sync new results for soup_S_100M_work_dir

.gitattributes

@@ -50,3 +50,6 @@ soup_XXL_100M_work_dir/wandb/run-20260629_164439-28bdlj4m/run-28bdlj4m.wandb fil
 soup_XXXL_100M_work_dir/wandb/run-20260629_001336-li3322kn/run-li3322kn.wandb filter=lfs diff=lfs merge=lfs -text
 soup_XXXL_100M_work_dir/wandb/run-20260629_164506-22igyjpv/run-22igyjpv.wandb filter=lfs diff=lfs merge=lfs -text
 soup_B_100M_work_dir/wandb/run-20260708_090206-4s2ksbsn/run-4s2ksbsn.wandb filter=lfs diff=lfs merge=lfs -text
+soup_S_100M_work_dir/wandb/run-20260708_114417-8pbrlu7w/run-8pbrlu7w.wandb filter=lfs diff=lfs merge=lfs -text
+soup_S_100M_work_dir/wandb/run-20260710_081650-8qh9ccql/run-8qh9ccql.wandb filter=lfs diff=lfs merge=lfs -text
+soup_S_100M_work_dir/wandb/run-20260710_200207-wzv4r5ph/run-wzv4r5ph.wandb filter=lfs diff=lfs merge=lfs -text

soup_S_100M_work_dir/ckpt/0.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:26e9503089cf22d27ab64bf78ca3c4834f7890caf66e43229a79c5ac714a6bcd
+size 33756331

soup_S_100M_work_dir/ckpt/10_000_000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:537f295080c96ffab8f7d3318432d1701b00be05b53c9bb95cb37d05200a6dea
+size 36227605

soup_S_100M_work_dir/ckpt/15_000_000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9d7253e7779022a72c46c18ce53f46448c2142428c46d1a25bc6e0f85c1ecac7
+size 36227605

soup_S_100M_work_dir/ckpt/20_000_000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:41c845efcf7114869ef5daf24f5d9185619db8916543edd852ce1ad3e2230ba0
+size 36227605

soup_S_100M_work_dir/ckpt/25_000_000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86e1860e27630fb1c3d9131ba2a10194dd2c6896804813bcb66e4518d364d78b
+size 36227605

soup_S_100M_work_dir/ckpt/5_000_000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba5978a255925988db58ef90b1594f1f9d10fad5b02894ad2346fea068e8b34f
+size 36227211

soup_S_100M_work_dir/wandb/debug-internal.log

@@ -0,0 +1,10 @@
+{"time":"2026-07-10T20:02:08.042270494+08:00","level":"INFO","msg":"stream: starting","core version":"0.22.1"}
+{"time":"2026-07-10T20:02:08.571791737+08:00","level":"INFO","msg":"stream: created new stream","id":"wzv4r5ph"}
+{"time":"2026-07-10T20:02:08.571857921+08:00","level":"INFO","msg":"handler: started","stream_id":"wzv4r5ph"}
+{"time":"2026-07-10T20:02:08.572352151+08:00","level":"INFO","msg":"stream: started","id":"wzv4r5ph"}
+{"time":"2026-07-10T20:02:08.572360008+08:00","level":"INFO","msg":"writer: started","stream_id":"wzv4r5ph"}
+{"time":"2026-07-10T20:02:08.572362278+08:00","level":"INFO","msg":"sender: started","stream_id":"wzv4r5ph"}
+{"time":"2026-07-10T20:12:29.33661486+08:00","level":"INFO","msg":"flowcontrol: backed up, offloading to disk","recordNumber":5608}
+{"time":"2026-07-10T20:12:29.799465299+08:00","level":"INFO","msg":"flowcontrol: unblocked","totalOffloaded":17}
+{"time":"2026-07-10T20:34:17.674522565+08:00","level":"INFO","msg":"flowcontrol: backed up, offloading to disk","recordNumber":17923}
+{"time":"2026-07-10T20:34:20.955546777+08:00","level":"INFO","msg":"flowcontrol: unblocked","totalOffloaded":31}

