Hugging Face's logo

shuohsuan
/
act_grasp_multi_base

Safetensors
Model card Files
xet
 Community
shuohsuan commited on
Commit
b98be23
·
verified ·
1 Parent(s): ec5bb4d

Upload folder using huggingface_hub

Browse files
Files changed (3) hide show
  1. config.json +64 -0
  2. model.safetensors +3 -0
  3. train_config.json +177 -0
config.json ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "act",
3
+ "n_obs_steps": 1,
4
+ "normalization_mapping": {
5
+ "VISUAL": "MEAN_STD",
6
+ "STATE": "MEAN_STD",
7
+ "ACTION": "MEAN_STD"
8
+ },
9
+ "input_features": {
10
+ "observation.state": {
11
+ "type": "STATE",
12
+ "shape": [
13
+ 6
14
+ ]
15
+ },
16
+ "observation.images.laptop": {
17
+ "type": "VISUAL",
18
+ "shape": [
19
+ 3,
20
+ 480,
21
+ 640
22
+ ]
23
+ },
24
+ "observation.images.phone": {
25
+ "type": "VISUAL",
26
+ "shape": [
27
+ 3,
28
+ 480,
29
+ 640
30
+ ]
31
+ }
32
+ },
33
+ "output_features": {
34
+ "action": {
35
+ "type": "ACTION",
36
+ "shape": [
37
+ 6
38
+ ]
39
+ }
40
+ },
41
+ "device": "cuda",
42
+ "use_amp": false,
43
+ "chunk_size": 100,
44
+ "n_action_steps": 100,
45
+ "vision_backbone": "resnet18",
46
+ "pretrained_backbone_weights": "ResNet18_Weights.IMAGENET1K_V1",
47
+ "replace_final_stride_with_dilation": false,
48
+ "pre_norm": false,
49
+ "dim_model": 512,
50
+ "n_heads": 8,
51
+ "dim_feedforward": 3200,
52
+ "feedforward_activation": "relu",
53
+ "n_encoder_layers": 4,
54
+ "n_decoder_layers": 1,
55
+ "use_vae": true,
56
+ "latent_dim": 32,
57
+ "n_vae_encoder_layers": 4,
58
+ "temporal_ensemble_coeff": null,
59
+ "dropout": 0.1,
60
+ "kl_weight": 10.0,
61
+ "optimizer_lr": 1e-05,
62
+ "optimizer_weight_decay": 0.0001,
63
+ "optimizer_lr_backbone": 1e-05
64
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:802e7ab84eca2a845c8372703422c4529097fec41f7534e8bfee68327214a1f1
3
+ size 206701072
train_config.json ADDED
@@ -0,0 +1,177 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": [
4
+ "shuohsuan/grasp_data_ts_20",
5
+ "shuohsuan/grasp_data_ts_30",
6
+ "shuohsuan/grasp_data_ts_50",
7
+ "shuohsuan/grasp_data_tg_10",
8
+ "shuohsuan/grasp_data_tg_50",
9
+ "shuohsuan/grasp_data_tg_51"
10
+ ],
11
+ "root": null,
12
+ "episodes": null,
13
+ "image_transforms": {
14
+ "enable": false,
15
+ "max_num_transforms": 3,
16
+ "random_order": false,
17
+ "tfs": {
18
+ "brightness": {
19
+ "weight": 1.0,
20
+ "type": "ColorJitter",
21
+ "kwargs": {
22
+ "brightness": [
23
+ 0.8,
24
+ 1.2
25
+ ]
26
+ }
27
+ },
28
+ "contrast": {
29
+ "weight": 1.0,
30
+ "type": "ColorJitter",
31
+ "kwargs": {
32
+ "contrast": [
33
+ 0.8,
34
+ 1.2
35
+ ]
36
+ }
37
+ },
38
+ "saturation": {
39
+ "weight": 1.0,
40
+ "type": "ColorJitter",
41
+ "kwargs": {
42
+ "saturation": [
43
+ 0.5,
