Hugging Face's logo

RobotLearningVLA
/
test_eval2_v2

Robotics
LeRobot
Safetensors
smolvla
Model card Files
xet
 Community
yukk1 commited on
Commit
471789e
·
verified ·
1 Parent(s): 8894851

Checkpoint step 1000/10000

Browse files
Files changed (7) hide show
  1. config.json +117 -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 +105 -0
  6. policy_preprocessor_step_5_normalizer_processor.safetensors +3 -0
  7. train_config.json +308 -0
config.json ADDED
@@ -0,0 +1,117 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "smolvla",
3
+ "n_obs_steps": 1,
4
+ "input_features": {
5
+ "observation.state": {
6
+ "type": "STATE",
7
+ "shape": [
8
+ 6
9
+ ]
10
+ },
11
+ "observation.images.camera1": {
12
+ "type": "VISUAL",
13
+ "shape": [
14
+ 3,
15
+ 256,
16
+ 256
17
+ ]
18
+ },
19
+ "observation.images.camera2": {
20
+ "type": "VISUAL",
21
+ "shape": [
22
+ 3,
23
+ 256,
24
+ 256
25
+ ]
26
+ },
27
+ "observation.images.camera3": {
28
+ "type": "VISUAL",
29
+ "shape": [
30
+ 3,
31
+ 256,
32
+ 256
33
+ ]
34
+ },
35
+ "observation.images.empty_camera_0": {
36
+ "type": "VISUAL",
37
+ "shape": [
38
+ 3,
39
+ 480,
40
+ 640
41
+ ]
42
+ },
43
+ "observation.images.empty_camera_1": {
44
+ "type": "VISUAL",
45
+ "shape": [
46
+ 3,
47
+ 480,
48
+ 640
49
+ ]
50
+ }
51
+ },
52
+ "output_features": {
53
+ "action": {
54
+ "type": "ACTION",
55
+ "shape": [
56
+ 6
57
+ ]
58
+ }
59
+ },
60
+ "device": "cuda",
61
+ "use_amp": true,
62
+ "use_peft": false,
63
+ "push_to_hub": true,
64
+ "repo_id": "RobotLearningVLA/test_eval2_v2",
65
+ "private": null,
66
+ "tags": null,
67
+ "license": null,
68
+ "pretrained_path": "lerobot/smolvla_base",
69
+ "chunk_size": 50,
70
+ "n_action_steps": 50,
71
+ "normalization_mapping": {
72
+ "VISUAL": "IDENTITY",
73
+ "STATE": "MEAN_STD",
74
+ "ACTION": "MEAN_STD"
75
+ },
76
+ "max_state_dim": 32,
77
+ "max_action_dim": 32,
78
+ "resize_imgs_with_padding": [
79
+ 512,
80
+ 512
81
+ ],
82
+ "empty_cameras": 2,
83
+ "adapt_to_pi_aloha": false,
84
+ "use_delta_joint_actions_aloha": false,
85
+ "tokenizer_max_length": 48,
86
+ "num_steps": 10,
87
+ "use_cache": true,
88
+ "freeze_vision_encoder": true,
89
+ "train_expert_only": true,
90
+ "train_state_proj": true,
91
+ "optimizer_lr": 0.0008,
92
+ "optimizer_betas": [
93
+ 0.9,
94
+ 0.95
95
+ ],
96
+ "optimizer_eps": 1e-08,
97
+ "optimizer_weight_decay": 1e-10,
98
+ "optimizer_grad_clip_norm": 10.0,
99
+ "scheduler_warmup_steps": 1000,
100
+ "scheduler_decay_steps": 9000,
101
+ "scheduler_decay_lr": 2.5e-06,
102
+ "vlm_model_name": "HuggingFaceTB/SmolVLM2-500M-Video-Instruct",
103
+ "load_vlm_weights": true,
104
+ "add_image_special_tokens": false,
105
+ "attention_mode": "cross_attn",
106
+ "prefix_length": 0,
107
+ "pad_language_to": "max_length",
108
+ "num_expert_layers": 0,
109
+ "num_vlm_layers": 16,
110
+ "self_attn_every_n_layers": 2,
111
