Robotics
LeRobot
Safetensors
pi052
pepijn223 HF Staff commited on
Commit
072a319
·
verified ·
1 Parent(s): 882de3d

Upload policy weights, train config and readme

Browse files
Files changed (4) hide show
  1. README.md +0 -0
  2. config.json +134 -0
  3. model.safetensors +3 -0
  4. train_config.json +290 -0
README.md ADDED
The diff for this file is too large to render. See raw diff
 
config.json ADDED
@@ -0,0 +1,134 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "pi052",
3
+ "n_obs_steps": 1,
4
+ "input_features": {
5
+ "observation.images.robot0_agentview_left": {
6
+ "type": "VISUAL",
7
+ "shape": [
8
+ 3,
9
+ 256,
10
+ 256
11
+ ]
12
+ },
13
+ "observation.images.robot0_agentview_right": {
14
+ "type": "VISUAL",
15
+ "shape": [
16
+ 3,
17
+ 256,
18
+ 256
19
+ ]
20
+ },
21
+ "observation.images.robot0_eye_in_hand": {
22
+ "type": "VISUAL",
23
+ "shape": [
24
+ 3,
25
+ 256,
26
+ 256
27
+ ]
28
+ },
29
+ "observation.state": {
30
+ "type": "STATE",
31
+ "shape": [
32
+ 16
33
+ ]
34
+ }
35
+ },
36
+ "output_features": {
37
+ "action": {
38
+ "type": "ACTION",
39
+ "shape": [
40
+ 12
41
+ ]
42
+ }
43
+ },
44
+ "device": "cuda",
45
+ "use_amp": false,
46
+ "use_peft": false,
47
+ "push_to_hub": true,
48
+ "repo_id": "pepijn223/pi052_robocasa_joint_flow1_30k_32gpu",
49
+ "private": null,
50
+ "tags": null,
51
+ "license": null,
52
+ "pretrained_path": "lerobot/pi05_base",
53
+ "pretrained_revision": null,
54
+ "paligemma_variant": "gemma_2b",
55
+ "action_expert_variant": "gemma_300m",
56
+ "dtype": "bfloat16",
57
+ "chunk_size": 20,
58
+ "n_action_steps": 20,
59
+ "max_state_dim": 32,
60
+ "max_action_dim": 32,
61
+ "num_inference_steps": 10,
62
+ "time_sampling_beta_alpha": 1.5,
63
+ "time_sampling_beta_beta": 1.0,
64
+ "time_sampling_scale": 0.999,
65
+ "time_sampling_offset": 0.001,
66
+ "min_period": 0.004,
67
+ "max_period": 4.0,
68
+ "use_relative_actions": false,
69
+ "relative_exclude_joints": [
70
+ "gripper"
71
+ ],
72
+ "action_feature_names": null,
73
+ "rtc_config": null,
74
+ "image_resolution": [
75
+ 224,
76
+ 224
77
+ ],
78
+ "empty_cameras": 0,
79
+ "tokenizer_max_length": 128,
80
+ "normalization_mapping": {
81
+ "VISUAL": "IDENTITY",
82
+ "STATE": "QUANTILES",
83
+ "ACTION": "QUANTILES"
84
+ },
85
+ "gradient_checkpointing": true,
86
+ "compile_model": true,
87
+ "compile_mode": "default",
88
+ "freeze_vision_encoder": false,
89
+ "train_expert_only": false,
90
+ "optimizer_lr": 2.5e-05,
91
+ "optimizer_betas": [
92
+ 0.9,
93
+ 0.95
94
+ ],
95
+ "optimizer_eps": 1e-08,
96
+ "optimizer_weight_decay": 1e-10,
97
+ "optimizer_grad_clip_norm": 1.0,
98
+ "scheduler_warmup_steps": 1000,
99
+ "scheduler_decay_steps": 30000,
100
+ "scheduler_decay_lr": 2.5e-06,
101
