Upload folder using huggingface_hub
Browse files
InternVLPI-0312-eep-pretrain_T2_sft_rel_90000/checkpoints/steps_90000_pytorch_model.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b0c2836e6c1352477d7ebbd44b050dfa74ef164ad084c8acd4dc6e6cfbd9e22c
|
| 3 |
+
size 2966606180
|
InternVLPI-0312-eep-pretrain_T2_sft_rel_90000/config.yaml
ADDED
|
@@ -0,0 +1,69 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
datasets:
|
| 2 |
+
vla_data:
|
| 3 |
+
action_mode: rel
|
| 4 |
+
data_mix: train_QwenPI
|
| 5 |
+
data_root_dir: /lumos-vePFS/shenzhen/data/data_warehouse/data_warehouse_output_clean
|
| 6 |
+
dataset_py: lerobot_datasets
|
| 7 |
+
image_size: 256
|
| 8 |
+
num_workers: 2
|
| 9 |
+
per_device_batch_size: 16
|
| 10 |
+
video_backend: pyav
|
| 11 |
+
framework:
|
| 12 |
+
action_model:
|
| 13 |
+
action_dim: 20
|
| 14 |
+
add_pos_embed: true
|
| 15 |
+
diffusion_model_cfg:
|
| 16 |
+
dropout: 0.2
|
| 17 |
+
final_dropout: true
|
| 18 |
+
interleave_self_attention: true
|
| 19 |
+
norm_type: ada_norm
|
| 20 |
+
num_layers: 28
|
| 21 |
+
output_dim: 2560
|
| 22 |
+
positional_embeddings: null
|
| 23 |
+
future_action_window_size: 29
|
| 24 |
+
hidden_dim: 1024
|
| 25 |
+
max_seq_len: 1024
|
| 26 |
+
noise_beta_alpha: 1.5
|
| 27 |
+
noise_beta_beta: 1.0
|
| 28 |
+
noise_s: 0.999
|
| 29 |
+
num_inference_timesteps: 8
|
| 30 |
+
num_target_vision_tokens: 32
|
| 31 |
+
num_timestep_buckets: 1000
|
| 32 |
+
past_action_window_size: 0
|
| 33 |
+
smoothness_loss_weight: 1
|
| 34 |
+
state_dim: 20
|
| 35 |
+
use_kv_cache_dit: true
|
| 36 |
+
use_scaled_noise: true
|
| 37 |
+
name: internvlPI
|
| 38 |
+
qwenvl:
|
| 39 |
+
attn_implementation: eager
|
| 40 |
+
base_vlm: /lumos-vePFS/shenzhen/models/InternVL3_5-1B-Instruct
|
| 41 |
+
num_vl_layers: 28
|
| 42 |
+
vl_hidden_dim: 1024
|
| 43 |
+
output_dir: /lumos-vePFS/shenzhen/models/lumos_pi/InternVLPI-0312-eep-pretrain_T2_sft_rel
|
| 44 |
+
run_id: InternVLPI-0312-eep-pretrain_T2_sft_rel
|
| 45 |
+
run_root_dir: /lumos-vePFS/shenzhen/models/lumos_pi
|
| 46 |
+
seed: 42
|
| 47 |
+
trainer:
|
| 48 |
+
eval_interval: 10000000
|
| 49 |
+
freeze_modules: null
|
| 50 |
+
gradient_accumulation_steps: 3
|
| 51 |
+
gradient_clipping: 1.0
|
| 52 |
+
learning_rate:
|
| 53 |
+
action_model: 0.0001
|
| 54 |
+
base: 1.0e-05
|
| 55 |
+
intern_vl_interface: 1.0e-05
|
| 56 |
+
logging_frequency: 10
|
| 57 |
+
lr_scheduler_type: cosine_with_min_lr
|
| 58 |
+
max_train_steps: 200000
|
| 59 |
+
num_warmup_steps: 5000
|
| 60 |
+
optimizer:
|
| 61 |
+
betas:
|
| 62 |
+
- 0.9
|
| 63 |
+
- 0.95
|
| 64 |
+
eps: 1.0e-08
|
| 65 |
+
weight_decay: 0
|
| 66 |
+
pretrained_checkpoint: /lumos-vePFS/shenzhen/models/lumos_pi/InternVLPI-0225-eep-pretrain_T2/checkpoints/steps_23000_pytorch_model.pt
|
| 67 |
+
save_interval: 1000
|
| 68 |
+
scheduler_specific_kwargs:
|
| 69 |
+
min_lr: 1.0e-06
|
InternVLPI-0312-eep-pretrain_T2_sft_rel_90000/dataset_statistics.json
ADDED
|
@@ -0,0 +1,218 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"new_embodiment": {
|
| 3 |
+
"action": {
|
| 4 |
+
"mean": [
|
| 5 |
+
-0.00011513745432074015,
|
| 6 |
+
0.0002766112522956312,
