Hugging Face's logo

xander2077
/
smolvla_arx5_pull_out_drawer_test2

Robotics
LeRobot
Safetensors
smolvla
Model card Files
xet
Community
xander2077 commited on
Commit
b9887a7
·
verified ·
1 Parent(s): cf4d4c6

Upload policy weights, train config and readme

Browse files
Files changed (4) hide show
  1. README.md +63 -0
  2. config.json +134 -0
  3. model.safetensors +3 -0
  4. train_config.json +263 -0
README.md ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: lerobot/smolvla_base
3
+ datasets: xander2077/test2
4
+ library_name: lerobot
5
+ license: apache-2.0
6
+ model_name: smolvla
7
+ pipeline_tag: robotics
8
+ tags:
9
+ - smolvla
10
+ - robotics
11
+ - lerobot
12
+ ---
13
+
14
+ # Model Card for smolvla
15
+
16
+ <!-- Provide a quick summary of what the model is/does. -->
17
+
18
+
19
+ [SmolVLA](https://huggingface.co/papers/2506.01844) is a compact, efficient vision-language-action model that achieves competitive performance at reduced computational costs and can be deployed on consumer-grade hardware.
20
+
21
+
22
+ This policy has been trained and pushed to the Hub using [LeRobot](https://github.com/huggingface/lerobot).
23
+ See the full documentation at [LeRobot Docs](https://huggingface.co/docs/lerobot/index).
24
+
25
+ ---
26
+
27
+ ## How to Get Started with the Model
28
+
29
+ For a complete walkthrough, see the [training guide](https://huggingface.co/docs/lerobot/il_robots#train-a-policy).
30
+ Below is the short version on how to train and run inference/eval:
31
+
32
+ ### Train from scratch
33
+
34
+ ```bash
35
+ lerobot-train \
36
+ --dataset.repo_id=${HF_USER}/<dataset> \
37
+ --policy.type=act \
38
+ --output_dir=outputs/train/<desired_policy_repo_id> \
39
+ --job_name=lerobot_training \
40
+ --policy.device=cuda \
41
+ --policy.repo_id=${HF_USER}/<desired_policy_repo_id>
42
+ --wandb.enable=true
43
+ ```
44
+
45
+ _Writes checkpoints to `outputs/train/<desired_policy_repo_id>/checkpoints/`._
46
+
47
+ ### Evaluate the policy/run inference
48
+
49
+ ```bash
50
+ lerobot-record \
51
+ --robot.type=so100_follower \
52
+ --dataset.repo_id=<hf_user>/eval_<dataset> \
53
+ --policy.path=<hf_user>/<desired_policy_repo_id> \
54
+ --episodes=10
55
+ ```
56
+
57
+ Prefix the dataset repo with **eval\_** and supply `--policy.path` pointing to a local or hub checkpoint.
58
+
59
+ ---
60
+
61
+ ## Model Details
62
+
63
+ - **License:** apache-2.0
config.json ADDED
@@ -0,0 +1,134 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "smolvla",
3
+ "n_obs_steps": 1,
4
+ "input_features": {
5
+ "observation.state": {
6
+ "type": "STATE",
7
+ "shape": [
8
+ 28
9
+ ]
10
+ },
11
+ "observation.robot_status.end_pos": {
12
+ "type": "STATE",
13
+ "shape": [
14
+ 6
15
+ ]
16
+ },
17
+ "observation.robot_status.joint_pos": {
18
+ "type": "STATE",
19
+ "shape": [
20
+ 7
21
+ ]
22
+ },
23
+ "observation.robot_status.joint_vel": {
24
+ "type": "STATE",
25
+ "shape": [
26
+ 7
27
+ ]
28
+ },
29
+ "observation.robot_status.joint_cur": {
30
+ "type": "STATE",
31
+ "shape": [
32
+ 7
33
+ ]
34
+ },
35
+ "observation.gripper": {
36
+ "type": "STATE",
37
+ "shape": [
38
+ 1
39
+ ]
40
+ },
41
+ "observation.images.d405_rgb": {
42
+ "type": "VISUAL",
43