+ "recipe_path": "recipes/subtask_joint.yaml",
102
+ "apply_chat_template": false,
103
+ "text_loss_weight": 1.0,
104
+ "flow_loss_weight": 1.0,
105
+ "unfreeze_lm_head": true,
106
+ "plan_dropout_prob": 0.0,
107
+ "memory_dropout_prob": 0.0,
108
+ "subtask_dropout_prob": 0.0,
109
+ "enable_fast_action_loss": true,
110
+ "action_tokenizer_name": "lerobot/fast-action-tokenizer",
111
+ "max_action_tokens": 384,
112
+ "fast_skip_tokens": 1152,
113
+ "fast_action_loss_weight": 1.0,
114
+ "subtask_replan_steps": 0,
115
+ "joint_subtask_conditioning": true,
116
+ "auto_fit_fast_tokenizer": true,
117
+ "fast_tokenizer_cache_dir": "~/.cache/lerobot/fast_tokenizers",
118
+ "fast_tokenizer_fit_samples": 8192,
119
+ "knowledge_insulation": true,
120
+ "use_flashrt_adarms": false,
121
+ "use_compiled_text_ce": true,
122
+ "use_compiled_vision": false,
123
+ "use_flex_attention": false,
124
+ "use_manual_attention": false,
125
+ "manual_attention_scope": "all",
126
+ "lm_head_lr_scale": 1.0,
127
+ "backbone_lr_scale": 1.0,
128
+ "action_expert_lr_scale": 1.0,
129
+ "flow_num_repeats": 5,
130
+ "text_ce_z_loss_weight": 0.0001,
131
+ "use_flashrt_fp8_mlp": false,
132
+ "optimizer_foreach": false,
133
+ "optimizer_fused": true
134
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e3f2008cedd476ace53b11aa561071667ee9ce01dc66970368c904637794c1f
3
+ size 9354050752
train_config.json ADDED
@@ -0,0 +1,290 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "pepijn223/robocasa_pretrain_human300_v4_annotated5",
4
+ "root": null,
5
+ "episodes": null,
6
+ "exclude_episodes": null,
7
+ "image_transforms": {
8
+ "enable": false,
9
+ "max_num_transforms": 3,
10
+ "random_order": false,
11
+ "tfs": {
12
+ "brightness": {
13
+ "weight": 1.0,
14
+ "type": "ColorJitter",
15
+ "kwargs": {
16
+ "brightness": [
17
+ 0.8,
18
+ 1.2
19
+ ]
20
+ }
21
+ },
22
+ "contrast": {
23
+ "weight": 1.0,
24
+ "type": "ColorJitter",
25
+ "kwargs": {
26
+ "contrast": [
27
+ 0.8,
28
+ 1.2
29
+ ]
30
+ }
31
+ },
32
+ "saturation": {
33
+ "weight": 1.0,
34
+ "type": "ColorJitter",
35
+ "kwargs": {
36
+ "saturation": [
37
+ 0.5,
38
+ 1.5
39
+ ]
40
+ }
41
+ },
42
+ "hue": {
43
+ "weight": 1.0,
44
+ "type": "ColorJitter",
45
+ "kwargs": {
46
+ "hue": [
47
+ -0.05,
48
+ 0.05
49
+ ]
50
+ }
51
+ },
52
+ "sharpness": {
53
+ "weight": 1.0,
54
+ "type": "SharpnessJitter",
55
+ "kwargs": {
56
+ "sharpness": [
57
+ 0.5,
58
+ 1.5
59
+ ]
60
+ }
61
+ },
62
+ "affine": {
63
+ "weight": 1.0,
64
+ "type": "RandomAffine",
65
+ "kwargs": {
66
+ "degrees": [
67
+ -5.0,
68
+ 5.0
69
+ ],
70
+ "translate": [
71
+ 0.05,
72
+ 0.05
73
+ ]
74
+ }
75
+ }
76
+ }
77
+ },
78
+ "revision": "main",
79
+ "use_imagenet_stats": true,
80
+ "video_backend": "pyav",
81
+ "return_uint8": false,
82
