Hugging Face's logo

andabi
/
M11

Robotics
Safetensors
lerobot
act
model_hub_mixin
pytorch_model_hub_mixin
Model card Files Community
andabi commited on
Commit
0b2f574
·
verified ·
1 Parent(s): 28d7e8a

Upload folder using huggingface_hub

Browse files
Files changed (4) hide show
  1. README.md +12 -0
  2. config.json +55 -0
  3. config.yaml +180 -0
  4. model.safetensors +3 -0
README.md ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: lerobot
3
+ tags:
4
+ - act
5
+ - model_hub_mixin
6
+ - pytorch_model_hub_mixin
7
+ - robotics
8
+ ---
9
+
10
+ This model has been pushed to the Hub using the [PytorchModelHubMixin](https://huggingface.co/docs/huggingface_hub/package_reference/mixins#huggingface_hub.PyTorchModelHubMixin) integration:
11
+ - Library: https://github.com/huggingface/lerobot
12
+ - Docs: [More Information Needed]
config.json ADDED
@@ -0,0 +1,55 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "chunk_size": 30,
3
+ "dim_feedforward": 3200,
4
+ "dim_model": 512,
5
+ "dropout": 0.1,
6
+ "feedforward_activation": "relu",
7
+ "input_normalization_modes": {
8
+ "observation.images.bird": "mean_std",
9
+ "observation.images.wrist_left": "mean_std",
10
+ "observation.images.wrist_right": "mean_std",
11
+ "observation.state": "mean_std"
12
+ },
13
+ "input_shapes": {
14
+ "observation.images.bird": [
15
+ 3,
16
+ 480,
17
+ 640
18
+ ],
19
+ "observation.images.wrist_left": [
20
+ 3,
21
+ 480,
22
+ 640
23
+ ],
24
+ "observation.images.wrist_right": [
25
+ 3,
26
+ 480,
27
+ 640
28
+ ],
29
+ "observation.state": [
30
+ 12
31
+ ]
32
+ },
33
+ "kl_weight": 10.0,
34
+ "latent_dim": 32,
35
+ "n_action_steps": 30,
36
+ "n_decoder_layers": 1,
37
+ "n_encoder_layers": 4,
38
+ "n_heads": 8,
39
+ "n_obs_steps": 1,
40
+ "n_vae_encoder_layers": 4,
41
+ "output_normalization_modes": {
42
+ "action": "mean_std"
43
+ },
44
+ "output_shapes": {
45
+ "action": [
46
+ 12
47
+ ]
48
+ },
49
+ "pre_norm": false,
50
+ "pretrained_backbone_weights": "ResNet18_Weights.IMAGENET1K_V1",
51
+ "replace_final_stride_with_dilation": false,
52
+ "temporal_ensemble_coeff": null,
53
+ "use_vae": true,
54
+ "vision_backbone": "resnet18"
55
+ }
config.yaml ADDED
@@ -0,0 +1,180 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ resume: false
2
+ device: cuda
3
+ use_amp: false
4
+ seed: 1000
5
+ dataset_repo_id:
6
+ - andabi/D8
7
+ - andabi/D12
8
+ video_backend: pyav
9
+ training:
10
+ offline_steps: 200000
11
+ num_workers: 4
12
+ batch_size: 16
13
+ eval_freq: -1
14
+ log_freq: 100
15
+ save_checkpoint: true
16
+ save_freq: 10000
17
+ online_steps: 0
18
+ online_rollout_n_episodes: 1
19
+ online_rollout_batch_size: 1
20
+ online_steps_between_rollouts: 1
21
+ online_sampling_ratio: 0.5
22
+ online_env_seed: null
23
+ online_buffer_capacity: null
24
+ online_buffer_seed_size: 0
25
+ do_online_rollout_async: false
26
+ image_transforms:
27
+ enable: false
28
+ max_num_transforms: 3
29
+ random_order: false
30
+ brightness:
31
+ weight: 1
32
+ min_max:
33
+ - 0.8
34
+ - 1.2
35
+ contrast:
36
+ weight: 1
37
+ min_max:
38
+ - 0.8
39
+ - 1.2
40
