Hugging Face's logo

jorgemunozl
/
pi05_unified

Safetensors
Model card Files
xet
 Community
jorgemunozl commited on
Commit
d6bf622
·
verified ·
1 Parent(s): 0e11f4c

Upload folder using huggingface_hub

Browse files
Files changed (7) hide show
  1. checkpoints/config.json +98 -0
  2. checkpoints/model.safetensors +3 -0
  3. checkpoints/policy_postprocessor.json +36 -0
  4. checkpoints/policy_postprocessor_step_0_unnormalizer_processor.safetensors +3 -0
  5. checkpoints/policy_preprocessor.json +91 -0
  6. checkpoints/policy_preprocessor_step_2_normalizer_processor.safetensors +3 -0
  7. checkpoints/train_config.json +282 -0
checkpoints/config.json ADDED
@@ -0,0 +1,98 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "pi05_ki",
3
+ "n_obs_steps": 1,
4
+ "input_features": {
5
+ "observation.state": {
6
+ "type": "STATE",
7
+ "shape": [
8
+ 14
9
+ ]
10
+ },
11
+ "observation.images.left": {
12
+ "type": "VISUAL",
13
+ "shape": [
14
+ 3,
15
+ 376,
16
+ 672
17
+ ]
18
+ },
19
+ "observation.images.top": {
20
+ "type": "VISUAL",
21
+ "shape": [
22
+ 3,
23
+ 376,
24
+ 672
25
+ ]
26
+ },
27
+ "observation.images.right": {
28
+ "type": "VISUAL",
29
+ "shape": [
30
+ 3,
31
+ 376,
32
+ 672
33
+ ]
34
+ }
35
+ },
36
+ "output_features": {
37
+ "action": {
38
+ "type": "ACTION",
39
+ "shape": [
40
+ 14
41
+ ]
42
+ }
43
+ },
44
+ "device": "cuda",
45
+ "use_amp": false,
46
+ "use_peft": false,
47
+ "push_to_hub": true,
48
+ "repo_id": "NONHUMAN-RESEARCH/pi05_unified",
49
+ "private": null,
50
+ "tags": null,
51
+ "license": null,
52
+ "pretrained_path": "lerobot/pi05_base",
53
+ "alpha": 10.0,
54
+ "max_decoding_steps": 100,
55
+ "temperature": 0.1,
56
+ "eos_token_id": 1,
57
+ "paligemma_variant": "gemma_2b",
58
+ "action_expert_variant": "gemma_300m",
59
+ "dtype": "bfloat16",
60
+ "chunk_size": 50,
61
+ "n_action_steps": 50,
62
+ "max_state_dim": 32,
63
+ "max_action_dim": 32,
64
+ "num_inference_steps": 5,
65
+ "time_sampling_beta_alpha": 1.5,
66
+ "time_sampling_beta_beta": 1.0,
67
+ "time_sampling_scale": 0.999,
68
+ "time_sampling_offset": 0.001,
69
+ "min_period": 0.004,
70
+ "max_period": 4.0,
71
+ "rtc_config": null,
72
+ "image_resolution": [
73
+ 224,
74
+ 224
75
+ ],
76
+ "empty_cameras": 0,
77
+ "tokenizer_max_length": 200,
78
+ "subtask_prediction_frequency": 100,
79
+ "normalization_mapping": {
80
+ "VISUAL": "IDENTITY",
81
+ "STATE": "QUANTILES",
82
+ "ACTION": "QUANTILES"
83
+ },
84
+ "gradient_checkpointing": true,
85
+ "compile_model": false,
86
+ "compile_mode": "max-autotune",
87
+ "optimizer_lr": 1.5e-05,
88
+ "optimizer_betas": [
89
+ 0.9,
90
+ 0.95
91
+ ],
92
+ "optimizer_eps": 1e-08,
93
+ "optimizer_weight_decay": 0.01,
94
+ "optimizer_grad_clip_norm": 1.0,
95
+ "scheduler_warmup_steps": 1000,
96
+ "scheduler_decay_steps": 14000,
97
+ "scheduler_decay_lr": 2.5e-06
98
+ }
