Hugging Face's logo

danielsanjosepro
/
cascaded_flow_stack_cake_v2

Robotics
LeRobot
Safetensors
cascaded_flow
Model card Files
xet
 Community
danielsanjosepro commited on
Commit
524653c
·
verified ·
1 Parent(s): 10075b2

Upload policy weights, train config and readme

Browse files
Files changed (4) hide show
  1. README.md +62 -0
  2. config.json +359 -0
  3. model.safetensors +3 -0
  4. train_config.json +469 -0
README.md ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ datasets: LSY-lab/stack_cake_v2
3
+ library_name: lerobot
4
+ license: apache-2.0
5
+ model_name: cascaded_flow
6
+ pipeline_tag: robotics
7
+ tags:
8
+ - lerobot
9
+ - robotics
10
+ - cascaded_flow
11
+ ---
12
+
13
+ # Model Card for cascaded_flow
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 +1,359 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "cascaded_flow",
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.tactile": {
19
+ "type": "VISUAL",
20
+ "shape": [
21
+ 3,
22
+ 256,
23
+ 256
24
+ ]
25
+ },
26
+ "observation.images.wrist": {
27
+ "type": "VISUAL",
28
+ "shape": [
29
+ 3,
30
+ 256,
31
+ 256
32
+ ]
33
+ },
34
+ "observation.state.cartesian": {
35
+ "type": "STATE",
36
+ "shape": [
37
+ 6
38
+ ]
39
+ },
40
+ "observation.state.gripper": {
41
+ "type": "STATE",
42
+ "shape": [
43
+ 1
44
+ ]
45
+ },
46
+ "observation.state.joints": {
47
+ "type": "STATE",
48
+ "shape": [
49
+ 7
50
+ ]
51
+ },
52
+ "observation.state.sensors_ft_sensor": {
53
+ "type": "STATE",
54
+ "shape": [
55
+ 6
56
+ ]
57
+ },
58
+ "observation.state.sensors_tactile_sensor": {
59
+ "type": "STATE",
60
+ "shape": [
61
+ 15
62
+ ]
63
+ },
64
+ "observation.state.target": {
65
+ "type": "STATE",
66
+ "shape": [
67
+ 6
68
+ ]
69
+ },
70
+ "observation.state": {
71
+ "type": "STATE",
72
+ "shape": [
73
+ 41
74
+ ]
75
+ }
76
+ },
77
+ "output_features": {
78
+ "action": {
79
+ "type": "ACTION",
80
+ "shape": [
81
+ 7
82
+ ]
83
+ }
84
+ },
85
+ "device": "cuda",
86
+ "use_amp": false,
87
+ "push_to_hub": true,
88
+ "repo_id": "danielsanjosepro/cascaded_flow_stack_cake_v2",
89
+ "private": null,
90
+ "tags": null,
91
+ "license": null,
92
+ "slow_policy_config": {
93
+ "type": "ditmeanflow",
94
+ "n_obs_steps": 2,
95
+ "normalization_mapping": {
96
+ "VISUAL": "MEAN_STD",
97
+ "STATE": "MIN_MAX",
98
+ "ACTION": "MEAN_STD"
99
+ },
100
+ "input_features": {
101
+ "observation.images.primary": {
102
+ "type": "VISUAL",
103
+ "shape": [
104
+ 3,
105
+ 256,
106
+ 256
107
+ ]
108
+ },
109
+ "observation.images.wrist": {
110
+ "type": "VISUAL",
111
+ "shape": [
112
+ 3,
113
+ 256,
114
+ 256
115
+ ]
116
+ },
117
+ "observation.state.cartesian": {
118
+ "type": "STATE",
119
+ "shape": [
120
+ 6
121
+ ]
122
+ },
123
+ "observation.state.gripper": {
124
+ "type": "STATE",
125
+ "shape": [
126
+ 1
127
+ ]
128
+ },
129
+ "observation.state.joints": {
130
+ "type": "STATE",