|
| 7 |
+
-0.0004226025693169917,
|
| 8 |
+
0.018302556783468888,
|
| 9 |
+
0.0024173898868453266,
|
| 10 |
+
0.004111765382781776,
|
| 11 |
+
-0.00010867678648725358,
|
| 12 |
+
-0.000400677792103207,
|
| 13 |
+
-0.0002734223110936992,
|
| 14 |
+
-0.0007054930959963136,
|
| 15 |
+
0.005622064506166102,
|
| 16 |
+
0.002259439455718331,
|
| 17 |
+
0.017533802272406654,
|
| 18 |
+
-9.481074571714535e-05
|
| 19 |
+
],
|
| 20 |
+
"std": [
|
| 21 |
+
0.039236366804014836,
|
| 22 |
+
0.045220609612901724,
|
| 23 |
+
0.04230235274100189,
|
| 24 |
+
1.8214571181145556,
|
| 25 |
+
0.12485401218607987,
|
| 26 |
+
1.5867366363284652,
|
| 27 |
+
0.018502423072173205,
|
| 28 |
+
0.035430127718394086,
|
| 29 |
+
0.032482596083518125,
|
| 30 |
+
0.032734586420464326,
|
| 31 |
+
1.7161009823789404,
|
| 32 |
+
0.10771976128383372,
|
| 33 |
+
1.4270776989997542,
|
| 34 |
+
0.017152386685708797
|
| 35 |
+
],
|
| 36 |
+
"max": [
|
| 37 |
+
0.34321898221969604,
|
| 38 |
+
0.5628479719161987,
|
| 39 |
+
0.5319600105285645,
|
| 40 |
+
6.282644271850586,
|
| 41 |
+
1.6204335689544678,
|
| 42 |
+
6.282644271850586,
|
| 43 |
+
0.08064000308513641,
|
| 44 |
+
0.36259499192237854,
|
| 45 |
+
0.40049299597740173,
|
| 46 |
+
0.40070098638534546,
|
| 47 |
+
6.282644271850586,
|
| 48 |
+
1.390242099761963,
|
| 49 |
+
6.282644271850586,
|
| 50 |
+
0.05985000357031822
|
| 51 |
+
],
|
| 52 |
+
"min": [
|
| 53 |
+
-0.35113799571990967,
|
| 54 |
+
-0.5106030106544495,
|
| 55 |
+
-0.3719419836997986,
|
| 56 |
+
-6.282644271850586,
|
| 57 |
+
-2.0788092613220215,
|
| 58 |
+
-6.282644271850586,
|
| 59 |
+
-0.06313999742269516,
|
| 60 |
+
-0.36908799409866333,
|
| 61 |
+
-0.3504599928855896,
|
| 62 |
+
-0.4005409777164459,
|
| 63 |
+
-6.282644271850586,
|
| 64 |
+
-1.5251384973526,
|
| 65 |
+
-6.282644271850586,
|
| 66 |
+
-0.06034000217914581
|
| 67 |
+
],
|
| 68 |
+
"q01": [
|
| 69 |
+
-0.15719900593161584,
|
| 70 |
+
-0.24724191382527352,
|
| 71 |
+
-0.1255129873752594,
|
| 72 |
+
-6.245469002723694,
|
| 73 |
+
-0.6122440099716187,
|
| 74 |
+
-6.242082595825195,
|
| 75 |
+
-0.05914999917149544,
|
| 76 |
+
-0.16179936677217482,
|
| 77 |
+
-0.1904055467247963,
|
| 78 |
+
-0.13128199443221092,
|
| 79 |
+
-6.257476806640625,
|
| 80 |
+
-0.5619439399242401,
|
| 81 |
+
-6.232770047187805,
|
| 82 |
+
-0.05936000123620033
|
| 83 |
+
],
|
| 84 |
+
"q99": [
|
| 85 |
+
0.1523151311278342,
|
| 86 |
+
0.3472435465455051,
|
| 87 |
+
0.19813139498233762,
|
| 88 |
+
6.251490116119385,
|
| 89 |
+
0.6815669369697557,
|
| 90 |
+
6.2439680099487305,
|
| 91 |
+
0.059220001101493835,
|
| 92 |
+
0.1795560048520565,
|
| 93 |
+
0.18499764487147297,
|
| 94 |
+
0.15433004498481684,
|
| 95 |
+
6.2573897171020505,
|
| 96 |
+
0.5667995488643682,
|
| 97 |
+
6.2453813552856445,
|
| 98 |
+
0.05950000137090683
|
| 99 |
+
],
|
| 100 |
+
"mask": [
|
| 101 |
+
true,
|
| 102 |
+
true,
|
| 103 |
+
true,
|
| 104 |
+
true,
|
| 105 |
+
true,
|
| 106 |
+
true,
|
| 107 |
+
false,
|
| 108 |
+
true,
|
| 109 |
+
true,
|
| 110 |
+
true,
|
| 111 |
+
true,
|
| 112 |
+
true,
|
| 113 |
+
true,
|
| 114 |
+
false
|
| 115 |
+
]
|
| 116 |
+
},
|
| 117 |
+
