Publish complete RSCI 40000 checkpoint: newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130
Browse files- newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/argv.txt +64 -0
- newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoint-manifest.json +99 -0
- newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/model.safetensors +3 -0
- newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/optimizer.bin +3 -0
- newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/random_states_0.pkl +3 -0
- newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/random_states_1.pkl +3 -0
- newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/random_states_2.pkl +3 -0
- newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/random_states_3.pkl +3 -0
- newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/random_states_4.pkl +3 -0
- newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/random_states_5.pkl +3 -0
- newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/random_states_6.pkl +3 -0
- newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/random_states_7.pkl +3 -0
- newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/scheduler.bin +3 -0
- newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/launch_contract.json +417 -0
- newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/run_config.json +434 -0
newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/argv.txt
ADDED
|
@@ -0,0 +1,64 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
/root/workspace/Haoxuan/code/Psi0/RSCI/feture/ARDY_PSCI/scripts/train.py
|
| 2 |
+
finetune_simple_psi0_config
|
| 3 |
+
--train.name=finetune
|
| 4 |
+
--train.data_parallel=ddp
|
| 5 |
+
--train.mixed_precision=bf16
|
| 6 |
+
--train.gradient_accumulation_steps=1
|
| 7 |
+
--train.learning_rate=1e-4
|
| 8 |
+
--train.max_training_steps=40000
|
| 9 |
+
--train.warmup_ratio=None
|
| 10 |
+
--train.warmup_steps=1000
|
| 11 |
+
--train.checkpointing_steps=1000
|
| 12 |
+
--train.checkpoint-anchor-steps=10000
|
| 13 |
+
--train.max_checkpoints_to_keep=4
|
| 14 |
+
--train.validation_steps=500
|
| 15 |
+
--train.val_num_batches=20
|
| 16 |
+
--train.max_grad_norm=1.0
|
| 17 |
+
--train.lr_scheduler_type=cosine
|
| 18 |
+
--train.lr_scheduler_kwargs.weight_decay=1e-6
|
| 19 |
+
--train.lr_scheduler_kwargs.betas 0.95 0.999
|
| 20 |
+
--data.transform.repack.pad-action-dim=36
|
| 21 |
+
--data.transform.repack.pad-state-dim=36
|
| 22 |
+
--data.transform.field.stat-path=meta/stats_psi0.json
|
| 23 |
+
--data.transform.field.stat-action-key=action
|
| 24 |
+
--data.transform.field.stat-state-key=states
|
| 25 |
+
--data.transform.field.action_norm_type=bounds
|
| 26 |
+
--data.transform.field.no-use-norm-mask
|
| 27 |
+
--data.transform.field.normalize-state
|
| 28 |
+
--data.transform.field.pad-action-dim=36
|
| 29 |
+
--data.transform.field.pad-state-dim=36
|
| 30 |
+
--data.transform.model.img-aug
|
| 31 |
+
--data.transform.model.resize.size 180 320
|
| 32 |
+
--data.transform.model.center_crop.size 180 320
|
| 33 |
+
--model.noise-scheduler=flow
|
| 34 |
+
--model.train-diffusion-steps=1000
|
| 35 |
+
--model.n_conditions=0
|
| 36 |
+
--model.action-chunk-size=30
|
| 37 |
+
--model.action-dim=36
|
| 38 |
+
--model.action-exec-horizon=30
|
| 39 |
+
--model.observation-horizon=1
|
| 40 |
+
--model.odim=36
|
| 41 |
+
--model.view_feature_dim=2048
|
| 42 |
+
--model.no-use_film
|
| 43 |
+
--model.no-combined_temb
|
| 44 |
+
--model.rtc
|
| 45 |
+
--model.max-delay=8
|
| 46 |
+
--model.psi0-ardy-aux-weight=0.1
|
| 47 |
+
--exp=RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1
|
| 48 |
+
--seed=292285
|
| 49 |
+
--train.output-dir=/root/workspace/Haoxuan/data/Psi0/RSCI/.runs/G1WholebodyHandoverTeleop-v0/VGQ-AE/frozen/seed-292285/handover100-newh264-v1
|
| 50 |
+
--train.train_batch_size=16
|
| 51 |
+
--train.num-workers=8
|
| 52 |
+
--log.report_to=None
|
| 53 |
+
--train.val-num-workers=8
|
| 54 |
+
--train.prefetch-factor=2
|
| 55 |
+
--data.root_dir=/root/workspace/Haoxuan/data/Psi0/data/simple-handover-new-training-h264-20261021
|
| 56 |
+
--data.train-repo-ids=G1WholebodyHandoverTeleop-v0
|
| 57 |
+
--model.model_name_or_path=/root/workspace/Haoxuan/data/Psi0/cache/checkpoints/psi0/pre.fast.1by1.2601091803.ckpt.ego200k.he30k
|
| 58 |
+
--model.pretrained-action-header-path=/root/workspace/Haoxuan/data/Psi0/cache/checkpoints/psi0/postpre.1by1.pad36.2601131206.ckpt.he30k
|
| 59 |
+
--model.rsci-variant=VGQ-AE
|
| 60 |
+
--model.rsci-vlm-mode=frozen
|
| 61 |
+
--data.v5-ardy-sidecar-root=/root/workspace/Haoxuan/data/Psi0/data/simple-handover-new-training-ardy-h264-20261021/G1WholebodyHandoverTeleop-v0
|
| 62 |
+
--train.optimizer-fused
|
| 63 |
+
--train.pin-memory
|
| 64 |
+
--train.non-blocking-transfer
|
newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoint-manifest.json
ADDED
|
@@ -0,0 +1,99 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"schema": 1,
|
| 3 |
+
