Hugging Face's logo

jasontchan
/
emg-spec-100

Safetensors
Model card Files
xet
 Community
jasontchan commited on
Commit
f16ab06
·
verified ·
1 Parent(s): f8a88fd

Upload folder using huggingface_hub

Browse files
Files changed (3) hide show
  1. config.json +107 -0
  2. model.safetensors +3 -0
  3. train_config.json +220 -0
config.json ADDED
@@ -0,0 +1,107 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "smolvla",
3
+ "n_obs_steps": 1,
4
+ "normalization_mapping": {
5
+ "VISUAL": "IDENTITY",
6
+ "STATE": "MEAN_STD",
7
+ "ACTION": "MEAN_STD",
8
+ "EMG": "MEAN_STD"
9
+ },
10
+ "input_features": {
11
+ "observation.state": {
12
+ "type": "STATE",
13
+ "shape": [
14
+ 6
15
+ ]
16
+ },
17
+ "observation.images.front": {
18
+ "type": "VISUAL",
19
+ "shape": [
20
+ 3,
21
+ 1080,
22
+ 1920
23
+ ]
24
+ },
25
+ "observation.images.top": {
26
+ "type": "VISUAL",
27
+ "shape": [
28
+ 3,
29
+ 720,
30
+ 1280
31
+ ]
32
+ },
33
+ "observation.images.gripper": {
34
+ "type": "VISUAL",
35
+ "shape": [
36
+ 3,
37
+ 1080,
38
+ 1920
39
+ ]
40
+ },
41
+ "observation.emg.lower": {
42
+ "type": "EMG",
43
+ "shape": [
44
+ 100,
45
+ 8
46
+ ]
47
+ }
48
+ },
49
+ "output_features": {
50
+ "action": {
51
+ "type": "ACTION",
52
+ "shape": [
53
+ 6
54
+ ]
55
+ }
56
+ },
57
+ "device": "cuda",
58
+ "use_amp": false,
59
+ "emg_mask_prob": 0.0,
60
+ "emg_spectrogram": true,
61
+ "n_fft": 100,
62
+ "use_camera_keys": {
63
+ "observation.images.top": true,
64
+ "observation.images.gripper": true,
65
+ "observation.images.front": true
66
+ },
67
+ "chunk_size": 50,
68
+ "n_action_steps": 50,
69
+ "max_state_dim": 32,
70
+ "max_action_dim": 32,
71
+ "resize_imgs_with_padding": [
72
+ 512,
73
+ 512
74
+ ],
75
+ "empty_cameras": 0,
76
+ "adapt_to_pi_aloha": false,
77
+ "use_delta_joint_actions_aloha": false,
78
+ "tokenizer_max_length": 48,
79
+ "num_steps": 10,
80
+ "use_cache": true,
81
+ "freeze_vision_encoder": true,
82
+ "train_expert_only": true,
83
+ "train_state_proj": true,
84
+ "optimizer_lr": 0.0001,
85
+ "optimizer_betas": [
86
+ 0.9,
87
+ 0.95
88
+ ],
89
+ "optimizer_eps": 1e-08,
90
+ "optimizer_weight_decay": 1e-10,
91
+ "optimizer_grad_clip_norm": 10.0,
92
+ "scheduler_warmup_steps": 1000,
93
+ "scheduler_decay_steps": 30000,
94
+ "scheduler_decay_lr": 2.5e-06,
95
+ "vlm_model_name": "HuggingFaceTB/SmolVLM2-500M-Video-Instruct",
96
+ "load_vlm_weights": true,
97
+ "add_image_special_tokens": false,
98
+ "attention_mode": "cross_attn",
99
+ "prefix_length": 0,
100
+ "pad_language_to": "max_length",
101
+ "num_expert_layers": 0,
102
+ "num_vlm_layers": 16,
103
+ "self_attn_every_n_layers": 2,
104
+ "expert_width_multiplier": 0.75,
105
+ "min_period": 0.004,
106
+ "max_period": 4.0
107
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eb7a38e1aa789cd46ecf6541510c5d320819a2712300a7f00d412622c46bfc55
3
+ size 907173952
train_config.json ADDED
@@ -0,0 +1,220 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "jasontchan/emg-gestures-pick-place-yellow-block",
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
+ }
62
+ },
63
+ "revision": null,
64
+ "use_imagenet_stats": true,
65
+ "video_backend": "torchcodec"
66
+ },
67
+ "env": null,
68
+ "policy": {
69
+ "type": "smolvla",
70
+ "n_obs_steps": 1,
71
+ "normalization_mapping": {
72
+ "VISUAL": "IDENTITY",
73
+ "STATE": "MEAN_STD",