+ "depth_output_unit": "mm",
83
+ "streaming": false,
84
+ "eval_split": 0.0
85
+ },
86
+ "env": null,
87
+ "policy": {
88
+ "type": "pi052",
89
+ "n_obs_steps": 1,
90
+ "input_features": {
91
+ "observation.images.robot0_agentview_left": {
92
+ "type": "VISUAL",
93
+ "shape": [
94
+ 3,
95
+ 256,
96
+ 256
97
+ ]
98
+ },
99
+ "observation.images.robot0_agentview_right": {
100
+ "type": "VISUAL",
101
+ "shape": [
102
+ 3,
103
+ 256,
104
+ 256
105
+ ]
106
+ },
107
+ "observation.images.robot0_eye_in_hand": {
108
+ "type": "VISUAL",
109
+ "shape": [
110
+ 3,
111
+ 256,
112
+ 256
113
+ ]
114
+ },
115
+ "observation.state": {
116
+ "type": "STATE",
117
+ "shape": [
118
+ 16
119
+ ]
120
+ }
121
+ },
122
+ "output_features": {
123
+ "action": {
124
+ "type": "ACTION",
125
+ "shape": [
126
+ 12
127
+ ]
128
+ }
129
+ },
130
+ "device": "cuda",
131
+ "use_amp": false,
132
+ "use_peft": false,
133
+ "push_to_hub": true,
134
+ "repo_id": "pepijn223/pi052_robocasa_joint_flow1_30k_32gpu",
135
+ "private": null,
136
+ "tags": null,
137
+ "license": null,
138
+ "pretrained_path": "lerobot/pi05_base",
139
+ "pretrained_revision": null,
140
+ "paligemma_variant": "gemma_2b",
141
+ "action_expert_variant": "gemma_300m",
142
+ "dtype": "bfloat16",
143
+ "chunk_size": 20,
144
+ "n_action_steps": 20,
145
+ "max_state_dim": 32,
146
+ "max_action_dim": 32,
147
+ "num_inference_steps": 10,
148
+ "time_sampling_beta_alpha": 1.5,
149
+ "time_sampling_beta_beta": 1.0,
150
+ "time_sampling_scale": 0.999,
151
+ "time_sampling_offset": 0.001,
152
+ "min_period": 0.004,
153
+ "max_period": 4.0,
154
+ "use_relative_actions": false,
155
+ "relative_exclude_joints": [
156
+ "gripper"
157
+ ],
158
+ "action_feature_names": null,
159
+ "rtc_config": null,
160
+ "image_resolution": [
161
+ 224,
162
+ 224
163
+ ],
164
+ "empty_cameras": 0,
165
+ "tokenizer_max_length": 128,
166
+ "normalization_mapping": {
167
+ "VISUAL": "IDENTITY",
168
+ "STATE": "QUANTILES",
169
+ "ACTION": "QUANTILES"
170
+ },
171
+ "gradient_checkpointing": true,
172
+ "compile_model": true,
173
+ "compile_mode": "default",
174
+ "freeze_vision_encoder": false,
175
+ "train_expert_only": false,
176
+ "optimizer_lr": 2.5e-05,
177
+ "optimizer_betas": [
178
+ 0.9,
179
+ 0.95
180
+ ],
181
+ "optimizer_eps": 1e-08,
182
+ "optimizer_weight_decay": 1e-10,
183
+ "optimizer_grad_clip_norm": 1.0,
184
+ "scheduler_warmup_steps": 1000,
185
+ "scheduler_decay_steps": 30000,
186
+ "scheduler_decay_lr": 2.5e-06,
187
+ "recipe_path": "recipes/subtask_joint.yaml",
188
+ "apply_chat_template": false,
189
+ "text_loss_weight": 1.0,
190
+ "flow_loss_weight": 1.0,
191
+ "unfreeze_lm_head": true,
192
+ "plan_dropout_prob": 0.0,
193
+ "memory_dropout_prob": 0.0,
194
+ "subtask_dropout_prob": 0.0,
