Kasra99 commited on
Commit
eff866a
·
verified ·
1 Parent(s): 3d3aa84

checkpoint 012000 (pretrained_model)

Browse files
.gitattributes CHANGED
@@ -38,3 +38,4 @@ checkpoints/004000/pretrained_model_ema/tokenizer/tokenizer.json filter=lfs diff
38
  checkpoints/008000/pretrained_model_ema/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
39
  checkpoints/008000/pretrained_model/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
40
  checkpoints/012000/pretrained_model_ema/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 
 
38
  checkpoints/008000/pretrained_model_ema/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
39
  checkpoints/008000/pretrained_model/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
40
  checkpoints/012000/pretrained_model_ema/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
41
+ checkpoints/012000/pretrained_model/tokenizer/tokenizer.json filter=lfs diff=lfs merge=lfs -text
checkpoints/012000/pretrained_model/config.json ADDED
@@ -0,0 +1,113 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "pi05",
3
+ "n_obs_steps": 1,
4
+ "input_features": {
5
+ "observation.images.head": {
6
+ "type": "VISUAL",
7
+ "shape": [
8
+ 3,
9
+ 480,
10
+ 640
11
+ ]
12
+ },
13
+ "observation.images.left_wrist": {
14
+ "type": "VISUAL",
15
+ "shape": [
16
+ 3,
17
+ 480,
18
+ 640
19
+ ]
20
+ },
21
+ "observation.images.right_wrist": {
22
+ "type": "VISUAL",
23
+ "shape": [
24
+ 3,
25
+ 480,
26
+ 640
27
+ ]
28
+ },
29
+ "observation.state": {
30
+ "type": "STATE",
31
+ "shape": [
32
+ 20
33
+ ]
34
+ }
35
+ },
36
+ "output_features": {
37
+ "action": {
38
+ "type": "ACTION",
39
+ "shape": [
40
+ 20
41
+ ]
42
+ }
43
+ },
44
+ "device": "cuda",
45
+ "use_amp": false,
46
+ "use_peft": false,
47
+ "push_to_hub": false,
48
+ "repo_id": "Kasra99/pi05_dex_warehouse",
49
+ "private": true,
50
+ "tags": [
51
+ "pi05",
52
+ "vla",
53
+ "dexmate-vega",
54
+ "lerobot"
55
+ ],
56
+ "license": null,
57
+ "pretrained_path": "outputs/pi05_dexwh_expert_20260818_2151/checkpoints/last/pretrained_model",
58
+ "pretrained_revision": null,
59
+ "paligemma_variant": "gemma_2b",
60
+ "action_expert_variant": "gemma_300m",
61
+ "dtype": "bfloat16",
62
+ "chunk_size": 50,
63
+ "n_action_steps": 25,
64
+ "use_visual_memory": false,
65
+ "use_proprioceptive_memory": false,
66
+ "memory_frames": 6,
67
+ "memory_stride": 30,
68
+ "memory_temporal_attention_every": 4,
69
+ "max_state_dim": 32,
70
+ "max_action_dim": 32,
71
+ "num_inference_steps": 10,
72
+ "time_sampling_beta_alpha": 1.5,
73
+ "time_sampling_beta_beta": 1.0,
74
+ "time_sampling_scale": 0.999,
75
+ "time_sampling_offset": 0.001,
76
+ "min_period": 0.004,
77
+ "max_period": 4.0,
78
+ "use_relative_actions": false,
79
+ "relative_exclude_joints": [
80
+ "gripper"
81
+ ],
82
+ "action_feature_names": [
83
+ "action"
84
+ ],
85
+ "rtc_config": null,
86
+ "image_resolution": [
87
+ 224,
88
+ 224
89