soup_S_100M_work_dir/wandb/run-20260708_114417-8pbrlu7w/files/output.log

@@ -0,0 +1,1487 @@
+Logs will be synced with wandb.
+Architecture: DDPWrapper(
+  (_module): DistributedDataParallel(
+    (module): Newt World Model
+    Encoder (132,608): ModuleDict(
+      (state): Sequential(
+        (0): NormedLinear(in_features=640, out_features=128, bias=True, act=Mish)
+        (1): NormedLinear(in_features=128, out_features=384, bias=True, act=SimNorm)
+      )
+    )
+    Dynamics (400,000): Sequential(
+      (0): NormedLinear(in_features=912, out_features=256, bias=True, act=Mish)
+      (1): NormedLinear(in_features=256, out_features=256, bias=True, act=Mish)
+      (2): NormedLinear(in_features=256, out_features=384, bias=True, act=SimNorm)
+    )
+    Reward (326,501): Sequential(
+      (0): NormedLinear(in_features=912, out_features=256, bias=True, act=Mish)
+      (1): NormedLinear(in_features=256, out_features=256, bias=True, act=Mish)
+      (2): Linear(in_features=256, out_features=101, bias=True)
+    )
+    Contrastive F (300,801): Sequential(
+      (0): NormedLinear(in_features=912, out_features=256, bias=True, act=Mish)
+      (1): NormedLinear(in_features=256, out_features=256, bias=True, act=Mish)
+      (2): Linear(in_features=256, out_features=1, bias=True)
+    )
+    Policy prior (304,672): Sequential(
+      (0): NormedLinear(in_features=896, out_features=256, bias=True, act=Mish)
+      (1): NormedLinear(in_features=256, out_features=256, bias=True, act=Mish)
+      (2): Linear(in_features=256, out_features=32, bias=True)
+    )
+    Q-functions (979,503): QEnsemble(
+      (_Qs): ModuleList(
+        (0-2): 3 x Sequential(
+          (0): NormedLinear(in_features=912, out_features=256, bias=True, act=Mish)
+          (1): NormedLinear(in_features=256, out_features=256, bias=True, act=Mish)
+          (2): Linear(in_features=256, out_features=101, bias=True)
+        )
+      )
+    )
+    Learnable parameters: 2,444,085
+  )
+)
+Update frequency: 200,000
+Episodes per update frequency: 1,933
+No checkpoint found, training from scratch.
+[Rank 0] Pretrain start
+Pretraining agent on demonstrations...
+[Rank 0] prior_coef is 10.0, setting to 1.0 for pretraining.
+Pretraining:   0%|          | 1/100000 [00:24<679:38:37, 24.47s/it][rank0]:V0708 11:44:46.470000 301 site-packages/torch/_dynamo/guards.py:3508] [0/1] [__recompiles] Recompiling function _loss_fn in /media/damoxing/che-liu-fileset/cxy_worldmodel/newt/tdmpc2/tdmpc2.py:488
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.02521
+ reward_loss           4.30745
+ value_loss            4.30745
+ total_loss            2.20881
+ bc_loss               0.48458
+ entropy_loss          0.00077
+ pi_prior_loss         0.14998
+ pi_entropy            0.45615
+ pi_scaled_entropy     -7.73570
+ pi_std                0.77084
+ pi_max_std            1.00000
+ contrastive_loss      0.69315
+ contrastive_pos_logit 0.00000
+ contrastive_neg_logit 0.00000
+ contrastive_mean      0.00000
+ contrastive_std       0.99000
+ grad_norm             2.81896
+ lr_enc                0.00000
+ lr                    0.00000
+ lr_pi                 0.00000
+------------------------------
+[rank0]:V0708 11:44:46.470000 301 site-packages/torch/_dynamo/guards.py:3508] [0/1] [__recompiles]     triggered by the following guard failure(s):
+[rank0]:V0708 11:44:46.470000 301 site-packages/torch/_dynamo/guards.py:3508] [0/1] [__recompiles]     - 0/0: len(G['__import_tensordict_dot_utils']._TENSORCLASS_MEMO) != 15
+Pretraining:   4%|▍         | 4000/100000 [04:42<2:03:16, 12.98it/s]
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00273
+ reward_loss           0.73688
+ value_loss            0.81584
+ total_loss            0.94457
+ bc_loss               0.31269
+ entropy_loss          -0.00016
+ pi_prior_loss         0.10026
+ pi_entropy            1.71875
+ pi_scaled_entropy     1.57193
+ pi_std                0.74871
+ pi_max_std            1.00000
+ contrastive_loss      0.63443
+ contrastive_pos_logit 0.23565
+ contrastive_neg_logit -0.24741
+ contrastive_mean      0.00116
+ contrastive_std       0.69165
+ grad_norm             3.10157
+ lr_enc                0.00004
+ lr                    0.00012
+ lr_pi                 0.00012
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00229
+ reward_loss           0.69771
+ value_loss            0.80274
+ total_loss            0.92310
+ bc_loss               0.27428
+ entropy_loss          -0.00066
+ pi_prior_loss         0.08870
+ pi_entropy            1.89145
+ pi_scaled_entropy     6.62537
+ pi_std                0.74833
+ pi_max_std            1.00000
+ contrastive_loss      0.63847
+ contrastive_pos_logit 0.43973
+ contrastive_neg_logit -0.10414
+ contrastive_mean      0.02803
+ contrastive_std       0.74187
+ grad_norm             3.10540
+ lr_enc                0.00007
+ lr                    0.00024
+ lr_pi                 0.00024
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00219
+ reward_loss           0.56483
+ value_loss            0.57832
+ total_loss            0.89189
+ bc_loss               0.26940
+ entropy_loss          -0.00074
+ pi_prior_loss         0.08640
+ pi_entropy            2.70065
+ pi_scaled_entropy     7.37485
+ pi_std                0.77493
+ pi_max_std            1.00000
+ contrastive_loss      0.64729
+ contrastive_pos_logit 0.42179
+ contrastive_neg_logit -0.09580
+ contrastive_mean      0.03044
+ contrastive_std       0.73179
+ grad_norm             2.87581
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00221
+ reward_loss           0.47316
+ value_loss            0.60535
+ total_loss            0.89205
+ bc_loss               0.26799
+ entropy_loss          -0.00080
+ pi_prior_loss         0.08537
+ pi_entropy            2.82074
+ pi_scaled_entropy     7.96440
+ pi_std                0.76962
+ pi_max_std            1.00000
+ contrastive_loss      0.65457
+ contrastive_pos_logit 0.40211
+ contrastive_neg_logit -0.09049
+ contrastive_mean      0.03336
+ contrastive_std       0.76136
+ grad_norm             2.20943
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00220
+ reward_loss           0.55349
+ value_loss            0.53544
+ total_loss            0.87441
+ bc_loss               0.24836
+ entropy_loss          -0.00085
+ pi_prior_loss         0.08071
+ pi_entropy            2.37867
+ pi_scaled_entropy     8.46805
+ pi_std                0.76538
+ pi_max_std            1.00000
+ contrastive_loss      0.64081
+ contrastive_pos_logit 0.36977
+ contrastive_neg_logit -0.11713
+ contrastive_mean      0.03109
+ contrastive_std       0.77046
+ grad_norm             1.63281
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00212
+ reward_loss           0.52657
+ value_loss            0.54138
+ total_loss            0.86344
+ bc_loss               0.24966
+ entropy_loss          -0.00073
+ pi_prior_loss         0.07663
+ pi_entropy            2.18751
+ pi_scaled_entropy     7.33219
+ pi_std                0.75425
+ pi_max_std            1.00000
+ contrastive_loss      0.63766
+ contrastive_pos_logit 0.36458
+ contrastive_neg_logit -0.18740
+ contrastive_mean      0.03447
+ contrastive_std       0.79517
+ grad_norm             1.47809
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00183
+ reward_loss           0.47468
+ value_loss            0.53324
+ total_loss            0.84602
+ bc_loss               0.20999
+ entropy_loss          -0.00085
+ pi_prior_loss         0.06863
+ pi_entropy            1.89196
+ pi_scaled_entropy     8.47269
+ pi_std                0.76286
+ pi_max_std            1.00000
+ contrastive_loss      0.64002
+ contrastive_pos_logit 0.19249
+ contrastive_neg_logit -0.22988
+ contrastive_mean      0.03106
+ contrastive_std       0.78956
+ grad_norm             1.59315
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00217
+ reward_loss           0.48962
+ value_loss            0.57187
+ total_loss            0.87775
+ bc_loss               0.22624
+ entropy_loss          -0.00100
+ pi_prior_loss         0.07300
+ pi_entropy            2.45651
+ pi_scaled_entropy     9.99224
+ pi_std                0.76919
+ pi_max_std            1.00000
+ contrastive_loss      0.65515
+ contrastive_pos_logit 0.17933
+ contrastive_neg_logit -0.23876
+ contrastive_mean      0.03345
+ contrastive_std       0.81210
+ grad_norm             1.89072
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00210
+ reward_loss           0.47250
+ value_loss            0.54662
+ total_loss            0.83244
+ bc_loss               0.23421
+ entropy_loss          -0.00125
+ pi_prior_loss         0.07469
+ pi_entropy            2.17788
+ pi_scaled_entropy     12.45328
+ pi_std                0.76042
+ pi_max_std            1.08264
+ contrastive_loss      0.61383
+ contrastive_pos_logit 0.33355
+ contrastive_neg_logit -0.30452
+ contrastive_mean      0.03305
+ contrastive_std       0.83245
+ grad_norm             1.19811
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00240
+ reward_loss           0.55856
+ value_loss            0.97694
+ total_loss            0.90722
+ bc_loss               0.25566
+ entropy_loss          -0.00086
+ pi_prior_loss         0.07845
+ pi_entropy            1.73890
+ pi_scaled_entropy     8.62198
+ pi_std                0.74759
+ pi_max_std            1.00000
+ contrastive_loss      0.62729
+ contrastive_pos_logit 0.35662
+ contrastive_neg_logit -0.29679
+ contrastive_mean      0.03428
+ contrastive_std       0.82575
+ grad_norm             1.64994
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00190
+ reward_loss           0.46923
+ value_loss            0.58825
+ total_loss            0.84705
+ bc_loss               0.22924
+ entropy_loss          -0.00104
+ pi_prior_loss         0.07392
+ pi_entropy            2.44861
+ pi_scaled_entropy     10.44510
+ pi_std                0.76906
+ pi_max_std            1.10454
