Hugging Face's logo

danielsanjosepro
/
ditflow_drawer_v2

Robotics
LeRobot
Safetensors
ditflow
Model card Files
xet
Community
danielsanjosepro commited on
Commit
c3ffc7a
·
verified ·
1 Parent(s): 11e682e

Upload policy weights, train config and readme

Browse files
Files changed (4) hide show
  1. README.md +62 -0
  2. config.json +128 -0
  3. model.safetensors +3 -0
  4. train_config.json +238 -0
README.md ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ datasets: LSY-lab/drawer_v2
3
+ library_name: lerobot
4
+ license: apache-2.0
5
+ model_name: ditflow
6
+ pipeline_tag: robotics
7
+ tags:
8
+ - ditflow
9
+ - lerobot
10
+ - robotics
11
+ ---
12
+
13
+ # Model Card for ditflow
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
+ python -m lerobot.scripts.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
+ python -m 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 +1,128 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "ditflow",
3
+ "n_obs_steps": 2,
4
+ "normalization_mapping": {
5
+ "VISUAL": "MEAN_STD",
6
+ "STATE": "MIN_MAX",
7
+ "ACTION": "MIN_MAX"
8
+ },
9
+ "input_features": {
10
+ "observation.images.primary": {
11
+ "type": "VISUAL",
12
+ "shape": [
13
+ 3,
14
+ 256,
15
+ 256
16
+ ]
17
+ },
18
+ "observation.images.wrist": {
19
+ "type": "VISUAL",
20
+ "shape": [
21
+ 3,
22
+ 256,
23
+ 256
24
+ ]
25
+ },
26
+ "observation.state.cartesian": {
27
+ "type": "STATE",
28
+ "shape": [
29
+ 6
30
+ ]
31
+ },
32
+ "observation.state.gripper": {
33
+ "type": "STATE",
34
+ "shape": [
35
+ 1
36
+ ]
37
+ },
38
+ "observation.state.joints": {
39
+ "type": "STATE",
40
+ "shape": [
41
+ 7
42
+ ]
43
+ },
44
+ "observation.state.sensors_ft_sensor": {
45
+ "type": "STATE",
46
+ "shape": [
47
+ 6
48
+ ]
49
+ },
50
+ "observation.state.sensors_tactile_sensor": {
51
+ "type": "STATE",
52
+ "shape": [
53
+ 15
54
+ ]
55
+ },
56
+ "observation.state.target": {
57
+ "type": "STATE",
58
+ "shape": [
59
+ 6
60
+ ]
61
+ },
62
+ "observation.state": {
63
+ "type": "STATE",
64
+ "shape": [
65
+ 41
66
+ ]
67
+ }
68
+ },
69
+ "output_features": {
70
+ "action": {
71
+ "type": "ACTION",
72
+ "shape": [
73
+ 7
74
+ ]
75
+ }
76
+ },
77
+ "device": "cuda",
78
+ "use_amp": false,
79
+ "push_to_hub": true,
80
+ "repo_id": "danielsanjosepro/ditflow_drawer_v2",
81
+ "private": null,
82
+ "tags": null,
83
+ "license": null,
84
+ "horizon": 16,
85
+ "n_action_steps": 8,
86
+ "use_proprioceptive": true,
87
+ "drop_n_last_frames": 7,
88
+ "vision_backbone": "resnet18",
89
+ "crop_shape": [
90
+ 224,
91
+ 224
92
+ ],
93
+ "crop_is_random": true,
94
+ "pretrained_backbone_weights": null,
95
+ "use_group_norm": true,
96
+ "spatial_softmax_num_keypoints": 32,
97
+ "use_separate_rgb_encoder_per_camera": true,
98
+ "use_mlp_for_state_encoding": false,
99
+ "mlp_state_encoding_dims": [
100
+ 256,
101
+ 128
102
+ ],
103
+ "frequency_embedding_dim": 256,
104
+ "hidden_dim": 512,
105
+ "num_blocks": 6,
106
+ "num_heads": 16,
107
+ "dropout": 0.1,
108
+ "dim_feedforward": 4096,
109
+ "activation": "gelu",
110
+ "training_noise_sampling": "uniform",
111
+ "clip_sample": true,
112
+ "clip_sample_range": 1.0,
113
+ "num_inference_steps": 100,
114
+ "do_mask_loss_for_padding": false,
115
+ "optimizer_lr": 0.0001,
116
+ "optimizer_betas": [
117
+ 0.95,
118
+ 0.999
119
+ ],
120
+ "optimizer_eps": 1e-08,
121
+ "optimizer_weight_decay": 1e-06,
122
+ "scheduler_name": "cosine",
123
+ "scheduler_warmup_steps": 500,
124
+ "features_to_exclude": [
125
+ "observation.images.tactile"
126
+ ],
127
+ "features_to_include": null
128
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1f3239b8f6e909c0692d98d84003a6284fb2ff96fa42b1b7ea220e88c632e66a
3
+ size 257916300
train_config.json ADDED
@@ -0,0 +1,238 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "LSY-lab/drawer_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
+ }
