Hugging Face's logo

ns69956
/
focal_loss

Safetensors
Model card Files
xet
Community
ns69956 commited on
Commit
a400afa
·
verified ·
1 Parent(s): 094b3ea

Upload 3 files

Browse files
Files changed (3) hide show
  1. config.json +131 -0
  2. model.safetensors +3 -0
  3. train_config.json +243 -0
config.json ADDED
@@ -0,0 +1,131 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "smolvla",
3
+ "n_obs_steps": 1,
4
+ "normalization_mapping": {
5
+ "VISUAL": "IDENTITY",
6
+ "STATE": "MEAN_STD",
7
+ "ACTION": "MEAN_STD"
8
+ },
9
+ "input_features": {
10
+ "observation.images.scene_image": {
11
+ "type": "VISUAL",
12
+ "shape": [
13
+ 3,
14
+ 480,
15
+ 640
16
+ ]
17
+ },
18
+ "observation.images.wrist_image": {
19
+ "type": "VISUAL",
20
+ "shape": [
21
+ 3,
22
+ 480,
23
+ 640
24
+ ]
25
+ },
26
+ "observation.state": {
27
+ "type": "STATE",
28
+ "shape": [
29
+ 8
30
+ ]
31
+ }
32
+ },
33
+ "output_features": {
34
+ "action": {
35
+ "type": "ACTION",
36
+ "shape": [
37
+ 8
38
+ ]
39
+ }
40
+ },
41
+ "device": "cuda",
42
+ "use_amp": false,
43
+ "scanner_state_pos_weight_mode": "mean",
44
+ "scanner_state_pos_weight": 1.0,
45
+ "chunk_size": 50,
46
+ "n_action_steps": 50,
47
+ "mean_std_normalization_stats": {
48
+ "observation.state": {
49
+ "mean": [
50
+ 0.27873939275741577,
51
+ -0.29261285066604614,
52
+ 0.4602305293083191,
53
+ 0.08445943892002106,
54
+ 2.2974722385406494,
55
+ -0.2794015109539032,
56
+ 0.0,
57
+ 0.0
58
+ ],
59
+ "std": [
60
+ 0.10405326634645462,
61
+ 0.1369001269340515,
62
+ 0.12905506789684296,
63
+ 1.0609674453735352,
64
+ 1.6164617538452148,
65
+ 0.2593933939933777,
66
+ 1.0,
67
+ 1.0
68
+ ]
69
+ },
70
+ "action": {
71
+ "mean": [
72
+ 0.27874284982681274,
73
+ -0.2925970256328583,
74
+ 0.4602431356906891,
75
+ 0.08444322645664215,
76
+ 2.2977328300476074,
77
+ -0.27943673729896545,
78
+ 0.0,
79
+ 0.0
80
+ ],
81
+ "std": [
82
+ 0.10405024886131287,
83
+ 0.1369035840034485,
84
+ 0.12906143069267273,
85
+ 1.060977578163147,
86
+ 1.6160578727722168,
87
+ 0.25939232110977173,
88
+ 1.0,
89
+ 1.0
90
+ ]
91
+ }
92
+ },
93
+ "max_state_dim": 32,
94
+ "max_action_dim": 32,
95
+ "resize_imgs_with_padding": [
96
+ 512,
97
+ 512
98
+ ],
99
+ "empty_cameras": 0,
100
+ "adapt_to_pi_aloha": false,
101
+ "use_delta_joint_actions_aloha": false,
102
+ "tokenizer_max_length": 48,
103
+ "num_steps": 10,
104
+ "use_cache": true,
105
+ "freeze_vision_encoder": false,
106
+ "train_expert_only": true,
107
+ "train_state_proj": true,
108
+ "optimizer_lr": 0.0001,
109
+ "optimizer_betas": [
110
+ 0.9,
111
+ 0.95
112
+ ],
113
+ "optimizer_eps": 1e-08,
114
+ "optimizer_weight_decay": 1e-10,
115
+ "optimizer_grad_clip_norm": 10,
116
+ "scheduler_warmup_steps": 1000,
117
+ "scheduler_decay_steps": 30000,
118
+ "scheduler_decay_lr": 2.5e-06,
119
+ "vlm_model_name": "HuggingFaceTB/SmolVLM2-500M-Video-Instruct",
120
+ "load_vlm_weights": false,
121
+ "add_image_special_tokens": false,
122
+ "attention_mode": "cross_attn",
123
+ "prefix_length": -1,
124
+ "pad_language_to": "longest",
125
+ "num_expert_layers": -1,
126
+ "num_vlm_layers": 16,
127
+ "self_attn_every_n_layers": 2,
128
+ "expert_width_multiplier": 0.75,
129
+ "min_period": 0.004,
130
+ "max_period": 4.0
131
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1cfc7f7bbd345dff8a7274dcea8ca28165869274288c0c98d27364c5e1abe79a
3
+ size 1197790064
train_config.json ADDED
@@ -0,0 +1,243 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "ns69956/ur5e_scanner2",
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": "pyav"
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
+ },
76
+ "input_features": {
77
+ "observation.images.scene_image": {
78
+ "type": "VISUAL",
79
+ "shape": [
80
+ 3,