+ "expert_width_multiplier": 0.75,
112
+ "min_period": 0.004,
113
+ "max_period": 4.0,
114
+ "rtc_config": null,
115
+ "compile_model": true,
116
+ "compile_mode": "max-autotune"
117
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b82e7089795da524c888630cdc815ee12374966791723f1ca134584862c5968e
3
+ size 906712520
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
+ 6
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
+ }
policy_postprocessor_step_0_unnormalizer_processor.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:58f412e40cc89f20f8b03a055196c3deeaf977642c1394fe2ab5bdca1407c85a
3
+ size 6560
policy_preprocessor.json ADDED
@@ -0,0 +1,105 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "name": "policy_preprocessor",
3
+ "steps": [
4
+ {
5
+ "registry_name": "rename_observations_processor",
6
+ "config": {
7
+ "rename_map": {
8
+ "observation.images.front": "observation.images.camera1"
9
+ }
10
+ }
11
+ },
12
+ {
13
+ "registry_name": "to_batch_processor",
14
+ "config": {}
15
+ },
16
+ {
17
+ "registry_name": "smolvla_new_line_processor",
18
+ "config": {}
19
+ },
20
+ {
21
+ "registry_name": "tokenizer_processor",
22
+ "config": {
23
+ "max_length": 48,
24
+ "task_key": "task",
25
+ "padding_side": "right",
26
+ "padding": "max_length",
27
+ "truncation": true,
28
+ "tokenizer_name": "HuggingFaceTB/SmolVLM2-500M-Video-Instruct"
29
+ }
30
+ },
31
+ {
32
+ "registry_name": "device_processor",
33
+ "config": {
34
+ "device": "cuda",
35
+ "float_dtype": null
36
+ }
37
+ },
38
+ {
39
+ "registry_name": "normalizer_processor",
40
+ "config": {
41
+ "eps": 1e-08,
42
+ "features": {
43
+ "observation.state": {
44
+ "type": "STATE",
45
+ "shape": [
46
+ 6
47
+ ]
48
+ },
49
+ "observation.images.camera1": {
50
+ "type": "VISUAL",
51
+ "shape": [
52
+ 3,
53
+ 256,
54
+ 256
55
+ ]
56
+ },
57
+ "observation.images.camera2": {
58
+ "type": "VISUAL",
59
+ "shape": [
60
+ 3,
61
+ 256,
62
+ 256
63
+ ]
64
+ },
65
+ "observation.images.camera3": {
66
+ "type": "VISUAL",
67
+ "shape": [
68
+ 3,
69
+ 256,
70
+ 256
71
+ ]
72
+ },
73
+ "observation.images.empty_camera_0": {
74
+ "type": "VISUAL",
75
+ "shape": [
76
+ 3,
77
+ 480,
78
+ 640
79
+ ]
80
+ },
81
+ "observation.images.empty_camera_1": {
82
+ "type": "VISUAL",
83
+ "shape": [
84
+ 3,
85
+ 480,
86
+ 640
87
+ ]
88
+ },
89
+ "action": {
90
+ "type": "ACTION",
91
+ "shape": [
92
+ 6
93
+ ]
94
+ }
95
+ },
96
+ "norm_map": {
97
+ "VISUAL": "IDENTITY",
98
+ "STATE": "MEAN_STD",
99
+ "ACTION": "MEAN_STD"
100
+ }
101
+ },
102
+ "state_file": "policy_preprocessor_step_5_normalizer_processor.safetensors"
103
+ }
104
+ ]
105
+ }
policy_preprocessor_step_5_normalizer_processor.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:58f412e40cc89f20f8b03a055196c3deeaf977642c1394fe2ab5bdca1407c85a
3
+ size 6560
train_config.json ADDED
@@ -0,0 +1,308 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "RobotLearningVLA/banana_blue_bowl_eval2_GreenBlueRed_v2",