74
+ "ACTION": "MEAN_STD",
75
+ "EMG": "MEAN_STD"
76
+ },
77
+ "input_features": {
78
+ "observation.state": {
79
+ "type": "STATE",
80
+ "shape": [
81
+ 6
82
+ ]
83
+ },
84
+ "observation.images.front": {
85
+ "type": "VISUAL",
86
+ "shape": [
87
+ 3,
88
+ 1080,
89
+ 1920
90
+ ]
91
+ },
92
+ "observation.images.top": {
93
+ "type": "VISUAL",
94
+ "shape": [
95
+ 3,
96
+ 720,
97
+ 1280
98
+ ]
99
+ },
100
+ "observation.images.gripper": {
101
+ "type": "VISUAL",
102
+ "shape": [
103
+ 3,
104
+ 1080,
105
+ 1920
106
+ ]
107
+ },
108
+ "observation.emg.lower": {
109
+ "type": "EMG",
110
+ "shape": [
111
+ 100,
112
+ 8
113
+ ]
114
+ }
115
+ },
116
+ "output_features": {
117
+ "action": {
118
+ "type": "ACTION",
119
+ "shape": [
120
+ 6
121
+ ]
122
+ }
123
+ },
124
+ "device": "cuda",
125
+ "use_amp": false,
126
+ "emg_mask_prob": 0.0,
127
+ "emg_spectrogram": true,
128
+ "n_fft": 100,
129
+ "use_camera_keys": {
130
+ "observation.images.top": true,
131
+ "observation.images.gripper": true,
132
+ "observation.images.front": true
133
+ },
134
+ "chunk_size": 50,
135
+ "n_action_steps": 50,
136
+ "max_state_dim": 32,
137
+ "max_action_dim": 32,
138
+ "resize_imgs_with_padding": [
139
+ 512,
140
+ 512
141
+ ],
142
+ "empty_cameras": 0,
143
+ "adapt_to_pi_aloha": false,
144
+ "use_delta_joint_actions_aloha": false,
145
+ "tokenizer_max_length": 48,
146
+ "num_steps": 10,
147
+ "use_cache": true,
148
+ "freeze_vision_encoder": true,
149
+ "train_expert_only": true,
150
+ "train_state_proj": true,
151
+ "optimizer_lr": 0.0001,
152
+ "optimizer_betas": [
153
+ 0.9,
154
+ 0.95
155
+ ],
156
+ "optimizer_eps": 1e-08,
157
+ "optimizer_weight_decay": 1e-10,
158
+ "optimizer_grad_clip_norm": 10.0,
159
+ "scheduler_warmup_steps": 1000,
160
+ "scheduler_decay_steps": 30000,
161
+ "scheduler_decay_lr": 2.5e-06,
162
+ "vlm_model_name": "HuggingFaceTB/SmolVLM2-500M-Video-Instruct",
163
+ "load_vlm_weights": true,
164
+ "add_image_special_tokens": false,
165
+ "attention_mode": "cross_attn",
166
+ "prefix_length": 0,
167
+ "pad_language_to": "max_length",
168
+ "num_expert_layers": 0,
169
+ "num_vlm_layers": 16,
170
+ "self_attn_every_n_layers": 2,
171
+ "expert_width_multiplier": 0.75,
172
+ "min_period": 0.004,
173
+ "max_period": 4.0
174
+ },
175
+ "output_dir": "outputs/train/emg-spec-100",
176
+ "job_name": "emg-spec-100",
177
+ "resume": false,
178
+ "seed": 1000,
179
+ "num_workers": 4,
180
+ "batch_size": 8,
181
+ "steps": 100000,
182
+ "eval_freq": 20000,
183
+ "log_freq": 200,
184
+ "save_checkpoint": true,
185
+ "save_freq": 20000,
186
+ "use_policy_training_preset": true,
187
+ "optimizer": {
188
+ "type": "adamw",
189
+ "lr": 0.0001,
190
+ "weight_decay": 1e-10,
191
+ "grad_clip_norm": 10.0,
192
+ "betas": [
193
+ 0.9,
194
+ 0.95
195
+ ],
196
+ "eps": 1e-08
197
+ },
198
+ "scheduler": {
199
+ "type": "cosine_decay_with_warmup",
200
+ "num_warmup_steps": 1000,
201
+ "num_decay_steps": 30000,
202
+ "peak_lr": 0.0001,
203
+ "decay_lr": 2.5e-06
204
+ },
205
+ "eval": {
206
+ "n_episodes": 50,
207
+ "batch_size": 50,
208
+ "use_async_envs": false
209
+ },
210
+ "wandb": {
211
+ "enable": true,
212
+ "disable_artifact": false,
213
+ "project": "lerobot",
214
+ "entity": null,
215
+ "notes": null,
216
+ "run_id": "4a7aa5ad",
217
+ "mode": null
218
+ },
219
+ "use_emg": true
220
+ }