"source_run": "/root/workspace/Haoxuan/data/Psi0/RSCI/.runs/G1WholebodyHandoverTeleop-v0/VGQ-AE/frozen/seed-292285/handover100-newh264-v1/finetune/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130",
|
| 4 |
+
"task": "G1WholebodyHandoverTeleop-v0",
|
| 5 |
+
"model": "VGQ-AE",
|
| 6 |
+
"vlm": "frozen",
|
| 7 |
+
"step": 40000,
|
| 8 |
+
"world_size": 8,
|
| 9 |
+
"prefix": "newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130",
|
| 10 |
+
"total_bytes": 10289936724,
|
| 11 |
+
"files": {
|
| 12 |
+
"checkpoints/ckpt_40000/model.safetensors": {
|
| 13 |
+
"bytes": 6292351720,
|
| 14 |
+
"sha256": "94d86163a94f30ef15e265c4d0dac9fce3b4f3e13002de616aab2e40c63b491f",
|
| 15 |
+
"git_blob_id": null,
|
| 16 |
+
"mtime_ns": 1792554453415160182
|
| 17 |
+
},
|
| 18 |
+
"checkpoints/ckpt_40000/optimizer.bin": {
|
| 19 |
+
"bytes": 3997422027,
|
| 20 |
+
"sha256": "fac9e89ac19d2809d02a808f51e56634abd82872a0e132ed75e347baacb5c74f",
|
| 21 |
+
"git_blob_id": null,
|
| 22 |
+
"mtime_ns": 1792554457702162591
|
| 23 |
+
},
|
| 24 |
+
"checkpoints/ckpt_40000/random_states_0.pkl": {
|
| 25 |
+
"bytes": 16577,
|
| 26 |
+
"sha256": "7263fa3c625381cb9bed89ad3af33e6999b691dd6996bfae103671924a4fb4a2",
|
| 27 |
+
"git_blob_id": "39d8d7b158eee4753c35ae87fd692a17b48f9a99",
|
| 28 |
+
"mtime_ns": 1792554457709162595
|
| 29 |
+
},
|
| 30 |
+
"checkpoints/ckpt_40000/random_states_1.pkl": {
|
| 31 |
+
"bytes": 16513,
|
| 32 |
+
"sha256": "8790ae4d9825175bdd934a9c18c27df076eb4fa6e3b7bc3dff25e5f1269d66a5",
|
| 33 |
+
"git_blob_id": "50019d7d76fe0f6ce7d82c2b8a88cf04d18bcf2e",
|
| 34 |
+
"mtime_ns": 1792554439336152267
|
| 35 |
+
},
|
| 36 |
+
"checkpoints/ckpt_40000/random_states_2.pkl": {
|
| 37 |
+
"bytes": 16513,
|
| 38 |
+
"sha256": "f68bcfded24c500e9b20d70ecae4f2bdffafa75ca6efec8d6f83841c015f31fd",
|
| 39 |
+
"git_blob_id": "1b6c63b9dbdec078b5092b888e82749944f641c0",
|
| 40 |
+
"mtime_ns": 1792554439336152267
|
| 41 |
+
},
|
| 42 |
+
"checkpoints/ckpt_40000/random_states_3.pkl": {
|
| 43 |
+
"bytes": 16513,
|
| 44 |
+
"sha256": "ca59cb4c2a7433ad42a391e085362f4efee4858da06b04e5a03c3a04893b6acd",
|
| 45 |
+
"git_blob_id": "e5836f1e5275e76e73d3ba8c8cf9f16a944bdbcf",
|
| 46 |
+
"mtime_ns": 1792554439336152267
|
| 47 |
+
},
|
| 48 |
+
"checkpoints/ckpt_40000/random_states_4.pkl": {
|
| 49 |
+
"bytes": 16513,
|
| 50 |
+
"sha256": "6dc6899a74ca29dfa1d8359ef83d132daf4cb2633b2a92563fa727343322fb56",
|
| 51 |
+
"git_blob_id": "23656a2683aa1cef7a694aa5051a76620d4690d7",
|
| 52 |
+
"mtime_ns": 1792554439336152267
|
| 53 |
+
},
|
| 54 |
+
"checkpoints/ckpt_40000/random_states_5.pkl": {
|
| 55 |
+
"bytes": 16513,
|
| 56 |
+
"sha256": "50af4c80de5025aafdb36cb7e8c100e3de68b8af4cd4d50578f6dc4b29e50610",
|
| 57 |
+
"git_blob_id": "1785fef0a19e0cfe41e04a7c973df401e1ad3e56",
|
| 58 |
+
"mtime_ns": 1792554439336152267
|
| 59 |
+
},
|
| 60 |
+
"checkpoints/ckpt_40000/random_states_6.pkl": {
|
| 61 |
+
"bytes": 16513,
|
| 62 |
+
"sha256": "3db2f8ce1e38cbbb81376fae36cc32248d3338cd990ced40f80d15248abea7cf",
|
| 63 |
+
"git_blob_id": "38b26553f2c981501b01cdc3123b7f56f606db21",
|
| 64 |
+
"mtime_ns": 1792554439336152267
|
| 65 |
+
},
|
| 66 |
+
"checkpoints/ckpt_40000/random_states_7.pkl": {
|
| 67 |
+
"bytes": 16513,
|
| 68 |
+
"sha256": "28d8967f9073429aca66666bd3affcb8f242364b70cbb01e7e3b10f05d185186",
|
| 69 |
+
"git_blob_id": "a04e55028c55ff29aebf0a3ad0ea727fd326a80f",
|
| 70 |
+
"mtime_ns": 1792554439336152267
|
| 71 |
+
},
|
| 72 |
+
"checkpoints/ckpt_40000/scheduler.bin": {
|
| 73 |
+
"bytes": 1401,
|
| 74 |
+
"sha256": "333a511cd62b9f69fddb60ba012421d8c02880f26090feecb0440de01e609e64",
|
| 75 |
+
"git_blob_id": "579b8a5952a2274f9241e922390e3fef6f18e8b8",
|
| 76 |
+
"mtime_ns": 1792554457704162592
|
| 77 |
+
},
|
| 78 |
+
"run_config.json": {
|
| 79 |
+
"bytes": 15088,
|
| 80 |
+
"sha256": "7efb5a0fb709bbc709ae47b7f5546ece001ec48a00ec0a064d010250bce5e7f9",
|
| 81 |
+
"git_blob_id": "358e6df5230f47990c3d7b9f3885627e81369aa1",
|
| 82 |
+
"mtime_ns": 1792517437881509645
|
| 83 |
+
},
|
| 84 |
+
"argv.txt": {
|
| 85 |
+
"bytes": 2594,
|
| 86 |
+
"sha256": "a5e4387cb6b97251d82ff2c7c0944adbfd99dfe9836dea761977a9efc4b1a8e1",
|
| 87 |
+
"git_blob_id": "119636f24688348e65e11a68f39ad26965eb8c09",
|
| 88 |
+
"mtime_ns": 1792517437883509646
|
| 89 |
+
},
|
| 90 |
+
"launch_contract.json": {
|
| 91 |
+
"bytes": 11726,
|
| 92 |
+
"sha256": "223894db70a34701f7cef09ddd96e9f3071732c1299144205e93c9732b643b2f",
|
| 93 |
+
"git_blob_id": "0602831abe84305afa24823cac5a42941a18c00a",
|
| 94 |
+
"mtime_ns": 1792517416873502672
|
| 95 |
+
}
|
| 96 |
+
},
|
| 97 |
+
"launch_source_sha256": "9a6e2b02c7c64f7942c3f5f02c2cd0333b6e46ea12038e0c41cdb6598df98cdb",
|
| 98 |
+
"contents": "full checkpoint including optimizer, scheduler and all rank RNG states; configs included"
|
| 99 |
+
}
|
newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:94d86163a94f30ef15e265c4d0dac9fce3b4f3e13002de616aab2e40c63b491f
|
| 3 |
+
size 6292351720
|
newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/optimizer.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fac9e89ac19d2809d02a808f51e56634abd82872a0e132ed75e347baacb5c74f
