Hugging Face's logo

Tsagkas
/
lafa_v2

Robotics
LeRobot
Safetensors
lafa
Model card Files
xet
 Community
Tsagkas commited on
Commit
8bd151a
·
verified ·
1 Parent(s): a8d62b9

Upload policy weights, train config and readme

Browse files
Files changed (4) hide show
  1. README.md +62 -0
  2. config.json +111 -0
  3. model.safetensors +3 -0
  4. train_config.json +234 -0
README.md ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ datasets: Tsagkas/dataset-test-v2
3
+ library_name: lerobot
4
+ license: apache-2.0
5
+ model_name: lafa
6
+ pipeline_tag: robotics
7
+ tags:
8
+ - robotics
9
+ - lerobot
10
+ - lafa
11
+ ---
12
+
13
+ # Model Card for lafa
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 +1,111 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "lafa",
3
+ "n_obs_steps": 1,
4
+ "input_features": {
5
+ "observation.state": {
6
+ "type": "STATE",
7
+ "shape": [
8
+ 7
9
+ ]
10
+ },
11
+ "observation.images.wrist": {
12
+ "type": "VISUAL",
13
+ "shape": [
14
+ 3,
15
+ 256,
16
+ 256
17
+ ]
18
+ },
19
+ "observation.images.front": {
20
+ "type": "VISUAL",
21
+ "shape": [
22
+ 3,
23
+ 256,
24
+ 256
25
+ ]
26
+ },
27
+ "observation.vfm_features.wrist": {
28
+ "type": "STATE",
29
+ "shape": [
30
+ 16,
31
+ 16,
32
+ 768
33
+ ]
34
+ },
35
+ "observation.vfm_features.front": {
36
+ "type": "STATE",
37
+ "shape": [
38
+ 16,
39
+ 16,
40
+ 768
41
+ ]
42
+ },
43
+ "observation.skill_feature": {
44
+ "type": "STATE",
45
+ "shape": [
46
+ 1024
47
+ ]
48
+ },
49
+ "observation.object_feature": {
50
+ "type": "STATE",
51
+ "shape": [
52
+ 1024
53
+ ]
54
+ },
55
+ "observation.vlm_features.wrist": {
56
+ "type": "STATE",
57
+ "shape": [
58
+ 16,
59
+ 16,
60
+ 1024
61
+ ]
62
+ },
63
+ "observation.vlm_features.front": {
64
+ "type": "STATE",
65
+ "shape": [
66
+ 16,
67
+ 16,
68
+ 1024
69
+ ]
70
+ }
71
+ },
72
+ "output_features": {
73
+ "action": {
74
+ "type": "ACTION",
75
+ "shape": [
76
+ 7
77
+ ]
78
+ }
79
+ },
80
+ "device": "cuda",
81
+ "use_amp": false,
82
+ "push_to_hub": true,
83
+ "repo_id": "Tsagkas/lafa_v2",
84
+ "private": null,
85
+ "tags": null,
86
+ "license": null,
87
+ "pretrained_path": null,
88
+ "feature_map_shape": [
89
+ 16,
90
+ 16,
91
+ 768
92
+ ],
93
+ "state_dim": 7,
94
+ "action_dim": 7,
95
+ "chunk_size": 100,
96
+ "n_action_steps": 1,
97
+ "temporal_ensemble_coeff": 0.01,
98
+ "normalization_mapping": {
99
+ "VISUAL": "IDENTITY",
100
+ "STATE": "MEAN_STD",
101
+ "ACTION": "MEAN_STD"
102
+ },
103
+ "obj_text_dim": 1024,
104
+ "dim_model": 768,
105
+ "n_heads": 8,
106
+ "mlp_hidden_dim": 512,
107
+ "mlp_num_layers": 5,
108
+ "dropout": 0.1,
109
+ "optimizer_lr": 0.0001,
110
+ "optimizer_weight_decay": 0.0001
111
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5f8a979ebb1ee1d36a562a3f674ef1c7aa07afc71ba4fe39a8f4750cdd45f0e4
3
+ size 20352256
train_config.json ADDED
@@ -0,0 +1,234 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "Tsagkas/dataset-test-v2",
