Hugging Face's logo

fracapuano
/
inhand-orienting

Robotics
LeRobot
Safetensors
act
Model card Files
xet
 Community
fracapuano commited on
Commit
8690041
·
verified ·
1 Parent(s): fff60b6

Upload policy weights, train config and readme

Browse files
Files changed (4) hide show
  1. README.md +62 -0
  2. config.json +83 -0
  3. model.safetensors +3 -0
  4. train_config.json +213 -0
README.md ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ datasets: fracapuano/inhand-orientation
3
+ library_name: lerobot
4
+ license: apache-2.0
5
+ model_name: act
6
+ pipeline_tag: robotics
7
+ tags:
8
+ - lerobot
9
+ - robotics
10
+ - act
11
+ ---
12
+
13
+ # Model Card for act
14
+
15
+ <!-- Provide a quick summary of what the model is/does. -->
16
+
17
+
18
+ [Action Chunking with Transformers (ACT)](https://huggingface.co/papers/2304.13705) is an imitation-learning method that predicts short action chunks instead of single steps. It learns from teleoperated data and often achieves high success rates.
19
+
20
+
21
+ This policy has been trained and pushed to the Hub using [LeRobot](https://github.com/huggingface/lerobot).
22
+ See the full documentation at [LeRobot Docs](https://huggingface.co/docs/lerobot/index).
23
+
24
+ ---
25
+
26
+ ## How to Get Started with the Model
27
+
28
+ For a complete walkthrough, see the [training guide](https://huggingface.co/docs/lerobot/il_robots#train-a-policy).
29
+ Below is the short version on how to train and run inference/eval:
30
+
31
+ ### Train from scratch
32
+
33
+ ```bash
34
+ lerobot-train \
35
+ --dataset.repo_id=${HF_USER}/<dataset> \
36
+ --policy.type=act \
37
+ --output_dir=outputs/train/<desired_policy_repo_id> \
38
+ --job_name=lerobot_training \
39
+ --policy.device=cuda \
40
+ --policy.repo_id=${HF_USER}/<desired_policy_repo_id>
41
+ --wandb.enable=true
42
+ ```
43
+
44
+ _Writes checkpoints to `outputs/train/<desired_policy_repo_id>/checkpoints/`._
45
+
46
+ ### Evaluate the policy/run inference
47
+
48
+ ```bash
49
+ lerobot-record \
50
+ --robot.type=so100_follower \
51
+ --dataset.repo_id=<hf_user>/eval_<dataset> \
52
+ --policy.path=<hf_user>/<desired_policy_repo_id> \
53
+ --episodes=10
54
+ ```
55
+
56
+ Prefix the dataset repo with **eval\_** and supply `--policy.path` pointing to a local or hub checkpoint.
57
+
58
+ ---
59
+
60
+ ## Model Details
61
+
62
+ - **License:** apache-2.0
config.json ADDED
@@ -0,0 +1,83 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "act",
3
+ "n_obs_steps": 1,
4
+ "input_features": {
5
+ "observation.state": {
6
+ "type": "STATE",
7
+ "shape": [
8
+ 17
9
+ ]
10
+ },
11
+ "observation.cube_pos": {
12
+ "type": "STATE",
13
+ "shape": [
14
+ 3
15
+ ]
16
+ },
17
+ "observation.cube_quat": {
18
+ "type": "STATE",
19
+ "shape": [
20
+ 4
21
+ ]
22
+ },
23
+ "observation.images.wrist_camera": {
24
+ "type": "VISUAL",
25
+ "shape": [
26
+ 3,
27
+ 240,
28
+ 320
29
+ ]
30
+ },
31
+ "observation.images.topdown": {
32
+ "type": "VISUAL",
33
+ "shape": [
34
+ 3,
35
+ 240,
36
+ 320
37
+ ]
38
+ }
39
+ },
40
+ "output_features": {
41
+ "action": {
42
+ "type": "ACTION",
43
+ "shape": [
44
+ 17
45
+ ]
46
+ }
47
+ },
48
+ "device": "mps",
49
+ "use_amp": false,
50
+ "use_peft": false,
51
+ "push_to_hub": true,
52
+ "repo_id": "fracapuano/inhand-orienting",
53
+ "private": null,
54
+ "tags": null,
55
+ "license": null,
56
+ "pretrained_path": null,
57
+ "chunk_size": 100,
58
+ "n_action_steps": 100,
59
+ "normalization_mapping": {
60
+ "VISUAL": "MEAN_STD",
61
+ "STATE": "MEAN_STD",
62
+ "ACTION": "MEAN_STD"
63
+ },
64
+ "vision_backbone": "resnet18",
65
+ "pretrained_backbone_weights": "ResNet18_Weights.IMAGENET1K_V1",
66
+ "replace_final_stride_with_dilation": false,
67
+ "pre_norm": false,
68
+ "dim_model": 512,
69
+ "n_heads": 8,
70
+ "dim_feedforward": 3200,
71
+ "feedforward_activation": "relu",
72
+ "n_encoder_layers": 4,
73