44
+ 1.5
45
+ ]
46
+ }
47
+ },
48
+ "hue": {
49
+ "weight": 1.0,
50
+ "type": "ColorJitter",
51
+ "kwargs": {
52
+ "hue": [
53
+ -0.05,
54
+ 0.05
55
+ ]
56
+ }
57
+ },
58
+ "sharpness": {
59
+ "weight": 1.0,
60
+ "type": "SharpnessJitter",
61
+ "kwargs": {
62
+ "sharpness": [
63
+ 0.5,
64
+ 1.5
65
+ ]
66
+ }
67
+ }
68
+ }
69
+ },
70
+ "revision": null,
71
+ "use_imagenet_stats": true,
72
+ "video_backend": "torchcodec"
73
+ },
74
+ "env": null,
75
+ "policy": {
76
+ "type": "act",
77
+ "n_obs_steps": 1,
78
+ "normalization_mapping": {
79
+ "VISUAL": "MEAN_STD",
80
+ "STATE": "MEAN_STD",
81
+ "ACTION": "MEAN_STD"
82
+ },
83
+ "input_features": {
84
+ "observation.state": {
85
+ "type": "STATE",
86
+ "shape": [
87
+ 6
88
+ ]
89
+ },
90
+ "observation.images.laptop": {
91
+ "type": "VISUAL",
92
+ "shape": [
93
+ 3,
94
+ 480,
95
+ 640
96
+ ]
97
+ },
98
+ "observation.images.phone": {
99
+ "type": "VISUAL",
100
+ "shape": [
101
+ 3,
102
+ 480,
103
+ 640
104
+ ]
105
+ }
106
+ },
107
+ "output_features": {
108
+ "action": {
109
+ "type": "ACTION",
110
+ "shape": [
111
+ 6
112
+ ]
113
+ }
114
+ },
115
+ "device": "cuda",
116
+ "use_amp": false,
117
+ "chunk_size": 100,
118
+ "n_action_steps": 100,
119
+ "vision_backbone": "resnet18",
120
+ "pretrained_backbone_weights": "ResNet18_Weights.IMAGENET1K_V1",
121
+ "replace_final_stride_with_dilation": false,
122
+ "pre_norm": false,
123
+ "dim_model": 512,
124
+ "n_heads": 8,
125
+ "dim_feedforward": 3200,
126
+ "feedforward_activation": "relu",
127
+ "n_encoder_layers": 4,
128
+ "n_decoder_layers": 1,
129
+ "use_vae": true,
130
+ "latent_dim": 32,
131
+ "n_vae_encoder_layers": 4,
132
+ "temporal_ensemble_coeff": null,
133
+ "dropout": 0.1,
134
+ "kl_weight": 10.0,
135
+ "optimizer_lr": 1e-05,
136
+ "optimizer_weight_decay": 0.0001,
137
+ "optimizer_lr_backbone": 1e-05
138
+ },
139
+ "output_dir": "/mnt/storage/outputs/train/act_grasp_multi_base",
140
+ "job_name": "act",
141
+ "resume": false,
142
+ "seed": 1000,
143
+ "num_workers": 4,
144
+ "batch_size": 8,
145
+ "steps": 200000,
146
+ "eval_freq": 20000,
147
+ "log_freq": 200,
148
+ "save_checkpoint": true,
149
+ "save_freq": 20000,
150
+ "use_policy_training_preset": true,
151
+ "optimizer": {
152
+ "type": "adamw",
153
+ "lr": 1e-05,
154
+ "weight_decay": 0.0001,
155
+ "grad_clip_norm": 10.0,
156
+ "betas": [
157
+ 0.9,
158
+ 0.999
159
+ ],
160
+ "eps": 1e-08
161
+ },
162
+ "scheduler": null,
163
+ "eval": {
164
+ "n_episodes": 50,
165
+ "batch_size": 50,
166
+ "use_async_envs": false
167
+ },
168
+ "wandb": {
169
+ "enable": false,
170
+ "disable_artifact": false,
171
+ "project": "lerobot",
172
+ "entity": null,
173
+ "notes": null,
174
+ "run_id": null,
175
+ "mode": null
176
+ }
177
+ }