|
| 3 |
+
size 3997422027
|
newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/random_states_0.pkl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7263fa3c625381cb9bed89ad3af33e6999b691dd6996bfae103671924a4fb4a2
|
| 3 |
+
size 16577
|
newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/random_states_1.pkl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8790ae4d9825175bdd934a9c18c27df076eb4fa6e3b7bc3dff25e5f1269d66a5
|
| 3 |
+
size 16513
|
newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/random_states_2.pkl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f68bcfded24c500e9b20d70ecae4f2bdffafa75ca6efec8d6f83841c015f31fd
|
| 3 |
+
size 16513
|
newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/random_states_3.pkl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ca59cb4c2a7433ad42a391e085362f4efee4858da06b04e5a03c3a04893b6acd
|
| 3 |
+
size 16513
|
newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/random_states_4.pkl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6dc6899a74ca29dfa1d8359ef83d132daf4cb2633b2a92563fa727343322fb56
|
| 3 |
+
size 16513
|
newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/random_states_5.pkl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:50af4c80de5025aafdb36cb7e8c100e3de68b8af4cd4d50578f6dc4b29e50610
|
| 3 |
+
size 16513
|
newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/random_states_6.pkl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3db2f8ce1e38cbbb81376fae36cc32248d3338cd990ced40f80d15248abea7cf
|
| 3 |
+
size 16513
|
newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/random_states_7.pkl
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:28d8967f9073429aca66666bd3affcb8f242364b70cbb01e7e3b10f05d185186
|
| 3 |
+
size 16513
|
newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/checkpoints/ckpt_40000/scheduler.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:333a511cd62b9f69fddb60ba012421d8c02880f26090feecb0440de01e609e64
|
| 3 |
+
size 1401
|
newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/launch_contract.json
ADDED
|
@@ -0,0 +1,417 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"schema": 1,
|
| 3 |
+
"model": {
|
| 4 |
+
"resnet_store_path": null,
|
| 5 |
+
"pretrained_action_header_path": "/root/workspace/Haoxuan/data/Psi0/cache/checkpoints/psi0/postpre.1by1.pad36.2601131206.ckpt.he30k",
|
| 6 |
+
"rtc": true,
|
| 7 |
+
"max_delay": 8,
|
| 8 |
+
"action_dim": 36,
|
| 9 |
+
"action_chunk_size": 30,
|
| 10 |
+
"action_exec_horizon": 30,
|
| 11 |
+
"observation_horizon": 1,
|
| 12 |
+
"img_chunk": 1,
|
| 13 |
+
"n_cams": 1,
|
| 14 |
+
"use_obs": "add_token",
|
| 15 |
+
"rsci_variant": "VGQ-AE",
|
| 16 |
+
"rsci_vlm_mode": "frozen",
|
| 17 |
+
"rsci_lora_rank": 8,
|
| 18 |
+
"rsci_lora_alpha": 16.0,
|
| 19 |
+
"psi0_ardy_aux": false,
|
| 20 |
+
"psi0_ardy_aux_weight": 0.1,
|
| 21 |
+
"dropout": 0.1,
|
| 22 |
+
"noise_scheduler": "flow",
|
| 23 |
+
"train_diffusion_steps": 1000,
|
| 24 |
+
"eval_diffusion_steps": 10,
|
| 25 |
+
"share_cam_features": false,
|
| 26 |
+
"early_fusion": false,
|
| 27 |
+
"odim": 36,
|
| 28 |
+
"n_conditions": 0,
|
| 29 |
+
"token_fusion": "concat",
|
| 30 |
+
"loss_w": [
|
| 31 |
+
0.1,
|
| 32 |
+
0.2,
|
| 33 |
+
0.1
|
| 34 |
+
],
|
| 35 |
+
"time_dim": 256,
|
| 36 |
+
"hidden_dim": 1536,
|
| 37 |
+
"num_blocks": 6,
|
| 38 |
+
"dim_feedforward": 2048,
|
| 39 |
+
"nhead": 24,
|
| 40 |
+
"activation": "gelu",
|
| 41 |
+
"view_feature_dim": 2048,
|
| 42 |
+
"use_film": false,
|
| 43 |
+
"combined_temb": false,
|
| 44 |
+
"use_dit": false,
|
| 45 |
+
"use_depth": false,
|
| 46 |
+
"depth_encoder_name": "defm_vit_s14",
|
| 47 |
+
"depth_encoder_path": null,
|
| 48 |
+
"depth_encoder_dim": 384,
|
| 49 |
+
"depth_encoder_input_size": [
|
| 50 |
+
196,
|
| 51 |
+
350
|
| 52 |
+
],
|
| 53 |
+
"depth_encoder_patch_size": 14,
|
| 54 |
+
"freeze_depth_encoder": true,
|
| 55 |
+
"depth_cross_attention_layers": [
|
| 56 |
+
0,
|
| 57 |
+
1,
|
| 58 |
+
2,
|
| 59 |
+
3,
|
| 60 |
+
4,
|
| 61 |
+
5
|
| 62 |
+
],
|
| 63 |
+
"depth_fusion_mode": "cross_attention",
|
| 64 |
+
"depth_query_count": 16,
|
| 65 |
+
"depth_query_target_count": 8.0,
|
| 66 |
+
"depth_query_score_dim": 384,
|
| 67 |
+
"depth_query_selection": true,
|
| 68 |
+
"depth_query_use_state": true,
|
| 69 |
+
"depth_query_fuse": false,
|
| 70 |
+
"depth_query_align_depth": false,
|
| 71 |
+
"depth_query_independent_rgb": false,
|
| 72 |
+
"depth_query_concat_streams": false,
|
| 73 |
+
"depth_query_cross_attention_fusion": false,
|
| 74 |
+
"depth_rgb_encoder_path": null,
|
| 75 |
+
"depth_rgb_encoder_type": "dinov2",
|
| 76 |
+
"depth_query_pre_orthogonality": false,
|
| 77 |
+
"depth_query_pre_orthogonality_weight": 0.01,
|
| 78 |
+
"depth_query_post_orthogonality": true,
|
| 79 |
+
"depth_query_resampler_max_tokens": 4096,
|
| 80 |
+
"depth_query_gate_init": 1.0,
|
| 81 |
+
"depth_query_sparsity_init": 0.01,
|
| 82 |
+
"depth_query_sparsity_rate": 0.01,
|
| 83 |
+
"depth_query_sparsity_warmup_steps": 1000,