81
+ 480,
82
+ 640
83
+ ]
84
+ },
85
+ "observation.images.wrist_image": {
86
+ "type": "VISUAL",
87
+ "shape": [
88
+ 3,
89
+ 480,
90
+ 640
91
+ ]
92
+ },
93
+ "observation.state": {
94
+ "type": "STATE",
95
+ "shape": [
96
+ 8
97
+ ]
98
+ }
99
+ },
100
+ "output_features": {
101
+ "action": {
102
+ "type": "ACTION",
103
+ "shape": [
104
+ 8
105
+ ]
106
+ }
107
+ },
108
+ "device": "cuda",
109
+ "use_amp": false,
110
+ "scanner_state_pos_weight_mode": "mean",
111
+ "scanner_state_pos_weight": 1.0,
112
+ "chunk_size": 50,
113
+ "n_action_steps": 50,
114
+ "mean_std_normalization_stats": {
115
+ "observation.state": {
116
+ "mean": [
117
+ 0.27873939275741577,
118
+ -0.29261285066604614,
119
+ 0.4602305293083191,
120
+ 0.08445943892002106,
121
+ 2.2974722385406494,
122
+ -0.2794015109539032,
123
+ 0.0,
124
+ 0.0
125
+ ],
126
+ "std": [
127
+ 0.10405326634645462,
128
+ 0.1369001269340515,
129
+ 0.12905506789684296,
130
+ 1.0609674453735352,
131
+ 1.6164617538452148,
132
+ 0.2593933939933777,
133
+ 1.0,
134
+ 1.0
135
+ ]
136
+ },
137
+ "action": {
138
+ "mean": [
139
+ 0.27874284982681274,
140
+ -0.2925970256328583,
141
+ 0.4602431356906891,
142
+ 0.08444322645664215,
143
+ 2.2977328300476074,
144
+ -0.27943673729896545,
145
+ 0.0,
146
+ 0.0
147
+ ],
148
+ "std": [
149
+ 0.10405024886131287,
150
+ 0.1369035840034485,
151
+ 0.12906143069267273,
152
+ 1.060977578163147,
153
+ 1.6160578727722168,
154
+ 0.25939232110977173,
155
+ 1.0,
156
+ 1.0
157
+ ]
158
+ }
159
+ },
160
+ "max_state_dim": 32,
161
+ "max_action_dim": 32,
162
+ "resize_imgs_with_padding": [
163
+ 512,
164
+ 512
165
+ ],
166
+ "empty_cameras": 0,
167
+ "adapt_to_pi_aloha": false,
168
+ "use_delta_joint_actions_aloha": false,
169
+ "tokenizer_max_length": 48,
170
+ "num_steps": 10,
171
+ "use_cache": true,
172
+ "freeze_vision_encoder": false,
173
+ "train_expert_only": true,
174
+ "train_state_proj": true,
175
+ "optimizer_lr": 0.0001,
176
+ "optimizer_betas": [
177
+ 0.9,
178
+ 0.95
179
+ ],
180
+ "optimizer_eps": 1e-08,
181
+ "optimizer_weight_decay": 1e-10,
182
+ "optimizer_grad_clip_norm": 10,
183
+ "scheduler_warmup_steps": 1000,
184
+ "scheduler_decay_steps": 30000,
185
+ "scheduler_decay_lr": 2.5e-06,
186
+ "vlm_model_name": "HuggingFaceTB/SmolVLM2-500M-Video-Instruct",
187
+ "load_vlm_weights": false,
188
+ "add_image_special_tokens": false,
189
+ "attention_mode": "cross_attn",
190
+ "prefix_length": -1,
191
+ "pad_language_to": "longest",
192
+ "num_expert_layers": -1,
193
+ "num_vlm_layers": 16,
194
+ "self_attn_every_n_layers": 2,
195
+ "expert_width_multiplier": 0.75,
196
+ "min_period": 0.004,
197
+ "max_period": 4.0
198
+ },
199
+ "output_dir": "outputs\\train\\2025-07-21\\13-41-16_smolvla",
200
+ "job_name": "smolvla",
201
+ "resume": false,
202
+ "seed": 1000,
203
+ "num_workers": 4,
204
+ "batch_size": 8,
205
+ "steps": 500,
206
+ "eval_freq": 250,
207
+ "log_freq": 10,
208
+ "save_checkpoint": true,
209
+ "save_freq": 250,
210
+ "use_policy_training_preset": true,
211
+ "optimizer": {
212
+ "type": "adamw",
213
+ "lr": 0.0001,
214
+ "weight_decay": 1e-10,
215
+ "grad_clip_norm": 10,
216
+ "betas": [
217
+ 0.9,
218
+ 0.95
219
+ ],
220
+ "eps": 1e-08
221
+ },
222
+ "scheduler": {
223
+ "type": "cosine_decay_with_warmup",
224
+ "num_warmup_steps": 1000,
225
+ "num_decay_steps": 30000,
226
+ "peak_lr": 0.0001,
227
+ "decay_lr": 2.5e-06
228
+ },
229
+ "eval": {
230
+ "n_episodes": 50,
231
+ "batch_size": 50,
232
+ "use_async_envs": false
233
+ },
234
+ "wandb": {
235
+ "enable": true,
236
+ "disable_artifact": false,
237
+ "project": "smolVLA_robotic_scanner",
238
+ "entity": null,
239
+ "notes": null,
240
+ "run_id": "ah4n50ar",
241
+ "mode": null
242
+ }
243
+ }