checkpoints/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9336c14cfd0fdf866b5f0770f299402eafaaee00ac33a039708bf5c9b43ac3be
3
+ size 7473096344
checkpoints/policy_postprocessor.json ADDED
@@ -0,0 +1,36 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "name": "policy_postprocessor",
3
+ "steps": [
4
+ {
5
+ "registry_name": "unnormalizer_processor",
6
+ "config": {
7
+ "eps": 1e-08,
8
+ "features": {
9
+ "action": {
10
+ "type": "ACTION",
11
+ "shape": [
12
+ 14
13
+ ]
14
+ }
15
+ },
16
+ "norm_map": {
17
+ "VISUAL": "IDENTITY",
18
+ "STATE": "QUANTILES",
19
+ "ACTION": "QUANTILES"
20
+ }
21
+ },
22
+ "state_file": "policy_postprocessor_step_0_unnormalizer_processor.safetensors"
23
+ },
24
+ {
25
+ "registry_name": "device_processor",
26
+ "config": {
27
+ "device": "cpu",
28
+ "float_dtype": null
29
+ }
30
+ },
31
+ {
32
+ "registry_name": "pi05_ki_detokenize_subtask_processor_step",
33
+ "config": {}
34
+ }
35
+ ]
36
+ }
checkpoints/policy_postprocessor_step_0_unnormalizer_processor.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6a0e0f26516dbf1e52cbdf9637f23e6093c821f58662138eadaa49c324855946
3
+ size 9168
checkpoints/policy_preprocessor.json ADDED
@@ -0,0 +1,91 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "name": "policy_preprocessor",
3
+ "steps": [
4
+ {
5
+ "registry_name": "rename_observations_processor",
6
+ "config": {
7
+ "rename_map": {}
8
+ }
9
+ },
10
+ {
11
+ "registry_name": "to_batch_processor",
12
+ "config": {}
13
+ },
14
+ {
15
+ "registry_name": "normalizer_processor",
16
+ "config": {
17
+ "eps": 1e-08,
18
+ "features": {
19
+ "observation.state": {
20
+ "type": "STATE",
21
+ "shape": [
22
+ 14
23
+ ]
24
+ },
25
+ "observation.images.left": {
26
+ "type": "VISUAL",
27
+ "shape": [
28
+ 3,
29
+ 376,
30
+ 672
31
+ ]
32
+ },
33
+ "observation.images.top": {
34
+ "type": "VISUAL",
35
+ "shape": [
36
+ 3,
37
+ 376,
38
+ 672
39
+ ]
40
+ },
41
+ "observation.images.right": {
42
+ "type": "VISUAL",
43
+ "shape": [
44
+ 3,
45
+ 376,
46
+ 672
47
+ ]
48
+ },
49
+ "action": {
50
+ "type": "ACTION",
51
+ "shape": [
52
+ 14
53
+ ]
54
+ }
55
+ },
56
+ "norm_map": {
57
+ "VISUAL": "IDENTITY",
58
+ "STATE": "QUANTILES",
59
+ "ACTION": "QUANTILES"
60
+ }
61
+ },
62
+ "state_file": "policy_preprocessor_step_2_normalizer_processor.safetensors"
63
+ },
64
+ {
65
+ "registry_name": "pi05_ki_prepare_state_tokenizer_processor_step",
66
+ "config": {}
67
+ },
68
+ {
69
+ "registry_name": "pi05_ki_tokenize_subtask_processor_step",
70
+ "config": {}
71
+ },
72
+ {
73
+ "registry_name": "tokenizer_processor",
74
+ "config": {
75
+ "max_length": 200,
76
+ "task_key": "task",
77
+ "padding_side": "right",
78
+ "padding": "max_length",
79
+ "truncation": true,
80
+ "tokenizer_name": "google/paligemma-3b-pt-224"
81
+ }
82
+ },
83
+ {
84
+ "registry_name": "device_processor",
85
+ "config": {
86
