Hugging Face's logo

Naomiihao
/
eval3-top50-fast15

Robotics
LeRobot
Safetensors
smolvla
Model card Files
xet
 Community
Naomiihao commited on
Commit
0d1c0d0
·
verified ·
1 Parent(s): e0da2c1

Upload policy weights, train config and readme

Browse files
Files changed (4) hide show
  1. README.md +63 -0
  2. config.json +109 -0
  3. model.safetensors +3 -0
  4. train_config.json +274 -0
README.md ADDED
@@ -0,0 +1,63 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: lerobot/smolvla_base
3
+ datasets: RobotLearningVLA/dataset_v4_taylor_left
4
+ library_name: lerobot
5
+ license: apache-2.0
6
+ model_name: smolvla
7
+ pipeline_tag: robotics
8
+ tags:
9
+ - lerobot
10
+ - robotics
11
+ - smolvla
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 +1,109 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ },
44
+ "output_features": {
45
+ "action": {
46
+ "type": "ACTION",
47
+ "shape": [
48
+ 6
49
+ ]
50
+ }
51
+ },
52
+ "device": "cuda",
53
+ "use_amp": false,
54
+ "use_peft": false,
55
+ "push_to_hub": true,
56
+ "repo_id": "Naomiihao/eval3-top50-fast15",
57
+ "private": null,
58
+ "tags": null,
59
+ "license": null,
60
+ "pretrained_path": "RobotLearningVLA/eval3-smolvla-v16-pinsv5-step5k",
61
+ "chunk_size": 50,
62
+ "n_action_steps": 50,
63
+ "normalization_mapping": {
64
+ "VISUAL": "IDENTITY",
65
+ "STATE": "MEAN_STD",
66
+ "ACTION": "MEAN_STD"
67
+ },
68
+ "max_state_dim": 32,
69
+ "max_action_dim": 32,
70
+ "resize_imgs_with_padding": [
71
+ 512,
72
+ 512
73
+ ],
74
+ "empty_cameras": 1,
75
+ "adapt_to_pi_aloha": false,
76
+ "use_delta_joint_actions_aloha": false,
77
+ "tokenizer_max_length": 48,
78
+ "num_steps": 10,
79
+ "use_cache": true,
80
+ "freeze_vision_encoder": true,
81
+ "train_expert_only": true,
82
+ "train_state_proj": true,
83
+ "optimizer_lr": 0.0001,
84
+ "optimizer_betas": [
85
+ 0.9,
86
+ 0.95
87
+ ],
88
+ "optimizer_eps": 1e-08,
89
+ "optimizer_weight_decay": 1e-10,
90
+ "optimizer_grad_clip_norm": 10.0,
91
+ "scheduler_warmup_steps": 1000,
92
+ "scheduler_decay_steps": 30000,
93
+ "scheduler_decay_lr": 2.5e-06,
94
+ "vlm_model_name": "HuggingFaceTB/SmolVLM2-500M-Video-Instruct",
95
+ "load_vlm_weights": true,
96
+ "add_image_special_tokens": false,
97
+ "attention_mode": "cross_attn",
98
+ "prefix_length": 0,
99
+ "pad_language_to": "max_length",
100
+ "num_expert_layers": 0,
101
+ "num_vlm_layers": 16,
102
+ "self_attn_every_n_layers": 2,
103
+ "expert_width_multiplier": 0.75,
104
+ "min_period": 0.004,
105
+ "max_period": 4.0,
106
+ "rtc_config": null,
107
+ "compile_model": false,
108
+ "compile_mode": "max-autotune"
109
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:756a38d933a49492a64ca84f39a3bd043ddf8770e7b3c7af79def07224b3ed8b
3
+ size 912702764
train_config.json ADDED
@@ -0,0 +1,274 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "RobotLearningVLA/dataset_v4_taylor_left",
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
+ 0.7,
37
+ 1.3
38
+ ]
39
+ }
40
+ },
41
+ "hue": {
42
+ "weight": 0.5,
43
+ "type": "ColorJitter",
44
+ "kwargs": {
45
+ "hue": [
46
+ -0.05,
47
+ 0.05
48
+ ]
49
+ }
50
+ },
51
+ "sharpness": {
52
+ "weight": 0.75,
53
+ "type": "SharpnessJitter",
54
+ "kwargs": {
55
+ "sharpness": [
56
+ 0.6,
57
+ 1.4
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
+ 0.04,
71
+ 0.04
72
+ ]
73
+ }
74
+ },
75
+ "perspective": {
76
+ "weight": 0.75,
77
+ "type": "RandomPerspective",
78
+ "kwargs": {
79
+ "distortion_scale": 0.18,
80
+ "p": 0.4
81
+ }
82
+ },
83
+ "random_erasing": {
84
+ "weight": 1.0,
85
+ "type": "RandomErasing",
86
+ "kwargs": {
87
+ "p": 0.4,
88
+ "scale": [
89
+ 0.02,
90
+ 0.08
91
+ ],
92
+ "ratio": [
93
+ 0.3,
94
+ 3.3
95
+ ],
96
+ "value": 0
97
+ }
98
+ }
99
+ }
100
+ },
101
+ "revision": null,
102
+ "use_imagenet_stats": true,
103
+ "video_backend": "pyav",
104
+ "streaming": false
105
+ },
106
+ "env": null,
107
+ "policy": {
108
+ "type": "smolvla",
109
+ "n_obs_steps": 1,
110
+ "input_features": {
111
+ "observation.state": {