4
+ "root": null,
5
+ "episodes": null,
6
+ "image_transforms": {
7
+ "enable": true,
8
+ "max_num_transforms": 4,
9
+ "random_order": false,
10
+ "tfs": {
11
+ "brightness": {
12
+ "weight": 2.0,
13
+ "type": "ColorJitter",
14
+ "kwargs": {
15
+ "brightness": [
16
+ 0.6,
17
+ 1.4
18
+ ]
19
+ }
20
+ },
21
+ "contrast": {
22
+ "weight": 2.0,
23
+ "type": "ColorJitter",
24
+ "kwargs": {
25
+ "contrast": [
26
+ 0.6,
27
+ 1.4
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
+ -0.02,
47
+ 0.02
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
+ -3.0,
67
+ 3.0
68
+ ],
69
+ "translate": [
70
+ 0.03,
71
+ 0.03
72
+ ]
73
+ }
74
+ },
75
+ "perspective": {
76
+ "weight": 1.5,
77
+ "type": "RandomPerspective",
78
+ "kwargs": {
79
+ "distortion_scale": 0.2,
80
+ "p": 0.5
81
+ }
82
+ },
83
+ "resized_crop": {
84
+ "weight": 1.0,
85
+ "type": "RandomResizedCrop",
86
+ "kwargs": {
87
+ "size": [
88
+ 480,
89
+ 640
90
+ ],
91
+ "scale": [
92
+ 0.75,
93
+ 1.0
94
+ ],
95
+ "ratio": [
96
+ 0.95,
97
+ 1.05
98
+ ]
99
+ }
100
+ },
101
+ "gaussian_blur": {
102
+ "weight": 0.5,
103
+ "type": "GaussianBlur",
104
+ "kwargs": {
105
+ "kernel_size": [
106
+ 5,
107
+ 9
108
+ ],
109
+ "sigma": [
110
+ 0.3,
111
+ 2.0
112
+ ]
113
+ }
114
+ },
115
+ "erase": {
116
+ "weight": 0.5,
117
+ "type": "RandomErasing",
118
+ "kwargs": {
119
+ "p": 0.3,
120
+ "scale": [
121
+ 0.02,
122
+ 0.1
123
+ ]
124
+ }
125
+ }
126
+ }
127
+ },
128
+ "revision": null,
129
+ "use_imagenet_stats": true,
130
+ "video_backend": "torchcodec",
131
+ "return_uint8": false,
132
+ "streaming": false
133
+ },
134
+ "env": null,
135
+ "policy": {
136
+ "type": "smolvla",
137
+ "n_obs_steps": 1,
138
+ "input_features": {
139
+ "observation.state": {
140
+ "type": "STATE",
141
+ "shape": [
142
+ 6
143
+ ]
144
+ },
145
+ "observation.images.camera1": {
146
+ "type": "VISUAL",
147
+ "shape": [
148
+ 3,
149
+ 256,
150
+ 256
151
+ ]
152
+ },
153
+ "observation.images.camera2": {
154
+ "type": "VISUAL",
155
+ "shape": [
156
+ 3,
157
+ 256,
158
+ 256
159
+ ]
160
+ },
161
+ "observation.images.camera3": {
162
+ "type": "VISUAL",
163
+ "shape": [
164
+ 3,
165
+ 256,
166
+ 256
167
+ ]
168
+ },
169
+ "observation.images.empty_camera_0": {
170
+ "type": "VISUAL",
171
+ "shape": [
172
+ 3,
173
+ 480,
174
+ 640
175
+ ]
176
+ },
177
+ "observation.images.empty_camera_1": {
178
+ "type": "VISUAL",
179
+ "shape": [
180
+ 3,
181
+ 480,
182
+ 640
183
+ ]
184
+ }
185
+ },
186
+ "output_features": {
187
+ "action": {
188
+ "type": "ACTION",
189
+ "shape": [
190
+ 6
191
+ ]
192
+ }
193
+ },
194
+ "device": "cuda",
195
+ "use_amp": true,
196
+ "use_peft": false,
197
+ "push_to_hub": true,
198
+ "repo_id": "RobotLearningVLA/test_eval2_v2",
199
+ "private": null,
200
+ "tags": null,
201
+ "license": null,
202
+ "pretrained_path": "lerobot/smolvla_base",