+ "shape": [
44
+ 3,
45
+ 480,
46
+ 640
47
+ ]
48
+ },
49
+ "observation.images.d435_rgb": {
50
+ "type": "VISUAL",
51
+ "shape": [
52
+ 3,
53
+ 480,
54
+ 640
55
+ ]
56
+ },
57
+ "observation.d405_depth_m": {
58
+ "type": "STATE",
59
+ "shape": [
60
+ 480,
61
+ 640
62
+ ]
63
+ },
64
+ "observation.d435_depth_m": {
65
+ "type": "STATE",
66
+ "shape": [
67
+ 480,
68
+ 640
69
+ ]
70
+ }
71
+ },
72
+ "output_features": {
73
+ "action": {
74
+ "type": "ACTION",
75
+ "shape": [
76
+ 7
77
+ ]
78
+ }
79
+ },
80
+ "device": "cuda",
81
+ "use_amp": false,
82
+ "push_to_hub": true,
83
+ "repo_id": "xander2077/smolvla_arx5_pull_out_drawer_test2",
84
+ "private": null,
85
+ "tags": null,
86
+ "license": null,
87
+ "pretrained_path": null,
88
+ "chunk_size": 50,
89
+ "n_action_steps": 50,
90
+ "normalization_mapping": {
91
+ "VISUAL": "IDENTITY",
92
+ "STATE": "MEAN_STD",
93
+ "ACTION": "MEAN_STD"
94
+ },
95
+ "max_state_dim": 32,
96
+ "max_action_dim": 32,
97
+ "resize_imgs_with_padding": [
98
+ 512,
99
+ 512
100
+ ],
101
+ "empty_cameras": 0,
102
+ "adapt_to_pi_aloha": false,
103
+ "use_delta_joint_actions_aloha": false,
104
+ "tokenizer_max_length": 48,
105
+ "num_steps": 10,
106
+ "use_cache": true,
107
+ "freeze_vision_encoder": true,
108
+ "train_expert_only": true,
109
+ "train_state_proj": true,
110
+ "optimizer_lr": 0.0001,
111
+ "optimizer_betas": [
112
+ 0.9,
113
+ 0.95
114
+ ],
115
+ "optimizer_eps": 1e-08,
116
+ "optimizer_weight_decay": 1e-10,
117
+ "optimizer_grad_clip_norm": 10,
118
+ "scheduler_warmup_steps": 1000,
119
+ "scheduler_decay_steps": 30000,
120
+ "scheduler_decay_lr": 2.5e-06,
121
+ "vlm_model_name": "HuggingFaceTB/SmolVLM2-500M-Video-Instruct",
122
+ "load_vlm_weights": false,
123
+ "add_image_special_tokens": false,
124
+ "attention_mode": "cross_attn",
125
+ "prefix_length": -1,
126
+ "pad_language_to": "longest",
127
+ "num_expert_layers": -1,
128
+ "num_vlm_layers": 16,
129
+ "self_attn_every_n_layers": 2,
130
+ "expert_width_multiplier": 0.75,
131
+ "min_period": 0.004,
132
+ "max_period": 4.0,
133
+ "rtc_config": null
134
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e5e2525359b43f322d7e2f9b92f82b23bdbd05b0aba6307b5aa2c6497d888c46
3
+ size 1197789224
train_config.json ADDED
@@ -0,0 +1,263 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "xander2077/test2",
4
+ "root": "/home/xander/Code/LocoManip/lerobot/datasets/pull_out_drawer",
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.state": {
88
+ "type": "STATE",
89
+ "shape": [
90
+ 28
91
+ ]
92
+ },
93
+ "observation.robot_status.end_pos": {
94
+ "type": "STATE",
95
+ "shape": [
96
+ 6
97
+ ]
98
+ },
99
+ "observation.robot_status.joint_pos": {
100
+ "type": "STATE",
101
+ "shape": [
102
+ 7
103
+ ]
104
+ },
105
+ "observation.robot_status.joint_vel": {
106
+ "type": "STATE",
107
+ "shape": [
108
+ 7
109
+ ]
110
+ },
111
+ "observation.robot_status.joint_cur": {
112
+ "type": "STATE",
113
+ "shape": [
114
+ 7
115
+ ]
116
+ },
117
+ "observation.gripper": {
118
+ "type": "STATE",
119
+ "shape": [
120
+ 1
121
+ ]
122
+ },
123
+ "observation.images.d405_rgb": {
124
+ "type": "VISUAL",
125
+ "shape": [
126
+ 3,
127
+ 480,
128
+ 640
129
+ ]
130
+ },
131