|
| 84 |
+
"depth_query_recovery_after_steps": 100,
|
| 85 |
+
"depth_query_recovery_weight": 0.01,
|
| 86 |
+
"depth_query_recovery_window_steps": 20,
|
| 87 |
+
"depth_query_recovery_sparsity_scale": 0.0,
|
| 88 |
+
"depth_query_alive_margin": 0.1,
|
| 89 |
+
"depth_query_alive_weight": 1.0,
|
| 90 |
+
"depth_query_selection_weight": 0.01,
|
| 91 |
+
"depth_query_orthogonality_weight": 0.01,
|
| 92 |
+
"v5_enabled": false,
|
| 93 |
+
"v5_film": true,
|
| 94 |
+
"v5_pooling": "attention",
|
| 95 |
+
"v5_ardy": true,
|
| 96 |
+
"v5_hand_state": false,
|
| 97 |
+
"v5_object": true,
|
| 98 |
+
"v5_time_encoding": true,
|
| 99 |
+
"v5_temporal_layers": 0,
|
| 100 |
+
"v5_ardy_weight": 0.1,
|
| 101 |
+
"v5_object_weight": 0.1,
|
| 102 |
+
"v5_intra_modal_dedup": false,
|
| 103 |
+
"v5_intra_modal_dedup_weight": 0.01,
|
| 104 |
+
"depth_gate_init": 0.001,
|
| 105 |
+
"depth_module_lr": 0.0001,
|
| 106 |
+
"action_expert_lr": 0.0001,
|
| 107 |
+
"weight_decay": 0.01,
|
| 108 |
+
"model_name_or_path": "/root/workspace/Haoxuan/data/Psi0/cache/checkpoints/psi0/pre.fast.1by1.2601091803.ckpt.ego200k.he30k",
|
| 109 |
+
"vlm_ckpt_step": null,
|
| 110 |
+
"tune_vlm": false,
|
| 111 |
+
"tune_mm_llm": false,
|
| 112 |
+
"tune_mm_vision": false,
|
| 113 |
+
"tune_mm_mlp": false,
|
| 114 |
+
"gradient_checkpointing": true,
|
| 115 |
+
"lang_backbone_lr": 1e-05,
|
| 116 |
+
"mm_projector_lr": 1e-05,
|
| 117 |
+
"vision_tower_lr": 1e-06,
|
| 118 |
+
"optim": "adamw_torch",
|
| 119 |
+
"model_max_length": 4096,
|
| 120 |
+
"data_flatten": true,
|
| 121 |
+
"data_packing": true,
|
| 122 |
+
"max_pixels": 451584,
|
| 123 |
+
"min_pixels": 12544
|
| 124 |
+
},
|
| 125 |
+
"data": {
|
| 126 |
+
"transform": {
|
| 127 |
+
"repack": {
|
| 128 |
+
"dataset_name": "simple",
|
| 129 |
+
"rsci_variant": "VGQ-AE",
|
| 130 |
+
"psi0_ardy_aux": false,
|
| 131 |
+
"v5_ardy": false,
|
| 132 |
+
"v5_object": false,
|
| 133 |
+
"num_past_frames": 0,
|
| 134 |
+
"action_chunk_size": 30,
|
| 135 |
+
"use_depth": false,
|
| 136 |
+
"depth_key": "observation.depth.egocentric",
|
| 137 |
+
"pad_action_dim": 36,
|
| 138 |
+
"pad_state_dim": 36
|
| 139 |
+
},
|
| 140 |
+
"model": {
|
| 141 |
+
"depth_query_independent_rgb": false,
|
| 142 |
+
"resize": {
|
| 143 |
+
"size": [
|
| 144 |
+
180,
|
| 145 |
+
320
|
| 146 |
+
]
|
| 147 |
+
},
|
| 148 |
+
"center_crop": {
|
| 149 |
+
"size": [
|
| 150 |
+
180,
|
| 151 |
+
320
|
| 152 |
+
]
|
| 153 |
+
},
|
| 154 |
+
"color_jitter": {
|
| 155 |
+
"brightness": 0.2,
|
| 156 |
+
"contrast": [
|
| 157 |
+
0.8,
|
| 158 |
+
1.2
|
| 159 |
+
],
|
| 160 |
+
"saturation": [
|
| 161 |
+
0.8,
|
| 162 |
+
1.2
|
| 163 |
+
],
|
| 164 |
+
"hue": 0.05
|
| 165 |
+
},
|
| 166 |
+
"gaussian_noise": {
|
| 167 |
+
"mean": 0.0,
|
| 168 |
+
"std": 3.0,
|
| 169 |
+
"prob_skip": 0.1
|
| 170 |
+
},
|
| 171 |
+
"img_aug": true,
|
| 172 |
+
"adaptive_resize": false,
|
| 173 |
+
"img_sizes": {
|
| 174 |
+
"egodex": [
|
| 175 |
+
270,
|
| 176 |
+
480
|
| 177 |
+
],
|
| 178 |
+
"he": [
|
| 179 |
+
240,
|
| 180 |
+
320
|
| 181 |
+
]
|
| 182 |
+
}
|
| 183 |
+
},
|
| 184 |
+
"field": {
|
| 185 |
+
"stat_path": "meta/stats_psi0.json",
|
| 186 |
+
"action_norm_type": "bounds",
|
| 187 |
+
"stat_action_key": "action",
|
| 188 |
+
"stat_state_key": "states",
|
| 189 |
+
"use_norm_mask": false,
|
| 190 |
+
"action_norm_masks": [
|
| 191 |
+
true,
|
| 192 |
+
true,
|
| 193 |
+
true,
|
| 194 |
+
true,
|
| 195 |
+
true,
|
| 196 |
+
true,
|
| 197 |
+
false
|
| 198 |
+
],
|
| 199 |
+
"action_min": [
|
| 200 |
+
-0.5,
|
| 201 |
+
-1.1095792134107943e-16,
|
| 202 |
+
-1.1095792134107943e-16,
|
| 203 |
+
-1.0,
|
| 204 |
+
-1.5,
|
| 205 |
+
-1.5,
|
| 206 |
+
-1.5,
|
| 207 |
+
-0.5,
|
| 208 |
+
-0.699999988079071,
|
| 209 |
+
-0.699999988079071,
|
| 210 |
+
-2.2166350627321588e-16,
|
| 211 |
+
-2.2166350627321588e-16,
|
| 212 |
+
0.0,
|
| 213 |
+
-2.2166350627321588e-16,
|
| 214 |
+
-0.47567468881607056,
|
| 215 |
+
0.1900009959936142,
|
| 216 |
+
-0.512170135974884,
|
| 217 |
+
-0.6265152096748352,
|
| 218 |
+
-0.5008617043495178,
|
| 219 |
+
-0.5746880769729614,
|
| 220 |
+
-0.9223371148109436,
|
| 221 |
+
-0.49507391452789307,
|
| 222 |
+
-0.3437551259994507,
|
| 223 |
+
-0.6871383190155029,
|
| 224 |
+
-0.8137853145599365,
|
| 225 |
+
-0.7568023204803467,
|
| 226 |
+
-0.4524245262145996,
|
| 227 |
+
-0.4588268995285034,
|
| 228 |
+
-0.13876836001873016,
|
| 229 |
+
-0.10360867530107498,
|
| 230 |
+
-0.47856518626213074,
|
| 231 |
+
0.7400000095367432,
|
| 232 |
+
-0.5,
|
| 233 |
+
-0.5,
|
| 234 |
+
-0.26161932945251465,
|
| 235 |
+
-0.06718750298023224
|
| 236 |
+
],
|
| 237 |
+
"action_max": [
|
| 238 |
+
0.0,
|
| 239 |
+
0.699999988079071,
|
| 240 |
+
0.699999988079071,
|
| 241 |
+
0.0,
|
| 242 |
+
2.2146225653890418e-16,
|
| 243 |
+
2.2146225653890418e-16,
|
| 244 |
+
2.2146225653890418e-16,
|
| 245 |
+
1.2266071310501902e-19,
|
| 246 |
+
1.1078670818917075e-16,
|
| 247 |
+
1.1078670818917075e-16,
|
| 248 |
+
1.5,
|
| 249 |
+
1.5,
|
| 250 |
+
0.6000000238418579,
|