+ "device": "cuda",
87
+ "float_dtype": null
88
+ }
89
+ }
90
+ ]
91
+ }
checkpoints/policy_preprocessor_step_2_normalizer_processor.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6a0e0f26516dbf1e52cbdf9637f23e6093c821f58662138eadaa49c324855946
3
+ size 9168
checkpoints/train_config.json ADDED
@@ -0,0 +1,282 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "NONHUMAN-RESEARCH/general-task-index",
4
+ "root": null,
5
+ "episodes": [
6
+ 110,
7
+ 79,
8
+ 402,
9
+ 174,
10
+ 421,
11
+ 357,
12
+ 81,
13
+ 388,
14
+ 3,
15
+ 142,
16
+ 301,
17
+ 229,
18
+ 112,
19
+ 214,
20
+ 424,
21
+ 359,
22
+ 332,
23
+ 366,
24
+ 101,
25
+ 287,
26
+ 13,
27
+ 308,
28
+ 258,
29
+ 119,
30
+ 111,
31
+ 47,
32
+ 15,
33
+ 16,
34
+ 216,
35
+ 302,
36
+ 44,
37
+ 279,
38
+ 433,
39
+ 346,
40
+ 52,
41
+ 377,
42
+ 71,
43
+ 114,
44
+ 125,
45
+ 140,
46
+ 379,
47
+ 12,
48
+ 57,
49
+ 327
50
+ ],
51
+ "image_transforms": {
52
+ "enable": false,
53
+ "max_num_transforms": 3,
54
+ "random_order": false,
55
+ "tfs": {
56
+ "brightness": {
57
+ "weight": 1.0,
58
+ "type": "ColorJitter",
59
+ "kwargs": {
60
+ "brightness": [
61
+ 0.8,
62
+ 1.2
63
+ ]
64
+ }
65
+ },
66
+ "contrast": {
67
+ "weight": 1.0,
68
+ "type": "ColorJitter",
69
+ "kwargs": {
70
+ "contrast": [
71
+ 0.8,
72
+ 1.2
73
+ ]
74
+ }
75
+ },
76
+ "saturation": {
77
+ "weight": 1.0,
78
+ "type": "ColorJitter",
79
+ "kwargs": {
80
+ "saturation": [
81
+ 0.5,
82
+ 1.5
83
+ ]
84
+ }
85
+ },
86
+ "hue": {
87
+ "weight": 1.0,
88
+ "type": "ColorJitter",
89
+ "kwargs": {
90
+ "hue": [
91
+ -0.05,
92
+ 0.05
93
+ ]
94
+ }
95
+ },
96
+ "sharpness": {
97
+ "weight": 1.0,
98
+ "type": "SharpnessJitter",
99
+ "kwargs": {
100
+ "sharpness": [
101
+ 0.5,
102
+ 1.5
103
+ ]
104
+ }
105
+ },
106
+ "affine": {
107
+ "weight": 1.0,
108
+ "type": "RandomAffine",
109
+ "kwargs": {
110
+ "degrees": [
111
+ -5.0,
112
+ 5.0
113
+ ],
114
+ "translate": [
115
+ 0.05,
116
+ 0.05
117
+ ]
118
+ }
119
+ }
120
+ }
121
+ },
122
+ "revision": null,
123
+ "use_imagenet_stats": true,
124
+ "video_backend": "torchcodec",
125
+ "streaming": false
126
+ },
127
+ "env": null,
128
+ "policy": {
129
+ "type": "pi05_ki",
130
+ "n_obs_steps": 1,
131
+ "input_features": {
132
+ "observation.state": {
133
+ "type": "STATE",
134
+ "shape": [
135
+ 14
136
+ ]
137
+ },
138
+ "observation.images.left": {
139
+ "type": "VISUAL",
140
+ "shape": [
141
+ 3,
142
+ 376,
143
+ 672
144
+ ]
145
+ },
146
+ "observation.images.top": {
147
+ "type": "VISUAL",
148
+ "shape": [
149
+ 3,
150
+ 376,
151
+ 672
152
+ ]
153
+ },
154
+ "observation.images.right": {
155
+ "type": "VISUAL",
156
+ "shape": [
157
+ 3,
158
+ 376,
159
+ 672
160
+ ]
161
+ }
162
+ },
163
+ "output_features": {
164
+ "action": {
165
+ "type": "ACTION",
166
+ "shape": [
167
+ 14
168
+ ]
169
+ }
170
+ },
171
+ "device": "cuda",
172
+ "use_amp": false,
173
+ "use_peft": false,