131
+ "shape": [
132
+ 7
133
+ ]
134
+ },
135
+ "observation.state.sensors_ft_sensor": {
136
+ "type": "STATE",
137
+ "shape": [
138
+ 6
139
+ ]
140
+ },
141
+ "observation.state.sensors_tactile_sensor": {
142
+ "type": "STATE",
143
+ "shape": [
144
+ 15
145
+ ]
146
+ },
147
+ "observation.state.target": {
148
+ "type": "STATE",
149
+ "shape": [
150
+ 6
151
+ ]
152
+ },
153
+ "observation.state": {
154
+ "type": "STATE",
155
+ "shape": [
156
+ 41
157
+ ]
158
+ }
159
+ },
160
+ "output_features": {
161
+ "action": {
162
+ "type": "ACTION",
163
+ "shape": [
164
+ 7
165
+ ]
166
+ }
167
+ },
168
+ "device": "cuda",
169
+ "use_amp": false,
170
+ "push_to_hub": true,
171
+ "repo_id": null,
172
+ "private": null,
173
+ "tags": null,
174
+ "license": null,
175
+ "horizon": 32,
176
+ "n_action_steps": 16,
177
+ "use_proprioceptive": true,
178
+ "drop_n_last_frames": 7,
179
+ "vision_backbone": "resnet18",
180
+ "crop_shape": [
181
+ 256,
182
+ 256
183
+ ],
184
+ "crop_is_random": true,
185
+ "pretrained_backbone_weights": null,
186
+ "use_group_norm": true,
187
+ "spatial_softmax_num_keypoints": 32,
188
+ "use_separate_rgb_encoder_per_camera": false,
189
+ "frequency_embedding_dim": 256,
190
+ "hidden_dim": 512,
191
+ "num_blocks": 6,
192
+ "num_heads": 16,
193
+ "dropout": 0.1,
194
+ "dim_feedforward": 4096,
195
+ "activation": "gelu",
196
+ "training_noise_sampling": "uniform",
197
+ "clip_sample": true,
198
+ "clip_sample_range": 1.0,
199
+ "num_inference_steps": 100,
200
+ "do_mask_loss_for_padding": false,
201
+ "optimizer_lr": 0.000141,
202
+ "optimizer_betas": [
203
+ 0.95,
204
+ 0.999
205
+ ],
206
+ "optimizer_eps": 1e-08,
207
+ "optimizer_weight_decay": 1e-06,
208
+ "scheduler_name": "cosine",
209
+ "scheduler_warmup_steps": 500,
210
+ "features_to_exclude": [
211
+ "observation.images.tactile"
212
+ ],
213
+ "features_to_include": null,
214
+ "use_meanflow": true,
215
+ "do_multi_step_sampling": false,
216
+ "inference_timesteps": 1,
217
+ "time_distribution": "uniform",
218
+ "log_norm_mu": -0.4,
219
+ "log_norm_sigma": 1.0,
220
+ "use_autograd_functional_jvp": false,
221
+ "use_adaptative_loss": true,
222
+ "flow_ratio": 0.5,
223
+ "cfg_prob": 0.1,
224
+ "cfg_omega": 2.0,
225
+ "cfg_w": 2.0,
226
+ "cfg_kappa": 0.0,
227
+ "optimizer_grad_clip_norm": 0.5
228
+ },
229
+ "fast_policy_config": {
230
+ "type": "ditmeanflow",
231
+ "n_obs_steps": 2,
232
+ "normalization_mapping": {
233
+ "VISUAL": "MEAN_STD",
234
+ "STATE": "MIN_MAX",
235
+ "ACTION": "MEAN_STD"
236
+ },
237
+ "input_features": {
238
+ "observation.state.cartesian": {
239
+ "type": "STATE",
240
+ "shape": [
241
+ 6
242
+ ]
243
+ },
244
+ "observation.state.gripper": {
245
+ "type": "STATE",
246
+ "shape": [
247
+ 1
248
+ ]
249
+ },
250
+ "observation.state.joints": {
251
+ "type": "STATE",
252
+ "shape": [
253
+ 7
254
+ ]
255
+ },
256
+ "observation.state.sensors_ft_sensor": {
