Hugging Face's logo

Spring14th
/
hybrid_150k

Safetensors
Model card Files
xet
Community
Spring14th commited on
Commit
0c7eef9
·
verified ·
1 Parent(s): 1276376

Upload folder using huggingface_hub

Browse files
Files changed (3) hide show
  1. config.json +165 -0
  2. model.safetensors +3 -0
  3. train_config.json +309 -0
config.json ADDED
@@ -0,0 +1,165 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "hybrid",
3
+ "n_obs_steps": 1,
4
+ "normalization_mapping": {},
5
+ "input_features": {
6
+ "observation.images.top": {
7
+ "type": "VISUAL",
8
+ "shape": [
9
+ 3,
10
+ 480,
11
+ 640
12
+ ]
13
+ },
14
+ "observation.state": {
15
+ "type": "STATE",
16
+ "shape": [
17
+ 14
18
+ ]
19
+ }
20
+ },
21
+ "output_features": {
22
+ "action": {
23
+ "type": "ACTION",
24
+ "shape": [
25
+ 14
26
+ ]
27
+ }
28
+ },
29
+ "device": "cuda",
30
+ "use_amp": false,
31
+ "act_config": {
32
+ "type": "act",
33
+ "n_obs_steps": 2,
34
+ "normalization_mapping": {
35
+ "VISUAL": "MEAN_STD",
36
+ "STATE": "MEAN_STD",
37
+ "ACTION": "MEAN_STD"
38
+ },
39
+ "input_features": {
40
+ "observation.state": {
41
+ "type": "STATE",
42
+ "shape": [
43
+ 14
44
+ ]
45
+ },
46
+ "observation.images.top": {
47
+ "type": "VISUAL",
48
+ "shape": [
49
+ 3,
50
+ 480,
51
+ 640
52
+ ]
53
+ }
54
+ },
55
+ "output_features": {
56
+ "action": {
57
+ "type": "ACTION",
58
+ "shape": [
59
+ 14
60
+ ]
61
+ }
62
+ },
63
+ "device": "cuda",
64
+ "use_amp": false,
65
+ "chunk_size": 400,
66
+ "n_action_steps": 400,
67
+ "vision_backbone": "resnet18",
68
+ "pretrained_backbone_weights": "ResNet18_Weights.IMAGENET1K_V1",
69
+ "replace_final_stride_with_dilation": false,
70
+ "pre_norm": false,
71
+ "dim_model": 512,
72
+ "n_heads": 8,
73
+ "dim_feedforward": 3200,
74
+ "feedforward_activation": "relu",
75
+ "n_encoder_layers": 4,
76
+ "n_decoder_layers": 1,
77
+ "use_vae": true,
78
+ "latent_dim": 32,
79
+ "n_vae_encoder_layers": 4,
80
+ "temporal_ensemble_coeff": null,
81
+ "dropout": 0.1,
82
+ "kl_weight": 10.0,
83
+ "smoothness_weight": 0.0,
84
+ "smoothness_type": "velocity",
85
+ "optimizer_lr": 1e-05,
86
+ "optimizer_weight_decay": 0.0001,
87
+ "optimizer_lr_backbone": 1e-05
88
+ },
89
+ "diffusion_config": {
90
+ "type": "diffusion",
91
+ "n_obs_steps": 2,
92
+ "normalization_mapping": {
93
+ "VISUAL": "MEAN_STD",
94
+ "STATE": "MEAN_STD",
95
+ "ACTION": "MEAN_STD"
96
+ },
97
+ "input_features": {
98
+ "observation.state": {
99
+ "type": "STATE",
100
+ "shape": [
101
+ 14
102
+ ]
103
+ },
104
+ "observation.images.top": {
105
+ "type": "VISUAL",
106
+ "shape": [
107
+ 3,
108
+ 480,
109
+ 640
110
+ ]
111
+ }
112
+ },
113
+ "output_features": {
114
+ "action": {
115
+ "type": "ACTION",
116
+ "shape": [
117
+ 14
118
+ ]
119
+ }
120
+ },
121
+ "device": "cuda",
122
+ "use_amp": false,
123
+ "horizon": 16,
124
+ "n_action_steps": 1,
125
+ "drop_n_last_frames": 7,
126
+ "vision_backbone": "resnet18",
127
+ "crop_shape": null,
128
+ "crop_is_random": true,
129
+ "pretrained_backbone_weights": null,
130
+ "use_group_norm": true,
131
+ "spatial_softmax_num_keypoints": 32,
132
+ "use_separate_rgb_encoder_per_camera": false,
133
+ "down_dims": [
134
+ 128,
135
+ 256,
136
+ 512
137
+ ],
138
+ "kernel_size": 5,
139