+ ],
90
+ "empty_cameras": 0,
91
+ "tokenizer_max_length": 200,
92
+ "normalization_mapping": {
93
+ "VISUAL": "IDENTITY",
94
+ "STATE": "QUANTILES",
95
+ "ACTION": "QUANTILES"
96
+ },
97
+ "gradient_checkpointing": true,
98
+ "compile_model": false,
99
+ "compile_mode": "max-autotune",
100
+ "freeze_vision_encoder": true,
101
+ "train_expert_only": true,
102
+ "optimizer_lr": 5e-05,
103
+ "optimizer_betas": [
104
+ 0.9,
105
+ 0.95
106
+ ],
107
+ "optimizer_eps": 1e-08,
108
+ "optimizer_weight_decay": 0.01,
109
+ "optimizer_grad_clip_norm": 1.0,
110
+ "scheduler_warmup_steps": 1000,
111
+ "scheduler_decay_steps": 28000,
112
+ "scheduler_decay_lr": 2.5e-06
113
+ }
checkpoints/012000/pretrained_model/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d0bc41e6b8106fab9a0c6753c5fd72af0d9ab2d7818868e36c616992b596f987
3
+ size 9354050784
checkpoints/012000/pretrained_model/policy_postprocessor.json ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ 20
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": "absolute_actions_processor",
26
+ "config": {
27
+ "enabled": false
28
+ }
29
+ },
30
+ {
31
+ "registry_name": "device_processor",
32
+ "config": {
33
+ "device": "cpu",
34
+ "float_dtype": null
35
+ }
36
+ }
37
+ ]
38
+ }
checkpoints/012000/pretrained_model/policy_postprocessor_step_0_unnormalizer_processor.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:de9c96f3f817a2acb784635f248e543a0a93fd40b90313a073208f3c964bc021
3
+ size 9768
checkpoints/012000/pretrained_model/policy_preprocessor.json ADDED
@@ -0,0 +1,100 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": "relative_actions_processor",
16
+ "config": {
17
+ "enabled": false,
18
+ "exclude_joints": [
19
+ "gripper"
20
+ ],
21
+ "action_names": null
22
+ }
23
+ },
24
+ {
25
+ "registry_name": "normalizer_processor",
26
+ "config": {
27
+ "eps": 1e-08,
28
+ "features": {
29
+ "observation.images.head": {
30
+ "type": "VISUAL",
31
+ "shape": [
32
+ 3,
33
+ 480,
34
+ 640
35
+ ]
36
+ },
37
+ "observation.images.left_wrist": {
38
+ "type": "VISUAL",
39
+ "shape": [
40
+ 3,
41
+ 480,
42
+ 640
43
+ ]
44
+ },
45
+ "observation.images.right_wrist": {
46
+ "type": "VISUAL",
47
+ "shape": [
48
+ 3,
49
+ 480,
50
+ 640
51
+ ]
52
+ },
53
+ "observation.state": {
54
+ "type": "STATE",
55
+ "shape": [
56
+ 20
57
+ ]
58
+ },
59
+ "action": {
60
+ "type": "ACTION",
61
+ "shape": [
62
+ 20
63
+ ]
64
+ }
65
+ },
66
+ "norm_map": {
67
+ "VISUAL": "IDENTITY",
68
+ "STATE": "QUANTILES",
69
+ "ACTION": "QUANTILES"
70
+ }
71
+ },
72
+ "state_file": "policy_preprocessor_step_3_normalizer_processor.safetensors"
73
+ },
74
+ {
75
+ "registry_name": "pi05_prepare_state_tokenizer_processor_step",
76
+ "config": {}
77
+ },
78
+ {
79
+ "registry_name": "tokenizer_processor",
80
+ "config": {
81
+ "max_length": 200,
82
+ "task_key": "task",
83
+ "padding_side": "right",
84
+ "padding": "max_length",
85
+ "truncation": true,