257
+ "type": "STATE",
258
+ "shape": [
259
+ 6
260
+ ]
261
+ },
262
+ "observation.state.sensors_tactile_sensor": {
263
+ "type": "STATE",
264
+ "shape": [
265
+ 15
266
+ ]
267
+ },
268
+ "observation.state.target": {
269
+ "type": "STATE",
270
+ "shape": [
271
+ 6
272
+ ]
273
+ },
274
+ "observation.state": {
275
+ "type": "STATE",
276
+ "shape": [
277
+ 41
278
+ ]
279
+ }
280
+ },
281
+ "output_features": {
282
+ "action": {
283
+ "type": "ACTION",
284
+ "shape": [
285
+ 7
286
+ ]
287
+ }
288
+ },
289
+ "device": "cuda",
290
+ "use_amp": false,
291
+ "push_to_hub": true,
292
+ "repo_id": null,
293
+ "private": null,
294
+ "tags": null,
295
+ "license": null,
296
+ "horizon": 8,
297
+ "n_action_steps": 1,
298
+ "use_proprioceptive": true,
299
+ "drop_n_last_frames": 7,
300
+ "vision_backbone": "resnet18",
301
+ "crop_shape": [
302
+ 256,
303
+ 256
304
+ ],
305
+ "crop_is_random": true,
306
+ "pretrained_backbone_weights": null,
307
+ "use_group_norm": true,
308
+ "spatial_softmax_num_keypoints": 32,
309
+ "use_separate_rgb_encoder_per_camera": false,
310
+ "frequency_embedding_dim": 256,
311
+ "hidden_dim": 512,
312
+ "num_blocks": 6,
313
+ "num_heads": 16,
314
+ "dropout": 0.1,
315
+ "dim_feedforward": 4096,
316
+ "activation": "gelu",
317
+ "training_noise_sampling": "uniform",
318
+ "clip_sample": true,
319
+ "clip_sample_range": 1.0,
320
+ "num_inference_steps": 100,
321
+ "do_mask_loss_for_padding": false,
322
+ "optimizer_lr": 0.000141,
323
+ "optimizer_betas": [
324
+ 0.95,
325
+ 0.999
326
+ ],
327
+ "optimizer_eps": 1e-08,
328
+ "optimizer_weight_decay": 1e-06,
329
+ "scheduler_name": "cosine",
330
+ "scheduler_warmup_steps": 500,
331
+ "features_to_exclude": [
332
+ "observation.images.primary",
333
+ "observation.images.wrist",
334
+ "observation.images.tactile"
335
+ ],
336
+ "features_to_include": null,
337
+ "use_meanflow": true,
338
+ "do_multi_step_sampling": false,
339
+ "inference_timesteps": 1,
340
+ "time_distribution": "uniform",
341
+ "log_norm_mu": -0.4,
342
+ "log_norm_sigma": 1.0,
343
+ "use_autograd_functional_jvp": false,
344
+ "use_adaptative_loss": true,
345
+ "flow_ratio": 0.5,
346
+ "cfg_prob": 0.1,
347
+ "cfg_omega": 2.0,
348
+ "cfg_w": 2.0,
349
+ "cfg_kappa": 0.0,
350
+ "optimizer_grad_clip_norm": 0.5
351
+ },
352
+ "action_batch_size": 20,
353
+ "sampling_strategy": "stochastic",
354
+ "sampling_temperature": 1.0,
355
+ "smooth_slow_policy_output": false,
356
+ "apply_slow_policy_output": false,
357
+ "use_image_features_for_fast_policy": true,
358
+ "training_stage": "slow_policy"
359
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:79edb872485cb532e95a4fc6f3bad84ad63a341c4c109160e89f789f546560c8
3
+ size 384728888
train_config.json ADDED
@@ -0,0 +1,469 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "LSY-lab/stack_cake_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": "cascaded_flow",
