Hugging Face's logo

joon-stack
/
smolvla_pick_place_bowl

Robotics
LeRobot
Safetensors
smolvla
Model card Files
xet
 Community
joon-stack commited on
Commit
2fc4ffb
·
verified ·
1 Parent(s): 8fcaa43

Upload folder using huggingface_hub

Browse files
Files changed (7) hide show
  1. checkpoints/040000/config.json +93 -0
  2. checkpoints/040000/model.safetensors +3 -0
  3. checkpoints/040000/policy_postprocessor.json +32 -0
  4. checkpoints/040000/policy_postprocessor_step_0_unnormalizer_processor.safetensors +3 -0
  5. checkpoints/040000/policy_preprocessor.json +79 -0
  6. checkpoints/040000/policy_preprocessor_step_5_normalizer_processor.safetensors +3 -0
  7. checkpoints/040000/train_config.json +232 -0
checkpoints/040000/config.json ADDED
@@ -0,0 +1,93 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "smolvla",
3
+ "n_obs_steps": 1,
4
+ "input_features": {
5
+ "observation.state": {
6
+ "type": "STATE",
7
+ "shape": [
8
+ 7
9
+ ]
10
+ },
11
+ "observation.images.wrist": {
12
+ "type": "VISUAL",
13
+ "shape": [
14
+ 3,
15
+ 480,
16
+ 640
17
+ ]
18
+ },
19
+ "observation.images.top": {
20
+ "type": "VISUAL",
21
+ "shape": [
22
+ 3,
23
+ 480,
24
+ 640
25
+ ]
26
+ }
27
+ },
28
+ "output_features": {
29
+ "action": {
30
+ "type": "ACTION",
31
+ "shape": [
32
+ 7
33
+ ]
34
+ }
35
+ },
36
+ "device": "cuda",
37
+ "use_amp": false,
38
+ "use_peft": false,
39
+ "push_to_hub": true,
40
+ "repo_id": "joon-stack/smolvla_pick_place_bowl_",
41
+ "private": null,
42
+ "tags": null,
43
+ "license": null,
44
+ "pretrained_path": "lerobot/smolvla_base",
45
+ "chunk_size": 50,
46
+ "n_action_steps": 50,
47
+ "normalization_mapping": {
48
+ "VISUAL": "IDENTITY",
49
+ "STATE": "MEAN_STD",
50
+ "ACTION": "MEAN_STD"
51
+ },
52
+ "max_state_dim": 32,
53
+ "max_action_dim": 32,
54
+ "resize_imgs_with_padding": [
55
+ 512,
56
+ 512
57
+ ],
58
+ "empty_cameras": 0,
59
+ "adapt_to_pi_aloha": false,
60
+ "use_delta_joint_actions_aloha": false,
61
+ "tokenizer_max_length": 48,
62
+ "num_steps": 10,
63
+ "use_cache": true,
64
+ "freeze_vision_encoder": true,
65
+ "train_expert_only": true,
66
+ "train_state_proj": true,
67
+ "optimizer_lr": 0.0001,
68
+ "optimizer_betas": [
69
+ 0.9,
70
+ 0.95
71
+ ],
72
+ "optimizer_eps": 1e-08,
73
+ "optimizer_weight_decay": 1e-10,
74
+ "optimizer_grad_clip_norm": 10.0,
75
+ "scheduler_warmup_steps": 1000,
76
+ "scheduler_decay_steps": 200000,
77
+ "scheduler_decay_lr": 2.5e-06,
78
+ "vlm_model_name": "HuggingFaceTB/SmolVLM2-500M-Video-Instruct",
79
+ "load_vlm_weights": true,
80
+ "add_image_special_tokens": false,
81
+ "attention_mode": "cross_attn",
82
+ "prefix_length": -1,
83
+ "pad_language_to": "longest",
84
+ "num_expert_layers": -1,
85
+ "num_vlm_layers": 16,
86
+ "self_attn_every_n_layers": 2,
87
+ "expert_width_multiplier": 0.75,
88
+ "min_period": 0.004,
89
+ "max_period": 4.0,
90
+ "rtc_config": null,
91
+ "compile_model": false,
92
+ "compile_mode": "max-autotune"
93
+ }
checkpoints/040000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a896764279eeaa81f01cc224fd88ba5e3d9517ba1399043d8a1bd65b63c61b10
3
+ size 906712520