112
+ "type": "STATE",
113
+ "shape": [
114
+ 6
115
+ ]
116
+ },
117
+ "observation.images.camera1": {
118
+ "type": "VISUAL",
119
+ "shape": [
120
+ 3,
121
+ 256,
122
+ 256
123
+ ]
124
+ },
125
+ "observation.images.camera2": {
126
+ "type": "VISUAL",
127
+ "shape": [
128
+ 3,
129
+ 256,
130
+ 256
131
+ ]
132
+ },
133
+ "observation.images.camera3": {
134
+ "type": "VISUAL",
135
+ "shape": [
136
+ 3,
137
+ 256,
138
+ 256
139
+ ]
140
+ },
141
+ "observation.images.empty_camera_0": {
142
+ "type": "VISUAL",
143
+ "shape": [
144
+ 3,
145
+ 480,
146
+ 640
147
+ ]
148
+ }
149
+ },
150
+ "output_features": {
151
+ "action": {
152
+ "type": "ACTION",
153
+ "shape": [
154
+ 6
155
+ ]
156
+ }
157
+ },
158
+ "device": "cuda",
159
+ "use_amp": false,
160
+ "use_peft": false,
161
+ "push_to_hub": true,
162
+ "repo_id": "Naomiihao/eval3-top50-fast15",
163
+ "private": null,
164
+ "tags": null,
165
+ "license": null,
166
+ "pretrained_path": "RobotLearningVLA/eval3-smolvla-v16-pinsv5-step5k",
167
+ "chunk_size": 50,
168
+ "n_action_steps": 50,
169
+ "normalization_mapping": {
170
+ "VISUAL": "IDENTITY",
171
+ "STATE": "MEAN_STD",
172
+ "ACTION": "MEAN_STD"
173
+ },
174
+ "max_state_dim": 32,
175
+ "max_action_dim": 32,
176
+ "resize_imgs_with_padding": [
177
+ 512,
178
+ 512
179
+ ],
180
+ "empty_cameras": 1,
181
+ "adapt_to_pi_aloha": false,
182
+ "use_delta_joint_actions_aloha": false,
183
+ "tokenizer_max_length": 48,
184
+ "num_steps": 10,
185
+ "use_cache": true,
186
+ "freeze_vision_encoder": true,
187
+ "train_expert_only": true,
188
+ "train_state_proj": true,
189
+ "optimizer_lr": 0.0001,
190
+ "optimizer_betas": [
191
+ 0.9,
192
+ 0.95
193
+ ],
194
+ "optimizer_eps": 1e-08,
195
+ "optimizer_weight_decay": 1e-10,
196
+ "optimizer_grad_clip_norm": 10.0,
197
+ "scheduler_warmup_steps": 1000,
198
+ "scheduler_decay_steps": 30000,
199
+ "scheduler_decay_lr": 2.5e-06,
200
+ "vlm_model_name": "HuggingFaceTB/SmolVLM2-500M-Video-Instruct",
201
+ "load_vlm_weights": true,
202
+ "add_image_special_tokens": false,
203
+ "attention_mode": "cross_attn",
204
+ "prefix_length": 0,
205
+ "pad_language_to": "max_length",
206
+ "num_expert_layers": 0,
207
+ "num_vlm_layers": 16,
208
+ "self_attn_every_n_layers": 2,
209
+ "expert_width_multiplier": 0.75,
210
+ "min_period": 0.004,
211
+ "max_period": 4.0,
212
+ "rtc_config": null,
213
+ "compile_model": false,
214
+ "compile_mode": "max-autotune"
215
+ },
216
+ "output_dir": "/ephemeral/outputs/train/eval3_top50_15EVAL3_JOB_NAME=eval3_top50_15",
217
+ "job_name": "eval3_v17_custom_ood_slot",
218
+ "resume": false,
219
+ "seed": 1000,
220
+ "cudnn_deterministic": false,
221
+ "num_workers": 4,
222
+ "batch_size": 128,
223
+ "steps": 5000,
224
+ "eval_freq": 20000,
225
+ "log_freq": 200,
226
+ "tolerance_s": 0.0001,
227
+ "save_checkpoint": true,
228
+ "save_freq": 1000,
229
+ "use_policy_training_preset": false,
230
+ "optimizer": {
231
+ "type": "adamw",
232
+ "lr": 0.0001,
233
+ "weight_decay": 1e-10,
234
+ "grad_clip_norm": 10.0,
235
+ "betas": [
236
+ 0.9,
237
+ 0.999
238
+ ],
239
+ "eps": 1e-08
240
+ },
241
+ "scheduler": {
242
+ "type": "cosine_decay_with_warmup",
243
+ "num_warmup_steps": 250,
244
+ "num_decay_steps": 4750,
245
+ "peak_lr": 0.0001,
246
+ "decay_lr": 1e-06
247
+ },
248
+ "eval": {
249
+ "n_episodes": 50,
250
+ "batch_size": 50,
251
+ "use_async_envs": false
252
+ },
253
+ "wandb": {
254
+ "enable": false,
255
+ "disable_artifact": false,
256
+ "project": "eval3-v16-real-data-slot",
257
+ "entity": null,
258
+ "notes": null,
259
+ "run_id": null,
260
+ "mode": null,
261
+ "add_tags": true
262
+ },
263
+ "peft": null,
264
+ "use_rabc": false,
265
+ "rabc_progress_path": null,
266
+ "rabc_kappa": 0.01,
267
+ "rabc_epsilon": 1e-06,
268
+ "rabc_head_mode": "sparse",
269
+ "rename_map": {
270
+ "observation.images.front": "observation.images.camera1",
271
+ "observation.images.front_frame0": "observation.images.camera2"
272
+ },
273
+ "checkpoint_path": null
274
+ }