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.tactile": {
86
+ "type": "VISUAL",
87
+ "shape": [
88
+ 3,
89
+ 256,
90
+ 256
91
+ ]
92
+ },
93
+ "observation.images.wrist": {
94
+ "type": "VISUAL",
95
+ "shape": [
96
+ 3,
97
+ 256,
98
+ 256
99
+ ]
100
+ },
101
+ "observation.state.cartesian": {
102
+ "type": "STATE",
103
+ "shape": [
104
+ 6
105
+ ]
106
+ },
107
+ "observation.state.gripper": {
108
+ "type": "STATE",
109
+ "shape": [
110
+ 1
111
+ ]
112
+ },
113
+ "observation.state.joints": {
114
+ "type": "STATE",
115
+ "shape": [
116
+ 7
117
+ ]
118
+ },
119
+ "observation.state.sensors_ft_sensor": {
120
+ "type": "STATE",
121
+ "shape": [
122
+ 6
123
+ ]
124
+ },
125
+ "observation.state.sensors_tactile_sensor": {
126
+ "type": "STATE",
127
+ "shape": [
128
+ 15
129
+ ]
130
+ },
131
+ "observation.state.target": {
132
+ "type": "STATE",
133
+ "shape": [
134
+ 6
135
+ ]
136
+ },
137
+ "observation.state": {
138
+ "type": "STATE",
139
+ "shape": [
140
+ 41
141
+ ]
142
+ }
143
+ },
144
+ "output_features": {
145
+ "action": {
146
+ "type": "ACTION",
147
+ "shape": [
148
+ 7
149
+ ]
150
+ }
151
+ },
152
+ "device": "cuda",
153
+ "use_amp": false,
154
+ "push_to_hub": true,
155
+ "repo_id": "danielsanjosepro/cascaded_flow_stack_cake_v2",
156
+ "private": null,
157
+ "tags": null,
158
+ "license": null,
159
+ "slow_policy_config": {
160
+ "type": "ditmeanflow",
161
+ "n_obs_steps": 2,
162
+ "normalization_mapping": {
163
+ "VISUAL": "MEAN_STD",
164
+ "STATE": "MIN_MAX",
165
+ "ACTION": "MEAN_STD"
166
+ },
167
+ "input_features": {
168
+ "observation.images.primary": {
169
+ "type": "VISUAL",
170
+ "shape": [
171
+ 3,
172
+ 256,
173
+ 256
174
+ ]
175
+ },
176
+ "observation.images.wrist": {
177
+ "type": "VISUAL",
178
+ "shape": [
179
+ 3,
180
+ 256,
181
+ 256
182
+ ]
183
+ },
184
+ "observation.state.cartesian": {
185
+ "type": "STATE",
186
+ "shape": [
187
+ 6
188
+ ]
189
+ },
190
+ "observation.state.gripper": {
191
+ "type": "STATE",
192
+ "shape": [
193
+ 1
194
+ ]
195
+ },
196
+ "observation.state.joints": {
197
+ "type": "STATE",
198
+ "shape": [
199
+ 7
200
+ ]
201
+ },
202
+ "observation.state.sensors_ft_sensor": {
203
+ "type": "STATE",
204
+ "shape": [
205
+ 6
206
+ ]
207
+ },
208
+ "observation.state.sensors_tactile_sensor": {
209
+ "type": "STATE",
210
+ "shape": [
211
+ 15
212
+ ]
213
+ },
214
+ "observation.state.target": {
215
+ "type": "STATE",
216
+ "shape": [
217
+ 6
218
+ ]
219
+ },
220
+ "observation.state": {
221
+ "type": "STATE",
222
+ "shape": [
223
+ 41
224
+ ]
225
+ }
226
+ },
227
+ "output_features": {
228
+ "action": {
229
+ "type": "ACTION",
230
+ "shape": [
231
+ 7
232
+ ]
233
+ }
234
+ },
235
+ "device": "cuda",
236
+ "use_amp": false,
237
+ "push_to_hub": true,
238
+ "repo_id": null,
239
+ "private": null,
240
+ "tags": null,
241
+ "license": null,
242