86
+ "tokenizer_name": "tokenizer"
87
+ },
88
+ "artifacts": {
89
+ "tokenizer_name": "tokenizer"
90
+ }
91
+ },
92
+ {
93
+ "registry_name": "device_processor",
94
+ "config": {
95
+ "device": "cuda",
96
+ "float_dtype": null
97
+ }
98
+ }
99
+ ]
100
+ }
checkpoints/012000/pretrained_model/policy_preprocessor_step_3_normalizer_processor.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:59ce08d9bedecacd8ead672e026dd638c1dc0d4d799a6c0b6440c3ce9dc55684
3
+ size 9712
checkpoints/012000/pretrained_model/tokenizer/tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7336bb4ce016bf8543de94e3a17c3d6ca502a6fc8628fec6ed88893639d7a296
3
+ size 34387384
checkpoints/012000/pretrained_model/tokenizer/tokenizer_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "backend": "tokenizers",
3
+ "bos_token": "<bos>",
4
+ "clean_up_tokenization_spaces": false,
5
+ "eos_token": "<eos>",
6
+ "extra_special_tokens": [
7
+ "<image>"
8
+ ],
9
+ "is_local": true,
10
+ "mask_token": "<mask>",
11
+ "max_length": 200,
12
+ "model_max_length": 1000000000000000019884624838656,
13
+ "pad_to_multiple_of": null,
14
+ "pad_token": "<pad>",
15
+ "pad_token_type_id": 0,
16
+ "padding_side": "right",
17
+ "processor_class": "PaliGemmaProcessor",
18
+ "sp_model_kwargs": {},
19
+ "spaces_between_special_tokens": false,
20
+ "stride": 0,
21
+ "tokenizer_class": "GemmaTokenizer",
22
+ "truncation_side": "right",
23
+ "truncation_strategy": "longest_first",
24
+ "unk_token": "<unk>",
25
+ "use_default_system_prompt": false
26
+ }
checkpoints/012000/pretrained_model/train_config.json ADDED
@@ -0,0 +1,329 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "Kasra99/dex-warehouse",
4
+ "repo_type": "dataset",
5
+ "root": "data/dex-warehouse-pi05",
6
+ "episodes": null,
7
+ "exclude_episodes": [
8
+ 39,
9
+ 73,
10
+ 92,
11
+ 108,
12
+ 111,
13
+ 134,
14
+ 145,
15
+ 157,
16
+ 168,
17
+ 171,
18
+ 173,
19
+ 176
20
+ ],
21
+ "image_transforms": {
22
+ "enable": false,
23
+ "max_num_transforms": 3,
24
+ "random_order": false,
25
+ "tfs": {
26
+ "brightness": {
27
+ "weight": 1.0,
28
+ "type": "ColorJitter",
29
+ "kwargs": {
30
+ "brightness": [
31
+ 0.8,
32
+ 1.2
33
+ ]
34
+ }
35
+ },
36
+ "contrast": {
37
+ "weight": 1.0,
38
+ "type": "ColorJitter",
39
+ "kwargs": {
40
+ "contrast": [
41
+ 0.8,
42
+ 1.2
43
+ ]
44
+ }
45
+ },
46
+ "saturation": {
47
+ "weight": 1.0,
48
+ "type": "ColorJitter",
49
+ "kwargs": {
50
+ "saturation": [
51
+ 0.5,
52
+ 1.5
53
+ ]
54
+ }
55
+ },
56
+ "hue": {
57
+ "weight": 1.0,
58
+ "type": "ColorJitter",
59
+ "kwargs": {
60
+ "hue": [
61
+ -0.05,
62
+ 0.05
63
+ ]
64
+ }
65
+ },
66
+ "sharpness": {
67
+ "weight": 1.0,
68
+ "type": "SharpnessJitter",
69
+ "kwargs": {
70
+ "sharpness": [
71
+ 0.5,
72
+ 1.5
73
+ ]
74
+ }
75
+ },
76
+ "affine": {
77
+ "weight": 1.0,
78
+ "type": "RandomAffine",
79
+ "kwargs": {
80
+ "degrees": [
81
+ -5.0,
82
+ 5.0
83
+ ],
84
+ "translate": [