| 251 |
+
1.5,
|
| 252 |
+
0.2540479004383087,
|
| 253 |
+
0.7282471060752869,
|
| 254 |
+
1.2709754705429077,
|
| 255 |
+
0.42311304807662964,
|
| 256 |
+
0.8564174771308899,
|
| 257 |
+
0.5002086162567139,
|
| 258 |
+
0.5172277092933655,
|
| 259 |
+
0.16140148043632507,
|
| 260 |
+
-0.1900009959936142,
|
| 261 |
+
0.5362864136695862,
|
| 262 |
+
0.395627498626709,
|
| 263 |
+
0.5115647315979004,
|
| 264 |
+
0.5737534165382385,
|
| 265 |
+
0.6410111784934998,
|
| 266 |
+
0.18312667310237885,
|
| 267 |
+
0.28792673349380493,
|
| 268 |
+
0.2582152187824249,
|
| 269 |
+
0.7400000095367432,
|
| 270 |
+
0.5,
|
| 271 |
+
0.5,
|
| 272 |
+
0.3454970121383667,
|
| 273 |
+
0.2899305522441864
|
| 274 |
+
],
|
| 275 |
+
"state_min": [
|
| 276 |
+
-0.5564982891082764,
|
| 277 |
+
-0.27445706725120544,
|
| 278 |
+
-0.0005447770818136632,
|
| 279 |
+
-0.8388738632202148,
|
| 280 |
+
-1.3655147552490234,
|
| 281 |
+
-0.564049482345581,
|
| 282 |
+
-1.4599460363388062,
|
| 283 |
+
-0.5904954671859741,
|
| 284 |
+
-0.5149835348129272,
|
| 285 |
+
-0.6775947213172913,
|
| 286 |
+
-0.001480442238971591,
|
| 287 |
+
-0.0002713006397243589,
|
| 288 |
+
-0.0011839715298265219,
|
| 289 |
+
-0.00019419840828049928,
|
| 290 |
+
-0.4206617772579193,
|
| 291 |
+
0.13972464203834534,
|
| 292 |
+
-0.546251654624939,
|
| 293 |
+
-0.5596316456794739,
|
| 294 |
+
-0.4764360189437866,
|
| 295 |
+
-0.5082273483276367,
|
| 296 |
+
-0.9443663954734802,
|
| 297 |
+
-0.4381798803806305,
|
| 298 |
+
-0.3338131606578827,
|
| 299 |
+
-0.667724072933197,
|
| 300 |
+
-0.7291237711906433,
|
| 301 |
+
-0.7544379830360413,
|
| 302 |
+
-0.3845277726650238,
|
| 303 |
+
-0.4484957158565521,
|
| 304 |
+
-0.13709338009357452,
|
| 305 |
+
-0.07451793551445007,
|
| 306 |
+
-0.4748336970806122,
|
| 307 |
+
0.7400000095367432,
|
| 308 |
+
0.0,
|
| 309 |
+
0.0,
|
| 310 |
+
0.0,
|
| 311 |
+
0.0
|
| 312 |
+
],
|
| 313 |
+
"state_max": [
|
| 314 |
+
2.9680453735636547e-05,
|
| 315 |
+
0.38201195001602173,
|
| 316 |
+
0.6575677990913391,
|
| 317 |
+
0.004060761071741581,
|
| 318 |
+
0.0005196159472689033,
|
| 319 |
+
0.0004725759499706328,
|
| 320 |
+
0.00010080631182063371,
|
| 321 |
+
1.310737025050912e-05,
|
| 322 |
+
0.21882089972496033,
|
| 323 |
+
0.00044666804024018347,
|
| 324 |
+
0.5350406169891357,
|
| 325 |
+
1.4410353899002075,
|
| 326 |
+
1.4605127573013306,
|
| 327 |
+
1.4595911502838135,
|
| 328 |
+
0.2758432626724243,
|
| 329 |
+
0.657910943031311,
|
| 330 |
+
1.2573398351669312,
|
| 331 |
+
0.502498209476471,
|
| 332 |
+
0.829430103302002,
|
| 333 |
+
0.5248894095420837,
|
| 334 |
+
0.4653257131576538,
|
| 335 |
+
0.18638382852077484,
|
| 336 |
+
-0.1659134179353714,
|
| 337 |
+
0.48569726943969727,
|
| 338 |
+
0.4635644555091858,
|
| 339 |
+
0.4788283109664917,
|
| 340 |
+
0.6258893013000488,
|
| 341 |
+
0.653800904750824,
|
| 342 |
+
0.16572202742099762,
|
| 343 |
+
0.3150315284729004,
|
| 344 |
+
0.26422709226608276,
|
| 345 |
+
0.7400000095367432,
|
| 346 |
+
0.0,
|
| 347 |
+
0.0,
|
| 348 |
+
0.0,
|
| 349 |
+
0.0
|
| 350 |
+
],
|
| 351 |
+
"normalize_state": true,
|
| 352 |
+
"pad_action_dim": 36,
|
| 353 |
+
"pad_state_dim": 36
|
| 354 |
+
}
|
| 355 |
+
},
|
| 356 |
+
"root_dir": "/root/workspace/Haoxuan/data/Psi0/data/simple-handover-new-training-h264-20261021",
|
| 357 |
+
"v5_ardy_sidecar_root": "/root/workspace/Haoxuan/data/Psi0/data/simple-handover-new-training-ardy-h264-20261021/G1WholebodyHandoverTeleop-v0",
|
| 358 |
+
"rgb_frame_cache_root": null,
|
| 359 |
+
"train_repo_ids": [
|
| 360 |
+
"G1WholebodyHandoverTeleop-v0"
|
| 361 |
+
],
|
| 362 |
+
"val_repo_ids": [
|
| 363 |
+
"G1WholebodyHandoverTeleop-v0"
|
| 364 |
+
]
|
| 365 |
+
},
|
| 366 |
+
"train": {
|
| 367 |
+
"num_workers": 8,
|
| 368 |
+
"overfit_single_batch": false,
|
| 369 |
+
"name": "finetune",
|
| 370 |
+
"hf_token": ".hf_token",
|
| 371 |
+
"lora": false,
|
| 372 |
+
"output_dir": "/root/workspace/Haoxuan/data/Psi0/RSCI/.runs/G1WholebodyHandoverTeleop-v0/VGQ-AE/frozen/seed-292285/handover100-newh264-v1",
|
| 373 |
+
"gradient_accumulation_steps": 1,
|
| 374 |
+
"mixed_precision": "bf16",
|
| 375 |
+
"max_grad_norm": 1.0,
|
| 376 |
+
"optimizer_foreach": null,
|
| 377 |
+
"optimizer_fused": true,
|
| 378 |
+
"pin_memory": true,
|
| 379 |
+
"non_blocking_transfer": true,
|
| 380 |
+
"compile_frozen_encoders": false,
|
| 381 |
+
"val_num_workers": 8,
|
| 382 |
+
"prefetch_factor": 2,
|
| 383 |
+
"timing_enabled": false,
|
| 384 |
+
"reuse_eval_features": false,
|
| 385 |
+
"train_batch_size": 16,
|
| 386 |
+
"val_batch_size": 16,
|
| 387 |
+
"val_num_batches": 20,
|
| 388 |
+
"checkpointing_steps": 1000,
|
| 389 |
+
"checkpoint_anchor_steps": 10000,
|
| 390 |
+
"max_checkpoints_to_keep": 4,
|
| 391 |
+
"validation_steps": 500,
|
| 392 |
+
"learning_rate": 0.0001,
|
| 393 |
+
"lr_scheduler_type": "cosine",
|
| 394 |
+
"lr_scheduler_kwargs": {
|
| 395 |
+
"betas": [
|
| 396 |
+
0.95,
|
| 397 |
+
0.999
|
| 398 |
+
],
|
| 399 |
+
"weight_decay": 1e-06,
|
| 400 |
+
"eps": 1e-08
|
| 401 |
+
},
|
| 402 |
+