+ contrastive_loss      0.62931
+ contrastive_pos_logit 0.26092
+ contrastive_neg_logit -0.26009
+ contrastive_mean      0.03492
+ contrastive_std       0.83998
+ grad_norm             1.30396
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00214
+ reward_loss           0.48103
+ value_loss            0.57262
+ total_loss            0.85669
+ bc_loss               0.23138
+ entropy_loss          -0.00133
+ pi_prior_loss         0.07232
+ pi_entropy            2.06099
+ pi_scaled_entropy     13.27265
+ pi_std                0.76157
+ pi_max_std            1.00000
+ contrastive_loss      0.63617
+ contrastive_pos_logit 0.26912
+ contrastive_neg_logit -0.27757
+ contrastive_mean      0.03577
+ contrastive_std       0.84821
+ grad_norm             1.37652
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00223
+ reward_loss           0.49712
+ value_loss            0.49257
+ total_loss            0.83504
+ bc_loss               0.21489
+ entropy_loss          -0.00095
+ pi_prior_loss         0.06854
+ pi_entropy            1.56783
+ pi_scaled_entropy     9.53431
+ pi_std                0.75405
+ pi_max_std            1.00000
+ contrastive_loss      0.62287
+ contrastive_pos_logit 0.31128
+ contrastive_neg_logit -0.23698
+ contrastive_mean      0.03434
+ contrastive_std       0.84147
+ grad_norm             1.14734
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00211
+ reward_loss           0.49948
+ value_loss            0.53062
+ total_loss            0.85816
+ bc_loss               0.23369
+ entropy_loss          -0.00117
+ pi_prior_loss         0.07388
+ pi_entropy            2.31661
+ pi_scaled_entropy     11.71799
+ pi_std                0.76682
+ pi_max_std            1.00000
+ contrastive_loss      0.63914
+ contrastive_pos_logit 0.37158
+ contrastive_neg_logit -0.19740
+ contrastive_mean      0.03458
+ contrastive_std       0.86313
+ grad_norm             1.65834
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00197
+ reward_loss           0.49435
+ value_loss            0.58107
+ total_loss            0.84753
+ bc_loss               0.22560
+ entropy_loss          -0.00107
+ pi_prior_loss         0.06999
+ pi_entropy            2.62426
+ pi_scaled_entropy     10.71816
+ pi_std                0.77505
+ pi_max_std            1.00000
+ contrastive_loss      0.63055
+ contrastive_pos_logit 0.33054
+ contrastive_neg_logit -0.19825
+ contrastive_mean      0.03587
+ contrastive_std       0.86288
+ grad_norm             1.23997
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00181
+ reward_loss           0.45691
+ value_loss            0.59176
+ total_loss            0.84152
+ bc_loss               0.20923
+ entropy_loss          -0.00101
+ pi_prior_loss         0.06803
+ pi_entropy            2.59903
+ pi_scaled_entropy     10.08895
+ pi_std                0.77989
+ pi_max_std            1.00000
+ contrastive_loss      0.63246
+ contrastive_pos_logit 0.27849
+ contrastive_neg_logit -0.24854
+ contrastive_mean      0.03822
+ contrastive_std       0.89365
+ grad_norm             1.09990
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00190
+ reward_loss           0.48853
+ value_loss            0.59212
+ total_loss            0.83372
+ bc_loss               0.21282
+ entropy_loss          -0.00097
+ pi_prior_loss         0.06613
+ pi_entropy            1.76489
+ pi_scaled_entropy     9.72047
+ pi_std                0.76253
+ pi_max_std            1.07610
+ contrastive_loss      0.62161
+ contrastive_pos_logit 0.37887
+ contrastive_neg_logit -0.32792
+ contrastive_mean      0.03664
+ contrastive_std       0.89867
+ grad_norm             1.11028
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00193
+ reward_loss           0.48886
+ value_loss            0.60991
+ total_loss            0.82927
+ bc_loss               0.21882
+ entropy_loss          -0.00096
+ pi_prior_loss         0.06843
+ pi_entropy            1.71882
+ pi_scaled_entropy     9.59138
+ pi_std                0.75810
+ pi_max_std            1.00000
+ contrastive_loss      0.61229
+ contrastive_pos_logit 0.42240
+ contrastive_neg_logit -0.31080
+ contrastive_mean      0.03751
+ contrastive_std       0.91459
+ grad_norm             1.47667
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00196
+ reward_loss           0.51903
+ value_loss            0.63513
+ total_loss            0.84469
+ bc_loss               0.23613
+ entropy_loss          -0.00099
+ pi_prior_loss         0.07387
+ pi_entropy            1.44428
+ pi_scaled_entropy     9.87240
+ pi_std                0.76388
+ pi_max_std            1.46006
+ contrastive_loss      0.61614
+ contrastive_pos_logit 0.36633
+ contrastive_neg_logit -0.35031
+ contrastive_mean      0.03540
+ contrastive_std       0.91298
+ grad_norm             1.96202
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00195
+ reward_loss           0.48514
+ value_loss            0.55571
+ total_loss            0.83539
+ bc_loss               0.22322
+ entropy_loss          -0.00103
+ pi_prior_loss         0.07068
+ pi_entropy            1.67680
+ pi_scaled_entropy     10.29625
+ pi_std                0.75268
+ pi_max_std            1.07386
+ contrastive_loss      0.62160
+ contrastive_pos_logit 0.45456
+ contrastive_neg_logit -0.28074
+ contrastive_mean      0.03664
+ contrastive_std       0.92362
+ grad_norm             1.03191
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00198
+ reward_loss           0.46844
+ value_loss            0.53886
+ total_loss            0.84156
+ bc_loss               0.21182
+ entropy_loss          -0.00095
+ pi_prior_loss         0.06720
+ pi_entropy            2.00899
+ pi_scaled_entropy     9.47758
+ pi_std                0.76027
+ pi_max_std            1.13343
+ contrastive_loss      0.63400
+ contrastive_pos_logit 0.36734
+ contrastive_neg_logit -0.26811
+ contrastive_mean      0.03685
+ contrastive_std       0.92128
+ grad_norm             1.45832
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00174
+ reward_loss           0.48662
+ value_loss            0.53879
+ total_loss            0.82614
+ bc_loss               0.19286
+ entropy_loss          -0.00117
+ pi_prior_loss         0.06157
+ pi_entropy            1.83995
+ pi_scaled_entropy     11.74927
+ pi_std                0.75983
+ pi_max_std            1.35816
+ contrastive_loss      0.62728
+ contrastive_pos_logit 0.38343
+ contrastive_neg_logit -0.25265
+ contrastive_mean      0.03530
+ contrastive_std       0.92869
+ grad_norm             1.19864
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00202
+ reward_loss           0.43461
+ value_loss            0.58218
+ total_loss            0.80958
+ bc_loss               0.21323
+ entropy_loss          -0.00126
+ pi_prior_loss         0.06631
+ pi_entropy            2.22129
+ pi_scaled_entropy     12.64654
+ pi_std                0.77008
+ pi_max_std            1.15293
+ contrastive_loss      0.60127
+ contrastive_pos_logit 0.48756
+ contrastive_neg_logit -0.30817
+ contrastive_mean      0.03804
+ contrastive_std       0.93538
+ grad_norm             1.24997
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00189
+ reward_loss           0.44861
+ value_loss            0.63508
+ total_loss            0.83189
+ bc_loss               0.19177
+ entropy_loss          -0.00113
+ pi_prior_loss         0.05957
+ pi_entropy            1.88113
+ pi_scaled_entropy     11.29756
+ pi_std                0.76127
+ pi_max_std            1.00000
+ contrastive_loss      0.62615
+ contrastive_pos_logit 0.34781
+ contrastive_neg_logit -0.30102
+ contrastive_mean      0.03364
+ contrastive_std       0.93544
+ grad_norm             1.42878
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00189
+ reward_loss           0.48444
+ value_loss            0.53060
+ total_loss            0.82980
+ bc_loss               0.22144
+ entropy_loss          -0.00116
+ pi_prior_loss         0.06814
+ pi_entropy            2.19987
+ pi_scaled_entropy     11.64948
+ pi_std                0.76990
+ pi_max_std            1.53339
+ contrastive_loss      0.62240
+ contrastive_pos_logit 0.28846
+ contrastive_neg_logit -0.35313
+ contrastive_mean      0.03339
+ contrastive_std       0.93395
+ grad_norm             1.30457
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00209
+ reward_loss           0.50824
+ value_loss            0.56685
+ total_loss            0.82649
+ bc_loss               0.21722
+ entropy_loss          -0.00122
+ pi_prior_loss         0.06888
+ pi_entropy            2.03730
+ pi_scaled_entropy     12.16272
+ pi_std                0.76473
+ pi_max_std            1.15473
+ contrastive_loss      0.60832
+ contrastive_pos_logit 0.51716
+ contrastive_neg_logit -0.27262
+ contrastive_mean      0.03841
+ contrastive_std       0.94565
+ grad_norm             1.08470
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00178
+ reward_loss           0.51498
+ value_loss            0.56682
+ total_loss            0.81710
+ bc_loss               0.21700
+ entropy_loss          -0.00125
+ pi_prior_loss         0.06898
+ pi_entropy            1.61752
+ pi_scaled_entropy     12.49030
+ pi_std                0.76385
+ pi_max_std            1.75492
+ contrastive_loss      0.60428
+ contrastive_pos_logit 0.44221
+ contrastive_neg_logit -0.33616
+ contrastive_mean      0.03368
+ contrastive_std       0.93008
+ grad_norm             1.06169
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00178
+ reward_loss           0.46298
+ value_loss            0.61950
+ total_loss            0.83754
+ bc_loss               0.22093
+ entropy_loss          -0.00136
+ pi_prior_loss         0.06943
+ pi_entropy            2.21030
+ pi_scaled_entropy     13.56577
+ pi_std                0.77595
+ pi_max_std            1.48110
+ contrastive_loss      0.62421
+ contrastive_pos_logit 0.36307
+ contrastive_neg_logit -0.29495
+ contrastive_mean      0.03489