85
+ 0.05,
86
+ 0.05
87
+ ]
88
+ }
89
+ }
90
+ }
91
+ },
92
+ "revision": null,
93
+ "use_imagenet_stats": true,
94
+ "video_backend": "torchcodec",
95
+ "return_uint8": true,
96
+ "depth_output_unit": "mm",
97
+ "streaming": false,
98
+ "eval_split": 0.0
99
+ },
100
+ "env": null,
101
+ "policy": {
102
+ "type": "pi05",
103
+ "n_obs_steps": 1,
104
+ "input_features": {
105
+ "observation.images.head": {
106
+ "type": "VISUAL",
107
+ "shape": [
108
+ 3,
109
+ 480,
110
+ 640
111
+ ]
112
+ },
113
+ "observation.images.left_wrist": {
114
+ "type": "VISUAL",
115
+ "shape": [
116
+ 3,
117
+ 480,
118
+ 640
119
+ ]
120
+ },
121
+ "observation.images.right_wrist": {
122
+ "type": "VISUAL",
123
+ "shape": [
124
+ 3,
125
+ 480,
126
+ 640
127
+ ]
128
+ },
129
+ "observation.state": {
130
+ "type": "STATE",
131
+ "shape": [
132
+ 20
133
+ ]
134
+ }
135
+ },
136
+ "output_features": {
137
+ "action": {
138
+ "type": "ACTION",
139
+ "shape": [
140
+ 20
141
+ ]
142
+ }
143
+ },
144
+ "device": "cuda",
145
+ "use_amp": false,
146
+ "use_peft": false,
147
+ "push_to_hub": false,
148
+ "repo_id": "Kasra99/pi05_dex_warehouse",
149
+ "private": true,
150
+ "tags": [
151
+ "pi05",
152
+ "vla",
153
+ "dexmate-vega",
154
+ "lerobot"
155
+ ],
156
+ "license": null,
157
+ "pretrained_path": "outputs/pi05_dexwh_expert_20260818_2151/checkpoints/last/pretrained_model",
158
+ "pretrained_revision": null,
159
+ "paligemma_variant": "gemma_2b",
160
+ "action_expert_variant": "gemma_300m",
161
+ "dtype": "bfloat16",
162
+ "chunk_size": 50,
163
+ "n_action_steps": 25,
164
+ "use_visual_memory": false,
165
+ "use_proprioceptive_memory": false,
166
+ "memory_frames": 6,
167
+ "memory_stride": 30,
168
+ "memory_temporal_attention_every": 4,
169
+ "max_state_dim": 32,
170
+ "max_action_dim": 32,
171
+ "num_inference_steps": 10,
172
+ "time_sampling_beta_alpha": 1.5,
173
+ "time_sampling_beta_beta": 1.0,
174
+ "time_sampling_scale": 0.999,
175
+ "time_sampling_offset": 0.001,
176
+ "min_period": 0.004,
177
+ "max_period": 4.0,
178
+ "use_relative_actions": false,
179
+ "relative_exclude_joints": [
180
+ "gripper"
181
+ ],
182
+ "action_feature_names": [
183
+ "action"
184
+ ],
185
+ "rtc_config": null,
186
+ "image_resolution": [
187
+ 224,
188
+ 224
189
+ ],
190
+ "empty_cameras": 0,
191
+ "tokenizer_max_length": 200,
192
+ "normalization_mapping": {
193
+ "VISUAL": "IDENTITY",
194
+ "STATE": "QUANTILES",
195
+ "ACTION": "QUANTILES"
196
+ },
197
+ "gradient_checkpointing": true,
198
+ "compile_model": false,
199
+ "compile_mode": "max-autotune",
200
+ "freeze_vision_encoder": true,
201
+ "train_expert_only": true,
202
+ "optimizer_lr": 5e-05,
203
+ "optimizer_betas": [
204
+ 0.9,
205
+ 0.95
206
+ ],
207
+ "optimizer_eps": 1e-08,
208
+ "optimizer_weight_decay": 0.01,
209
+ "optimizer_grad_clip_norm": 1.0,
210
+ "scheduler_warmup_steps": 1000,
211
+ "scheduler_decay_steps": 28000,
212