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": "lafa",
85
+ "n_obs_steps": 1,
86
+ "input_features": {
87
+ "observation.state": {
88
+ "type": "STATE",
89
+ "shape": [
90
+ 7
91
+ ]
92
+ },
93
+ "observation.images.wrist": {
94
+ "type": "VISUAL",
95
+ "shape": [
96
+ 3,
97
+ 256,
98
+ 256
99
+ ]
100
+ },
101
+ "observation.images.front": {
102
+ "type": "VISUAL",
103
+ "shape": [
104
+ 3,
105
+ 256,
106
+ 256
107
+ ]
108
+ },
109
+ "observation.vfm_features.wrist": {
110
+ "type": "STATE",
111
+ "shape": [
112
+ 16,
113
+ 16,
114
+ 768
115
+ ]
116
+ },
117
+ "observation.vfm_features.front": {
118
+ "type": "STATE",
119
+ "shape": [
120
+ 16,
121
+ 16,
122
+ 768
123
+ ]
124
+ },
125
+ "observation.skill_feature": {
126
+ "type": "STATE",
127
+ "shape": [
128
+ 1024
129
+ ]
130
+ },
131
+ "observation.object_feature": {
132
+ "type": "STATE",
133
+ "shape": [
134
+ 1024
135
+ ]
136
+ },
137
+ "observation.vlm_features.wrist": {
138
+ "type": "STATE",
139
+ "shape": [
140
+ 16,
141
+ 16,
142
+ 1024
143
+ ]
144
+ },
145
+ "observation.vlm_features.front": {
146
+ "type": "STATE",
147
+ "shape": [
148
+ 16,
149
+ 16,
150
+ 1024
151
+ ]
152
+ }
153
+ },
154
+ "output_features": {
155
+ "action": {
156
+ "type": "ACTION",
157
+ "shape": [
158
+ 7
159
+ ]
160
+ }
161
+ },
162
+ "device": "cuda",
163
+ "use_amp": false,
164
+ "push_to_hub": true,
165
+ "repo_id": "Tsagkas/lafa_v2",
166
+ "private": null,
167
+ "tags": null,
168
+ "license": null,
169
+ "pretrained_path": null,
170
+ "feature_map_shape": [
171
+ 16,
172
+ 16,
173
+ 768
174
+ ],
175
+ "state_dim": 7,
176
+ "action_dim": 7,
177
+ "chunk_size": 100,
178
+ "n_action_steps": 1,
179
+ "temporal_ensemble_coeff": 0.01,
180
+ "normalization_mapping": {
181
+ "VISUAL": "IDENTITY",
182
+ "STATE": "MEAN_STD",
183
+ "ACTION": "MEAN_STD"
184
+ },
185
+ "obj_text_dim": 1024,
186
+ "dim_model": 768,
187
+ "n_heads": 8,
188
+ "mlp_hidden_dim": 512,
189
+ "mlp_num_layers": 5,
190
+ "dropout": 0.1,
191
+ "optimizer_lr": 0.0001,
192
+ "optimizer_weight_decay": 0.0001
193
+ },
194
+ "output_dir": "outputs/train/lafa_v2",
195
+ "job_name": "lafa_v2",
196
+ "resume": false,
197
+ "seed": 1000,
198
+ "num_workers": 16,
199
+ "batch_size": 8,
200
+ "steps": 60000,
201
+ "eval_freq": 20000,
202
+ "log_freq": 10,
203
+ "save_checkpoint": true,
204
+ "save_freq": 30000,
205
+ "use_policy_training_preset": true,
206
+ "optimizer": {
207
+ "type": "adamw",
208
+ "lr": 0.0001,
209
+ "weight_decay": 0.0001,
210
+ "grad_clip_norm": 10.0,
211
+ "betas": [
212
+ 0.9,
213
+ 0.999
214
+ ],
215
+ "eps": 1e-08
216
+ },
217
+ "scheduler": null,
218
+ "eval": {
219
+ "n_episodes": 50,
220
+ "batch_size": 50,
221
+ "use_async_envs": false
222
+ },
223
+ "wandb": {
224
+ "enable": true,
225
+ "disable_artifact": false,
226
+ "project": "lerobot",
227
+ "entity": null,
228
+ "notes": null,
229
+ "run_id": "iqpmd1gj",
230
+ "mode": null
231
+ },
232
+ "checkpoint_path": null,
233
+ "rename_map": {}
234
+ }