Hugging Face's logo

continuallearning
/
ditflow_image_sort_red_blocks_100_60000

Safetensors
Model card Files
xet
Community
continuallearning commited on
Commit
3bbba78
·
verified ·
1 Parent(s): 5e28f6b

Upload folder using huggingface_hub

Browse files
Files changed (4) hide show
  1. config.json +90 -0
  2. dataset_info.json +173 -0
  3. model.safetensors +3 -0
  4. train_config.json +307 -0
config.json ADDED
@@ -0,0 +1,90 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "ditflow_image",
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.state": {
11
+ "type": "STATE",
12
+ "shape": [
13
+ 18
14
+ ]
15
+ },
16
+ "observation.images.rgb_static": {
17
+ "type": "VISUAL",
18
+ "shape": [
19
+ 3,
20
+ 256,
21
+ 256
22
+ ]
23
+ },
24
+ "observation.images.rgb_left": {
25
+ "type": "VISUAL",
26
+ "shape": [
27
+ 3,
28
+ 256,
29
+ 256
30
+ ]
31
+ },
32
+ "observation.images.rgb_right": {
33
+ "type": "VISUAL",
34
+ "shape": [
35
+ 3,
36
+ 256,
37
+ 256
38
+ ]
39
+ }
40
+ },
41
+ "output_features": {
42
+ "action": {
43
+ "type": "ACTION",
44
+ "shape": [
45
+ 17
46
+ ]
47
+ }
48
+ },
49
+ "device": "cuda",
50
+ "use_amp": false,
51
+ "push_to_hub": false,
52
+ "repo_id": null,
53
+ "private": null,
54
+ "tags": null,
55
+ "license": null,
56
+ "horizon": 16,
57
+ "n_action_steps": 8,
58
+ "drop_n_last_frames": 7,
59
+ "vision_backbone": "resnet18",
60
+ "crop_shape": [
61
+ 224,
62
+ 224
63
+ ],
64
+ "crop_is_random": true,
65
+ "pretrained_backbone_weights": null,
66
+ "use_group_norm": true,
67
+ "spatial_softmax_num_keypoints": 32,
68
+ "use_separate_rgb_encoder_per_camera": true,
69
+ "frequency_embedding_dim": 256,
70
+ "hidden_dim": 512,
71
+ "num_blocks": 6,
72
+ "num_heads": 16,
73
+ "dropout": 0.1,
74
+ "dim_feedforward": 4096,
75
+ "activation": "gelu",
76
+ "training_noise_sampling": "uniform",
77
+ "clip_sample": true,
78
+ "clip_sample_range": 1.0,
79
+ "num_inference_steps": 100,
80
+ "do_mask_loss_for_padding": false,
81
+ "optimizer_lr": 0.0002,
82
+ "optimizer_betas": [
83
+ 0.95,
84
+ 0.999
85
+ ],
86
+ "optimizer_eps": 1e-08,
87
+ "optimizer_weight_decay": 1e-06,
88
+ "scheduler_name": "cosine",
89
+ "scheduler_warmup_steps": 500
90
+ }
dataset_info.json ADDED
@@ -0,0 +1,173 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "codebase_version": "v2.1",
3
+ "robot_type": null,
4
+ "total_episodes": 106,
5
+ "total_frames": 26800,
6
+ "total_tasks": 1,
7
+ "total_videos": 318,
8
+ "total_chunks": 1,
9
+ "chunks_size": 1000,
10
+ "fps": 10,
11
+ "splits": {
12
+ "train": "0:106"
13
+ },
14
+ "data_path": "data/chunk-{episode_chunk:03d}/episode_{episode_index:06d}.parquet",
15
+ "video_path": "videos/chunk-{episode_chunk:03d}/{video_key}/episode_{episode_index:06d}.mp4",
16