+ "scheduler_decay_lr": 2.5e-06
213
+ },
214
+ "reward_model": null,
215
+ "output_dir": "outputs/pi05_dexwh_expert_20260818_2151",
216
+ "job_name": "pi05_dexwh_expert_20260818_2151",
217
+ "resume": true,
218
+ "seed": 1000,
219
+ "cudnn_deterministic": false,
220
+ "num_workers": 10,
221
+ "batch_size": 32,
222
+ "prefetch_factor": 4,
223
+ "persistent_workers": true,
224
+ "dataloader_multiprocessing_context": "spawn",
225
+ "steps": 28000,
226
+ "env_eval_freq": 20000,
227
+ "log_freq": 50,
228
+ "eval_steps": 0,
229
+ "max_eval_samples": 0,
230
+ "tolerance_s": 0.0001,
231
+ "save_checkpoint": true,
232
+ "save_freq": 4000,
233
+ "checkpoint_format": "safetensors",
234
+ "use_policy_training_preset": true,
235
+ "optimizer": {
236
+ "type": "adamw",
237
+ "lr": 5e-05,
238
+ "weight_decay": 0.01,
239
+ "grad_clip_norm": 1.0,
240
+ "betas": [
241
+ 0.9,
242
+ 0.95
243
+ ],
244
+ "eps": 1e-08
245
+ },
246
+ "scheduler": {
247
+ "type": "cosine_decay_with_warmup",
248
+ "num_warmup_steps": 1000,
249
+ "num_decay_steps": 28000,
250
+ "peak_lr": 5e-05,
251
+ "decay_lr": 2.5e-06
252
+ },
253
+ "parallelism": {
254
+ "dp_replicate": 1,
255
+ "dp_shard": 1,
256
+ "context_parallel": {
257
+ "ring_degree": 1,
258
+ "ulysses_degree": 1
259
+ },
260
+ "cfg_parallel": 1
261
+ },
262
+ "accelerator": {
263
+ "mixed_precision": "no",
264
+ "gradient_accumulation": {
265
+ "steps": 1
266
+ },
267
+ "fsdp": {
268
+ "reshard_after_forward": true,
269
+ "wrap_modules": null,
270
+ "min_num_params": null,
271
+ "cpu_offload": false,
272
+ "ignored_modules": null
273
+ },
274
+ "ddp": {
275
+ "find_unused_parameters": true,
276
+ "gradient_as_bucket_view": false,
277
+ "static_graph": false
278
+ },
279
+ "compile": {
280
+ "enabled": false,
281
+ "backend": "inductor",
282
+ "mode": null,
283
+ "regional": true
284
+ },
285
+ "activation_checkpointing": {
286
+ "mode": "none"
287
+ }
288
+ },
289
+ "eval": {
290
+ "n_episodes": 50,
291
+ "batch_size": 11,
292
+ "use_async_envs": true,
293
+ "recording": false,
294
+ "recording_repo_id": null,
295
+ "recording_private": false
296
+ },
297
+ "ema": {
298
+ "enable": true,
299
+ "decay": 0.99,
300
+ "update_after_step": 0,
301
+ "inv_gamma": 1.0,
302
+ "power": 0.75,
303
+ "min_decay": 0.0,
304
+ "max_decay": 0.9999,
305
+ "use_for_eval": true
306
+ },
307
+ "wandb": {
308
+ "enable": false,
309
+ "disable_artifact": false,
310
+ "project": "lerobot",
311
+ "entity": null,
312
+ "notes": null,
313
+ "run_id": null,
314
+ "mode": null,
315
+ "add_tags": true
316
+ },
317
+ "peft": null,
318
+ "job": {
319
+ "target": null,
320
+ "image": "huggingface/lerobot-gpu:latest",
321
+ "timeout": "2d",
322
+ "detach": false,
323
+ "tags": []
324
+ },
325
+ "save_checkpoint_to_hub": false,
326
+ "sample_weighting": null,
327
+ "rename_map": {},
328
+ "checkpoint_path": "outputs/pi05_dexwh_expert_20260818_2151/checkpoints/last"
329
+ }