+ contrastive_std       0.95441
+ grad_norm             1.11925
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00193
+ reward_loss           0.48780
+ value_loss            0.56859
+ total_loss            0.84233
+ bc_loss               0.21236
+ entropy_loss          -0.00107
+ pi_prior_loss         0.06650
+ pi_entropy            1.93426
+ pi_scaled_entropy     10.73226
+ pi_std                0.76327
+ pi_max_std            1.45223
+ contrastive_loss      0.63163
+ contrastive_pos_logit 0.42430
+ contrastive_neg_logit -0.23197
+ contrastive_mean      0.03725
+ contrastive_std       0.95625
+ grad_norm             1.07309
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00180
+ reward_loss           0.48706
+ value_loss            0.57084
+ total_loss            0.84509
+ bc_loss               0.19670
+ entropy_loss          -0.00118
+ pi_prior_loss         0.05999
+ pi_entropy            1.83125
+ pi_scaled_entropy     11.77449
+ pi_std                0.75785
+ pi_max_std            1.59319
+ contrastive_loss      0.64322
+ contrastive_pos_logit 0.22798
+ contrastive_neg_logit -0.26173
+ contrastive_mean      0.03308
+ contrastive_std       0.94541
+ grad_norm             1.08960
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00184
+ reward_loss           0.48619
+ value_loss            0.57111
+ total_loss            0.83819
+ bc_loss               0.20929
+ entropy_loss          -0.00098
+ pi_prior_loss         0.06715
+ pi_entropy            1.79620
+ pi_scaled_entropy     9.79338
+ pi_std                0.76337
+ pi_max_std            1.19188
+ contrastive_loss      0.62843
+ contrastive_pos_logit 0.30796
+ contrastive_neg_logit -0.25265
+ contrastive_mean      0.03258
+ contrastive_std       0.95640
+ grad_norm             1.26344
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00192
+ reward_loss           0.44963
+ value_loss            0.58371
+ total_loss            0.82882
+ bc_loss               0.21238
+ entropy_loss          -0.00109
+ pi_prior_loss         0.06634
+ pi_entropy            2.05001
+ pi_scaled_entropy     10.91898
+ pi_std                0.76084
+ pi_max_std            1.51610
+ contrastive_loss      0.62072
+ contrastive_pos_logit 0.37664
+ contrastive_neg_logit -0.34430
+ contrastive_mean      0.03316
+ contrastive_std       0.96358
+ grad_norm             1.12115
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00182
+ reward_loss           0.45907
+ value_loss            0.56105
+ total_loss            0.82716
+ bc_loss               0.21606
+ entropy_loss          -0.00126
+ pi_prior_loss         0.06958
+ pi_entropy            1.61315
+ pi_scaled_entropy     12.56694
+ pi_std                0.75959
+ pi_max_std            1.37671
+ contrastive_loss      0.61920
+ contrastive_pos_logit 0.37899
+ contrastive_neg_logit -0.33232
+ contrastive_mean      0.03530
+ contrastive_std       0.96744
+ grad_norm             1.19130
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00180
+ reward_loss           0.49088
+ value_loss            0.56396
+ total_loss            0.84985
+ bc_loss               0.22241
+ entropy_loss          -0.00101
+ pi_prior_loss         0.07027
+ pi_entropy            1.90320
+ pi_scaled_entropy     10.05456
+ pi_std                0.76784
+ pi_max_std            1.18648
+ contrastive_loss      0.63800
+ contrastive_pos_logit 0.25942
+ contrastive_neg_logit -0.30772
+ contrastive_mean      0.03415
+ contrastive_std       0.97110
+ grad_norm             1.11897
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00204
+ reward_loss           0.44169
+ value_loss            0.54891
+ total_loss            0.84611
+ bc_loss               0.23387
+ entropy_loss          -0.00087
+ pi_prior_loss         0.07360
+ pi_entropy            2.26853
+ pi_scaled_entropy     8.65495
+ pi_std                0.76387
+ pi_max_std            1.38145
+ contrastive_loss      0.63258
+ contrastive_pos_logit 0.30647
+ contrastive_neg_logit -0.27218
+ contrastive_mean      0.03378
+ contrastive_std       0.97436
+ grad_norm             0.99350
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00174
+ reward_loss           0.44890
+ value_loss            0.55717
+ total_loss            0.81921
+ bc_loss               0.19542
+ entropy_loss          -0.00100
+ pi_prior_loss         0.05995
+ pi_entropy            1.80315
+ pi_scaled_entropy     10.00733
+ pi_std                0.75814
+ pi_max_std            1.17022
+ contrastive_loss      0.62385
+ contrastive_pos_logit 0.44102
+ contrastive_neg_logit -0.22147
+ contrastive_mean      0.03286
+ contrastive_std       0.98168
+ grad_norm             1.07398
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00172
+ reward_loss           0.46107
+ value_loss            0.47388
+ total_loss            0.82764
+ bc_loss               0.19297
+ entropy_loss          -0.00140
+ pi_prior_loss         0.06003
+ pi_entropy            1.97732
+ pi_scaled_entropy     13.95624
+ pi_std                0.76764
+ pi_max_std            1.58140
+ contrastive_loss      0.63963
+ contrastive_pos_logit 0.30612
+ contrastive_neg_logit -0.27955
+ contrastive_mean      0.03544
+ contrastive_std       0.99206
+ grad_norm             0.98440
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00207
+ reward_loss           0.51147
+ value_loss            0.61787
+ total_loss            0.84680
+ bc_loss               0.21224
+ entropy_loss          -0.00122
+ pi_prior_loss         0.06839
+ pi_entropy            2.42979
+ pi_scaled_entropy     12.23696
+ pi_std                0.77032
+ pi_max_std            1.42016
+ contrastive_loss      0.62400
+ contrastive_pos_logit 0.41256
+ contrastive_neg_logit -0.20732
+ contrastive_mean      0.03589
+ contrastive_std       0.99547
+ grad_norm             0.88115
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00184
+ reward_loss           0.51034
+ value_loss            0.64624
+ total_loss            0.83893
+ bc_loss               0.19248
+ entropy_loss          -0.00113
+ pi_prior_loss         0.06016
+ pi_entropy            1.67299
+ pi_scaled_entropy     11.31397
+ pi_std                0.76078
+ pi_max_std            1.58833
+ contrastive_loss      0.62629
+ contrastive_pos_logit 0.33628
+ contrastive_neg_logit -0.30474
+ contrastive_mean      0.03222
+ contrastive_std       0.98548
+ grad_norm             1.31592
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00202
+ reward_loss           0.48543
+ value_loss            0.55676
+ total_loss            0.82658
+ bc_loss               0.20813
+ entropy_loss          -0.00114
+ pi_prior_loss         0.06311
+ pi_entropy            1.28784
+ pi_scaled_entropy     11.35652
+ pi_std                0.75505
+ pi_max_std            1.79255
+ contrastive_loss      0.61893
+ contrastive_pos_logit 0.30243
+ contrastive_neg_logit -0.40313
+ contrastive_mean      0.02980
+ contrastive_std       0.99635
+ grad_norm             1.08564
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00178
+ reward_loss           0.45595
+ value_loss            0.55784
+ total_loss            0.83084
+ bc_loss               0.21626
+ entropy_loss          -0.00127
+ pi_prior_loss         0.06749
+ pi_entropy            1.91412
+ pi_scaled_entropy     12.72243
+ pi_std                0.76863
+ pi_max_std            1.07008
+ contrastive_loss      0.62634
+ contrastive_pos_logit 0.37665
+ contrastive_neg_logit -0.33003
+ contrastive_mean      0.03157
+ contrastive_std       1.00571
+ grad_norm             0.98169
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00157
+ reward_loss           0.47415
+ value_loss            0.56397
+ total_loss            0.83035
+ bc_loss               0.20022
+ entropy_loss          -0.00121
+ pi_prior_loss         0.06207
+ pi_entropy            1.83645
+ pi_scaled_entropy     12.11595
+ pi_std                0.76561
+ pi_max_std            1.09832
+ contrastive_loss      0.63308
+ contrastive_pos_logit 0.31519
+ contrastive_neg_logit -0.31345
+ contrastive_mean      0.03274
+ contrastive_std       1.00609
+ grad_norm             1.01128
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00175
+ reward_loss           0.44619
+ value_loss            0.52502
+ total_loss            0.81798
+ bc_loss               0.19852
+ entropy_loss          -0.00151
+ pi_prior_loss         0.06061
+ pi_entropy            2.03370
+ pi_scaled_entropy     15.06222
+ pi_std                0.77284
+ pi_max_std            1.90922
+ contrastive_loss      0.62519
+ contrastive_pos_logit 0.34300
+ contrastive_neg_logit -0.39344
+ contrastive_mean      0.03262
+ contrastive_std       1.01829
+ grad_norm             1.52425
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00190
+ reward_loss           0.46445
+ value_loss            0.57271
+ total_loss            0.82252
+ bc_loss               0.18744
+ entropy_loss          -0.00136
+ pi_prior_loss         0.05968
+ pi_entropy            2.02639
+ pi_scaled_entropy     13.55574
+ pi_std                0.77643
+ pi_max_std            1.25032
+ contrastive_loss      0.62122
+ contrastive_pos_logit 0.40349
+ contrastive_neg_logit -0.32023
+ contrastive_mean      0.03089
+ contrastive_std       0.99638
+ grad_norm             1.09843
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00166
+ reward_loss           0.47068
+ value_loss            0.53200
+ total_loss            0.80650
+ bc_loss               0.18577
+ entropy_loss          -0.00135
+ pi_prior_loss         0.05806
+ pi_entropy            1.58484
+ pi_scaled_entropy     13.47677
+ pi_std                0.76522
+ pi_max_std            1.12116
+ contrastive_loss      0.61502
+ contrastive_pos_logit 0.37818
+ contrastive_neg_logit -0.44344
+ contrastive_mean      0.03111
+ contrastive_std       1.02627
+ grad_norm             0.98707
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00196