checkpoints/040000/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
+ 7
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/040000/policy_postprocessor_step_0_unnormalizer_processor.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:30397da3ad706d6c2bfba30605a07de25108d2a39fc922c07fd5b0208a6461b9
3
+ size 7608
checkpoints/040000/policy_preprocessor.json ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": "max_length",
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.state": {
42
+ "type": "STATE",
43
+ "shape": [
44
+ 7
45
+ ]
46
+ },
47
+ "observation.images.wrist": {
48
+ "type": "VISUAL",
49
+ "shape": [
50
+ 3,
51
+ 480,
52
+ 640
53
+ ]
54
+ },
55
+ "observation.images.top": {
56
+ "type": "VISUAL",
57
+ "shape": [
58
+ 3,
59
+ 480,
60
+ 640
61
+ ]
62
+ },
63
+ "action": {
64
+ "type": "ACTION",
65
+ "shape": [
66
+ 7
67
+ ]
68
+ }
69
+ },
70
+ "norm_map": {
71
+ "VISUAL": "IDENTITY",
72
+ "STATE": "MEAN_STD",
73
+ "ACTION": "MEAN_STD"
74
+ }
75
+ },
76
+ "state_file": "policy_preprocessor_step_5_normalizer_processor.safetensors"
77
+ }
78
+ ]
79
+ }
checkpoints/040000/policy_preprocessor_step_5_normalizer_processor.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:30397da3ad706d6c2bfba30605a07de25108d2a39fc922c07fd5b0208a6461b9
3
+ size 7608
checkpoints/040000/train_config.json ADDED
@@ -0,0 +1,232 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "nikriz/pick_place_bowl",
4
+ "root": null,
5
+ "episodes": null,
6
+ "image_transforms": {
7
+ "enable": false,
8
+ "max_num_transforms": 3,
9
+ "random_order": false,
10
+ "backend": "auto",
11
+ "tfs": {
12
+ "brightness": {
13
+ "weight": 1.0,
14
+ "type": "ColorJitter",
15
+ "kwargs": {
16
+ "brightness": [
17
+ 0.8,
18
+ 1.2
19
+ ]
20
+ }
21
+ },
22
+ "contrast": {
23
+ "weight": 1.0,
24
+ "type": "ColorJitter",
25
+ "kwargs": {
26
+ "contrast": [
27
+ 0.8,
28
+ 1.2
29
+ ]
30
+ }
31
+ },
32
+ "saturation": {
33
+ "weight": 1.0,
34
+ "type": "ColorJitter",
35
+ "kwargs": {
36
+ "saturation": [
37
+ 0.5,
38
+ 1.5
39
+ ]
40
+ }
41
+ },
42
+ "hue": {
43
+ "weight": 1.0,
44
+ "type": "ColorJitter",
45
+ "kwargs": {
46
+ "hue": [
47
+ -0.05,
48
+ 0.05
49
+ ]
50
+ }
51
+ },
52
+ "sharpness": {
53
+ "weight": 1.0,
54
+ "type": "SharpnessJitter",
55
+ "kwargs": {
56
+ "sharpness": [
57
+ 0.5,
58
+ 1.5
59
+ ]
60
+ }
61
+ },
62
+ "affine": {
63
+ "weight": 1.0,
64
+ "type": "RandomAffine",
65
+ "kwargs": {
66
+ "degrees": [
67
+ -5.0,
68
+ 5.0
69
+ ],
70
+ "translate": [
71
+ 0.05,
72
+ 0.05
73
+ ]
74
+ }
75
+ }
76
+ }
77
+ },
78
+ "revision": "main",
79
+ "use_imagenet_stats": true,
80
+ "video_backend": "torchcodec",
81
+ "streaming": false
82
+ },
83
+ "env": null,
84
+ "policy": {
85
+ "type": "smolvla",
86
+ "n_obs_steps": 1,
87
+ "input_features": {
88
+ "observation.state": {
89
+ "type": "STATE",
90
+ "shape": [
91
+ 7
92
+ ]
93
+ },
94
+ "observation.images.wrist": {
95
+ "type": "VISUAL",
96
+ "shape": [
97
+ 3,
98
+ 480,
99
+ 640
100
+ ]
101
+ },
102
+ "observation.images.top": {
103
+ "type": "VISUAL",
104
+ "shape": [
105
+ 3,
106
+ 480,
107
+ 640
108