"scheduler_specific_kwargs": {},
|
| 403 |
+
"data_parallel": "ddp",
|
| 404 |
+
"sharding_strategy": "full-shard",
|
| 405 |
+
"deepspeed_config": "/root/workspace/Haoxuan/code/Psi0/RSCI/feture/ARDY_PSCI/src/InternVLA/config/deepseeds/zero3.json",
|
| 406 |
+
"enable_gradient_checkpointing": true,
|
| 407 |
+
"enable_mixed_precision_training": true,
|
| 408 |
+
"reduce_in_full_precision": true,
|
| 409 |
+
"max_training_steps": 40000,
|
| 410 |
+
"num_train_epochs": null,
|
| 411 |
+
"warmup_steps": 1000,
|
| 412 |
+
"warmup_ratio": null
|
| 413 |
+
},
|
| 414 |
+
"seed": 292285,
|
| 415 |
+
"world_size": 8,
|
| 416 |
+
"source_sha256": "9a6e2b02c7c64f7942c3f5f02c2cd0333b6e46ea12038e0c41cdb6598df98cdb"
|
| 417 |
+
}
|
newcheckpoint/Psi0/VGQ-AE/handover/frozen/RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1.simple.flow1000.cosine.lr1.0e-04.b128.gpus8.2610210130/run_config.json
ADDED
|
@@ -0,0 +1,434 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"exp": "RSCI-G1WholebodyHandoverTeleop-v0-VGQ-AE-frozen-handover100-newh264-v1",
|
| 3 |
+
"seed": 292285,
|
| 4 |
+
"auto_tag_run": false,
|
| 5 |
+
"eval": false,
|
| 6 |
+
"debug": false,
|
| 7 |
+
"timestamp": "2610210130",
|
| 8 |
+
"log": {
|
| 9 |
+
"logging_dir": "logs",
|
| 10 |
+
"report_to": null,
|
| 11 |
+
"log_freq": 100
|
| 12 |
+
},
|
| 13 |
+
"wandb": {
|
| 14 |
+
"project": "psi",
|
| 15 |
+
"entity": null,
|
| 16 |
+
"group": null,
|
| 17 |
+
"id": null,
|
| 18 |
+
"name": null,
|
| 19 |
+
"resume": "allow"
|
| 20 |
+
},
|
| 21 |
+
"train": {
|
| 22 |
+
"num_workers": 8,
|
| 23 |
+
"overfit_single_batch": false,
|
| 24 |
+
"name": "finetune",
|
| 25 |
+
"resume_from_checkpoint": null,
|
| 26 |
+
"skip_resumed_steps": false,
|
| 27 |
+
"hf_token": ".hf_token",
|
| 28 |
+
"lora": false,
|
| 29 |
+
"output_dir": "/root/workspace/Haoxuan/data/Psi0/RSCI/.runs/G1WholebodyHandoverTeleop-v0/VGQ-AE/frozen/seed-292285/handover100-newh264-v1",
|
| 30 |
+
"gradient_accumulation_steps": 1,
|
| 31 |
+
"mixed_precision": "bf16",
|
| 32 |
+
"max_grad_norm": 1.0,
|
| 33 |
+
"optimizer_foreach": null,
|
| 34 |
+
"optimizer_fused": true,
|
| 35 |
+
"pin_memory": true,
|
| 36 |
+
"non_blocking_transfer": true,
|
| 37 |
+
"compile_frozen_encoders": false,
|
| 38 |
+
"val_num_workers": 8,
|
| 39 |
+
"prefetch_factor": 2,
|
| 40 |
+
"timing_enabled": false,
|
| 41 |
+
"reuse_eval_features": false,
|
| 42 |
+
"train_batch_size": 16,
|
| 43 |
+
"val_batch_size": 16,
|
| 44 |
+
"val_num_batches": 20,
|
| 45 |
+
"checkpointing_steps": 1000,
|
| 46 |
+
"checkpoint_anchor_steps": 10000,
|
| 47 |
+
"max_checkpoints_to_keep": 4,
|
| 48 |
+
"validation_steps": 500,
|
| 49 |
+
"learning_rate": 0.0001,
|
| 50 |
+
"lr_scheduler_type": "cosine",
|
| 51 |
+
"lr_scheduler_kwargs": {
|
| 52 |
+
"betas": [
|
| 53 |
+
0.95,
|
| 54 |
+
0.999
|
| 55 |
+
],
|
| 56 |
+
"weight_decay": 1e-6,
|
| 57 |
+
"eps": 1e-8
|
| 58 |
+
},
|
| 59 |
+
"scheduler_specific_kwargs": {},
|
| 60 |
+
"data_parallel": "ddp",
|
| 61 |
+
"sharding_strategy": "full-shard",
|
| 62 |
+
"deepspeed_config": "/root/workspace/Haoxuan/code/Psi0/RSCI/feture/ARDY_PSCI/src/InternVLA/config/deepseeds/zero3.json",
|
| 63 |
+
"enable_gradient_checkpointing": true,
|
| 64 |
+
"enable_mixed_precision_training": true,
|
| 65 |
+
"reduce_in_full_precision": true,
|
| 66 |
+
"max_training_steps": 40000,
|
| 67 |
+
"num_train_epochs": null,
|
| 68 |
+
"warmup_steps": 1000,
|
| 69 |
+
"warmup_ratio": null
|
| 70 |
+
},
|
| 71 |
+
"data": {
|
| 72 |
+
"transform": {
|
| 73 |
+
"repack": {
|
| 74 |
+
"dataset_name": "simple",
|
| 75 |
+
"rsci_variant": null,
|
| 76 |
+
"psi0_ardy_aux": false,
|
| 77 |
+
"v5_ardy": false,
|
| 78 |
+
"v5_object": false,
|
| 79 |
+
"num_past_frames": 0,
|
| 80 |
+
"action_chunk_size": 30,
|
| 81 |
+
"use_depth": false,
|
| 82 |
+
"depth_key": "observation.depth.egocentric",
|
| 83 |
+
"pad_action_dim": 36,
|
| 84 |
+
"pad_state_dim": 36
|
| 85 |
+
},
|
| 86 |
+
"model": {
|
| 87 |
+
"depth_query_independent_rgb": false,
|
| 88 |
+
"resize": {
|
| 89 |
+
"size": [
|
| 90 |
+
180,
|
| 91 |
+
320
|
| 92 |
+
]
|
| 93 |
+
},
|
| 94 |
+
"center_crop": {
|
| 95 |
+
"size": [
|
| 96 |
+
180,
|
| 97 |
+
320
|
| 98 |
+
]
|
| 99 |
+
},
|
| 100 |
+
"color_jitter": {
|
| 101 |
+
"brightness": 0.2,
|
| 102 |
+
"contrast": [
|
| 103 |
+
0.8,
|
| 104 |
+
1.2
|
| 105 |
+
],
|
| 106 |
+
"saturation": [
|
| 107 |
+
0.8,
|
| 108 |
+
1.2
|
| 109 |
+
],
|
| 110 |
+
"hue": 0.05
|
| 111 |
+
},
|
| 112 |
+
"gaussian_noise": {
|
| 113 |
+
"mean": 0.0,
|
| 114 |
+
"std": 3.0,
|
| 115 |
+
"prob_skip": 0.1
|
| 116 |
+
},
|
| 117 |
+
"img_aug": true,
|
| 118 |
+
"adaptive_resize": false,
|
| 119 |
+
"img_sizes": {
|
| 120 |
+
"egodex": [
|
| 121 |
+
270,
|
| 122 |
+
480
|
| 123 |
+
],
|
| 124 |
+
"he": [
|
| 125 |
+
240,
|
| 126 |
+
320
|
| 127 |
+
]
|
| 128 |
+
}
|
| 129 |
+
},
|
| 130 |
+
"field": {
|
| 131 |
+
"stat_path": "meta/stats_psi0.json",
|
| 132 |
+
"action_norm_type": "bounds",
|
| 133 |
+
"stat_action_key": "action",
|
| 134 |
+