+ "horizon": 32,
243
+ "n_action_steps": 16,
244
+ "use_proprioceptive": true,
245
+ "drop_n_last_frames": 7,
246
+ "vision_backbone": "resnet18",
247
+ "crop_shape": [
248
+ 256,
249
+ 256
250
+ ],
251
+ "crop_is_random": true,
252
+ "pretrained_backbone_weights": null,
253
+ "use_group_norm": true,
254
+ "spatial_softmax_num_keypoints": 32,
255
+ "use_separate_rgb_encoder_per_camera": false,
256
+ "frequency_embedding_dim": 256,
257
+ "hidden_dim": 512,
258
+ "num_blocks": 6,
259
+ "num_heads": 16,
260
+ "dropout": 0.1,
261
+ "dim_feedforward": 4096,
262
+ "activation": "gelu",
263
+ "training_noise_sampling": "uniform",
264
+ "clip_sample": true,
265
+ "clip_sample_range": 1.0,
266
+ "num_inference_steps": 100,
267
+ "do_mask_loss_for_padding": false,
268
+ "optimizer_lr": 0.000141,
269
+ "optimizer_betas": [
270
+ 0.95,
271
+ 0.999
272
+ ],
273
+ "optimizer_eps": 1e-08,
274
+ "optimizer_weight_decay": 1e-06,
275
+ "scheduler_name": "cosine",
276
+ "scheduler_warmup_steps": 500,
277
+ "features_to_exclude": [
278
+ "observation.images.tactile"
279
+ ],
280
+ "features_to_include": null,
281
+ "use_meanflow": true,
282
+ "do_multi_step_sampling": false,
283
+ "inference_timesteps": 1,
284
+ "time_distribution": "uniform",
285
+ "log_norm_mu": -0.4,
286
+ "log_norm_sigma": 1.0,
287
+ "use_autograd_functional_jvp": false,
288
+ "use_adaptative_loss": true,
289
+ "flow_ratio": 0.5,
290
+ "cfg_prob": 0.1,
291
+ "cfg_omega": 2.0,
292
+ "cfg_w": 2.0,
293
+ "cfg_kappa": 0.0,
294
+ "optimizer_grad_clip_norm": 0.5
295
+ },
296
+ "fast_policy_config": {
297
+ "type": "ditmeanflow",
298
+ "n_obs_steps": 2,
299
+ "normalization_mapping": {
300
+ "VISUAL": "MEAN_STD",
301
+ "STATE": "MIN_MAX",
302
+ "ACTION": "MEAN_STD"
303
+ },
304
+ "input_features": {
305
+ "observation.state.cartesian": {
306
+ "type": "STATE",
307
+ "shape": [
308
+ 6
309
+ ]
310
+ },
311
+ "observation.state.gripper": {
312
+ "type": "STATE",
313
+ "shape": [
314
+ 1
315
+ ]
316
+ },
317
+ "observation.state.joints": {
318
+ "type": "STATE",
319
+ "shape": [
320
+ 7
321
+ ]
322
+ },
323
+ "observation.state.sensors_ft_sensor": {
324
+ "type": "STATE",
325
+ "shape": [
326
+ 6
327
+ ]
328
+ },
329
+ "observation.state.sensors_tactile_sensor": {
330
+ "type": "STATE",
331
+ "shape": [
332
+ 15
333
+ ]
334
+ },
335
+ "observation.state.target": {
336
+ "type": "STATE",
337
+ "shape": [
338
+ 6
339
+ ]
340
+ },
341
+ "observation.state": {
342
+ "type": "STATE",
343
+ "shape": [
344
+ 41
345
+ ]
346
+ }
347
+ },
348
+ "output_features": {
349
+ "action": {
350
+ "type": "ACTION",
351
+ "shape": [
352
+ 7
353
+ ]
354
+ }
355
+ },
356
+ "device": "cuda",
357
+ "use_amp": false,
358
+ "push_to_hub": true,
359
+ "repo_id": null,
360
+ "private": null,
361
+ "tags": null,
362
+ "license": null,
363
+ "horizon": 8,
364
+ "n_action_steps": 1,
365
+ "use_proprioceptive": true,