+ reward_loss           0.42062
+ value_loss            0.56692
+ total_loss            0.82177
+ bc_loss               0.20526
+ entropy_loss          -0.00139
+ pi_prior_loss         0.06268
+ pi_entropy            2.38136
+ pi_scaled_entropy     13.87939
+ pi_std                0.77896
+ pi_max_std            1.73011
+ contrastive_loss      0.62118
+ contrastive_pos_logit 0.37449
+ contrastive_neg_logit -0.35895
+ contrastive_mean      0.03186
+ contrastive_std       1.01661
+ grad_norm             1.14766
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00180
+ reward_loss           0.43142
+ value_loss            0.55713
+ total_loss            0.81771
+ bc_loss               0.20674
+ entropy_loss          -0.00118
+ pi_prior_loss         0.06430
+ pi_entropy            2.47706
+ pi_scaled_entropy     11.77662
+ pi_std                0.76930
+ pi_max_std            2.04396
+ contrastive_loss      0.61865
+ contrastive_pos_logit 0.41767
+ contrastive_neg_logit -0.33589
+ contrastive_mean      0.03138
+ contrastive_std       1.01659
+ grad_norm             1.11658
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00180
+ reward_loss           0.43909
+ value_loss            0.51838
+ total_loss            0.82245
+ bc_loss               0.20604
+ entropy_loss          -0.00118
+ pi_prior_loss         0.06527
+ pi_entropy            1.61344
+ pi_scaled_entropy     11.81373
+ pi_std                0.76469
+ pi_max_std            1.62377
+ contrastive_loss      0.62536
+ contrastive_pos_logit 0.27371
+ contrastive_neg_logit -0.38160
+ contrastive_mean      0.03154
+ contrastive_std       1.02334
+ grad_norm             1.35422
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00165
+ reward_loss           0.42722
+ value_loss            0.54942
+ total_loss            0.79575
+ bc_loss               0.18971
+ entropy_loss          -0.00112
+ pi_prior_loss         0.05915
+ pi_entropy            2.50066
+ pi_scaled_entropy     11.15401
+ pi_std                0.77839
+ pi_max_std            1.02945
+ contrastive_loss      0.60596
+ contrastive_pos_logit 0.46050
+ contrastive_neg_logit -0.33440
+ contrastive_mean      0.02994
+ contrastive_std       1.02729
+ grad_norm             0.83265
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+[Rank 0] Set prior_coef to 10.0 after pretraining.
+Pretraining complete.
+[Rank 0] Pretrain end
+[Rank 0] Entering barrier: post_pretrain_updates
+[Rank 0] Exited barrier: post_pretrain_updates
+[Rank 0] Checkpoint save start: 0
+Saved checkpoint to /media/datasets/cheliu21/cxy_worldmodel/newt/soup_S_100M_work_dir/ckpt/0.pt (0.1s)
+[Rank 0] Checkpoint save end: 0
+[Rank 0] Entering barrier: post_pretrain_checkpoint
+[Rank 0] Exited barrier: post_pretrain_checkpoint
+Training agent for 100,000,000 steps...
+[Rank 0] Train env reset start
+[Rank 0] Train env reset end; entering train loop
+[EnsembleBuffer] Buffer capacity: 10,000,000
+[EnsembleBuffer] Storage required: 5.85 GB
+[EnsembleBuffer] Using cuda:0 memory for storage.
+ train            E: 1,933                 I: 200,000               R: 128.877               S: 0.080                 T: 1:46:11
+ train            E: 3,866                 I: 400,000               R: 130.777               S: 0.061                 T: 1:47:18
+ train            E: 5,799                 I: 600,000               R: 118.233               S: 0.071                 T: 1:48:23
+ train            E: 7,732                 I: 800,000               R: 122.216               S: 0.076                 T: 1:49:27
+ train            E: 9,665                 I: 1,000,000             R: 133.473               S: 0.069                 T: 1:50:33
+ train            E: 11,598                I: 1,200,000             R: 82.598                S: 0.037                 T: 2:10:33
+ train            E: 13,531                I: 1,400,000             R: 129.421               S: 0.067                 T: 2:26:26
+ train            E: 15,464                I: 1,600,000             R: 177.013               S: 0.112                 T: 2:42:25
+ train            E: 17,397                I: 1,800,000             R: 165.174               S: 0.105                 T: 2:58:18
+ train            E: 19,330                I: 2,000,000             R: 169.386               S: 0.139                 T: 3:14:14
+ train            E: 21,263                I: 2,200,000             R: 172.571               S: 0.132                 T: 3:30:22
+ train            E: 23,196                I: 2,400,000             R: 175.754               S: 0.154                 T: 3:46:25
+ train            E: 25,129                I: 2,600,000             R: 193.269               S: 0.172                 T: 4:02:22
+ train            E: 27,062                I: 2,800,000             R: 192.745               S: 0.166                 T: 4:18:16
+ train            E: 28,995                I: 3,000,000             R: 222.397               S: 0.199                 T: 4:34:20
+ train            E: 30,928                I: 3,200,000             R: 227.890               S: 0.196                 T: 4:50:18
+ train            E: 32,861                I: 3,400,000             R: 223.845               S: 0.208                 T: 5:06:27
+ train            E: 34,794                I: 3,600,000             R: 227.811               S: 0.195                 T: 5:22:28
+ train            E: 36,727                I: 3,800,000             R: 236.767               S: 0.178                 T: 5:38:56
+ train            E: 38,660                I: 4,000,000             R: 245.584               S: 0.191                 T: 5:55:07
+ train            E: 40,593                I: 4,200,000             R: 251.005               S: 0.205                 T: 6:11:13
+ train            E: 42,526                I: 4,400,000             R: 229.957               S: 0.186                 T: 6:27:01
+ train            E: 44,459                I: 4,600,000             R: 231.339               S: 0.179                 T: 6:43:03
+ train            E: 46,392                I: 4,800,000             R: 236.672               S: 0.217                 T: 6:59:35
+ train            E: 48,325                I: 5,000,000             R: 235.485               S: 0.216                 T: 7:16:25
+[Rank 0] Eval start at step=5000000
+[Rank 0] Entering barrier: eval_complete score_source=analytic
+[Rank 0] Exited barrier: eval_complete score_source=analytic
+[Rank 0] Entering gather_object: eval_results score_source=analytic
+[rank0]:W0708 19:05:52.807000 301 site-packages/torch/distributed/distributed_c10d.py:3070] _object_to_tensor size: 36331 hash value: 15486816564237258985
+[rank0]:W0708 19:05:52.877000 301 site-packages/torch/distributed/distributed_c10d.py:3085] _tensor_to_object size: 38213 hash value: 1359154625463019240
+[rank0]:W0708 19:05:52.879000 301 site-packages/torch/distributed/distributed_c10d.py:3085] _tensor_to_object size: 38213 hash value: 1359154625463019240
+[rank0]:W0708 19:05:52.882000 301 site-packages/torch/distributed/distributed_c10d.py:3085] _tensor_to_object size: 38213 hash value: 1359154625463019240
+[rank0]:W0708 19:05:52.884000 301 site-packages/torch/distributed/distributed_c10d.py:3085] _tensor_to_object size: 38213 hash value: 1359154625463019240
+[Rank 0] Exited gather_object: eval_results score_source=analytic
+[Rank 0] Eval end at step=5000000
+Evaluated agent on 200 tasks:
+  walker-stand                      	S: 0.118
+  walker-walk                       	S: 0.038
+  walker-run                        	S: 0.085
+  cheetah-run                       	S: 0.039
+  reacher-easy                      	S: 0.000
+  reacher-hard                      	S: 0.000
+  acrobot-swingup                   	S: 0.010
+  pendulum-swingup                  	S: 0.118
+  cartpole-balance                  	S: 0.281
+  cartpole-balance-sparse           	S: 0.081
+  cartpole-swingup                  	S: 0.167
+  cartpole-swingup-sparse           	S: 0.000
+  cup-catch                         	S: 0.250
+  finger-spin                       	S: 0.078
+  finger-turn-easy                  	S: 0.000
+  finger-turn-hard                  	S: 0.000
+  fish-swim                         	S: 0.045
+  hopper-stand                      	S: 0.001
+  hopper-hop                        	S: 0.004
+  quadruped-walk                    	S: 0.224
+  quadruped-run                     	S: 0.075
+  walker-walk-backward              	S: 0.022
+  walker-run-backward               	S: 0.059
+  cheetah-run-backward              	S: 0.150
+  cheetah-run-front                 	S: 0.054
+  cheetah-run-back                  	S: 0.018
+  cheetah-jump                      	S: 0.375
+  hopper-hop-backward               	S: 0.000
+  reacher-three-easy                	S: 0.146
+  reacher-three-hard                	S: 0.000
+  cup-spin                          	S: 0.712
+  pendulum-spin                     	S: 0.461
+  jumper-jump                       	S: 0.041
+  spinner-spin                      	S: 0.021
+  spinner-spin-backward             	S: 0.023
+  spinner-jump                      	S: 0.030
+  giraffe-run                       	S: 0.162
+  mw-assembly                       	S: 0.000
+  mw-basketball                     	S: 0.000
+  mw-button-press-topdown           	S: 0.050
+  mw-button-press-topdown-wall      	S: 0.050
+  mw-button-press                   	S: 0.800
+  mw-button-press-wall              	S: 0.750
+  mw-coffee-button                  	S: 0.950
+  mw-coffee-pull                    	S: 0.100
+  mw-coffee-push                    	S: 0.100
+  mw-dial-turn                      	S: 0.000
+  mw-disassemble                    	S: 0.000
+  mw-door-open                      	S: 0.150
+  mw-door-close                     	S: 0.900
+  mw-drawer-close                   	S: 0.500
+  mw-drawer-open                    	S: 0.250
+  mw-faucet-open                    	S: 0.000
