Hugging Face's logo

PL2011
/
act_sjadj_lebai-suction-plate_side

Robotics
LeRobot
Safetensors
act_romoya
Model card Files
xet
 Community
PL2011 commited on
Commit
5e826c1
·
verified ·
1 Parent(s): a2640e5

Upload policy weights, train config and readme

Browse files
Files changed (4) hide show
  1. README.md +62 -0
  2. config.json +136 -0
  3. model.safetensors +3 -0
  4. train_config.json +254 -0
README.md ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ datasets: PL2011/lebai-suction-plate
3
+ library_name: lerobot
4
+ license: apache-2.0
5
+ model_name: act_romoya
6
+ pipeline_tag: robotics
7
+ tags:
8
+ - act_romoya
9
+ - lerobot
10
+ - robotics
11
+ ---
12
+
13
+ # Model Card for act_romoya
14
+
15
+ <!-- Provide a quick summary of what the model is/does. -->
16
+
17
+
18
+ _Model type not recognized — please update this template._
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 +1,136 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "act_romoya",
3
+ "n_obs_steps": 1,
4
+ "input_features": {
5
+ "observation.state": {
6
+ "type": "STATE",
7
+ "shape": [
8
+ 8
9
+ ]
10
+ },
11
+ "observation.images.side": {
12
+ "type": "VISUAL",
13
+ "shape": [
14
+ 3,
15
+ 360,
16
+ 640
17
+ ]
18
+ }
19
+ },
20
+ "output_features": {
21
+ "action": {
22
+ "type": "ACTION",
23
+ "shape": [
24
+ 8
25
+ ]
26
+ }
27
+ },
28
+ "device": "cuda",
29
+ "use_amp": false,
30
+ "use_peft": false,
31
+ "push_to_hub": true,
32
+ "repo_id": "PL2011/act_sjadj_lebai-suction-plate_side",
33
+ "private": null,
34
+ "tags": null,
35
+ "license": null,
36
+ "pretrained_path": null,
37
+ "chunk_size": 60,
38
+ "n_action_steps": 5,
39
+ "normalization_mapping": {
40
+ "VISUAL": "MEAN_STD",
41
+ "STATE": "MEAN_STD",
42
+ "ACTION": "MEAN_STD"
43
+ },
44
+ "vision_backbone": "resnet18",
45
+ "pretrained_backbone_weights": "ResNet18_Weights.IMAGENET1K_V1",
46
+ "replace_final_stride_with_dilation": false,
47
+ "pre_norm": false,
48
+ "dim_model": 512,
49
+ "n_heads": 8,
50
+ "dim_feedforward": 3200,
51
+ "feedforward_activation": "relu",
52
+ "n_encoder_layers": 4,
53
+ "n_decoder_layers": 1,
54
+ "use_vae": true,
55
+ "latent_dim": 32,
56
+ "n_vae_encoder_layers": 4,
57
+ "temporal_ensemble_coeff": null,
58
+ "dropout": 0.1,
59
+ "kl_weight": 10.0,
60
+ "optimizer_lr": 1e-05,
61
+ "optimizer_weight_decay": 0.0001,
62
+ "optimizer_lr_backbone": 1e-05,
63
+ "state_feature_names_to_keep": [
64
+ "joint1.pos",
65
+ "joint2.pos",
66
+ "joint3.pos",
67
+ "joint4.pos",
68
+ "joint5.pos",
69
+ "joint6.pos",
70
+ "gripper.pos",
71
+ "DO_1"
72
+ ],
73
+ "action_mode": "delta_joint_gripper_do1",
74
+ "joint_action_names": [
75
+ "joint1.pos",
76
+ "joint2.pos",
77
+ "joint3.pos",
78
+ "joint4.pos",
79
+ "joint5.pos",
80
+ "joint6.pos"
81
+ ],
82
+ "gripper_action_name": "gripper.pos",
83
+ "do_action_names": [
84
+ "DO_1"
85
+ ],
86
+ "raw_observation_state_feature_names": [
87
+ "joint1.pos",
88
+ "joint2.pos",
89
+ "joint3.pos",
90
+ "joint4.pos",
91
+ "joint5.pos",
92
+ "joint6.pos",
93
+ "gripper.pos",
94
+ "gripper.force",
95
+ "DO_0",
96
+ "DO_1",
97
+ "joint1.effort",
98
+ "joint2.effort",
99