+ "features": {
17
+ "observation.state": {
18
+ "dtype": "float32",
19
+ "shape": [
20
+ 18
21
+ ],
22
+ "names": [
23
+ "tcp_left_pos_x",
24
+ "tcp_left_pos_y",
25
+ "tcp_left_pos_z",
26
+ "tcp_left_quat_x",
27
+ "tcp_left_quat_y",
28
+ "tcp_left_quat_z",
29
+ "tcp_left_quat_w",
30
+ "tcp_right_pos_x",
31
+ "tcp_right_pos_y",
32
+ "tcp_right_pos_z",
33
+ "tcp_right_quat_x",
34
+ "tcp_right_quat_y",
35
+ "tcp_right_quat_z",
36
+ "tcp_right_quat_w",
37
+ "gripper_left_0",
38
+ "gripper_left_1",
39
+ "gripper_right_0",
40
+ "gripper_right_1"
41
+ ]
42
+ },
43
+ "action": {
44
+ "dtype": "float32",
45
+ "shape": [
46
+ 17
47
+ ],
48
+ "names": [
49
+ "tcp_left_x",
50
+ "tcp_left_y",
51
+ "tcp_left_z",
52
+ "tcp_left_quat_x",
53
+ "tcp_left_quat_y",
54
+ "tcp_left_quat_z",
55
+ "tcp_left_quat_w",
56
+ "tcp_right_x",
57
+ "tcp_right_y",
58
+ "tcp_right_z",
59
+ "tcp_right_quat_x",
60
+ "tcp_right_quat_y",
61
+ "tcp_right_quat_z",
62
+ "tcp_right_quat_w",
63
+ "gripper_left",
64
+ "gripper_right",
65
+ "termination_signal"
66
+ ]
67
+ },
68
+ "observation.images.rgb_static": {
69
+ "dtype": "video",
70
+ "shape": [
71
+ 256,
72
+ 256,
73
+ 3
74
+ ],
75
+ "names": [
76
+ "height",
77
+ "width",
78
+ "channel"
79
+ ],
80
+ "info": {
81
+ "video.height": 256,
82
+ "video.width": 256,
83
+ "video.codec": "av1",
84
+ "video.pix_fmt": "yuv420p",
85
+ "video.is_depth_map": false,
86
+ "video.fps": 10,
87
+ "video.channels": 3,
88
+ "has_audio": false
89
+ }
90
+ },
91
+ "observation.images.rgb_left": {
92
+ "dtype": "video",
93
+ "shape": [
94
+ 256,
95
+ 256,
96
+ 3
97
+ ],
98
+ "names": [
99
+ "height",
100
+ "width",
101
+ "channel"
102
+ ],
103
+ "info": {
104
+ "video.height": 256,
105
+ "video.width": 256,
106
+ "video.codec": "av1",
107
+ "video.pix_fmt": "yuv420p",
108
+ "video.is_depth_map": false,
109
+ "video.fps": 10,
110
+ "video.channels": 3,
111
+ "has_audio": false
112
+ }
113
+ },
114
+ "observation.images.rgb_right": {
115
+ "dtype": "video",
116
+ "shape": [
117
+ 256,
118
+ 256,
119
+ 3
120
+ ],
121
+ "names": [
122
+ "height",
123
+ "width",
124
+ "channel"
125
+ ],
126
+ "info": {
127
+ "video.height": 256,
128
+ "video.width": 256,
129
+ "video.codec": "av1",
130
+ "video.pix_fmt": "yuv420p",
131
+ "video.is_depth_map": false,
132
+ "video.fps": 10,
133
+ "video.channels": 3,
134
+ "has_audio": false
135
+ }
136
+ },
137
+ "timestamp": {
138
+ "dtype": "float32",
139
+ "shape": [
140
+ 1
141
+ ],
142
+ "names": null
143
+ },
144
+ "frame_index": {
145
+ "dtype": "int64",
146
+ "shape": [
147
+ 1
148
+ ],
149
+ "names": null
150
+ },
151
+ "episode_index": {
152
+ "dtype": "int64",
153
+ "shape": [
154
+ 1
155
+ ],
156
+ "names": null
157
+ },
158
+ "index": {
159
+ "dtype": "int64",
160
+ "shape": [