174
+ "push_to_hub": true,
175
+ "repo_id": "NONHUMAN-RESEARCH/pi05_unified",
176
+ "private": null,
177
+ "tags": null,
178
+ "license": null,
179
+ "pretrained_path": "lerobot/pi05_base",
180
+ "alpha": 10.0,
181
+ "max_decoding_steps": 100,
182
+ "temperature": 0.1,
183
+ "eos_token_id": 1,
184
+ "paligemma_variant": "gemma_2b",
185
+ "action_expert_variant": "gemma_300m",
186
+ "dtype": "bfloat16",
187
+ "chunk_size": 50,
188
+ "n_action_steps": 50,
189
+ "max_state_dim": 32,
190
+ "max_action_dim": 32,
191
+ "num_inference_steps": 5,
192
+ "time_sampling_beta_alpha": 1.5,
193
+ "time_sampling_beta_beta": 1.0,
194
+ "time_sampling_scale": 0.999,
195
+ "time_sampling_offset": 0.001,
196
+ "min_period": 0.004,
197
+ "max_period": 4.0,
198
+ "rtc_config": null,
199
+ "image_resolution": [
200
+ 224,
201
+ 224
202
+ ],
203
+ "empty_cameras": 0,
204
+ "tokenizer_max_length": 200,
205
+ "subtask_prediction_frequency": 100,
206
+ "normalization_mapping": {
207
+ "VISUAL": "IDENTITY",
208
+ "STATE": "QUANTILES",
209
+ "ACTION": "QUANTILES"
210
+ },
211
+ "gradient_checkpointing": true,
212
+ "compile_model": false,
213
+ "compile_mode": "max-autotune",
214
+ "optimizer_lr": 1.5e-05,
215
+ "optimizer_betas": [
216
+ 0.9,
217
+ 0.95
218
+ ],
219
+ "optimizer_eps": 1e-08,
220
+ "optimizer_weight_decay": 0.01,
221
+ "optimizer_grad_clip_norm": 1.0,
222
+ "scheduler_warmup_steps": 1000,
223
+ "scheduler_decay_steps": 14000,
224
+ "scheduler_decay_lr": 2.5e-06
225
+ },
226
+ "output_dir": "outputs/pi05_unified",
227
+ "job_name": "pi05_unified",
228
+ "resume": false,
229
+ "seed": 1000,
230
+ "num_workers": 4,
231
+ "batch_size": 4,
232
+ "steps": 14000,
233
+ "eval_freq": 200,
234
+ "log_freq": 200,
235
+ "tolerance_s": 0.0001,
236
+ "save_checkpoint": true,
237
+ "save_freq": 500,
238
+ "use_policy_training_preset": true,
239
+ "optimizer": {
240
+ "type": "adamw",
241
+ "lr": 1.5e-05,
242
+ "weight_decay": 0.01,
243
+ "grad_clip_norm": 1.0,
244
+ "betas": [
245
+ 0.9,
246
+ 0.95
247
+ ],
248
+ "eps": 1e-08
249
+ },
250
+ "scheduler": {
251
+ "type": "cosine_decay_with_warmup",
252
+ "num_warmup_steps": 1000,
253
+ "num_decay_steps": 14000,
254
+ "peak_lr": 1.5e-05,
255
+ "decay_lr": 2.5e-06
256
+ },
257
+ "eval": {
258
+ "n_episodes": 50,
259
+ "batch_size": 50,
260
+ "use_async_envs": false
261
+ },
262
+ "wandb": {
263
+ "enable": true,
264
+ "disable_artifact": false,
265
+ "project": "XHUMAN",
266
+ "entity": null,
267
+ "notes": null,
268
+ "run_id": "aigngdcw",
269
+ "mode": null
270
+ },
271
+ "peft": null,
272
+ "use_rabc": false,
273
+ "rabc_progress_path": null,
274
+ "rabc_kappa": 0.01,
275
+ "rabc_epsilon": 1e-06,
276
+ "rabc_head_mode": "sparse",
277
+ "rename_map": {},
278
+ "checkpoint_path": null,
279
+ "split_ratio": 0.9,
280
+ "only_last_best": true,
281
+ "num_eval_batches": 50
282
+ }