Sakits commited on
Commit
30ea1d5
·
verified ·
1 Parent(s): 29fd837

Upload folder using huggingface_hub

Browse files
Files changed (3) hide show
  1. config.json +108 -0
  2. model.safetensors +3 -0
  3. train_config.json +260 -0
config.json ADDED
@@ -0,0 +1,108 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "groot",
3
+ "n_obs_steps": 1,
4
+ "input_features": {
5
+ "observation.state": {
6
+ "type": "STATE",
7
+ "shape": [
8
+ 6
9
+ ]
10
+ },
11
+ "observation.images.wrist": {
12
+ "type": "VISUAL",
13
+ "shape": [
14
+ 3,
15
+ 480,
16
+ 640
17
+ ]
18
+ }
19
+ },
20
+ "output_features": {
21
+ "action": {
22
+ "type": "ACTION",
23
+ "shape": [
24
+ 6
25
+ ]
26
+ }
27
+ },
28
+ "device": "cuda",
29
+ "use_amp": false,
30
+ "push_to_hub": false,
31
+ "repo_id": null,
32
+ "private": null,
33
+ "tags": null,
34
+ "license": null,
35
+ "pretrained_path": "nvidia/GR00T-N1.5-3B",
36
+ "base_model_path": "nvidia/GR00T-N1.5-3B",
37
+ "eagle_path": null,
38
+ "tokenizer_assets_repo": "lerobot/eagle2hg-processor-groot-n1p5",
39
+ "tune_llm": false,
40
+ "tune_visual": true,
41
+ "eagle_select_layer": 12,
42
+ "eagle_project_to_dim": null,
43
+ "tune_projector": true,
44
+ "tune_diffusion_model": true,
45
+ "chunk_size": 16,
46
+ "n_action_steps": 16,
47
+ "max_state_dim": 64,
48
+ "max_action_dim": 32,
49
+ "noise_beta_alpha": 1.5,
50
+ "noise_beta_beta": 1.0,
51
+ "noise_s": 0.999,
52
+ "num_timestep_buckets": 1000,
53
+ "num_inference_steps": 4,
54
+ "max_num_embodiments": 32,
55
+ "embodiment_tag": "new_embodiment",
56
+ "action_head_hidden_size": 1024,
57
+ "action_head_input_embedding_dim": 1536,
58
+ "action_head_backbone_embedding_dim": 2048,
59
+ "add_pos_embed": true,
60
+ "max_seq_len": 1024,
61
+ "num_target_vision_tokens": 32,
62
+ "use_vlln": true,
63
+ "diffusion_model_cfg": {
64
+ "attention_head_dim": 48,
65
+ "cross_attention_dim": 2048,
66
+ "dropout": 0.2,
67
+ "final_dropout": true,
68
+ "interleave_self_attention": true,
69
+ "norm_type": "ada_norm",
70
+ "num_attention_heads": 32,
71
+ "num_layers": 16,
72
+ "output_dim": 1024,
73
+ "positional_embeddings": null
74
+ },
75
+ "vl_self_attention_cfg": {
76
+ "attention_head_dim": 64,
77
+ "dropout": 0.2,
78
+ "final_dropout": true,
79
+ "num_attention_heads": 32,
80
+ "num_layers": 4,
81
+ "positional_embeddings": null
82
+ },
83
+ "image_size": [
84
+ 448,
85
+ 448
86
+ ],
87
+ "tokenizer_max_length": 200,
88
+ "use_bf16": true,
89
+ "compile_model": false,
90
+ "compile_mode": "max-autotune",
91
+ "dtype": "bfloat16",
92
+ "normalization_mapping": {
93
+ "VISUAL": "IDENTITY",
94
+ "STATE": "MEAN_STD",
95
+ "ACTION": "MEAN_STD"
96
+ },
97
+ "optimizer_lr": 0.0001,
98
+ "optimizer_betas": [
99
+ 0.95,
100
+ 0.999
101
+ ],
102
+ "optimizer_eps": 1e-08,
103
+ "optimizer_weight_decay": 1e-05,
104
+ "optimizer_grad_clip_norm": 1.0,
105
+ "scheduler_warmup_steps": 500,
106
+ "scheduler_decay_steps": 10000,
107
+ "scheduler_decay_lr": 1e-05
108
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9bc7643e9070224ee07a5db6669168b013ed69924bcacdfd035298b32f2a725f
3
+ size 4827185160
train_config.json ADDED
@@ -0,0 +1,260 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "Sakits/so101_pickplace_20260217",
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
+ "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": true,
79
+ "video_backend": "torchcodec",
80
+ "streaming": false
81
+ },
82
+ "env": null,
83
+ "policy": {
84
+ "type": "groot",
85
+ "n_obs_steps": 1,
86
+ "input_features": {
87
+ "observation.state": {
88
+ "type": "STATE",
89
+ "shape": [
90
+ 6
91
+ ]
92
+ },
93
+ "observation.images.wrist": {
94
+ "type": "VISUAL",
95
+ "shape": [
96
+ 3,
97
+ 480,
98
+ 640
99
+ ]
100
+ }
101
+ },
102