203
+ "chunk_size": 50,
204
+ "n_action_steps": 50,
205
+ "normalization_mapping": {
206
+ "VISUAL": "IDENTITY",
207
+ "STATE": "MEAN_STD",
208
+ "ACTION": "MEAN_STD"
209
+ },
210
+ "max_state_dim": 32,
211
+ "max_action_dim": 32,
212
+ "resize_imgs_with_padding": [
213
+ 512,
214
+ 512
215
+ ],
216
+ "empty_cameras": 2,
217
+ "adapt_to_pi_aloha": false,
218
+ "use_delta_joint_actions_aloha": false,
219
+ "tokenizer_max_length": 48,
220
+ "num_steps": 10,
221
+ "use_cache": true,
222
+ "freeze_vision_encoder": true,
223
+ "train_expert_only": true,
224
+ "train_state_proj": true,
225
+ "optimizer_lr": 0.0008,
226
+ "optimizer_betas": [
227
+ 0.9,
228
+ 0.95
229
+ ],
230
+ "optimizer_eps": 1e-08,
231
+ "optimizer_weight_decay": 1e-10,
232
+ "optimizer_grad_clip_norm": 10.0,
233
+ "scheduler_warmup_steps": 1000,
234
+ "scheduler_decay_steps": 9000,
235
+ "scheduler_decay_lr": 2.5e-06,
236
+ "vlm_model_name": "HuggingFaceTB/SmolVLM2-500M-Video-Instruct",
237
+ "load_vlm_weights": true,
238
+ "add_image_special_tokens": false,
239
+ "attention_mode": "cross_attn",
240
+ "prefix_length": 0,
241
+ "pad_language_to": "max_length",
242
+ "num_expert_layers": 0,
243
+ "num_vlm_layers": 16,
244
+ "self_attn_every_n_layers": 2,
245
+ "expert_width_multiplier": 0.75,
246
+ "min_period": 0.004,
247
+ "max_period": 4.0,
248
+ "rtc_config": null,
249
+ "compile_model": true,
250
+ "compile_mode": "max-autotune"
251
+ },
252
+ "reward_model": null,
253
+ "output_dir": "outputs/train/2026-05-20/03-07-49_smolvla",
254
+ "job_name": "smolvla",
255
+ "resume": false,
256
+ "seed": 1000,
257
+ "cudnn_deterministic": false,
258
+ "num_workers": 16,
259
+ "batch_size": 256,
260
+ "prefetch_factor": 4,
261
+ "persistent_workers": true,
262
+ "steps": 10000,
263
+ "eval_freq": 20000,
264
+ "log_freq": 25,
265
+ "tolerance_s": 0.0001,
266
+ "save_checkpoint": true,
267
+ "save_freq": 1000,
268
+ "use_policy_training_preset": true,
269
+ "optimizer": {
270
+ "type": "adamw",
271
+ "lr": 0.0008,
272
+ "weight_decay": 1e-10,
273
+ "grad_clip_norm": 10.0,
274
+ "betas": [
275
+ 0.9,
276
+ 0.95
277
+ ],
278
+ "eps": 1e-08
279
+ },
280
+ "scheduler": {
281
+ "type": "cosine_decay_with_warmup",
282
+ "num_warmup_steps": 1000,
283
+ "num_decay_steps": 9000,
284
+ "peak_lr": 0.0008,
285
+ "decay_lr": 2.5e-06
286
+ },
287
+ "eval": {
288
+ "n_episodes": 50,
289
+ "batch_size": 19,
290
+ "use_async_envs": true
291
+ },
292
+ "wandb": {
293
+ "enable": true,
294
+ "disable_artifact": false,
295
+ "project": "smolvla_banana_eval1",
296
+ "entity": null,
297
+ "notes": null,
298
+ "run_id": "eval1_3bowls_h100_bf16_20260520_030744",
299
+ "mode": null,
300
+ "add_tags": true
301
+ },
302
+ "peft": null,
303
+ "sample_weighting": null,
304
+ "rename_map": {
305
+ "observation.images.front": "observation.images.camera1"
306
+ },
307
+ "checkpoint_path": null
308
+ }