Instructions to use mcobzarenco/bijou-checkpoints with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- LeRobot
How to use mcobzarenco/bijou-checkpoints with LeRobot:
- Notebooks
- Google Colab
- Kaggle
Upload folder using huggingface_hub
Browse files
bijou_ft_rig_ar_armF/step_000500/backbone.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fc7805ab236ff9ed8a9d03484dfd43bb6f859b5ccbd14650fe6bccd6fb68b6f8
|
| 3 |
+
size 4295565124
|
bijou_ft_rig_ar_armF/step_000500/bijou_config.json
ADDED
|
@@ -0,0 +1,281 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"format": 2,
|
| 3 |
+
"encoder": {
|
| 4 |
+
"kind": "gemma4",
|
| 5 |
+
"backbone": "google/gemma-4-e2b-it",
|
| 6 |
+
"exports": [
|
| 7 |
+
4,
|
| 8 |
+
9,
|
| 9 |
+
14
|
| 10 |
+
],
|
| 11 |
+
"max_soft_tokens": 140
|
| 12 |
+
},
|
| 13 |
+
"decoder": {
|
| 14 |
+
"kind": "ar_fast",
|
| 15 |
+
"hidden_size": 1024,
|
| 16 |
+
"num_attention_heads": 8,
|
| 17 |
+
"intermediate_size": 4096,
|
| 18 |
+
"hidden_activation": "gelu_pytorch_tanh",
|
| 19 |
+
"rms_norm_eps": 1e-06,
|
| 20 |
+
"self_attention_rope_theta": 10000.0,
|
| 21 |
+
"cross_attention_heads": 8,
|
| 22 |
+
"schedule": [
|
| 23 |
+
"kv4",
|
| 24 |
+
"kv4",
|
| 25 |
+
"kv4",
|
| 26 |
+
"kv4",
|
| 27 |
+
"kv9",
|
| 28 |
+
"kv9",
|
| 29 |
+
"kv9",
|
| 30 |
+
"kv9",
|
| 31 |
+
"kv14",
|
| 32 |
+
"kv14",
|
| 33 |
+
"kv14",
|
| 34 |
+
"kv14",
|
| 35 |
+
"kv14",
|
| 36 |
+
"kv14",
|
| 37 |
+
"kv14",
|
| 38 |
+
"kv14"
|
| 39 |
+
],
|
| 40 |
+
"tokenizer": "mcobzarenco/bijou-checkpoints/fast_tokenizer_v2",
|
| 41 |
+
"vocab_total": 1026,
|
| 42 |
+
"state_dim": 6,
|
| 43 |
+
"chunk_size": 50,
|
| 44 |
+
"action_dim": 6
|
| 45 |
+
},
|
| 46 |
+
"step": 500,
|
| 47 |
+
"train_args": {
|
| 48 |
+
"train_data": [
|
| 49 |
+
"/home/ubuntu/datasets/mcobzarenco/so101_pick_place_v2",
|
| 50 |
+
"/home/ubuntu/datasets/mcobzarenco/so101_pick_place_clean"
|
| 51 |
+
],
|
| 52 |
+
"exclude": [],
|
| 53 |
+
"fps": [
|
| 54 |
+
30.0
|
| 55 |
+
],
|
| 56 |
+
"holdout_episodes": 0.1,
|
| 57 |
+
"split_seed": 0,
|
| 58 |
+
"backbone": "google/gemma-4-e2b-it",
|
| 59 |
+
"save_dir": "outputs/train/bijou_ft_rig_ar_armF",
|
| 60 |
+
"init_from": "outputs/train/bijou_ar_fast_v2_unftext2_50k_ddp2/step_050000",
|
| 61 |
+
"resume": null,
|
| 62 |
+
"instruction": null,
|
| 63 |
+
"cameras": null,
|
| 64 |
+
"max_cameras": null,
|
| 65 |
+
"max_soft_tokens": 140,
|
| 66 |
+
"stream_counts": [
|
| 67 |
+
4,
|
| 68 |
+
4,
|
| 69 |
+
8
|
| 70 |
+
],
|
| 71 |
+
"self_attention_mode": "causal_actions",
|
| 72 |
+
"time_conditioning": "additive",
|
| 73 |
+
"decoder": "ar_fast",
|
| 74 |
+
"fast_tokenizer": "mcobzarenco/bijou-checkpoints/fast_tokenizer_v2",
|
| 75 |
+
"expert_hidden": 1024,
|
| 76 |
+
"expert_heads": 8,
|
| 77 |
+
"expert_intermediate": 4096,
|
| 78 |
+
"expert_cross_heads": 8,
|
| 79 |
+
"chunk_size": 50,
|
| 80 |
+
"batch_size": 32,
|