+  mw-faucet-close                   	S: 0.000
+  mw-hammer                         	S: 0.050
+  mw-handle-press-side              	S: 0.350
+  mw-handle-press                   	S: 0.400
+  mw-handle-pull-side               	S: 0.100
+  mw-handle-pull                    	S: 0.000
+  mw-lever-pull                     	S: 0.000
+  mw-peg-insert-side                	S: 0.000
+  mw-peg-unplug-side                	S: 0.000
+  mw-pick-out-of-hole               	S: 0.000
+  mw-pick-place                     	S: 0.050
+  mw-pick-place-wall                	S: 0.000
+  mw-plate-slide                    	S: 0.050
+  mw-plate-slide-side               	S: 0.100
+  mw-plate-slide-back               	S: 0.050
+  mw-plate-slide-back-side          	S: 0.250
+  mw-push-back                      	S: 0.050
+  mw-push                           	S: 0.000
+  mw-push-wall                      	S: 0.000
+  mw-reach                          	S: 0.000
+  mw-reach-wall                     	S: 0.100
+  mw-soccer                         	S: 0.300
+  mw-stick-push                     	S: 0.000
+  mw-stick-pull                     	S: 0.000
+  mw-sweep-into                     	S: 0.300
+  mw-sweep                          	S: 0.000
+  mw-window-open                    	S: 0.700
+  mw-window-close                   	S: 0.050
+  mw-bin-picking                    	S: 0.000
+  mw-box-close                      	S: 0.000
+  mw-door-lock                      	S: 0.450
+  mw-door-unlock                    	S: 0.250
+  mw-hand-insert                    	S: 0.000
+  ms-ant-walk                       	S: 0.000
+  ms-ant-run                        	S: 0.000
+  ms-cartpole-balance               	S: 0.197
+  ms-cartpole-swingup               	S: 0.187
+  ms-hopper-stand                   	S: 0.006
+  ms-hopper-hop                     	S: 0.002
+  ms-pick-cube                      	S: 0.000
+  ms-pick-cube-eepose               	S: 0.000
+  ms-pick-cube-so                   	S: 0.013
+  ms-poke-cube                      	S: 0.075
+  ms-push-cube                      	S: 0.300
+  ms-pull-cube                      	S: 0.050
+  ms-pull-cube-tool                 	S: 0.000
+  ms-stack-cube                     	S: 0.000
+  ms-place-sphere                   	S: 0.000
+  ms-lift-peg                       	S: 0.000
+  ms-pick-apple                     	S: 0.000
+  ms-pick-banana                    	S: 0.087
+  ms-pick-can                       	S: 0.000
+  ms-pick-hammer                    	S: 0.025
+  ms-pick-fork                      	S: 0.062
+  ms-pick-knife                     	S: 0.062
+  ms-pick-mug                       	S: 0.025
+  ms-pick-orange                    	S: 0.025
+  ms-pick-screwdriver               	S: 0.037
+  ms-pick-spoon                     	S: 0.037
+  ms-pick-tennis-ball               	S: 0.062
+  ms-pick-baseball                  	S: 0.050
+  ms-pick-cube-xarm6                	S: 0.000
+  ms-pick-sponge                    	S: 0.000
+  ms-anymal-reach                   	S: 0.000
+  ms-reach                          	S: 0.950
+  ms-reach-eepose                   	S: 0.812
+  ms-reach-xarm6                    	S: 0.762
+  ms-cartpole-balance-sparse        	S: 0.363
+  ms-cartpole-swingup-sparse        	S: 0.304
+  mujoco-ant                        	S: 0.000
+  mujoco-halfcheetah                	S: 0.000
+  mujoco-hopper                     	S: 0.000
+  mujoco-inverted-pendulum          	S: 0.002
+  mujoco-reacher                    	S: 0.765
+  mujoco-walker                     	S: 0.000
+  bipedal-walker-flat               	S: 0.000
+  bipedal-walker-uneven             	S: 0.000
+  bipedal-walker-rugged             	S: 0.000
+  bipedal-walker-hills              	S: 0.001
+  bipedal-walker-obstacles          	S: 0.000
+  lunarlander-land                  	S: 0.000
+  lunarlander-hover                 	S: 0.103
+  lunarlander-takeoff               	S: 0.191
+  rd-push-red                       	S: 0.400
+  rd-push-green                     	S: 0.700
+  rd-push-blue                      	S: 1.000
+  rd-open-slide                     	S: 0.200
+  rd-open-drawer                    	S: 0.000
+  rd-flat-block-in-bin              	S: 0.050
+  og-ant                            	S: 0.013
+  og-antball                        	S: 0.000
+  og-point-arena                    	S: 0.950
+  og-point-maze                     	S: 0.800
+  og-point-bottleneck               	S: 0.750
+  og-point-circle                   	S: 0.500
+  og-point-spiral                   	S: 0.450
+  og-ant-arena                      	S: 0.000
+  og-ant-maze                       	S: 0.000
+  og-ant-bottleneck                 	S: 0.000
+  og-ant-circle                     	S: 0.000
+  og-ant-spiral                     	S: 0.000
+  pygame-cowboy                     	S: 0.964
+  pygame-coinrun                    	S: 0.805
+  pygame-spaceship                  	S: 0.000
+  pygame-pong                       	S: 0.083
+  pygame-bird-attack                	S: 0.000
+  pygame-highway                    	S: 0.050
+  pygame-landing                    	S: 0.725
+  pygame-air-hockey                 	S: 0.000
+  pygame-rocket-collect             	S: 0.048
+  pygame-chase-evade                	S: 0.002
+  pygame-coconut-dodge              	S: 0.400
+  pygame-cartpole-balance           	S: 0.044
+  pygame-cartpole-swingup           	S: 0.000
+  pygame-cartpole-balance-sparse    	S: 0.016
+  pygame-cartpole-swingup-sparse    	S: 0.000
+  pygame-cartpole-tremor            	S: 0.082
+  pygame-point-maze-var1            	S: 0.000
+  pygame-point-maze-var2            	S: 0.000
+  pygame-point-maze-var3            	S: 0.000
+  atari-alien                       	S: 0.140
+  atari-assault                     	S: 0.021
+  atari-asterix                     	S: 0.030
+  atari-atlantis                    	S: 0.086
+  atari-bank-heist                  	S: 0.000
+  atari-battle-zone                 	S: 0.100
+  atari-beamrider                   	S: 0.044
+  atari-boxing                      	S: 0.403
+  atari-chopper-command             	S: 0.150
+  atari-crazy-climber               	S: 0.072
+  atari-double-dunk                 	S: 0.000
+  atari-gopher                      	S: 0.025
+  atari-ice-hockey                  	S: 0.350
+  atari-jamesbond                   	S: 0.000
+  atari-kangaroo                    	S: 0.020
+  atari-krull                       	S: 0.000
+  atari-ms-pacman                   	S: 0.195
+  atari-name-this-game              	S: 0.233
+  atari-phoenix                     	S: 0.110
+  atari-pong                        	S: 0.000
+  atari-road-runner                 	S: 0.000
+  atari-robotank                    	S: 0.060
+  atari-seaquest                    	S: 0.020
+  atari-space-invaders              	S: 0.084
+  atari-tutankham                   	S: 0.010
+  atari-upndown                     	S: 0.156
+  atari-yars-revenge                	S: 0.000
+dmcontrol                         	S: 0.077
+dmcontrol-ext                     	S: 0.142
+metaworld                         	S: 0.167
+maniskill                         	S: 0.125
+mujoco                            	S: 0.128
+box2d                             	S: 0.037
+robodesk                          	S: 0.392
+ogbench                           	S: 0.289
+pygame                            	S: 0.169
+atari                             	S: 0.086
+unweighted score                  	S: 0.145
+weighted score                    	S: 0.161
+ eval             E: 48,325                I: 5,000,000             R: 178.726               S: 0.145                 T: 7:21:30
+[Rank 0] Checkpoint save start: 5_000_000
+Saved checkpoint to /media/datasets/cheliu21/cxy_worldmodel/newt/soup_S_100M_work_dir/ckpt/5_000_000.pt (0.1s)
+[Rank 0] Checkpoint save end: 5_000_000
+ train            E: 50,258                I: 5,200,000             R: 246.475               S: 0.222                 T: 7:37:08
+ train            E: 52,191                I: 5,400,000             R: 241.769               S: 0.221                 T: 7:52:32
+ train            E: 54,124                I: 5,600,000             R: 251.324               S: 0.242                 T: 8:08:15
+ train            E: 56,057                I: 5,800,000             R: 244.048               S: 0.231                 T: 8:24:36
+ train            E: 57,990                I: 6,000,000             R: 243.871               S: 0.240                 T: 8:41:42
+ train            E: 59,923                I: 6,200,000             R: 242.704               S: 0.242                 T: 8:57:31
+ train            E: 61,856                I: 6,400,000             R: 255.024               S: 0.238                 T: 9:14:19
+ train            E: 63,789                I: 6,600,000             R: 268.194               S: 0.218                 T: 9:32:30
+ train            E: 65,722                I: 6,800,000             R: 252.502               S: 0.240                 T: 9:48:13
+ train            E: 67,655                I: 7,000,000             R: 253.174               S: 0.234                 T: 10:04:00
+ train            E: 69,588                I: 7,200,000             R: 243.322               S: 0.214                 T: 10:19:56
+ train            E: 71,521                I: 7,400,000             R: 239.445               S: 0.232                 T: 10:35:40
+ train            E: 73,454                I: 7,600,000             R: 246.966               S: 0.258                 T: 10:51:12
+ train            E: 75,387                I: 7,800,000             R: 265.834               S: 0.248                 T: 11:06:57
+ train            E: 77,320                I: 8,000,000             R: 255.116               S: 0.266                 T: 11:22:48
+ train            E: 79,253                I: 8,200,000             R: 264.979               S: 0.270                 T: 11:39:20