+ "n_groups": 8,
140
+ "diffusion_step_embed_dim": 128,
141
+ "use_film_scale_modulation": true,
142
+ "noise_scheduler_type": "DDIM",
143
+ "num_train_timesteps": 100,
144
+ "beta_schedule": "squaredcos_cap_v2",
145
+ "beta_start": 0.0001,
146
+ "beta_end": 0.02,
147
+ "prediction_type": "epsilon",
148
+ "clip_sample": true,
149
+ "clip_sample_range": 1.0,
150
+ "num_inference_steps": 80,
151
+ "do_mask_loss_for_padding": false,
152
+ "optimizer_lr": 0.0001,
153
+ "optimizer_betas": [
154
+ 0.95,
155
+ 0.999
156
+ ],
157
+ "optimizer_eps": 1e-08,
158
+ "optimizer_weight_decay": 1e-06,
159
+ "scheduler_name": "cosine",
160
+ "scheduler_warmup_steps": 500
161
+ },
162
+ "act_loss_weight": 1.0,
163
+ "diffusion_loss_weight": 4.0,
164
+ "action_blend_alpha": 0.7
165
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:de14cc66053de076676f0a8ecb13fb2f368e5672fe2e0e2a3679c8deebb879b0
3
+ size 322278544
train_config.json ADDED
@@ -0,0 +1,309 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "lerobot/aloha_sim_insertion_human",
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": "pyav"
66
+ },
67
+ "env": {
68
+ "type": "aloha",
69
+ "task": "AlohaInsertion-v0",
70
+ "fps": 50,
71
+ "features": {
72
+ "action": {
73
+ "type": "ACTION",
74
+ "shape": [
75
+ 14
76
+ ]
77
+ },
78
+ "agent_pos": {
79
+ "type": "STATE",
80
+ "shape": [
81
+ 14
82
+ ]
83
+ },
84
+ "pixels/top": {
85
+ "type": "VISUAL",
86
+ "shape": [
87
+ 480,
88
+ 640,
89
+ 3
90
+ ]
91
+ }
92
+ },
93
+ "features_map": {
94
+ "action": "action",
95
+ "agent_pos": "observation.state",
96
+ "top": "observation.image.top",
97
+ "pixels/top": "observation.images.top"
98
+ },
99
+ "episode_length": 400,
100
+ "obs_type": "pixels_agent_pos",
101
+ "render_mode": "rgb_array"
102
+ },
103
+ "policy": {
104
+ "type": "hybrid",
105
+ "n_obs_steps": 1,
106
+ "normalization_mapping": {},
107
+ "input_features": {
108
+ "observation.images.top": {
109
+ "type": "VISUAL",
110
+ "shape": [
111
+ 3,
112
+ 480,
113
+ 640
114
+ ]
115
+ },
116
+ "observation.state": {
117
+ "type": "STATE",
118
+ "shape": [
119
+ 14
120
+ ]
121
+ }
122
+ },
123
+ "output_features": {
124
+ "action": {
125
+ "type": "ACTION",
126
+ "shape": [
127
+ 14
128
+ ]
129
+ }
130
+ },
131
+ "device": "cuda",
132
+ "use_amp": false,
133
+ "act_config": {
134
+ "type": "act",
135
+ "n_obs_steps": 2,
136
+ "normalization_mapping": {
137
+ "VISUAL": "MEAN_STD",
138
+ "STATE": "MEAN_STD",
139
+ "ACTION": "MEAN_STD"
140
+ },
141
+ "input_features": {
142
+ "observation.state": {
143
+ "type": "STATE",
144
+ "shape": [
145
+ 14
146
+ ]
147
+ },
148
+ "observation.images.top": {
149
+ "type": "VISUAL",
150
+ "shape": [
151
+ 3,
152
+ 480,
153
+ 640
154
+ ]
155
+ }
156
+ },
157
+ "output_features": {
158
+ "action": {
159
+ "type": "ACTION",
160
+ "shape": [
161
+ 14
162
+ ]
163
+ }
164
+ },
165
+ "device": "cuda",
166
+ "use_amp": false,
167
+ "chunk_size": 400,
168
+ "n_action_steps": 400,
169
+ "vision_backbone": "resnet18",
170
+ "pretrained_backbone_weights": "ResNet18_Weights.IMAGENET1K_V1",
171
+ "replace_final_stride_with_dilation": false,
172
+ "pre_norm": false,
173
+ "dim_model": 512,
174
+ "n_heads": 8,
175
+ "dim_feedforward": 3200,
176