| 81 |
+
"steps": 5000,
|
| 82 |
+
"expert_lr": 2e-05,
|
| 83 |
+
"text_lr": null,
|
| 84 |
+
"vision_lr": null,
|
| 85 |
+
"warmup_steps": 200,
|
| 86 |
+
"weight_decay": 1e-05,
|
| 87 |
+
"grad_clip": 10.0,
|
| 88 |
+
"log_every": 10,
|
| 89 |
+
"eval_every": 500,
|
| 90 |
+
"save_every": 500,
|
| 91 |
+
"num_workers": 8,
|
| 92 |
+
"prefetch_factor": 4,
|
| 93 |
+
"video_decoder_cache": 8,
|
| 94 |
+
"device": "cuda",
|
| 95 |
+
"seed": 19,
|
| 96 |
+
"eval_samples": 256,
|
| 97 |
+
"eval_seed": 0,
|
| 98 |
+
"wandb_project": "bijou-dev",
|
| 99 |
+
"wandb_run_name": "ft-rig-ar-unftext2-armF"
|
| 100 |
+
},
|
| 101 |
+
"normalization": {
|
| 102 |
+
"action": {
|
| 103 |
+
"mean": [
|
| 104 |
+
4.046886920928955,
|
| 105 |
+
-33.75446319580078,
|
| 106 |
+
55.77268981933594,
|
| 107 |
+
30.67245864868164,
|
| 108 |
+
53.6524658203125,
|
| 109 |
+
10.152764320373535
|
| 110 |
+
],
|
| 111 |
+
"std": [
|
| 112 |
+
18.56857681274414,
|
| 113 |
+
45.41136932373047,
|
| 114 |
+
39.47644805908203,
|
| 115 |
+
51.71804428100586,
|
| 116 |
+
68.10092163085938,
|
| 117 |
+
10.490633010864258
|
| 118 |
+
]
|
| 119 |
+
},
|
| 120 |
+
"observation.state": {
|
| 121 |
+
"mean": [
|
| 122 |
+
3.978522777557373,
|
| 123 |
+
-32.19367599487305,
|
| 124 |
+
57.57185745239258,
|
| 125 |
+
31.095197677612305,
|
| 126 |
+
53.569984436035156,
|
| 127 |
+
12.818948745727539
|
| 128 |
+
],
|
| 129 |
+
"std": [
|
| 130 |
+
18.546083450317383,
|
| 131 |
+
46.240440368652344,
|
| 132 |
+
38.361663818359375,
|
| 133 |
+
51.23649215698242,
|
| 134 |
+
67.94969177246094,
|
| 135 |
+
9.193977355957031
|
| 136 |
+
]
|
| 137 |
+
}
|
| 138 |
+
},
|
| 139 |
+
"per_dataset_normalization": {
|
| 140 |
+
"mcobzarenco/so101_pick_place_clean": {
|
| 141 |
+
"action": {
|
| 142 |
+
"mean": [
|
| 143 |
+
1.4819716215133667,
|
| 144 |
+
-36.11656188964844,
|
| 145 |
+
43.98069763183594,
|
| 146 |
+
53.50352096557617,
|
| 147 |
+
66.2265396118164,
|
| 148 |
+
10.923566818237305
|
| 149 |
+
],
|
| 150 |
+
"std": [
|
| 151 |
+
10.158525466918945,
|
| 152 |
+
44.32512283325195,
|
| 153 |
+
39.60983657836914,
|
| 154 |
+
24.64327621459961,
|
| 155 |
+
40.66754150390625,
|
| 156 |
+
9.115854263305664
|
| 157 |
+
],
|
| 158 |
+
"q01": [
|
| 159 |
+
-17.367910385131836,
|
| 160 |
+
-103.51648712158203,
|
| 161 |
+
-23.120878219604492,
|
| 162 |
+
-17.2764835357666,
|
| 163 |
+
-51.912086486816406,
|
| 164 |
+
2.566037654876709
|
| 165 |
+
],
|
| 166 |
+
"q99": [
|
| 167 |
+
19.560440063476562,
|
| 168 |
+
30.19780158996582,
|
| 169 |
+
97.75823974609375,
|
| 170 |
+
83.03296661376953,
|
| 171 |
+
127.51824188232422,
|
| 172 |
+
30.64301872253418
|
| 173 |
+
]
|
| 174 |
+
},
|
| 175 |
+
"observation.state": {
|
| 176 |
+
"mean": [
|
| 177 |
+
1.376397967338562,
|
| 178 |
+
-34.254249572753906,
|
| 179 |
+
46.027835845947266,
|
| 180 |
+
53.97844314575195,
|
| 181 |
+
66.22583770751953,
|
| 182 |