161
+ 1
162
+ ],
163
+ "names": null
164
+ },
165
+ "task_index": {
166
+ "dtype": "int64",
167
+ "shape": [
168
+ 1
169
+ ],
170
+ "names": null
171
+ }
172
+ }
173
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:89845015c58f2abaac65bc3e2eaa10c02aa3d96b9c089c046be5b607788c53e7
3
+ size 302847324
train_config.json ADDED
@@ -0,0 +1,307 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "sort_red_blocks_100",
4
+ "root": "sort_red_blocks_100",
5
+ "episodes": [
6
+ 0,
7
+ 1,
8
+ 2,
9
+ 3,
10
+ 4,
11
+ 5,
12
+ 6,
13
+ 7,
14
+ 8,
15
+ 9,
16
+ 10,
17
+ 11,
18
+ 12,
19
+ 13,
20
+ 14,
21
+ 15,
22
+ 16,
23
+ 17,
24
+ 18,
25
+ 19,
26
+ 20,
27
+ 21,
28
+ 22,
29
+ 23,
30
+ 24,
31
+ 25,
32
+ 26,
33
+ 27,
34
+ 28,
35
+ 29,
36
+ 30,
37
+ 31,
38
+ 32,
39
+ 33,
40
+ 34,
41
+ 35,
42
+ 36,
43
+ 37,
44
+ 38,
45
+ 39,
46
+ 40,
47
+ 41,
48
+ 42,
49
+ 43,
50
+ 44,
51
+ 45,
52
+ 46,
53
+ 47,
54
+ 48,
55
+ 49,
56
+ 50,
57
+ 51,
58
+ 52,
59
+ 53,
60
+ 54,
61
+ 55,
62
+ 56,
63
+ 57,
64
+ 58,
65
+ 59,
66
+ 60,
67
+ 61,
68
+ 62,
69
+ 63,
70
+ 64,
71
+ 65,
72
+ 66,
73
+ 67,
74
+ 68,
75
+ 69,
76
+ 70,
77
+ 71,
78
+ 72,
79
+ 73,
80
+ 74,
81
+ 75,
82
+ 76,
83
+ 77,
84
+ 78,
85
+ 79,
86
+ 80,
87
+ 81,
88
+ 82,
89
+ 83,
90
+ 84,
91
+ 85,
92
+ 86,
93
+ 87,
94
+ 88,
95
+ 89,
96
+ 90,
97
+ 91,
98
+ 92,
99
+ 93,
100
+ 94,
101
+ 95,
102
+ 96,
103
+ 97,
104
+ 98,
105
+ 99,
106
+ 100,
107
+ 101,
108
+ 102,
109
+ 103,
110
+ 104,
111
+ 105
112
+ ],
113
+ "image_transforms": {
114
+ "enable": false,
115
+ "max_num_transforms": 3,
116
+ "random_order": false,
117
+ "tfs": {
118
+ "brightness": {
119
+ "weight": 1.0,
120
+ "type": "ColorJitter",
121
+ "kwargs": {
122
+ "brightness": [
123
+ 0.8,
124
+ 1.2
125
+ ]
126
+ }
127
+ },
128
+ "contrast": {
129
+ "weight": 1.0,
130
+ "type": "ColorJitter",
131
+ "kwargs": {
132
+ "contrast": [
133
+ 0.8,
134
+ 1.2
135
+ ]
136
+ }
137
+ },
138
+ "saturation": {
139
+ "weight": 1.0,
140
+ "type": "ColorJitter",
141
+ "kwargs": {
142
+ "saturation": [
143
+ 0.5,
144
+ 1.5
145
+ ]
146
+ }
147
+ },
148
+ "hue": {
149
+ "weight": 1.0,
150
+ "type": "ColorJitter",
151
+ "kwargs": {
152
+ "hue": [
153
+ -0.05,
154
+ 0.05
155
+ ]
156
+ }
157
+ },
158
+ "sharpness": {
159
+ "weight": 1.0,
160
+ "type": "SharpnessJitter",
161
+ "kwargs": {
162
+ "sharpness": [
163
+ 0.5,
164
+ 1.5
165
+ ]
166
+ }
167
+ }
168
+ }
169
+ },
170
+ "revision": null,
171
+ "use_imagenet_stats": true,
172
+ "video_backend": "pyav"
173
+ },
174
+ "env": null,
175
+ "policy": {
176
+ "type": "ditflow_image",
177
+ "n_obs_steps": 2,
178
+ "normalization_mapping": {
179
+ "VISUAL": "MEAN_STD",
180
+ "STATE": "MIN_MAX",
181
+ "ACTION": "MIN_MAX"
182
+ },
183
+ "input_features": {
184
+ "observation.state": {
185
+ "type": "STATE",