195
+ "enable_fast_action_loss": true,
196
+ "action_tokenizer_name": "lerobot/fast-action-tokenizer",
197
+ "max_action_tokens": 384,
198
+ "fast_skip_tokens": 1152,
199
+ "fast_action_loss_weight": 1.0,
200
+ "subtask_replan_steps": 0,
201
+ "joint_subtask_conditioning": true,
202
+ "auto_fit_fast_tokenizer": true,
203
+ "fast_tokenizer_cache_dir": "~/.cache/lerobot/fast_tokenizers",
204
+ "fast_tokenizer_fit_samples": 8192,
205
+ "knowledge_insulation": true,
206
+ "use_flashrt_adarms": false,
207
+ "use_compiled_text_ce": true,
208
+ "use_compiled_vision": false,
209
+ "use_flex_attention": false,
210
+ "use_manual_attention": false,
211
+ "manual_attention_scope": "all",
212
+ "lm_head_lr_scale": 1.0,
213
+ "backbone_lr_scale": 1.0,
214
+ "action_expert_lr_scale": 1.0,
215
+ "flow_num_repeats": 5,
216
+ "text_ce_z_loss_weight": 0.0001,
217
+ "use_flashrt_fp8_mlp": false,
218
+ "optimizer_foreach": false,
219
+ "optimizer_fused": true
220
+ },
221
+ "reward_model": null,
222
+ "output_dir": "/fsx/pepijn/outputs/train/pi052-robocasa-joint-flow1-30k_22374833",
223
+ "job_name": "pi052-robocasa-joint-flow1-30k",
224
+ "resume": false,
225
+ "seed": 1000,
226
+ "cudnn_deterministic": false,
227
+ "num_workers": 4,
228
+ "batch_size": 32,
229
+ "prefetch_factor": 4,
230
+ "persistent_workers": true,
231
+ "steps": 30000,
232
+ "env_eval_freq": 30000,
233
+ "log_freq": 100,
234
+ "eval_steps": 0,
235
+ "max_eval_samples": 0,
236
+ "tolerance_s": 0.0001,
237
+ "save_checkpoint": true,
238
+ "save_freq": 5000,
239
+ "use_policy_training_preset": true,
240
+ "optimizer": {
241
+ "type": "adamw",
242
+ "lr": 2.5e-05,
243
+ "weight_decay": 1e-10,
244
+ "grad_clip_norm": 1.0,
245
+ "betas": [
246
+ 0.9,
247
+ 0.95
248
+ ],
249
+ "eps": 1e-08,
250
+ "foreach": false,
251
+ "fused": true
252
+ },
253
+ "scheduler": {
254
+ "type": "cosine_decay_with_warmup",
255
+ "num_warmup_steps": 1000,
256
+ "num_decay_steps": 30000,
257
+ "peak_lr": 2.5e-05,
258
+ "decay_lr": 2.5e-06
259
+ },
260
+ "eval": {
261
+ "n_episodes": 50,
262
+ "batch_size": 50,
263
+ "use_async_envs": true,
264
+ "recording": false,
265
+ "recording_repo_id": null,
266
+ "recording_private": false
267
+ },
268
+ "wandb": {
269
+ "enable": true,
270
+ "disable_artifact": true,
271
+ "project": "hirobot",
272
+ "entity": null,
273
+ "notes": null,
274
+ "run_id": "2pmqwdxm",
275
+ "mode": null,
276
+ "add_tags": true
277
+ },
278
+ "peft": null,
279
+ "job": {
280
+ "target": null,
281
+ "image": "huggingface/lerobot-gpu:latest",
282
+ "timeout": "2d",
283
+ "detach": false,
284
+ "tags": []
285
+ },
286
+ "save_checkpoint_to_hub": false,
287
+ "sample_weighting": null,
288
+ "rename_map": {},
289
+ "checkpoint_path": null
290
+ }