+
12.550904273986816
|
| 183 |
+
],
|
| 184 |
+
"std": [
|
| 185 |
+
10.14680004119873,
|
| 186 |
+
45.543617248535156,
|
| 187 |
+
38.693115234375,
|
| 188 |
+
24.173934936523438,
|
| 189 |
+
40.70050811767578,
|
| 190 |
+
8.099103927612305
|
| 191 |
+
],
|
| 192 |
+
"q01": [
|
| 193 |
+
-17.18857192993164,
|
| 194 |
+
-103.20879364013672,
|
| 195 |
+
-19.47252655029297,
|
| 196 |
+
-15.692307472229004,
|
| 197 |
+
-52.17582321166992,
|
| 198 |
+
3.3715012073516846
|
| 199 |
+
],
|
| 200 |
+
"q99": [
|
| 201 |
+
19.648351669311523,
|
| 202 |
+
32.263736724853516,
|
| 203 |
+
97.36264038085938,
|
| 204 |
+
82.76923370361328,
|
| 205 |
+
127.86813354492188,
|
| 206 |
+
30.47073745727539
|
| 207 |
+
]
|
| 208 |
+
}
|
| 209 |
+
},
|
| 210 |
+
"mcobzarenco/so101_pick_place_v2": {
|
| 211 |
+
"action": {
|
| 212 |
+
"mean": [
|
| 213 |
+
4.313668251037598,
|
| 214 |
+
-33.50877380371094,
|
| 215 |
+
56.99919509887695,
|
| 216 |
+
28.297758102416992,
|
| 217 |
+
52.3446159362793,
|
| 218 |
+
10.072591781616211
|
| 219 |
+
],
|
| 220 |
+
"std": [
|
| 221 |
+
19.21367835998535,
|
| 222 |
+
45.51582717895508,
|
| 223 |
+
39.25971603393555,
|
| 224 |
+
53.19718551635742,
|
| 225 |
+
70.2135238647461,
|
| 226 |
+
10.62020492553711
|
| 227 |
+
],
|
| 228 |
+
"q01": [
|
| 229 |
+
-48.39560317993164,
|
| 230 |
+
-103.69230651855469,
|
| 231 |
+
-44.41494369506836,
|
| 232 |
+
-100.96703338623047,
|
| 233 |
+
-119.69230651855469,
|
| 234 |
+
2.4150943756103516
|
| 235 |
+
],
|
| 236 |
+
"q99": [
|
| 237 |
+
47.16483688354492,
|
| 238 |
+
50.5054931640625,
|
| 239 |
+
97.5824203491211,
|
| 240 |
+
97.27472686767578,
|
| 241 |
+
171.1402130126953,
|
| 242 |
+
40.22641372680664
|
| 243 |
+
]
|
| 244 |
+
},
|
| 245 |
+
"observation.state": {
|
| 246 |
+
"mean": [
|
| 247 |
+
4.249174118041992,
|
| 248 |
+
-31.979351043701172,
|
| 249 |
+
58.77256774902344,
|
| 250 |
+
28.715070724487305,
|
| 251 |
+
52.253623962402344,
|
| 252 |
+
12.846829414367676
|
| 253 |
+
],
|
| 254 |
+
"std": [
|
| 255 |
+
19.189748764038086,
|
| 256 |
+
46.30705261230469,
|
| 257 |
+
38.12686538696289,
|
| 258 |
+
52.70024108886719,
|
| 259 |
+
70.04790496826172,
|
| 260 |
+
9.300016403198242
|
| 261 |
+
],
|
| 262 |
+
"q01": [
|
| 263 |
+
-48.131866455078125,
|
| 264 |
+
-103.20879364013672,
|
| 265 |
+
-41.53845977783203,
|
| 266 |
+
-98.43692016601562,
|
| 267 |
+
-119.34066009521484,
|
| 268 |
+
3.180661678314209
|
| 269 |
+
],
|
| 270 |
+
"q99": [
|
| 271 |
+
47.60439682006836,
|
| 272 |
+
53.27472686767578,
|
| 273 |
+
97.27472686767578,
|
| 274 |
+
96.3956069946289,
|
| 275 |
+
168.04396057128906,
|
| 276 |
+
40.139949798583984
|
| 277 |
+
]
|
| 278 |
+
}
|
| 279 |
+
}
|
| 280 |
+
}
|
| 281 |
+
}
|
bijou_ft_rig_ar_armF/step_000500/expert.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:04c441a937a840ae2c1f777c48f4656d7fb361cb80a5b1f1a7de1a386bb496e3
|
| 3 |
+
size 1619524960
|