+ "observation.images.d435_rgb": {
132
+ "type": "VISUAL",
133
+ "shape": [
134
+ 3,
135
+ 480,
136
+ 640
137
+ ]
138
+ },
139
+ "observation.d405_depth_m": {
140
+ "type": "STATE",
141
+ "shape": [
142
+ 480,
143
+ 640
144
+ ]
145
+ },
146
+ "observation.d435_depth_m": {
147
+ "type": "STATE",
148
+ "shape": [
149
+ 480,
150
+ 640
151
+ ]
152
+ }
153
+ },
154
+ "output_features": {
155
+ "action": {
156
+ "type": "ACTION",
157
+ "shape": [
158
+ 7
159
+ ]
160
+ }
161
+ },
162
+ "device": "cuda",
163
+ "use_amp": false,
164
+ "push_to_hub": true,
165
+ "repo_id": "xander2077/smolvla_arx5_pull_out_drawer_test2",
166
+ "private": null,
167
+ "tags": null,
168
+ "license": null,
169
+ "pretrained_path": null,
170
+ "chunk_size": 50,
171
+ "n_action_steps": 50,
172
+ "normalization_mapping": {
173
+ "VISUAL": "IDENTITY",
174
+ "STATE": "MEAN_STD",
175
+ "ACTION": "MEAN_STD"
176
+ },
177
+ "max_state_dim": 32,
178
+ "max_action_dim": 32,
179
+ "resize_imgs_with_padding": [
180
+ 512,
181
+ 512
182
+ ],
183
+ "empty_cameras": 0,
184
+ "adapt_to_pi_aloha": false,
185
+ "use_delta_joint_actions_aloha": false,
186
+ "tokenizer_max_length": 48,
187
+ "num_steps": 10,
188
+ "use_cache": true,
189
+ "freeze_vision_encoder": true,
190
+ "train_expert_only": true,
191
+ "train_state_proj": true,
192
+ "optimizer_lr": 0.0001,
193
+ "optimizer_betas": [
194
+ 0.9,
195
+ 0.95
196
+ ],
197
+ "optimizer_eps": 1e-08,
198
+ "optimizer_weight_decay": 1e-10,
199
+ "optimizer_grad_clip_norm": 10,
200
+ "scheduler_warmup_steps": 1000,
201
+ "scheduler_decay_steps": 30000,
202
+ "scheduler_decay_lr": 2.5e-06,
203
+ "vlm_model_name": "HuggingFaceTB/SmolVLM2-500M-Video-Instruct",
204
+ "load_vlm_weights": false,
205
+ "add_image_special_tokens": false,
206
+ "attention_mode": "cross_attn",
207
+ "prefix_length": -1,
208
+ "pad_language_to": "longest",
209
+ "num_expert_layers": -1,
210
+ "num_vlm_layers": 16,
211
+ "self_attn_every_n_layers": 2,
212
+ "expert_width_multiplier": 0.75,
213
+ "min_period": 0.004,
214
+ "max_period": 4.0,
215
+ "rtc_config": null
216
+ },
217
+ "output_dir": "outputs/train/smolvla_arx5_pull_out_drawer_test2",
218
+ "job_name": "smolvla_pull_out_drawer_test2",
219
+ "resume": false,
220
+ "seed": 1000,
221
+ "num_workers": 4,
222
+ "batch_size": 8,
223
+ "steps": 100000,
224
+ "eval_freq": 20000,
225
+ "log_freq": 200,
226
+ "save_checkpoint": true,
227
+ "save_freq": 20000,
228
+ "use_policy_training_preset": true,
229
+ "optimizer": {
230
+ "type": "adamw",
231
+ "lr": 0.0001,
232
+ "weight_decay": 1e-10,
233
+ "grad_clip_norm": 10,
234
+ "betas": [
235
+ 0.9,
236
+ 0.95
237
+ ],
238
+ "eps": 1e-08
239
+ },
240
+ "scheduler": {
241
+ "type": "cosine_decay_with_warmup",
242
+ "num_warmup_steps": 1000,
243
+ "num_decay_steps": 30000,
244
+ "peak_lr": 0.0001,
245
+ "decay_lr": 2.5e-06
246
+ },
247
+ "eval": {
248
+ "n_episodes": 50,
249
+ "batch_size": 50,
250
+ "use_async_envs": false
251
+ },
252
+ "wandb": {
253
+ "enable": true,
254
+ "disable_artifact": false,
255
+ "project": "lerobot",
256
+ "entity": null,
257
+ "notes": null,
258
+ "run_id": "wf6e0z94",
259
+ "mode": null
260
+ },
261
+ "checkpoint_path": null,
262
+ "rename_map": {}
263
+ }