+ "feedforward_activation": "relu",
177
+ "n_encoder_layers": 4,
178
+ "n_decoder_layers": 1,
179
+ "use_vae": true,
180
+ "latent_dim": 32,
181
+ "n_vae_encoder_layers": 4,
182
+ "temporal_ensemble_coeff": null,
183
+ "dropout": 0.1,
184
+ "kl_weight": 10.0,
185
+ "smoothness_weight": 0.0,
186
+ "smoothness_type": "velocity",
187
+ "optimizer_lr": 1e-05,
188
+ "optimizer_weight_decay": 0.0001,
189
+ "optimizer_lr_backbone": 1e-05
190
+ },
191
+ "diffusion_config": {
192
+ "type": "diffusion",
193
+ "n_obs_steps": 2,
194
+ "normalization_mapping": {
195
+ "VISUAL": "MEAN_STD",
196
+ "STATE": "MEAN_STD",
197
+ "ACTION": "MEAN_STD"
198
+ },
199
+ "input_features": {
200
+ "observation.state": {
201
+ "type": "STATE",
202
+ "shape": [
203
+ 14
204
+ ]
205
+ },
206
+ "observation.images.top": {
207
+ "type": "VISUAL",
208
+ "shape": [
209
+ 3,
210
+ 480,
211
+ 640
212
+ ]
213
+ }
214
+ },
215
+ "output_features": {
216
+ "action": {
217
+ "type": "ACTION",
218
+ "shape": [
219
+ 14
220
+ ]
221
+ }
222
+ },
223
+ "device": "cuda",
224
+ "use_amp": false,
225
+ "horizon": 16,
226
+ "n_action_steps": 1,
227
+ "drop_n_last_frames": 7,
228
+ "vision_backbone": "resnet18",
229
+ "crop_shape": null,
230
+ "crop_is_random": true,
231
+ "pretrained_backbone_weights": null,
232
+ "use_group_norm": true,
233
+ "spatial_softmax_num_keypoints": 32,
234
+ "use_separate_rgb_encoder_per_camera": false,
235
+ "down_dims": [
236
+ 128,
237
+ 256,
238
+ 512
239
+ ],
240
+ "kernel_size": 5,
241
+ "n_groups": 8,
242
+ "diffusion_step_embed_dim": 128,
243
+ "use_film_scale_modulation": true,
244
+ "noise_scheduler_type": "DDIM",
245
+ "num_train_timesteps": 100,
246
+ "beta_schedule": "squaredcos_cap_v2",
247
+ "beta_start": 0.0001,
248
+ "beta_end": 0.02,
249
+ "prediction_type": "epsilon",
250
+ "clip_sample": true,
251
+ "clip_sample_range": 1.0,
252
+ "num_inference_steps": 80,
253
+ "do_mask_loss_for_padding": false,
254
+ "optimizer_lr": 0.0001,
255
+ "optimizer_betas": [
256
+ 0.95,
257
+ 0.999
258
+ ],
259
+ "optimizer_eps": 1e-08,
260
+ "optimizer_weight_decay": 1e-06,
261
+ "scheduler_name": "cosine",
262
+ "scheduler_warmup_steps": 500
263
+ },
264
+ "act_loss_weight": 1.0,
265
+ "diffusion_loss_weight": 4.0,
266
+ "action_blend_alpha": 0.7
267
+ },
268
+ "output_dir": "/root/autodl-tmp/.autodl/output_hybrid",
269
+ "job_name": "aloha_hybrid",
270
+ "resume": true,
271
+ "seed": 1000,
272
+ "num_workers": 8,
273
+ "batch_size": 32,
274
+ "steps": 150000,
275
+ "eval_freq": 500000,
276
+ "log_freq": 400,
277
+ "save_checkpoint": true,
278
+ "save_freq": 10000,
279
+ "use_policy_training_preset": true,
280
+ "optimizer": {
281
+ "type": "adam",
282
+ "lr": 0.0001,
283
+ "weight_decay": 1e-06,
284
+ "grad_clip_norm": 10.0,
285
+ "betas": [
286
+ 0.95,
287
+ 0.999
288
+ ],
289
+ "eps": 1e-08
290
+ },
291
+ "scheduler": {
292
+ "type": "diffuser",
293
+ "num_warmup_steps": 500,
294
+ "name": "cosine"
295
+ },
296
+ "eval": {
297
+ "n_episodes": 50,
298
+ "batch_size": 50,
299
+ "use_async_envs": false
300
+ },
301
+ "wandb": {
302
+ "enable": true,
303
+ "disable_artifact": false,
304
+ "project": "hybrid_50k",
305
+ "entity": null,
306
+ "notes": null,
307
+ "run_id": "hybrid-resume"
308
+ }
309
+ }