+ saturation:
41
+ weight: 1
42
+ min_max:
43
+ - 0.5
44
+ - 1.5
45
+ hue:
46
+ weight: 1
47
+ min_max:
48
+ - -0.05
49
+ - 0.05
50
+ sharpness:
51
+ weight: 1
52
+ min_max:
53
+ - 0.8
54
+ - 1.2
55
+ lr: 1.0e-05
56
+ lr_backbone: 1.0e-05
57
+ weight_decay: 0.0001
58
+ grad_clip_norm: 10
59
+ delta_timestamps:
60
+ action:
61
+ - 0.0
62
+ - 0.06666666666666667
63
+ - 0.13333333333333333
64
+ - 0.2
65
+ - 0.26666666666666666
66
+ - 0.3333333333333333
67
+ - 0.4
68
+ - 0.4666666666666667
69
+ - 0.5333333333333333
70
+ - 0.6
71
+ - 0.6666666666666666
72
+ - 0.7333333333333333
73
+ - 0.8
74
+ - 0.8666666666666667
75
+ - 0.9333333333333333
76
+ - 1.0
77
+ - 1.0666666666666667
78
+ - 1.1333333333333333
79
+ - 1.2
80
+ - 1.2666666666666666
81
+ - 1.3333333333333333
82
+ - 1.4
83
+ - 1.4666666666666666
84
+ - 1.5333333333333334
85
+ - 1.6
86
+ - 1.6666666666666667
87
+ - 1.7333333333333334
88
+ - 1.8
89
+ - 1.8666666666666667
90
+ - 1.9333333333333333
91
+ eval:
92
+ n_episodes: 50
93
+ batch_size: 50
94
+ use_async_envs: false
95
+ wandb:
96
+ enable: false
97
+ disable_artifact: false
98
+ project: lerobot
99
+ notes: ''
100
+ fps: 15
101
+ env:
102
+ name: real_world
103
+ task: null
104
+ state_dim: 12
105
+ action_dim: 12
106
+ fps: ${fps}
107
+ override_dataset_stats:
108
+ observation.images.bird:
109
+ mean:
110
+ - - - 0.485
111
+ - - - 0.456
112
+ - - - 0.406
113
+ std:
114
+ - - - 0.229
115
+ - - - 0.224
116
+ - - - 0.225
117
+ observation.images.wrist_left:
118
+ mean:
119
+ - - - 0.485
120
+ - - - 0.456
121
+ - - - 0.406
122
+ std:
123
+ - - - 0.229
124
+ - - - 0.224
125
+ - - - 0.225
126
+ observation.images.wrist_right:
127
+ mean:
128
+ - - - 0.485
129
+ - - - 0.456
130
+ - - - 0.406
131
+ std:
132
+ - - - 0.229
133
+ - - - 0.224
134
+ - - - 0.225
135
+ policy:
136
+ name: act
137
+ n_obs_steps: 1
138
+ chunk_size: 30
139
+ n_action_steps: 30
140
+ input_shapes:
141
+ observation.images.bird:
142
+ - 3
143
+ - 480
144
+ - 640
145
+ observation.images.wrist_left:
146
+ - 3
147
+ - 480
148
+ - 640
149
+ observation.images.wrist_right:
150
+ - 3
151
+ - 480
152
+ - 640
153
+ observation.state:
154
+ - ${env.state_dim}
155
+ output_shapes:
156
+ action:
157
+ - ${env.action_dim}
158
+ input_normalization_modes:
159
+ observation.images.bird: mean_std
160
+ observation.images.wrist_left: mean_std
161
+ observation.images.wrist_right: mean_std
162
+ observation.state: mean_std
163
+ output_normalization_modes:
164
+ action: mean_std
165
+ vision_backbone: resnet18
166
+ pretrained_backbone_weights: ResNet18_Weights.IMAGENET1K_V1
167
+ replace_final_stride_with_dilation: false
168
+ pre_norm: false
169
+ dim_model: 512
170
+ n_heads: 8
171
+ dim_feedforward: 3200
172
+ feedforward_activation: relu
173
+ n_encoder_layers: 4
174
+ n_decoder_layers: 1
175
+ use_vae: true
176
+ latent_dim: 32
177
+ n_vae_encoder_layers: 4
178
+ temporal_ensemble_coeff: null
179
+ dropout: 0.1
180
+ kl_weight: 10.0
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bedab516953d15b9b2ecb6298c21ba517c4e2aa0fde9812868c3acb81bfd02ac
3
+ size 206463968