Hugging Face's logo

nnh-pbbb
/
vlash

Safetensors
Model card Files
xet
Community
nnh-pbbb commited on
Commit
06bb7f7
·
verified ·
1 Parent(s): e0861d4

Upload folder using huggingface_hub

Browse files
Files changed (3) hide show
  1. 010000_0109_new/config.json +108 -0
  2. 010000_0109_new/model.safetensors +3 -0
  3. 010000_0109_new/train_config.json +260 -0
010000_0109_new/config.json ADDED
@@ -0,0 +1,108 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "pi05",
3
+ "n_obs_steps": 1,
4
+ "input_features": {
5
+ "observation.state": {
6
+ "type": "STATE",
7
+ "shape": [
8
+ 20
9
+ ]
10
+ },
11
+ "observation.state_eepose": {
12
+ "type": "STATE",
13
+ "shape": [
14
+ 14
15
+ ]
16
+ },
17
+ "observation.images.cam_high": {
18
+ "type": "VISUAL",
19
+ "shape": [
20
+ 3,
21
+ 360,
22
+ 480
23
+ ]
24
+ },
25
+ "observation.images.cam_left_wrist": {
26
+ "type": "VISUAL",
27
+ "shape": [
28
+ 3,
29
+ 360,
30
+ 480
31
+ ]
32
+ },
33
+ "observation.images.cam_right_wrist": {
34
+ "type": "VISUAL",
35
+ "shape": [
36
+ 3,
37
+ 360,
38
+ 480
39
+ ]
40
+ }
41
+ },
42
+ "output_features": {
43
+ "action": {
44
+ "type": "ACTION",
45
+ "shape": [
46
+ 20
47
+ ]
48
+ },
49
+ "action_eepose": {
50
+ "type": "ACTION",
51
+ "shape": [
52
+ 14
53
+ ]
54
+ }
55
+ },
56
+ "device": "cuda",
57
+ "use_amp": false,
58
+ "push_to_hub": false,
59
+ "repo_id": null,
60
+ "private": null,
61
+ "tags": null,
62
+ "license": null,
63
+ "pretrained_path": "/data/LumosVLA/vlash/checkpoints/pi05_base",
64
+ "paligemma_variant": "gemma_2b",
65
+ "action_expert_variant": "gemma_300m",
66
+ "dtype": "bfloat16",
67
+ "chunk_size": 50,
68
+ "n_action_steps": 50,
69
+ "max_state_dim": 32,
70
+ "max_action_dim": 32,
71
+ "state_cond": true,
72
+ "num_inference_steps": 10,
73
+ "time_sampling_beta_alpha": 1.5,
74
+ "time_sampling_beta_beta": 1.0,
75
+ "time_sampling_scale": 0.999,
76
+ "time_sampling_offset": 0.001,
77
+ "min_period": 0.004,
78
+ "max_period": 4.0,
79
+ "image_resolution": [
80
+ 224,
81
+ 224
82
+ ],
83
+ "empty_cameras": 0,
84
+ "tokenizer_max_length": 200,
85
+ "normalization_mapping": {
86
+ "VISUAL": "IDENTITY",
87
+ "STATE": "MEAN_STD",
88
+ "ACTION": "MEAN_STD"
89
+ },
90
+ "gradient_checkpointing": false,
91
+ "compile_model": false,
92
+ "compile_mode": "max-autotune",
93
+ "fuse_qkv": false,
94
+ "fuse_gate_up": false,
95
+ "optimizer_lr": 2.5e-05,
96
+ "optimizer_betas": [
97
+ 0.9,
98
+ 0.95
99
+ ],
100
+ "optimizer_eps": 1e-08,
101
+ "optimizer_weight_decay": 0.01,
102
+ "optimizer_grad_clip_norm": 1.0,
103
+ "scheduler_warmup_steps": 1000,
104
+ "scheduler_decay_steps": 30000,
105
+ "scheduler_decay_lr": 2.5e-06,
106
+ "vlm_config": {},
107
+ "action_expert_config": {}
108
+ }
010000_0109_new/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cef4142fb80cc1f8cc604c590c06c4947a65844dc05f9d0cf14faa15529c3a94
3
+ size 7481487056
010000_0109_new/train_config.json ADDED
@@ -0,0 +1,260 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "data/pickup0102_eepose",
4
+ "root": "/data/LumosVLA/LumosVLA/data/9in1_846_0106_eepose",
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
+ "affine": {
62
+ "weight": 1.0,
63
+ "type": "RandomAffine",
64
+ "kwargs": {
65
+ "degrees": [
66
+ -5.0,
67
+ 5.0
68
+ ],
69
+ "translate": [
70
+ 0.05,
71
+ 0.05
72
+ ]
73
+ }
74
+ }
75
+ }
76
+ },
77
+ "revision": null,
78
+ "use_imagenet_stats": false,
79
+ "video_backend": "torchcodec",
80
+ "streaming": false
81
+ },
82
+ "env": null,
83
+ "policy": {
84
+ "type": "pi05",
85
+ "n_obs_steps": 1,
86
+ "input_features": {
87
+ "observation.state": {
88
+ "type": "STATE",
89
+ "shape": [
90
+ 20
91
+ ]
92
+ },
93
+ "observation.state_eepose": {
94
+ "type": "STATE",
95
+ "shape": [
96
+ 14
97
+ ]