+ ]
109
+ }
110
+ },
111
+ "output_features": {
112
+ "action": {
113
+ "type": "ACTION",
114
+ "shape": [
115
+ 7
116
+ ]
117
+ }
118
+ },
119
+ "device": "cuda",
120
+ "use_amp": false,
121
+ "use_peft": false,
122
+ "push_to_hub": true,
123
+ "repo_id": "joon-stack/smolvla_pick_place_bowl_",
124
+ "private": null,
125
+ "tags": null,
126
+ "license": null,
127
+ "pretrained_path": "lerobot/smolvla_base",
128
+ "chunk_size": 50,
129
+ "n_action_steps": 50,
130
+ "normalization_mapping": {
131
+ "VISUAL": "IDENTITY",
132
+ "STATE": "MEAN_STD",
133
+ "ACTION": "MEAN_STD"
134
+ },
135
+ "max_state_dim": 32,
136
+ "max_action_dim": 32,
137
+ "resize_imgs_with_padding": [
138
+ 512,
139
+ 512
140
+ ],
141
+ "empty_cameras": 0,
142
+ "adapt_to_pi_aloha": false,
143
+ "use_delta_joint_actions_aloha": false,
144
+ "tokenizer_max_length": 48,
145
+ "num_steps": 10,
146
+ "use_cache": true,
147
+ "freeze_vision_encoder": true,
148
+ "train_expert_only": true,
149
+ "train_state_proj": true,
150
+ "optimizer_lr": 0.0001,
151
+ "optimizer_betas": [
152
+ 0.9,
153
+ 0.95
154
+ ],
155
+ "optimizer_eps": 1e-08,
156
+ "optimizer_weight_decay": 1e-10,
157
+ "optimizer_grad_clip_norm": 10.0,
158
+ "scheduler_warmup_steps": 1000,
159
+ "scheduler_decay_steps": 200000,
160
+ "scheduler_decay_lr": 2.5e-06,
161
+ "vlm_model_name": "HuggingFaceTB/SmolVLM2-500M-Video-Instruct",
162
+ "load_vlm_weights": true,
163
+ "add_image_special_tokens": false,
164
+ "attention_mode": "cross_attn",
165
+ "prefix_length": -1,
166
+ "pad_language_to": "longest",
167
+ "num_expert_layers": -1,
168
+ "num_vlm_layers": 16,
169
+ "self_attn_every_n_layers": 2,
170
+ "expert_width_multiplier": 0.75,
171
+ "min_period": 0.004,
172
+ "max_period": 4.0,
173
+ "rtc_config": null,
174
+ "compile_model": false,
175
+ "compile_mode": "max-autotune"
176
+ },
177
+ "output_dir": "outputs/smolvla_pick_place_bowl_",
178
+ "job_name": "smolvla_pick_place_bowl",
179
+ "resume": false,
180
+ "seed": 1000,
181
+ "cudnn_deterministic": false,
182
+ "num_workers": 4,
183
+ "batch_size": 64,
184
+ "steps": 200000,
185
+ "eval_freq": 20000,
186
+ "log_freq": 200,
187
+ "tolerance_s": 0.0001,
188
+ "save_checkpoint": true,
189
+ "save_freq": 20000,
190
+ "use_policy_training_preset": true,
191
+ "optimizer": {
192
+ "type": "adamw",
193
+ "lr": 0.0001,
194
+ "weight_decay": 1e-10,
195
+ "grad_clip_norm": 10.0,
196
+ "betas": [
197
+ 0.9,
198
+ 0.95
199
+ ],
200
+ "eps": 1e-08
201
+ },
202
+ "scheduler": {
203
+ "type": "cosine_decay_with_warmup",
204
+ "num_warmup_steps": 1000,
205
+ "num_decay_steps": 200000,
206
+ "peak_lr": 0.0001,
207
+ "decay_lr": 2.5e-06
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": true,
217
+ "project": "pick_place_bowl",
218
+ "entity": null,
219
+ "notes": null,
220
+ "run_id": "e0yh4szy",
221
+ "mode": null,
222
+ "add_tags": true
223
+ },
224
+ "peft": null,
225
+ "use_rabc": false,
226
+ "rabc_progress_path": null,
227
+ "rabc_kappa": 0.01,
228
+ "rabc_epsilon": 1e-06,
229
+ "rabc_head_mode": "sparse",
230
+ "rename_map": {},
231
+ "checkpoint_path": null
232
+ }