+ "joint3.effort",
100
+ "joint4.effort",
101
+ "joint5.effort",
102
+ "joint6.effort",
103
+ "joint1.vel",
104
+ "joint2.vel",
105
+ "joint3.vel",
106
+ "joint4.vel",
107
+ "joint5.vel",
108
+ "joint6.vel",
109
+ "joint1.acc",
110
+ "joint2.acc",
111
+ "joint3.acc",
112
+ "joint4.acc",
113
+ "joint5.acc",
114
+ "joint6.acc",
115
+ "tcp.x",
116
+ "tcp.y",
117
+ "tcp.z",
118
+ "tcp.rx",
119
+ "tcp.ry",
120
+ "tcp.rz"
121
+ ],
122
+ "raw_action_feature_names": [
123
+ "joint1.pos",
124
+ "joint2.pos",
125
+ "joint3.pos",
126
+ "joint4.pos",
127
+ "joint5.pos",
128
+ "joint6.pos",
129
+ "gripper.pos",
130
+ "gripper.force",
131
+ "DO_0",
132
+ "DO_1"
133
+ ],
134
+ "do_threshold": 0.5,
135
+ "sigmoid_do_outputs": false
136
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5bff410352872927175314c04ba54febd04899d6583929f84a4ed38a39018093
3
+ size 206552280
train_config.json ADDED
@@ -0,0 +1,254 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "PL2011/lebai-suction-plate",
4
+ "root": null,
5
+ "episodes": null,
6
+ "image_transforms": {
7
+ "enable": true,
8
+ "max_num_transforms": 2,
9
+ "random_order": false,
10
+ "tfs": {
11
+ "brightness": {
12
+ "weight": 1.0,
13
+ "type": "ColorJitter",
14
+ "kwargs": {
15
+ "brightness": [
16
+ 0.9,
17
+ 1.1
18
+ ]
19
+ }
20
+ },
21
+ "contrast": {
22
+ "weight": 1.0,
23
+ "type": "ColorJitter",
24
+ "kwargs": {
25
+ "contrast": [
26
+ 0.9,
27
+ 1.1
28
+ ]
29
+ }
30
+ },
31
+ "saturation": {
32
+ "weight": 0.5,
33
+ "type": "ColorJitter",
34
+ "kwargs": {
35
+ "saturation": [
36
+ 0.9,
37
+ 1.1
38
+ ]
39
+ }
40
+ },
41
+ "hue": {
42
+ "weight": 0.25,
43
+ "type": "ColorJitter",
44
+ "kwargs": {
45
+ "hue": [
46
+ -0.02,
47
+ 0.02
48
+ ]
49
+ }
50
+ },
51
+ "sharpness": {
52
+ "weight": 0.5,
53
+ "type": "SharpnessJitter",
54
+ "kwargs": {
55
+ "sharpness": [
56
+ 0.9,
57
+ 1.1
58
+ ]
59
+ }
60
+ }
61
+ }
62
+ },
63
+ "revision": null,
64
+ "use_imagenet_stats": true,
65
+ "video_backend": "torchcodec",
66
+ "streaming": false
67
+ },
68
+ "env": null,
69
+ "policy": {
70
+ "type": "act_romoya",
71
+ "n_obs_steps": 1,
72
+ "input_features": {
73
+ "observation.state": {
74
+ "type": "STATE",
75
+ "shape": [
76
+ 8
77
+ ]
78
+ },
79
+ "observation.images.side": {
80
+ "type": "VISUAL",
81
+ "shape": [
82
+ 3,
83
+ 360,
84
+ 640
85
+ ]
86
+ }
87
+ },
88
+ "output_features": {
89
+ "action": {
90
+ "type": "ACTION",
91
+ "shape": [
92
+ 8
93
+ ]
94
+ }
95
+ },
96
+ "device": "cuda",
97
+ "use_amp": false,
98
+ "use_peft": false,
99
+ "push_to_hub": true,
100
+ "repo_id": "PL2011/act_sjadj_lebai-suction-plate_side",
101
+ "private": null,
102
+ "tags": null,
103
+ "license": null,
104
+ "pretrained_path": null,
105
+ "chunk_size": 60,
106
+ "n_action_steps": 5,
107
+ "normalization_mapping": {
108
+ "VISUAL": "MEAN_STD",
109
+ "STATE": "MEAN_STD",
110
+ "ACTION": "MEAN_STD"
111
+ },
112
+ "vision_backbone": "resnet18",
113
+ "pretrained_backbone_weights": "ResNet18_Weights.IMAGENET1K_V1",
114
+ "replace_final_stride_with_dilation": false,
115
+ "pre_norm": false,
116
+ "dim_model": 512,
117
+ "n_heads": 8,
118
+ "dim_feedforward": 3200,