+ "n_decoder_layers": 1,
74
+ "use_vae": true,
75
+ "latent_dim": 32,
76
+ "n_vae_encoder_layers": 4,
77
+ "temporal_ensemble_coeff": null,
78
+ "dropout": 0.1,
79
+ "kl_weight": 10.0,
80
+ "optimizer_lr": 1e-05,
81
+ "optimizer_weight_decay": 0.0001,
82
+ "optimizer_lr_backbone": 1e-05
83
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4d2e2c5469d1a511397ea5a054aa23b2273dd42cdb8f55565cc98db3ffa778a8
3
+ size 206789892
train_config.json ADDED
@@ -0,0 +1,213 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "fracapuano/inhand-orientation",
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
+ "affine": {
62
+ "weight": 1.0,
63
+ "type": "RandomAffine",
64
+ "kwargs": {
65
+ "degrees": [
66
+ -5.0,
67
+ 5.0
68
+ ],
69
+ "translate": [
70
+ 0.05,
71
+ 0.05
72
+ ]
73
+ }
74
+ }
75
+ }
76
+ },
77
+ "revision": null,
78
+ "use_imagenet_stats": true,
79
+ "video_backend": "torchcodec",
80
+ "streaming": false
81
+ },
82
+ "env": null,
83
+ "policy": {
84
+ "type": "act",
85
+ "n_obs_steps": 1,
86
+ "input_features": {
87
+ "observation.state": {
88
+ "type": "STATE",
89
+ "shape": [
90
+ 17
91
+ ]
92
+ },
93
+ "observation.cube_pos": {
94
+ "type": "STATE",
95
+ "shape": [
96
+ 3
97
+ ]
98
+ },
99
+ "observation.cube_quat": {
100
+ "type": "STATE",
101
+ "shape": [
102
+ 4
103
+ ]
104
+ },
105
+ "observation.images.wrist_camera": {
106
+ "type": "VISUAL",
107
+ "shape": [
108
+ 3,
109
+ 240,
110
+ 320
111
+ ]
112
+ },
113
+ "observation.images.topdown": {
114
+ "type": "VISUAL",
115
+ "shape": [
116
+ 3,
117
+ 240,
118
+ 320
119
+ ]
120
+ }
121
+ },
122
+ "output_features": {
123
+ "action": {
124
+ "type": "ACTION",
125
+ "shape": [
126
+ 17
127
+ ]
128
+ }
129
+ },
130
+ "device": "mps",
131
+ "use_amp": false,
132
+ "use_peft": false,
133
+ "push_to_hub": true,
134
+ "repo_id": "fracapuano/inhand-orienting",
135
+ "private": null,
136
+ "tags": null,
137
+ "license": null,
138
+ "pretrained_path": null,
139
+ "chunk_size": 100,
140
+ "n_action_steps": 100,
141
+ "normalization_mapping": {
142
+ "VISUAL": "MEAN_STD",
143
+ "STATE": "MEAN_STD",
144
+ "ACTION": "MEAN_STD"
145
+ },
146
+ "vision_backbone": "resnet18",
147
+ "pretrained_backbone_weights": "ResNet18_Weights.IMAGENET1K_V1",
148
+ "replace_final_stride_with_dilation": false,
149
+ "pre_norm": false,
150
+ "dim_model": 512,
151
+ "n_heads": 8,
152
+ "dim_feedforward": 3200,
153
+ "feedforward_activation": "relu",
154
+ "n_encoder_layers": 4,
155
+ "n_decoder_layers": 1,
156
+ "use_vae": true,
157
+ "latent_dim": 32,
158
+ "n_vae_encoder_layers": 4,
159
+ "temporal_ensemble_coeff": null,
160
+ "dropout": 0.1,
161
+ "kl_weight": 10.0,
162
+ "optimizer_lr": 1e-05,
163
+ "optimizer_weight_decay": 0.0001,
164
+ "optimizer_lr_backbone": 1e-05
165
+ },
166
+ "output_dir": "runs/inhand-act-v1",
167
+ "job_name": "act",
168
+ "resume": false,
169
+ "seed": 1000,
170
+ "num_workers": 4,
171
+ "batch_size": 8,
172
+ "steps": 50000,
173
+ "eval_freq": 20000,
174
+ "log_freq": 200,
175
+ "tolerance_s": 0.0001,
176
+ "save_checkpoint": true,
177
+ "save_freq": 20000,
178
+ "use_policy_training_preset": true,
179
+ "optimizer": {
180
+ "type": "adamw",
181
+ "lr": 1e-05,
182
+ "weight_decay": 0.0001,
183
+ "grad_clip_norm": 10.0,
184
+ "betas": [
185
+ 0.9,
186
+ 0.999
187
+ ],
188
+ "eps": 1e-08
189
+ },
190
+ "scheduler": null,
191
+ "eval": {
192
+ "n_episodes": 50,
193
+ "batch_size": 50,
194
+ "use_async_envs": false
195
+ },
196
+ "wandb": {
197
+ "enable": true,
198
+ "disable_artifact": false,
199
+ "project": "lerobot",
200
+ "entity": null,
201
+ "notes": null,
202
+ "run_id": "8sn0yfcr",
203
+ "mode": null
204
+ },
205
+ "peft": null,
206
+ "use_rabc": false,
207
+ "rabc_progress_path": null,
208
+ "rabc_kappa": 0.01,
209
+ "rabc_epsilon": 1e-06,
210
+ "rabc_head_mode": "sparse",
211
+ "rename_map": {},
212
+ "checkpoint_path": null
213
+ }