"stat_state_key": "states",
|
| 135 |
+
"use_norm_mask": false,
|
| 136 |
+
"action_norm_masks": [
|
| 137 |
+
true,
|
| 138 |
+
true,
|
| 139 |
+
true,
|
| 140 |
+
true,
|
| 141 |
+
true,
|
| 142 |
+
true,
|
| 143 |
+
false
|
| 144 |
+
],
|
| 145 |
+
"action_min": [
|
| 146 |
+
-0.5,
|
| 147 |
+
-1.1095792134107943e-16,
|
| 148 |
+
-1.1095792134107943e-16,
|
| 149 |
+
-1.0,
|
| 150 |
+
-1.5,
|
| 151 |
+
-1.5,
|
| 152 |
+
-1.5,
|
| 153 |
+
-0.5,
|
| 154 |
+
-0.699999988079071,
|
| 155 |
+
-0.699999988079071,
|
| 156 |
+
-2.2166350627321588e-16,
|
| 157 |
+
-2.2166350627321588e-16,
|
| 158 |
+
0.0,
|
| 159 |
+
-2.2166350627321588e-16,
|
| 160 |
+
-0.47567468881607056,
|
| 161 |
+
0.1900009959936142,
|
| 162 |
+
-0.512170135974884,
|
| 163 |
+
-0.6265152096748352,
|
| 164 |
+
-0.5008617043495178,
|
| 165 |
+
-0.5746880769729614,
|
| 166 |
+
-0.9223371148109436,
|
| 167 |
+
-0.49507391452789307,
|
| 168 |
+
-0.3437551259994507,
|
| 169 |
+
-0.6871383190155029,
|
| 170 |
+
-0.8137853145599365,
|
| 171 |
+
-0.7568023204803467,
|
| 172 |
+
-0.4524245262145996,
|
| 173 |
+
-0.4588268995285034,
|
| 174 |
+
-0.13876836001873016,
|
| 175 |
+
-0.10360867530107498,
|
| 176 |
+
-0.47856518626213074,
|
| 177 |
+
0.7400000095367432,
|
| 178 |
+
-0.5,
|
| 179 |
+
-0.5,
|
| 180 |
+
-0.26161932945251465,
|
| 181 |
+
-0.06718750298023224
|
| 182 |
+
],
|
| 183 |
+
"action_max": [
|
| 184 |
+
0.0,
|
| 185 |
+
0.699999988079071,
|
| 186 |
+
0.699999988079071,
|
| 187 |
+
0.0,
|
| 188 |
+
2.2146225653890418e-16,
|
| 189 |
+
2.2146225653890418e-16,
|
| 190 |
+
2.2146225653890418e-16,
|
| 191 |
+
1.2266071310501902e-19,
|
| 192 |
+
1.1078670818917075e-16,
|
| 193 |
+
1.1078670818917075e-16,
|
| 194 |
+
1.5,
|
| 195 |
+
1.5,
|
| 196 |
+
0.6000000238418579,
|
| 197 |
+
1.5,
|
| 198 |
+
0.2540479004383087,
|
| 199 |
+
0.7282471060752869,
|
| 200 |
+
1.2709754705429077,
|
| 201 |
+
0.42311304807662964,
|
| 202 |
+
0.8564174771308899,
|
| 203 |
+
0.5002086162567139,
|
| 204 |
+
0.5172277092933655,
|
| 205 |
+
0.16140148043632507,
|
| 206 |
+
-0.1900009959936142,
|
| 207 |
+
0.5362864136695862,
|
| 208 |
+
0.395627498626709,
|
| 209 |
+
0.5115647315979004,
|
| 210 |
+
0.5737534165382385,
|
| 211 |
+
0.6410111784934998,
|
| 212 |
+
0.18312667310237885,
|
| 213 |
+
0.28792673349380493,
|
| 214 |
+
0.2582152187824249,
|
| 215 |
+
0.7400000095367432,
|
| 216 |
+
0.5,
|
| 217 |
+
0.5,
|
| 218 |
+
0.3454970121383667,
|
| 219 |
+
0.2899305522441864
|
| 220 |
+
],
|
| 221 |
+
"state_min": [
|
| 222 |
+
-0.5564982891082764,
|
| 223 |
+
-0.27445706725120544,
|
| 224 |
+
-0.0005447770818136632,
|
| 225 |
+
-0.8388738632202148,
|
| 226 |
+
-1.3655147552490234,
|
| 227 |
+
-0.564049482345581,
|
| 228 |
+
-1.4599460363388062,
|
| 229 |
+
-0.5904954671859741,
|
| 230 |
+
-0.5149835348129272,
|
| 231 |
+
-0.6775947213172913,
|
| 232 |
+
-0.001480442238971591,
|
| 233 |
+
-0.0002713006397243589,
|
| 234 |
+
-0.0011839715298265219,
|
| 235 |
+
-0.00019419840828049928,
|
| 236 |
+
-0.4206617772579193,
|
| 237 |
+
0.13972464203834534,
|
| 238 |
+
-0.546251654624939,
|
| 239 |
+
-0.5596316456794739,
|
| 240 |
+
-0.4764360189437866,
|
| 241 |
+
-0.5082273483276367,
|
| 242 |
+
-0.9443663954734802,
|
| 243 |
+
-0.4381798803806305,
|
| 244 |
+
-0.3338131606578827,
|
| 245 |
+
-0.667724072933197,
|
| 246 |
+
-0.7291237711906433,
|
| 247 |
+
-0.7544379830360413,
|
| 248 |
+
-0.3845277726650238,
|
| 249 |
+
-0.4484957158565521,
|
| 250 |
+
-0.13709338009357452,
|
| 251 |
+
-0.07451793551445007,
|
| 252 |
+
-0.4748336970806122,
|
| 253 |
+
0.7400000095367432,
|
| 254 |
+
0.0,
|
| 255 |
+
0.0,
|
| 256 |
+
0.0,
|
| 257 |
+
0.0
|
| 258 |
+
],
|
| 259 |
+
"state_max": [
|
| 260 |
+
0.000029680453735636547,
|
| 261 |
+
0.38201195001602173,
|
| 262 |
+
0.6575677990913391,
|
| 263 |
+
0.004060761071741581,
|
| 264 |
+
0.0005196159472689033,
|
| 265 |
+
0.0004725759499706328,
|
| 266 |
+
0.00010080631182063371,
|
| 267 |
+
0.00001310737025050912,
|
| 268 |
+
0.21882089972496033,
|
| 269 |
+
0.00044666804024018347,
|
| 270 |
+
0.5350406169891357,
|
| 271 |
+
1.4410353899002075,
|
| 272 |
+
1.4605127573013306,
|
| 273 |
+
1.4595911502838135,
|
| 274 |
+
0.2758432626724243,
|
| 275 |
+
0.657910943031311,
|
| 276 |
+
1.2573398351669312,
|
| 277 |
+
0.502498209476471,
|
| 278 |
+
0.829430103302002,
|
| 279 |
+
0.5248894095420837,
|
| 280 |
+
0.4653257131576538,
|
| 281 |
+
0.18638382852077484,
|
| 282 |
+
-0.1659134179353714,
|
| 283 |
+
0.48569726943969727,
|
| 284 |
+
0.4635644555091858,
|
| 285 |
+
0.4788283109664917,
|
| 286 |
+
0.6258893013000488,
|
| 287 |
+
0.653800904750824,
|
| 288 |
+
0.16572202742099762,
|
| 289 |
+
0.3150315284729004,
|
| 290 |
+
0.26422709226608276,
|
| 291 |
+
0.7400000095367432,
|
| 292 |
+
0.0,
|
| 293 |
+
0.0,
|
| 294 |
+
0.0,
|
| 295 |
+
0.0
|
| 296 |
+
],
|
| 297 |
+
"normalize_state": true,
|
| 298 |
+
"pad_action_dim": 36,
|
| 299 |
+
"pad_state_dim": 36
|
| 300 |
+
}
|
| 301 |
+
},
|
| 302 |
+
"root_dir": "/root/workspace/Haoxuan/data/Psi0/data/simple-handover-new-training-h264-20261021",