119
+ "feedforward_activation": "relu",
120
+ "n_encoder_layers": 4,
121
+ "n_decoder_layers": 1,
122
+ "use_vae": true,
123
+ "latent_dim": 32,
124
+ "n_vae_encoder_layers": 4,
125
+ "temporal_ensemble_coeff": null,
126
+ "dropout": 0.1,
127
+ "kl_weight": 10.0,
128
+ "optimizer_lr": 1e-05,
129
+ "optimizer_weight_decay": 0.0001,
130
+ "optimizer_lr_backbone": 1e-05,
131
+ "state_feature_names_to_keep": [
132
+ "joint1.pos",
133
+ "joint2.pos",
134
+ "joint3.pos",
135
+ "joint4.pos",
136
+ "joint5.pos",
137
+ "joint6.pos",
138
+ "gripper.pos",
139
+ "DO_1"
140
+ ],
141
+ "action_mode": "delta_joint_gripper_do1",
142
+ "joint_action_names": [
143
+ "joint1.pos",
144
+ "joint2.pos",
145
+ "joint3.pos",
146
+ "joint4.pos",
147
+ "joint5.pos",
148
+ "joint6.pos"
149
+ ],
150
+ "gripper_action_name": "gripper.pos",
151
+ "do_action_names": [
152
+ "DO_1"
153
+ ],
154
+ "raw_observation_state_feature_names": [
155
+ "joint1.pos",
156
+ "joint2.pos",
157
+ "joint3.pos",
158
+ "joint4.pos",
159
+ "joint5.pos",
160
+ "joint6.pos",
161
+ "gripper.pos",
162
+ "gripper.force",
163
+ "DO_0",
164
+ "DO_1",
165
+ "joint1.effort",
166
+ "joint2.effort",
167
+ "joint3.effort",
168
+ "joint4.effort",
169
+ "joint5.effort",
170
+ "joint6.effort",
171
+ "joint1.vel",
172
+ "joint2.vel",
173
+ "joint3.vel",
174
+ "joint4.vel",
175
+ "joint5.vel",
176
+ "joint6.vel",
177
+ "joint1.acc",
178
+ "joint2.acc",
179
+ "joint3.acc",
180
+ "joint4.acc",
181
+ "joint5.acc",
182
+ "joint6.acc",
183
+ "tcp.x",
184
+ "tcp.y",
185
+ "tcp.z",
186
+ "tcp.rx",
187
+ "tcp.ry",
188
+ "tcp.rz"
189
+ ],
190
+ "raw_action_feature_names": [
191
+ "joint1.pos",
192
+ "joint2.pos",
193
+ "joint3.pos",
194
+ "joint4.pos",
195
+ "joint5.pos",
196
+ "joint6.pos",
197
+ "gripper.pos",
198
+ "gripper.force",
199
+ "DO_0",
200
+ "DO_1"
201
+ ],
202
+ "do_threshold": 0.5,
203
+ "sigmoid_do_outputs": false
204
+ },
205
+ "output_dir": "outputs/train/act_sjadj_lebai-suction-plate_side",
206
+ "job_name": "act_sjadj_lebai-suction-plate_side",
207
+ "resume": false,
208
+ "seed": 1000,
209
+ "cudnn_deterministic": false,
210
+ "num_workers": 12,
211
+ "batch_size": 96,
212
+ "steps": 40000,
213
+ "eval_freq": 20000,
214
+ "log_freq": 200,
215
+ "tolerance_s": 0.0001,
216
+ "save_checkpoint": true,
217
+ "save_freq": 20000,
218
+ "use_policy_training_preset": true,
219
+ "optimizer": {
220
+ "type": "adamw",
221
+ "lr": 1e-05,
222
+ "weight_decay": 0.0001,
223
+ "grad_clip_norm": 10.0,
224
+ "betas": [
225
+ 0.9,
226
+ 0.999
227
+ ],
228
+ "eps": 1e-08
229
+ },
230
+ "scheduler": null,
231
+ "eval": {
232
+ "n_episodes": 50,
233
+ "batch_size": 50,
234
+ "use_async_envs": false
235
+ },
236
+ "wandb": {
237
+ "enable": true,
238
+ "disable_artifact": false,
239
+ "project": "lerobot",
240
+ "entity": null,
241
+ "notes": null,
242
+ "run_id": "fbyi6f0l",
243
+ "mode": null,
244
+ "add_tags": true
245
+ },
246
+ "peft": null,
247
+ "use_rabc": false,
248
+ "rabc_progress_path": null,
249
+ "rabc_kappa": 0.01,
250
+ "rabc_epsilon": 1e-06,
251
+ "rabc_head_mode": "sparse",
252
+ "rename_map": {},
253
+ "checkpoint_path": null
254
+ }