98
+ },
99
+ "observation.images.cam_high": {
100
+ "type": "VISUAL",
101
+ "shape": [
102
+ 3,
103
+ 360,
104
+ 480
105
+ ]
106
+ },
107
+ "observation.images.cam_left_wrist": {
108
+ "type": "VISUAL",
109
+ "shape": [
110
+ 3,
111
+ 360,
112
+ 480
113
+ ]
114
+ },
115
+ "observation.images.cam_right_wrist": {
116
+ "type": "VISUAL",
117
+ "shape": [
118
+ 3,
119
+ 360,
120
+ 480
121
+ ]
122
+ }
123
+ },
124
+ "output_features": {
125
+ "action": {
126
+ "type": "ACTION",
127
+ "shape": [
128
+ 20
129
+ ]
130
+ },
131
+ "action_eepose": {
132
+ "type": "ACTION",
133
+ "shape": [
134
+ 14
135
+ ]
136
+ }
137
+ },
138
+ "device": "cuda",
139
+ "use_amp": false,
140
+ "push_to_hub": false,
141
+ "repo_id": null,
142
+ "private": null,
143
+ "tags": null,
144
+ "license": null,
145
+ "pretrained_path": "/data/LumosVLA/vlash/checkpoints/pi05_base",
146
+ "paligemma_variant": "gemma_2b",
147
+ "action_expert_variant": "gemma_300m",
148
+ "dtype": "bfloat16",
149
+ "chunk_size": 50,
150
+ "n_action_steps": 50,
151
+ "max_state_dim": 32,
152
+ "max_action_dim": 32,
153
+ "state_cond": true,
154
+ "num_inference_steps": 10,
155
+ "time_sampling_beta_alpha": 1.5,
156
+ "time_sampling_beta_beta": 1.0,
157
+ "time_sampling_scale": 0.999,
158
+ "time_sampling_offset": 0.001,
159
+ "min_period": 0.004,
160
+ "max_period": 4.0,
161
+ "image_resolution": [
162
+ 224,
163
+ 224
164
+ ],
165
+ "empty_cameras": 0,
166
+ "tokenizer_max_length": 200,
167
+ "normalization_mapping": {
168
+ "VISUAL": "IDENTITY",
169
+ "STATE": "MEAN_STD",
170
+ "ACTION": "MEAN_STD"
171
+ },
172
+ "gradient_checkpointing": false,
173
+ "compile_model": false,
174
+ "compile_mode": "max-autotune",
175
+ "fuse_qkv": false,
176
+ "fuse_gate_up": false,
177
+ "optimizer_lr": 2.5e-05,
178
+ "optimizer_betas": [
179
+ 0.9,
180
+ 0.95
181
+ ],
182
+ "optimizer_eps": 1e-08,
183
+ "optimizer_weight_decay": 0.01,
184
+ "optimizer_grad_clip_norm": 1.0,
185
+ "scheduler_warmup_steps": 1000,
186
+ "scheduler_decay_steps": 30000,
187
+ "scheduler_decay_lr": 2.5e-06,
188
+ "vlm_config": {},
189
+ "action_expert_config": {}
190
+ },
191
+ "output_dir": "/data/LumosVLA/vlash/outputs/train/pi05_async_eepose_nolora",
192
+ "job_name": "pi05_async_",
193
+ "resume": false,
194
+ "seed": 1000,
195
+ "num_workers": 1,
196
+ "batch_size": 4,
197
+ "steps": 50000,
198
+ "eval_freq": 20000,
199
+ "log_freq": 50,
200
+ "save_checkpoint": true,
201
+ "save_freq": 5000,
202
+ "use_policy_training_preset": false,
203
+ "optimizer": {
204
+ "type": "adamw",
205
+ "lr": 5e-05,
206
+ "weight_decay": 1e-10,
207
+ "grad_clip_norm": 10.0,
208
+ "betas": [
209
+ 0.9,
210
+ 0.95
211
+ ],
212
+ "eps": 1e-08
213
+ },
214
+ "scheduler": {
215
+ "type": "cosine_decay_with_warmup",
216
+ "num_warmup_steps": 1000,
217
+ "num_decay_steps": 50000,
218
+ "peak_lr": 5e-05,
219
+ "decay_lr": 2.5e-06
220
+ },
221
+ "eval": {
222
+ "n_episodes": 50,
223
+ "batch_size": 50,
224
+ "use_async_envs": false
225
+ },
226
+ "wandb": {
227
+ "enable": false,
228
+ "disable_artifact": true,
229
+ "project": "vlash",
230
+ "entity": null,
231
+ "notes": null,
232
+ "run_id": null,
233
+ "mode": null
234
+ },
235
+ "checkpoint_path": null,
236
+ "rename_map": {},
237
+ "max_delay_steps": 8,
238
+ "grad_accum_steps": 4,
239
+ "shared_observation": true,
240
+ "lora": {
241
+ "enable": false,
242
+ "backend": "peft",
243
+ "r": 16,
244
+ "alpha": 16,
245
+ "dropout": 0.0,
246
+ "extra_trainable_modules": [],
247
+ "target_modules": [
248
+ "q_proj",
249
+ "k_proj",
250
+ "v_proj",
251
+ "o_proj",
252
+ "up_proj",
253
+ "down_proj",
254
+ "gate_proj"
255
+ ],
256
+ "use_qlora": false,
257
+ "qlora_quant_type": "nf4",
258
+ "qlora_compute_dtype": "bfloat16"
259
+ }
260
+ }