"state": {
|
| 118 |
+
"mean": [
|
| 119 |
+
-0.00011513745432074015,
|
| 120 |
+
0.0002766112522956312,
|
| 121 |
+
-0.0004226025693169917,
|
| 122 |
+
0.018302556783468888,
|
| 123 |
+
0.0024173898868453266,
|
| 124 |
+
0.004111765382781776,
|
| 125 |
+
-0.00010867678648725358,
|
| 126 |
+
-0.000400677792103207,
|
| 127 |
+
-0.0002734223110936992,
|
| 128 |
+
-0.0007054930959963136,
|
| 129 |
+
0.005622064506166102,
|
| 130 |
+
0.002259439455718331,
|
| 131 |
+
0.017533802272406654,
|
| 132 |
+
-9.481074571714535e-05
|
| 133 |
+
],
|
| 134 |
+
"std": [
|
| 135 |
+
0.039236366804014836,
|
| 136 |
+
0.045220609612901724,
|
| 137 |
+
0.04230235274100189,
|
| 138 |
+
1.8214571181145556,
|
| 139 |
+
0.12485401218607987,
|
| 140 |
+
1.5867366363284652,
|
| 141 |
+
0.018502423072173205,
|
| 142 |
+
0.035430127718394086,
|
| 143 |
+
0.032482596083518125,
|
| 144 |
+
0.032734586420464326,
|
| 145 |
+
1.7161009823789404,
|
| 146 |
+
0.10771976128383372,
|
| 147 |
+
1.4270776989997542,
|
| 148 |
+
0.017152386685708797
|
| 149 |
+
],
|
| 150 |
+
"max": [
|
| 151 |
+
0.34321898221969604,
|
| 152 |
+
0.5628479719161987,
|
| 153 |
+
0.5319600105285645,
|
| 154 |
+
6.282644271850586,
|
| 155 |
+
1.6204335689544678,
|
| 156 |
+
6.282644271850586,
|
| 157 |
+
0.08064000308513641,
|
| 158 |
+
0.36259499192237854,
|
| 159 |
+
0.40049299597740173,
|
| 160 |
+
0.40070098638534546,
|
| 161 |
+
6.282644271850586,
|
| 162 |
+
1.390242099761963,
|
| 163 |
+
6.282644271850586,
|
| 164 |
+
0.05985000357031822
|
| 165 |
+
],
|
| 166 |
+
"min": [
|
| 167 |
+
-0.35113799571990967,
|
| 168 |
+
-0.5106030106544495,
|
| 169 |
+
-0.3719419836997986,
|
| 170 |
+
-6.282644271850586,
|
| 171 |
+
-2.0788092613220215,
|
| 172 |
+
-6.282644271850586,
|
| 173 |
+
-0.06313999742269516,
|
| 174 |
+
-0.36908799409866333,
|
| 175 |
+
-0.3504599928855896,
|
| 176 |
+
-0.4005409777164459,
|
| 177 |
+
-6.282644271850586,
|
| 178 |
+
-1.5251384973526,
|
| 179 |
+
-6.282644271850586,
|
| 180 |
+
-0.06034000217914581
|
| 181 |
+
],
|
| 182 |
+
"q01": [
|
| 183 |
+
-0.15719900593161584,
|
| 184 |
+
-0.24724191382527352,
|
| 185 |
+
-0.1255129873752594,
|
| 186 |
+
-6.245469002723694,
|
| 187 |
+
-0.6122440099716187,
|
| 188 |
+
-6.242082595825195,
|
| 189 |
+
-0.05914999917149544,
|
| 190 |
+
-0.16179936677217482,
|
| 191 |
+
-0.1904055467247963,
|
| 192 |
+
-0.13128199443221092,
|
| 193 |
+
-6.257476806640625,
|
| 194 |
+
-0.5619439399242401,
|
| 195 |
+
-6.232770047187805,
|
| 196 |
+
-0.05936000123620033
|
| 197 |
+
],
|
| 198 |
+
"q99": [
|
| 199 |
+
0.1523151311278342,
|
| 200 |
+
0.3472435465455051,
|
| 201 |
+
0.19813139498233762,
|
| 202 |
+
6.251490116119385,
|
| 203 |
+
0.6815669369697557,
|
| 204 |
+
6.2439680099487305,
|
| 205 |
+
0.059220001101493835,
|
| 206 |
+
0.1795560048520565,
|
| 207 |
+
0.18499764487147297,
|
| 208 |
+
0.15433004498481684,
|
| 209 |
+
6.2573897171020505,
|
| 210 |
+
0.5667995488643682,
|
| 211 |
+
6.2453813552856445,
|
| 212 |
+
0.05950000137090683
|
| 213 |
+
]
|
| 214 |
+
},
|
| 215 |
+
"num_transitions": 2557645,
|
| 216 |
+
"num_trajectories": 2411
|
| 217 |
+
}
|
| 218 |
+
}
|