soup_S_100M_work_dir/wandb/run-20260708_114417-8pbrlu7w/files/requirements.txt

@@ -0,0 +1,174 @@
+prometheus_client==0.25.0
+nvidia-curand-cu12==10.3.9.90
+hf-xet==1.5.0
+ml_dtypes==0.5.4
+Farama-Notifications==0.0.6
+etils==1.14.0
+tensorstore==0.1.84
+AutoROM==0.6.1
+jax-cuda12-pjrt==0.7.1
+sentry-sdk==2.61.1
+opt_einsum==3.4.0
+imageio-ffmpeg==0.6.0
+wandb==0.22.1
+pytorch-kinematics==0.7.5
+nvidia-nvshmem-cu12==3.6.5
+tqdm==4.67.1
+Pygments==2.20.0
+humanize==4.15.0
+regex==2026.5.9
+safetensors==0.7.0
+numpy==1.26.4
+aiofiles==25.1.0
+kornia==0.8.1
+nvidia-nccl-cu12==2.27.3
+moviepy==1.0.3
+pandas==3.0.3
+antlr4-python3-runtime==4.9.3
+zipp==4.1.0
+transformers==4.56.2
+setuptools==69.5.1
+mpmath==1.3.0
+cloudpickle==3.1.2
+nvidia-cuda-runtime-cu12==12.8.90
+typeguard==4.5.2
+pexpect==4.9.0
+mdurl==0.1.2
+platformdirs==4.10.0
+dm-env==1.6
+pillow==12.2.0
+nvidia-ml-py==13.610.43
+sympy==1.14.0
+gpustat==1.1.1
+annotated-types==0.7.0
+nvidia-cuda-cccl-cu12==12.9.27
+nvidia-cusolver-cu12==11.7.3.90
+dacite==1.9.2
+proglog==0.1.12
+pydantic==2.13.4
+h5py==3.14.0
+matplotlib-inline==0.2.2
+imageio==2.37.0
+asttokens==3.0.1
+importlib_resources==7.1.0
+pyvers==0.1.0
+python-dateutil==2.9.0.post0
+msgpack==1.1.2
+treescope==0.1.10
+pygame==2.6.1
+scipy==1.17.1
+tabulate==0.10.0
+decorator==4.4.2
+hydra-submitit-launcher==1.2.0
+trimesh==4.12.2
+smmap==5.0.3
+torchvision==0.23.0+cu128
+nvidia-cusparselt-cu12==0.7.1
+kiwisolver==1.5.0
+transforms3d==0.4.2
+torch==2.8.0+cu128
+docstring_parser==0.18.0
+gitdb==4.0.12
+arm_pytorch_utilities==0.5.0
+pydantic_core==2.46.4
+Jinja2==3.1.6
+typing-inspection==0.4.2
+requests==2.34.2
+toppra==0.6.3
+mani-skill-nightly==2025.9.19.39
+termcolor==3.1.0
+gymnasium==0.29.1
+flax==0.12.0
+contourpy==1.3.3
+orjson==3.11.9
+prompt_toolkit==3.0.52
+ipython_pygments_lexers==1.1.1
+urllib3==2.7.0
+packaging==25.0
+nvidia-cudnn-cu12==9.10.2.21
+psutil==7.2.2
+markdown-it-py==4.2.0
+sapien==3.0.3
+MarkupSafe==3.0.3
+certifi==2026.5.20
+click==8.4.1
+parso==0.8.7
+pyperclip==1.11.0
+torchaudio==2.8.0+cu128
+nvidia-cusparse-cu12==12.5.8.93
+six==1.17.0
+AutoROM.accept-rom-license==0.6.1
+nvidia-cuda-cupti-cu12==12.8.90
+executing==2.2.1
+wcwidth==0.7.0
+lxml==6.1.1
+charset-normalizer==3.4.7
+pure_eval==0.2.3
+ale-py==0.10.0
+ipython==8.37.0
+mplib==0.1.1
+torchrl==0.10.0
+traitlets==5.15.1
+glfw==2.10.0
+nvidia-cublas-cu12==12.8.4.1
+ptyprocess==0.7.0
+jinxed==2.0.4
+labmaze==1.0.6
+jax-cuda12-plugin==0.7.1
+rich==15.0.0
+idna==3.18
+nvidia-nvtx-cu12==12.8.90
+nvidia-cufile-cu12==1.13.1.3
+triton==3.4.0
+jedi==0.20.0
+swig==4.4.1
+absl-py==2.4.0
+gym==0.26.2
+nvidia-cufft-cu12==11.3.3.83
+fsspec==2026.4.0
+pynvml==13.0.1
+jaxlib==0.7.1
+networkx==3.6.1
+mujoco==3.3.6
+nvidia-cuda-nvcc-cu12==12.9.86
+nvidia-cuda-nvrtc-cu12==12.8.93
+pyparsing==3.3.2
+metaworld==2.0.0
+box2d-py==2.3.5
+attrs==26.1.0
+protobuf==5.29.6
+fonttools==4.63.0
+simplejson==4.1.1
+submitit==1.5.3
+PyYAML==6.0.3
+blessed==1.44.0
+filelock==3.29.0
+jax==0.7.1
+omegaconf==2.3.0
+opencv-python==4.11.0.86
+nvidia-nvjitlink-cu12==12.8.93
+robodesk==1.0.0
+uvloop==0.22.1
+pytorch-seed==0.2.0
+tyro==1.0.13
+typing_extensions==4.15.0
+kornia_rs==0.1.14
+hydra-core==1.3.2
+importlib_metadata==9.0.0
+wheel==0.45.1
+ogbench==1.1.5
+fast_kinematics==0.2.2
+cycler==0.12.1
+dm-tree==0.1.10
+dm_control==1.0.34
+huggingface_hub==0.36.2
+optax==0.2.8
+tokenizers==0.22.2
+PyOpenGL==3.1.10
+tensordict==0.10.0
+orbax-checkpoint==0.12.0
+stack-data==0.6.3
+GitPython==3.1.50
+pip==25.2
+matplotlib==3.10.9
+wrapt==2.2.1

soup_S_100M_work_dir/wandb/run-20260708_114417-8pbrlu7w/files/wandb-metadata.json

