a2770f
3
+ size 10924
checkpoints/step_020000/train_config.json ADDED
@@ -0,0 +1,286 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "ywu67/2025-challenge-demos-task0000",
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
+ "affine": {
62
+ "weight": 1.0,
63
+ "type": "RandomAffine",
64
+ "kwargs": {
65
+ "degrees": [
66
+ -5.0,
67
+ 5.0
68
+ ],
69
+ "translate": [
70
+ 0.05,
71
+ 0.05
72
+ ]
73
+ }
74
+ }
75
+ }
76
+ },
77
+ "revision": null,
78
+ "use_imagenet_stats": true,
79
+ "video_backend": "torchcodec",
80
+ "streaming": false
81
+ },
82
+ "env": null,
83
+ "policy": {
84
+ "type": "smolvla",
85
+ "n_obs_steps": 1,
86
+ "input_features": {
87
+ "observation.images.rgb.left_wrist": {
88
+ "type": "VISUAL",
89
+ "shape": [
90
+ 480,
91
+ 480,
92
+ 3
93
+ ]
94
+ },
95
+ "observation.images.rgb.right_wrist": {
96
+ "type": "VISUAL",
97
+ "shape": [
98
+ 480,
99
+ 480,
100
+ 3
101
+ ]
102
+ },
103
+ "observation.images.rgb.head": {
104
+ "type": "VISUAL",
105
+ "shape": [
106
+ 720,
107
+ 720,
108
+ 3
109
+ ]
110
+ },
111
+ "observation.images.depth.left_wrist": {
112
+ "type": "VISUAL",
113
+ "shape": [
114
+ 480,
115
+ 480,
116
+ 3
117
+ ]
118
+ },
119
+ "observation.images.depth.right_wrist": {
120
+ "type": "VISUAL",
121
+ "shape": [
122
+ 480,
123
+ 480,
124
+ 3
125
+ ]
126
+ },
127
+ "observation.images.depth.head": {
128
+ "type": "VISUAL",
129
+ "shape": [
130
+ 720,
131
+ 720,
132
+ 3
133
+ ]
134
+ },
135
+ "observation.images.seg_instance_id.left_wrist": {
136
+ "type": "VISUAL",
137
+ "shape": [
138
+ 480,
139
+ 480,
140
+ 3
141
+ ]
142
+ },
143
+ "observation.images.seg_instance_id.right_wrist": {
144
+ "type": "VISUAL",
145
+ "shape": [
146
+ 480,
147
+ 480,
148
+ 3
149
+ ]
150
+ },
151
+ "observation.images.seg_instance_id.head": {
152
+ "type": "VISUAL",
153
+ "shape": [
154
+ 720,
155
+ 720,
156
+ 3
157
+ ]
158
+ },
159
+ "observation.cam_rel_poses": {
160
+ "type": "STATE",
161
+ "shape": [
162
+ 21
163
+ ]
164
+ },
165
+ "observation.state": {
166
+ "type": "STATE",
167
+ "shape": [
168
+ 256
169
+ ]
170
+ },
171
+ "observation.task_info": {
172
+ "type": "STATE",
173
+ "shape": [
174
+ 46
175
+ ]
176
+ }
177
+ },
178
+ "output_features": {
179
+ "action": {
180
+ "type": "ACTION",
181
+ "shape": [
182
+ 23
183
+ ]
184
+ }
185
+ },
186
+ "device": "cuda",
187
+ "use_amp": false,
188
+ "push_to_hub": true,
189
+ "repo_id": "ywu67/smolvla-task0000",
190
+ "private": null,
191
+ "tags": null,
192
+ "license": null,
193
+ "pretrained_path": null,
194
+ "chunk_size": 50,
195
+ "n_action_steps": 50,
196
+ "normalization_mapping": {
197
+ "VISUAL": "IDENTITY",
198
+ "STATE": "MEAN_STD",
199
+ "ACTION": "MEAN_STD"
200
+ },
201
+ "max_state_dim": 256,
202
+ "max_action_dim": 32,
203
+ "resize_imgs_with_padding": [
204
+ 512,
205
+ 512
206
+ ],
207
+ "empty_cameras": 0,
208
+ "adapt_to_pi_aloha": false,
209
+ "use_delta_joint_actions_aloha": false,
210
+ "tokenizer_max_length": 48,
211
+ "num_steps": 10,
212
+ "use_cache": true,
213
+ "freeze_vision_encoder": true,
214
+ "train_expert_only": true,
215
+ "train_state_proj": true,
216
+ "optimizer_lr": 0.0001,
217
+ "optimizer_betas": [
218
+ 0.9,
219
+ 0.95
220
+ ],
221
+ "optimizer_eps": 1e-08,
222
+ "optimizer_weight_decay": 1e-10,
223
+ "optimizer_grad_clip_norm": 10,
224
+ "scheduler_warmup_steps": 1000,
225
+ "scheduler_decay_steps": 30000,
226
+ "scheduler_decay_lr": 2.5e-06,
227
+ "vlm_model_name": "HuggingFaceTB/SmolVLM2-500M-Video-Instruct",
228
+ "load_vlm_weights": false,
229
+ "add_image_special_tokens": false,
230
+ "attention_mode": "cross_attn",
231
+ "prefix_length": -1,
232
+ "pad_language_to": "longest",
233
+ "num_expert_layers": -1,
234
+ "num_vlm_layers": 16,
235
+ "self_attn_every_n_layers": 2,
236
+ "expert_width_multiplier": 0.75,
237
+ "min_period": 0.004,
238
+ "max_period": 4.0
239
+ },
240
+ "output_dir": "/data/output/smolvla_task0000",
241
+ "job_name": "smolvla_task0000_100k",
242
+ "resume": false,
243
+ "seed": 1000,
244
+ "num_workers": 16,
245
+ "batch_size": 16,
246
+ "steps": 100000,
247
+ "eval_freq": 20000,
248
+ "log_freq": 200,
249
+ "save_checkpoint": true,
250
+ "save_freq": 5000,
251
+ "use_policy_training_preset": true,
252
+ "optimizer": {
253
+ "type": "adamw",
254
+ "lr": 0.0001,
255
+ "weight_decay": 1e-10,
256
+ "grad_clip_norm": 10,
257
+ "betas": [
258
+ 0.9,
259
+ 0.95
260
+ ],
261
+ "eps": 1e-08
262
+ },
263
+ "scheduler": {
264
+ "type": "cosine_decay_with_warmup",
265
+ "num_warmup_steps": 1000,
266
+ "num_decay_steps": 30000,
267
+ "peak_lr": 0.0001,
268
+ "decay_lr": 2.5e-06
269
+ },
270
+ "eval": {
271
+ "n_episodes": 50,
272
+ "batch_size": 50,
273
+ "use_async_envs": false
274
+ },
275
+ "wandb": {
276
+ "enable": true,
277
+ "disable_artifact": true,
278
+ "project": "lerobot",
279
+ "entity": null,
280
+ "notes": null,
281
+ "run_id": "a66omu5w",
282
+ "mode": null
283
+ },
284
+ "checkpoint_path": null,
285
+ "rename_map": {}
286
+ }