|
| 303 |
+
"v5_ardy_sidecar_root": "/root/workspace/Haoxuan/data/Psi0/data/simple-handover-new-training-ardy-h264-20261021/G1WholebodyHandoverTeleop-v0",
|
| 304 |
+
"rgb_frame_cache_root": null,
|
| 305 |
+
"train_repo_ids": [
|
| 306 |
+
"G1WholebodyHandoverTeleop-v0"
|
| 307 |
+
],
|
| 308 |
+
"val_repo_ids": [
|
| 309 |
+
"G1WholebodyHandoverTeleop-v0"
|
| 310 |
+
]
|
| 311 |
+
},
|
| 312 |
+
"model": {
|
| 313 |
+
"resnet_store_path": null,
|
| 314 |
+
"pretrained_action_header_path": "/root/workspace/Haoxuan/data/Psi0/cache/checkpoints/psi0/postpre.1by1.pad36.2601131206.ckpt.he30k",
|
| 315 |
+
"rtc": true,
|
| 316 |
+
"max_delay": 8,
|
| 317 |
+
"action_dim": 36,
|
| 318 |
+
"action_chunk_size": 30,
|
| 319 |
+
"action_exec_horizon": 30,
|
| 320 |
+
"observation_horizon": 1,
|
| 321 |
+
"img_chunk": 1,
|
| 322 |
+
"n_cams": 1,
|
| 323 |
+
"use_obs": "add_token",
|
| 324 |
+
"rsci_variant": "VGQ-AE",
|
| 325 |
+
"rsci_vlm_mode": "frozen",
|
| 326 |
+
"rsci_lora_rank": 8,
|
| 327 |
+
"rsci_lora_alpha": 16.0,
|
| 328 |
+
"psi0_ardy_aux": false,
|
| 329 |
+
"psi0_ardy_aux_weight": 0.1,
|
| 330 |
+
"dropout": 0.1,
|
| 331 |
+
"noise_scheduler": "flow",
|
| 332 |
+
"train_diffusion_steps": 1000,
|
| 333 |
+
"eval_diffusion_steps": 10,
|
| 334 |
+
"share_cam_features": false,
|
| 335 |
+
"early_fusion": false,
|
| 336 |
+
"odim": 36,
|
| 337 |
+
"n_conditions": 0,
|
| 338 |
+
"token_fusion": "concat",
|
| 339 |
+
"loss_w": [
|
| 340 |
+
0.1,
|
| 341 |
+
0.2,
|
| 342 |
+
0.1
|
| 343 |
+
],
|
| 344 |
+
"time_dim": 256,
|
| 345 |
+
"hidden_dim": 1536,
|
| 346 |
+
"num_blocks": 6,
|
| 347 |
+
"dim_feedforward": 2048,
|
| 348 |
+
"nhead": 24,
|
| 349 |
+
"activation": "gelu",
|
| 350 |
+
"view_feature_dim": 2048,
|
| 351 |
+
"use_film": false,
|
| 352 |
+
"combined_temb": false,
|
| 353 |
+
"use_dit": false,
|
| 354 |
+
"use_depth": false,
|
| 355 |
+
"depth_encoder_name": "defm_vit_s14",
|
| 356 |
+
"depth_encoder_path": null,
|
| 357 |
+
"depth_encoder_dim": 384,
|
| 358 |
+
"depth_encoder_input_size": [
|
| 359 |
+
196,
|
| 360 |
+
350
|
| 361 |
+
],
|
| 362 |
+
"depth_encoder_patch_size": 14,
|
| 363 |
+
"freeze_depth_encoder": true,
|
| 364 |
+
"depth_cross_attention_layers": [
|
| 365 |
+
0,
|
| 366 |
+
1,
|
| 367 |
+
2,
|
| 368 |
+
3,
|
| 369 |
+
4,
|
| 370 |
+
5
|
| 371 |
+
],
|
| 372 |
+
"depth_fusion_mode": "cross_attention",
|
| 373 |
+
"depth_query_count": 16,
|
| 374 |
+
"depth_query_target_count": 8.0,
|
| 375 |
+
"depth_query_score_dim": 384,
|
| 376 |
+
"depth_query_selection": true,
|
| 377 |
+
"depth_query_use_state": true,
|
| 378 |
+
"depth_query_fuse": false,
|
| 379 |
+
"depth_query_align_depth": false,
|
| 380 |
+
"depth_query_independent_rgb": false,
|
| 381 |
+
"depth_query_concat_streams": false,
|
| 382 |
+
"depth_query_cross_attention_fusion": false,
|
| 383 |
+
"depth_rgb_encoder_path": null,
|
| 384 |
+
"depth_rgb_encoder_type": "dinov2",
|
| 385 |
+
"depth_query_pre_orthogonality": false,
|
| 386 |
+
"depth_query_pre_orthogonality_weight": 0.01,
|
| 387 |
+
"depth_query_post_orthogonality": true,
|
| 388 |
+
"depth_query_resampler_max_tokens": 4096,
|
| 389 |
+
"depth_query_gate_init": 1.0,
|
| 390 |
+
"depth_query_sparsity_init": 0.01,
|
| 391 |
+
"depth_query_sparsity_rate": 0.01,
|
| 392 |
+
"depth_query_sparsity_warmup_steps": 1000,
|
| 393 |
+
"depth_query_recovery_after_steps": 100,
|
| 394 |
+
"depth_query_recovery_weight": 0.01,
|
| 395 |
+
"depth_query_recovery_window_steps": 20,
|
| 396 |
+
"depth_query_recovery_sparsity_scale": 0.0,
|
| 397 |
+
"depth_query_alive_margin": 0.1,
|
| 398 |
+
"depth_query_alive_weight": 1.0,
|
| 399 |
+
"depth_query_selection_weight": 0.01,
|
| 400 |
+
"depth_query_orthogonality_weight": 0.01,
|
| 401 |
+
"v5_enabled": false,
|
| 402 |
+
"v5_film": true,
|
| 403 |
+
"v5_pooling": "attention",
|
| 404 |
+
"v5_ardy": true,
|
| 405 |
+
"v5_hand_state": false,
|
| 406 |
+
"v5_object": true,
|
| 407 |
+
"v5_time_encoding": true,
|
| 408 |
+
"v5_temporal_layers": 0,
|
| 409 |
+
"v5_ardy_weight": 0.1,
|
| 410 |
+
"v5_object_weight": 0.1,
|
| 411 |
+
"v5_intra_modal_dedup": false,
|
| 412 |
+
"v5_intra_modal_dedup_weight": 0.01,
|
| 413 |
+
"depth_gate_init": 0.001,
|
| 414 |
+
"depth_module_lr": 0.0001,
|
| 415 |
+
"action_expert_lr": 0.0001,
|
| 416 |
+
"weight_decay": 0.01,
|
| 417 |
+
"model_name_or_path": "/root/workspace/Haoxuan/data/Psi0/cache/checkpoints/psi0/pre.fast.1by1.2601091803.ckpt.ego200k.he30k",
|
| 418 |
+
"vlm_ckpt_step": null,
|
| 419 |
+
"tune_vlm": false,
|
| 420 |
+
"tune_mm_llm": false,
|
| 421 |
+
"tune_mm_vision": false,
|
| 422 |
+
"tune_mm_mlp": false,
|
| 423 |
+
"gradient_checkpointing": true,
|
| 424 |
+
"lang_backbone_lr": 0.00001,
|
| 425 |
+
"mm_projector_lr": 0.00001,
|
| 426 |
+
"vision_tower_lr": 1e-6,
|
| 427 |
+
"optim": "adamw_torch",
|
| 428 |
+
"model_max_length": 4096,
|
| 429 |
+
"data_flatten": true,
|
| 430 |
+
"data_packing": true,
|
| 431 |
+
"max_pixels": 451584,
|
| 432 |
+
"min_pixels": 12544
|
| 433 |
+
}
|
| 434 |
+
}
|