62
+ },
63
+ "revision": null,
64
+ "use_imagenet_stats": true,
65
+ "video_backend": "torchcodec"
66
+ },
67
+ "env": null,
68
+ "policy": {
69
+ "type": "ditflow",
70
+ "n_obs_steps": 2,
71
+ "normalization_mapping": {
72
+ "VISUAL": "MEAN_STD",
73
+ "STATE": "MIN_MAX",
74
+ "ACTION": "MIN_MAX"
75
+ },
76
+ "input_features": {
77
+ "observation.images.primary": {
78
+ "type": "VISUAL",
79
+ "shape": [
80
+ 3,
81
+ 256,
82
+ 256
83
+ ]
84
+ },
85
+ "observation.images.wrist": {
86
+ "type": "VISUAL",
87
+ "shape": [
88
+ 3,
89
+ 256,
90
+ 256
91
+ ]
92
+ },
93
+ "observation.state.cartesian": {
94
+ "type": "STATE",
95
+ "shape": [
96
+ 6
97
+ ]
98
+ },
99
+ "observation.state.gripper": {
100
+ "type": "STATE",
101
+ "shape": [
102
+ 1
103
+ ]
104
+ },
105
+ "observation.state.joints": {
106
+ "type": "STATE",
107
+ "shape": [
108
+ 7
109
+ ]
110
+ },
111
+ "observation.state.sensors_ft_sensor": {
112
+ "type": "STATE",
113
+ "shape": [
114
+ 6
115
+ ]
116
+ },
117
+ "observation.state.sensors_tactile_sensor": {
118
+ "type": "STATE",
119
+ "shape": [
120
+ 15
121
+ ]
122
+ },
123
+ "observation.state.target": {
124
+ "type": "STATE",
125
+ "shape": [
126
+ 6
127
+ ]
128
+ },
129
+ "observation.state": {
130
+ "type": "STATE",
131
+ "shape": [
132
+ 41
133
+ ]
134
+ }
135
+ },
136
+ "output_features": {
137
+ "action": {
138
+ "type": "ACTION",
139
+ "shape": [
140
+ 7
141
+ ]
142
+ }
143
+ },
144
+ "device": "cuda",
145
+ "use_amp": false,
146
+ "push_to_hub": true,
147
+ "repo_id": "danielsanjosepro/ditflow_drawer_v2",
148
+ "private": null,
149
+ "tags": null,
150
+ "license": null,
151
+ "horizon": 16,
152
+ "n_action_steps": 8,
153
+ "use_proprioceptive": true,
154
+ "drop_n_last_frames": 7,
155
+ "vision_backbone": "resnet18",
156
+ "crop_shape": [
157
+ 224,
158
+ 224
159
+ ],
160
+ "crop_is_random": true,
161
+ "pretrained_backbone_weights": null,
162
+ "use_group_norm": true,
163
+ "spatial_softmax_num_keypoints": 32,
164
+ "use_separate_rgb_encoder_per_camera": true,
165
+ "use_mlp_for_state_encoding": false,
166
+ "mlp_state_encoding_dims": [
167
+ 256,
168
+ 128
169
+ ],
170
+ "frequency_embedding_dim": 256,
171
+ "hidden_dim": 512,
172
+ "num_blocks": 6,
173
+ "num_heads": 16,
174
+ "dropout": 0.1,
175
+ "dim_feedforward": 4096,
176
+ "activation": "gelu",
177
+ "training_noise_sampling": "uniform",
178
+ "clip_sample": true,
179
+ "clip_sample_range": 1.0,
180
+ "num_inference_steps": 100,
181
+ "do_mask_loss_for_padding": false,
182
+ "optimizer_lr": 0.0001,
183
+ "optimizer_betas": [
184
+ 0.95,
185
+ 0.999
186
+ ],
187
+ "optimizer_eps": 1e-08,
188
+ "optimizer_weight_decay": 1e-06,
189
+ "scheduler_name": "cosine",
190
+ "scheduler_warmup_steps": 500,
191
+ "features_to_exclude": [
192
+ "observation.images.tactile"
193
+ ],
194
+ "features_to_include": null
195
+ },
196
+ "output_dir": "outputs/train/2025-12-25/18-38-13_config/train_ditflow_drawer.json",
197
+ "job_name": "config/train_ditflow_drawer.json",
198
+ "resume": false,
199
+ "seed": 1000,
200
+ "num_workers": 4,
201
+ "batch_size": 128,
202
+ "steps": 20000,
203
+ "eval_freq": 1000,
204
+ "log_freq": 100,
205
+ "save_checkpoint": true,
206
+ "save_freq": 5000,
207
+ "use_policy_training_preset": true,
208
+ "optimizer": {
209
+ "type": "adam",
210
+ "lr": 0.0001,
211
+ "weight_decay": 1e-06,
212
+ "grad_clip_norm": 10.0,
213
+ "betas": [
214
+ 0.95,
215
+ 0.999
216
+ ],
217
+ "eps": 1e-08
218
+ },
219
+ "scheduler": {
220
+ "type": "diffuser",
221
+ "num_warmup_steps": 500,
222
+ "name": "cosine"
223
+ },
224
+ "eval": {
225
+ "n_episodes": 50,
226
+ "batch_size": 50,
227
+ "use_async_envs": false
228
+ },
229
+ "wandb": {
230
+ "enable": true,
231
+ "disable_artifact": false,
232
+ "project": "policies",
233
+ "entity": null,
234
+ "notes": null,
235
+ "run_id": "l7exp4kq",
236
+ "mode": null
237
+ }
238
+ }