+ "output_features": {
103
+ "action": {
104
+ "type": "ACTION",
105
+ "shape": [
106
+ 6
107
+ ]
108
+ }
109
+ },
110
+ "device": "cuda",
111
+ "use_amp": false,
112
+ "push_to_hub": false,
113
+ "repo_id": null,
114
+ "private": null,
115
+ "tags": null,
116
+ "license": null,
117
+ "pretrained_path": "nvidia/GR00T-N1.5-3B",
118
+ "base_model_path": "nvidia/GR00T-N1.5-3B",
119
+ "eagle_path": null,
120
+ "tokenizer_assets_repo": "lerobot/eagle2hg-processor-groot-n1p5",
121
+ "tune_llm": false,
122
+ "tune_visual": true,
123
+ "eagle_select_layer": 12,
124
+ "eagle_project_to_dim": null,
125
+ "tune_projector": true,
126
+ "tune_diffusion_model": true,
127
+ "chunk_size": 16,
128
+ "n_action_steps": 16,
129
+ "max_state_dim": 64,
130
+ "max_action_dim": 32,
131
+ "noise_beta_alpha": 1.5,
132
+ "noise_beta_beta": 1.0,
133
+ "noise_s": 0.999,
134
+ "num_timestep_buckets": 1000,
135
+ "num_inference_steps": 4,
136
+ "max_num_embodiments": 32,
137
+ "embodiment_tag": "new_embodiment",
138
+ "action_head_hidden_size": 1024,
139
+ "action_head_input_embedding_dim": 1536,
140
+ "action_head_backbone_embedding_dim": 2048,
141
+ "add_pos_embed": true,
142
+ "max_seq_len": 1024,
143
+ "num_target_vision_tokens": 32,
144
+ "use_vlln": true,
145
+ "diffusion_model_cfg": {
146
+ "attention_head_dim": 48,
147
+ "cross_attention_dim": 2048,
148
+ "dropout": 0.2,
149
+ "final_dropout": true,
150
+ "interleave_self_attention": true,
151
+ "norm_type": "ada_norm",
152
+ "num_attention_heads": 32,
153
+ "num_layers": 16,
154
+ "output_dim": 1024,
155
+ "positional_embeddings": null
156
+ },
157
+ "vl_self_attention_cfg": {
158
+ "attention_head_dim": 64,
159
+ "dropout": 0.2,
160
+ "final_dropout": true,
161
+ "num_attention_heads": 32,
162
+ "num_layers": 4,
163
+ "positional_embeddings": null
164
+ },
165
+ "image_size": [
166
+ 448,
167
+ 448
168
+ ],
169
+ "tokenizer_max_length": 200,
170
+ "use_bf16": true,
171
+ "compile_model": false,
172
+ "compile_mode": "max-autotune",
173
+ "dtype": "bfloat16",
174
+ "normalization_mapping": {
175
+ "VISUAL": "IDENTITY",
176
+ "STATE": "MEAN_STD",
177
+ "ACTION": "MEAN_STD"
178
+ },
179
+ "optimizer_lr": 0.0001,
180
+ "optimizer_betas": [
181
+ 0.95,
182
+ 0.999
183
+ ],
184
+ "optimizer_eps": 1e-08,
185
+ "optimizer_weight_decay": 1e-05,
186
+ "optimizer_grad_clip_norm": 1.0,
187
+ "scheduler_warmup_steps": 500,
188
+ "scheduler_decay_steps": 10000,
189
+ "scheduler_decay_lr": 1e-05
190
+ },
191
+ "output_dir": "outputs/train/groot_async8_pickplace_20260217",
192
+ "job_name": "groot_async8_pickplace_20260217",
193
+ "resume": false,
194
+ "seed": 1000,
195
+ "num_workers": 4,
196
+ "batch_size": 8,
197
+ "steps": 50000,
198
+ "eval_freq": 20000,
199
+ "log_freq": 200,
200
+ "save_checkpoint": true,
201
+ "save_freq": 10000,
202
+ "use_policy_training_preset": false,
203
+ "optimizer": {
204
+ "type": "adamw",
205
+ "lr": 0.0001,
206
+ "weight_decay": 1e-05,
207
+ "grad_clip_norm": 10.0,
208
+ "betas": [
209
+ 0.95,
210
+ 0.999
211
+ ],
212
+ "eps": 1e-08
213
+ },
214
+ "scheduler": {
215
+ "type": "cosine_decay_with_warmup",
216
+ "num_warmup_steps": 500,
217
+ "num_decay_steps": 50000,
218
+ "peak_lr": 0.0001,
219
+ "decay_lr": 1e-05
220
+ },
221
+ "eval": {
222
+ "n_episodes": 50,
223
+ "batch_size": 50,
224
+ "use_async_envs": false
225
+ },
226
+ "wandb": {
227
+ "enable": true,
228
+ "disable_artifact": true,
229
+ "project": "vlash",
230
+ "entity": null,
231
+ "notes": null,
232
+ "run_id": "g7zwoyrt",
233
+ "mode": null
234
+ },
235
+ "checkpoint_path": null,
236
+ "rename_map": {},
237
+ "max_delay_steps": 8,
238
+ "grad_accum_steps": 1,
239
+ "shared_observation": false,
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
+ }