366
+ "drop_n_last_frames": 7,
367
+ "vision_backbone": "resnet18",
368
+ "crop_shape": [
369
+ 256,
370
+ 256
371
+ ],
372
+ "crop_is_random": true,
373
+ "pretrained_backbone_weights": null,
374
+ "use_group_norm": true,
375
+ "spatial_softmax_num_keypoints": 32,
376
+ "use_separate_rgb_encoder_per_camera": false,
377
+ "frequency_embedding_dim": 256,
378
+ "hidden_dim": 512,
379
+ "num_blocks": 6,
380
+ "num_heads": 16,
381
+ "dropout": 0.1,
382
+ "dim_feedforward": 4096,
383
+ "activation": "gelu",
384
+ "training_noise_sampling": "uniform",
385
+ "clip_sample": true,
386
+ "clip_sample_range": 1.0,
387
+ "num_inference_steps": 100,
388
+ "do_mask_loss_for_padding": false,
389
+ "optimizer_lr": 0.000141,
390
+ "optimizer_betas": [
391
+ 0.95,
392
+ 0.999
393
+ ],
394
+ "optimizer_eps": 1e-08,
395
+ "optimizer_weight_decay": 1e-06,
396
+ "scheduler_name": "cosine",
397
+ "scheduler_warmup_steps": 500,
398
+ "features_to_exclude": [
399
+ "observation.images.primary",
400
+ "observation.images.wrist",
401
+ "observation.images.tactile"
402
+ ],
403
+ "features_to_include": null,
404
+ "use_meanflow": true,
405
+ "do_multi_step_sampling": false,
406
+ "inference_timesteps": 1,
407
+ "time_distribution": "uniform",
408
+ "log_norm_mu": -0.4,
409
+ "log_norm_sigma": 1.0,
410
+ "use_autograd_functional_jvp": false,
411
+ "use_adaptative_loss": true,
412
+ "flow_ratio": 0.5,
413
+ "cfg_prob": 0.1,
414
+ "cfg_omega": 2.0,
415
+ "cfg_w": 2.0,
416
+ "cfg_kappa": 0.0,
417
+ "optimizer_grad_clip_norm": 0.5
418
+ },
419
+ "action_batch_size": 20,
420
+ "sampling_strategy": "stochastic",
421
+ "sampling_temperature": 1.0,
422
+ "smooth_slow_policy_output": false,
423
+ "apply_slow_policy_output": false,
424
+ "use_image_features_for_fast_policy": true,
425
+ "training_stage": "slow_policy"
426
+ },
427
+ "output_dir": "outputs/train/2025-11-21/08-55-42_config/train_cascaded_stack_cake_v2.json_slow",
428
+ "job_name": "config/train_cascaded_stack_cake_v2.json_slow",
429
+ "resume": false,
430
+ "seed": 1000,
431
+ "num_workers": 4,
432
+ "batch_size": 128,
433
+ "steps": 20000,
434
+ "eval_freq": 1000,
435
+ "log_freq": 100,
436
+ "save_checkpoint": true,
437
+ "save_freq": 5000,
438
+ "use_policy_training_preset": true,
439
+ "optimizer": {
440
+ "type": "adam",
441
+ "lr": 0.000141,
442
+ "weight_decay": 1e-06,
443
+ "grad_clip_norm": 10.0,
444
+ "betas": [
445
+ 0.95,
446
+ 0.999
447
+ ],
448
+ "eps": 1e-08
449
+ },
450
+ "scheduler": {
451
+ "type": "diffuser",
452
+ "num_warmup_steps": 500,
453
+ "name": "cosine"
454
+ },
455
+ "eval": {
456
+ "n_episodes": 50,
457
+ "batch_size": 50,
458
+ "use_async_envs": false
459
+ },
460
+ "wandb": {
461
+ "enable": true,
462
+ "disable_artifact": false,
463
+ "project": "policies",
464
+ "entity": null,
465
+ "notes": null,
466
+ "run_id": "d1kt65ao",
467
+ "mode": null
468
+ }
469
+ }