186
+ "shape": [
187
+ 18
188
+ ]
189
+ },
190
+ "observation.images.rgb_static": {
191
+ "type": "VISUAL",
192
+ "shape": [
193
+ 3,
194
+ 256,
195
+ 256
196
+ ]
197
+ },
198
+ "observation.images.rgb_left": {
199
+ "type": "VISUAL",
200
+ "shape": [
201
+ 3,
202
+ 256,
203
+ 256
204
+ ]
205
+ },
206
+ "observation.images.rgb_right": {
207
+ "type": "VISUAL",
208
+ "shape": [
209
+ 3,
210
+ 256,
211
+ 256
212
+ ]
213
+ }
214
+ },
215
+ "output_features": {
216
+ "action": {
217
+ "type": "ACTION",
218
+ "shape": [
219
+ 17
220
+ ]
221
+ }
222
+ },
223
+ "device": "cuda",
224
+ "use_amp": false,
225
+ "push_to_hub": false,
226
+ "repo_id": null,
227
+ "private": null,
228
+ "tags": null,
229
+ "license": null,
230
+ "horizon": 16,
231
+ "n_action_steps": 8,
232
+ "drop_n_last_frames": 7,
233
+ "vision_backbone": "resnet18",
234
+ "crop_shape": [
235
+ 224,
236
+ 224
237
+ ],
238
+ "crop_is_random": true,
239
+ "pretrained_backbone_weights": null,
240
+ "use_group_norm": true,
241
+ "spatial_softmax_num_keypoints": 32,
242
+ "use_separate_rgb_encoder_per_camera": true,
243
+ "frequency_embedding_dim": 256,
244
+ "hidden_dim": 512,
245
+ "num_blocks": 6,
246
+ "num_heads": 16,
247
+ "dropout": 0.1,
248
+ "dim_feedforward": 4096,
249
+ "activation": "gelu",
250
+ "training_noise_sampling": "uniform",
251
+ "clip_sample": true,
252
+ "clip_sample_range": 1.0,
253
+ "num_inference_steps": 100,
254
+ "do_mask_loss_for_padding": false,
255
+ "optimizer_lr": 0.0002,
256
+ "optimizer_betas": [
257
+ 0.95,
258
+ 0.999
259
+ ],
260
+ "optimizer_eps": 1e-08,
261
+ "optimizer_weight_decay": 1e-06,
262
+ "scheduler_name": "cosine",
263
+ "scheduler_warmup_steps": 500
264
+ },
265
+ "output_dir": "/root/autodl-tmp/outputs/ditflow_image_sort_red_blocks_100_long",
266
+ "job_name": "ditflow_image_sort_red_blocks_100_long",
267
+ "resume": true,
268
+ "seed": 1000,
269
+ "num_workers": 16,
270
+ "batch_size": 128,
271
+ "steps": 83600,
272
+ "eval_freq": 20000,
273
+ "log_freq": 100,
274
+ "save_checkpoint": true,
275
+ "save_freq": 5000,
276
+ "use_policy_training_preset": true,
277
+ "optimizer": {
278
+ "type": "adam",
279
+ "lr": 0.0002,
280
+ "weight_decay": 1e-06,
281
+ "grad_clip_norm": 10.0,
282
+ "betas": [
283
+ 0.95,
284
+ 0.999
285
+ ],
286
+ "eps": 1e-08
287
+ },
288
+ "scheduler": {
289
+ "type": "diffuser",
290
+ "num_warmup_steps": 500,
291
+ "name": "cosine"
292
+ },
293
+ "eval": {
294
+ "n_episodes": 50,
295
+ "batch_size": 50,
296
+ "use_async_envs": false
297
+ },
298
+ "wandb": {
299
+ "enable": true,
300
+ "disable_artifact": true,
301
+ "project": "paper",
302
+ "entity": "470620104-technical-university-of-munich",
303
+ "notes": null,
304
+ "run_id": "2kpstz8d",
305
+ "mode": null
306
+ }
307
+ }