Hugging Face's logo

continuallearning
/
ditflow_image_stack_red_blocks_85_40000

Safetensors
Model card Files
xet
Community
continuallearning commited on
Commit
f92aee8
·
verified ·
1 Parent(s): a4a912f

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 +288 -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": 87,
5
+ "total_frames": 34795,
6
+ "total_tasks": 1,
7
+ "total_videos": 261,
8
+ "total_chunks": 1,
9
+ "chunks_size": 1000,
10
+ "fps": 10,
11
+ "splits": {
12
+ "train": "0:87"
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 sha256:0e92e21f5a20a5de3272b892d18104d218a19c232d184af859046e0c12c82bc2
3
+ size 302847324
train_config.json ADDED
@@ -0,0 +1,288 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "stack_red_blocks_85",
4
+ "root": "/home/yizhang/Projects/hackathon-ki-fabrik/data/stack_red_blocks_85",
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
+ ],
94
+ "image_transforms": {
95
+ "enable": false,
96
+ "max_num_transforms": 3,
97
+ "random_order": false,
98
+ "tfs": {
99
+ "brightness": {
100
+ "weight": 1.0,
101
+ "type": "ColorJitter",
102
+ "kwargs": {
103
+ "brightness": [
104
+ 0.8,
105
+ 1.2
106
+ ]
107
+ }
108
+ },
109
+ "contrast": {
110
+ "weight": 1.0,
111
+ "type": "ColorJitter",
112
+ "kwargs": {
113
+ "contrast": [
114
+ 0.8,
115
+ 1.2
116
+ ]
117
+ }
118
+ },
119
+ "saturation": {
120
+ "weight": 1.0,
121
+ "type": "ColorJitter",
122
+ "kwargs": {
123
+ "saturation": [
124
+ 0.5,
125
+ 1.5
126
+ ]
127
+ }
128
+ },
129
+ "hue": {
130
+ "weight": 1.0,
131
+ "type": "ColorJitter",
132
+ "kwargs": {
133
+ "hue": [
134
+ -0.05,
135
+ 0.05
136
+ ]
137
+ }
138
+ },
139
+ "sharpness": {
140
+ "weight": 1.0,
141
+ "type": "SharpnessJitter",
142
+ "kwargs": {
143
+ "sharpness": [
144
+ 0.5,
145
+ 1.5
146
+ ]
147
+ }
148
+ }
149
+ }
150
+ },
151
+ "revision": null,
152
+ "use_imagenet_stats": true,
153
+ "video_backend": "pyav"
154
+ },
155
+ "env": null,
156
+ "policy": {
157
+ "type": "ditflow_image",
158
+ "n_obs_steps": 2,
159
+ "normalization_mapping": {
160
+ "VISUAL": "MEAN_STD",
161
+ "STATE": "MIN_MAX",
162
+ "ACTION": "MIN_MAX"
163
+ },
164
+ "input_features": {
165
+ "observation.state": {
166
+ "type": "STATE",
167
+ "shape": [
168
+ 18
169
+ ]
170
+ },
171
+ "observation.images.rgb_static": {
172
+ "type": "VISUAL",
173
+ "shape": [
174
+ 3,
175
+ 256,
176
+ 256
177
+ ]
178
+ },
179
+ "observation.images.rgb_left": {
180
+ "type": "VISUAL",
181
+ "shape": [
182
+ 3,
183
+ 256,
184
+ 256
185
+ ]
186
+ },
187
+ "observation.images.rgb_right": {
188
+ "type": "VISUAL",
189
+ "shape": [
190
+ 3,
191
+ 256,
192
+ 256
193
+ ]
194
+ }
195
+ },
196
+ "output_features": {
197
+ "action": {
198
+ "type": "ACTION",
199
+ "shape": [
200
+ 17
201
+ ]
202
+ }
203
+ },
204
+ "device": "cuda",
205
+ "use_amp": false,
206
+ "push_to_hub": false,
207
+ "repo_id": null,
208
+ "private": null,
209
+ "tags": null,
210
+ "license": null,
211
+ "horizon": 16,
212
+ "n_action_steps": 8,
213
+ "drop_n_last_frames": 7,
214
+ "vision_backbone": "resnet18",
215
+ "crop_shape": [
216
+ 224,
217
+ 224
218
+ ],
219
+ "crop_is_random": true,
220
+ "pretrained_backbone_weights": null,
221
+ "use_group_norm": true,
222
+ "spatial_softmax_num_keypoints": 32,
223
+ "use_separate_rgb_encoder_per_camera": true,
224
+ "frequency_embedding_dim": 256,
225
+ "hidden_dim": 512,
226
+ "num_blocks": 6,
227
+ "num_heads": 16,
228
+ "dropout": 0.1,
229
+ "dim_feedforward": 4096,
230
+ "activation": "gelu",
231
+ "training_noise_sampling": "uniform",
232
+ "clip_sample": true,
233
+ "clip_sample_range": 1.0,
234
+ "num_inference_steps": 100,
235
+ "do_mask_loss_for_padding": false,
236
+ "optimizer_lr": 0.0002,
237
+ "optimizer_betas": [
238
+ 0.95,
239
+ 0.999
240
+ ],
241
+ "optimizer_eps": 1e-08,
242
+ "optimizer_weight_decay": 1e-06,
243
+ "scheduler_name": "cosine",
244
+ "scheduler_warmup_steps": 500
245
+ },
246
+ "output_dir": "/home/yizhang/Projects/hackathon-ki-fabrik/outputs/ditflow_image_stack_red_blocks_85",
247
+ "job_name": "ditflow_image_stack_red_blocks_85",
248
+ "resume": false,
249
+ "seed": 1000,
250
+ "num_workers": 0,
251
+ "batch_size": 64,
252
+ "steps": 20000,
253
+ "eval_freq": 20000,
254
+ "log_freq": 200,
255
+ "save_checkpoint": true,
256
+ "save_freq": 5000,
257
+ "use_policy_training_preset": true,
258
+ "optimizer": {
259
+ "type": "adam",
260
+ "lr": 0.0002,
261
+ "weight_decay": 1e-06,
262
+ "grad_clip_norm": 10.0,
263
+ "betas": [
264
+ 0.95,
265
+ 0.999
266
+ ],
267
+ "eps": 1e-08
268
+ },
269
+ "scheduler": {
270
+ "type": "diffuser",
271
+ "num_warmup_steps": 500,
272
+ "name": "cosine"
273
+ },
274
+ "eval": {
275
+ "n_episodes": 50,
276
+ "batch_size": 50,
277
+ "use_async_envs": false
278
+ },
279
+ "wandb": {
280
+ "enable": true,
281
+ "disable_artifact": true,
282
+ "project": "paper",
283
+ "entity": "470620104-technical-university-of-munich",
284
+ "notes": null,
285
+ "run_id": "7g2q64na",
286
+ "mode": null
287
+ }
288
+ }