@@ -0,0 +1,80 @@
+{
+  "os":  "Linux-5.15.0-72-generic-x86_64-with-glibc2.35",
+  "python":  "CPython 3.11.15",
+  "startedAt":  "2026-07-08T03:44:17.489266Z",
+  "args":  [
+    "task=soup",
+    "model_size=S",
+    "steps=100000000",
+    "demo_steps=100000",
+    "train_eval_freq=5000000",
+    "checkpoint_save_freq=5000000",
+    "diffusion_final_rerank=True",
+    "compile=True",
+    "diffusion_compile=True",
+    "multiproc=True",
+    "use_score_network=False",
+    "eval_at_start=False",
+    "env_mode=async",
+    "planner_type=diffusion",
+    "exp_name=soup_S_100M",
+    "seed=5",
+    "work_dir=/media/datasets/cheliu21/cxy_worldmodel/newt/soup_S_100M_work_dir"
+  ],
+  "program":  "/media/damoxing/che-liu-fileset/cxy_worldmodel/newt/tdmpc2/train.py",
+  "codePath":  "tdmpc2/train.py",
+  "codePathLocal":  "train.py",
+  "git":  {
+    "remote":  "git@github.com:Wenxuan52/newt.git",
+    "commit":  "0c15f2e94dcb5661c338ee4d4a9852c1f74447a2"
+  },
+  "email":  "wenxuan.yuan@qq.com",
+  "root":  "/media/datasets/cheliu21/cxy_worldmodel/newt/soup_S_100M_work_dir",
+  "host":  "job-riatxnkdjtdz-master-0",
+  "executable":  "/opt/conda/envs/newt-jax/bin/python",
+  "cpu_count":  64,
+  "cpu_count_logical":  128,
+  "gpu":  "NVIDIA A800-SXM4-80GB",
+  "gpu_count":  4,
+  "disk":  {
+    "/":  {
+      "total":  "3779301933056",
+      "used":  "1195615617024"
+    }
+  },
+  "memory":  {
+    "total":  "2162270277632"
+  },
+  "gpu_nvidia":  [
+    {
+      "name":  "NVIDIA A800-SXM4-80GB",
+      "memoryTotal":  "85899345920",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-0946b577-dd86-9fdd-b845-b830f151a4a4"
+    },
+    {
+      "name":  "NVIDIA A800-SXM4-80GB",
+      "memoryTotal":  "85899345920",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-4b859b9b-1c27-c84b-af46-654d6fdb9330"
+    },
+    {
+      "name":  "NVIDIA A800-SXM4-80GB",
+      "memoryTotal":  "85899345920",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-3acf3f48-50e7-75e8-1345-c53abab4d192"
+    },
+    {
+      "name":  "NVIDIA A800-SXM4-80GB",
+      "memoryTotal":  "85899345920",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-26641a8e-19d0-c39c-c6cc-a6ceede2ec23"
+    }
+  ],
+  "cudaVersion":  "12.4",
+  "writerId":  "vhekgjh5p6c19gv35r7qqqx8dh5ws77e"
+}

soup_S_100M_work_dir/wandb/run-20260708_114417-8pbrlu7w/logs/debug-internal.log

@@ -0,0 +1,15 @@
+{"time":"2026-07-08T11:44:17.762005454+08:00","level":"INFO","msg":"stream: starting","core version":"0.22.1"}
+{"time":"2026-07-08T11:44:18.293434205+08:00","level":"INFO","msg":"stream: created new stream","id":"8pbrlu7w"}
+{"time":"2026-07-08T11:44:18.293488394+08:00","level":"INFO","msg":"handler: started","stream_id":"8pbrlu7w"}
+{"time":"2026-07-08T11:44:18.293951428+08:00","level":"INFO","msg":"stream: started","id":"8pbrlu7w"}
+{"time":"2026-07-08T11:44:18.293964726+08:00","level":"INFO","msg":"writer: started","stream_id":"8pbrlu7w"}
+{"time":"2026-07-08T11:44:18.293965625+08:00","level":"INFO","msg":"sender: started","stream_id":"8pbrlu7w"}
+{"time":"2026-07-08T12:03:07.15738093+08:00","level":"INFO","msg":"flowcontrol: backed up, offloading to disk","recordNumber":8374}
+{"time":"2026-07-08T12:03:07.363887642+08:00","level":"INFO","msg":"flowcontrol: unblocked","totalOffloaded":13}
+{"time":"2026-07-08T12:49:52.066736366+08:00","level":"INFO","msg":"flowcontrol: backed up, offloading to disk","recordNumber":29629}
+{"time":"2026-07-08T12:49:52.330148908+08:00","level":"INFO","msg":"flowcontrol: unblocked","totalOffloaded":13}
+{"time":"2026-07-08T14:26:13.723243203+08:00","level":"INFO","msg":"api: retrying HTTP error","status":429,"url":"https://api.wandb.ai/files/wenxuan-yuan-imperial-college-london/newt/8pbrlu7w/file_stream","body":"{\"error\":\"rate limit exceeded: per_run limit on filestream requests\"}"}
+{"time":"2026-07-08T14:26:22.571413986+08:00","level":"INFO","msg":"api: retrying HTTP error","status":429,"url":"https://api.wandb.ai/files/wenxuan-yuan-imperial-college-london/newt/8pbrlu7w/file_stream","body":"{\"error\":\"rate limit exceeded: per_run limit on filestream requests\"}"}
+{"time":"2026-07-08T14:26:33.278724117+08:00","level":"INFO","msg":"api: retrying HTTP error","status":429,"url":"https://api.wandb.ai/files/wenxuan-yuan-imperial-college-london/newt/8pbrlu7w/file_stream","body":"{\"error\":\"rate limit exceeded: per_run limit on filestream requests\"}"}
+{"time":"2026-07-08T14:26:43.697230612+08:00","level":"INFO","msg":"api: retrying HTTP error","status":429,"url":"https://api.wandb.ai/files/wenxuan-yuan-imperial-college-london/newt/8pbrlu7w/file_stream","body":"{\"error\":\"rate limit exceeded: per_run limit on filestream requests\"}"}
+{"time":"2026-07-08T14:27:00.624505338+08:00","level":"INFO","msg":"api: retrying HTTP error","status":429,"url":"https://api.wandb.ai/files/wenxuan-yuan-imperial-college-london/newt/8pbrlu7w/file_stream","body":"{\"error\":\"rate limit exceeded: per_run limit on filestream requests\"}"}

soup_S_100M_work_dir/wandb/run-20260708_114417-8pbrlu7w/run-8pbrlu7w.wandb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fdfa7bcc28ffa2c1a0515eb392affd15eed70d5316789ae940f493dd0b85ee7
+size 14188544

soup_S_100M_work_dir/wandb/run-20260710_081650-8qh9ccql/files/output.log

@@ -0,0 +1,1208 @@
+Logs will be synced with wandb.
+Architecture: DDPWrapper(
+  (_module): DistributedDataParallel(
+    (module): Newt World Model
+    Encoder (132,608): ModuleDict(
+      (state): Sequential(
+        (0): NormedLinear(in_features=640, out_features=128, bias=True, act=Mish)
+        (1): NormedLinear(in_features=128, out_features=384, bias=True, act=SimNorm)
+      )
+    )
+    Dynamics (400,000): Sequential(
+      (0): NormedLinear(in_features=912, out_features=256, bias=True, act=Mish)
+      (1): NormedLinear(in_features=256, out_features=256, bias=True, act=Mish)
+      (2): NormedLinear(in_features=256, out_features=384, bias=True, act=SimNorm)
+    )
+    Reward (326,501): Sequential(
+      (0): NormedLinear(in_features=912, out_features=256, bias=True, act=Mish)
+      (1): NormedLinear(in_features=256, out_features=256, bias=True, act=Mish)
+      (2): Linear(in_features=256, out_features=101, bias=True)
+    )
+    Contrastive F (300,801): Sequential(
+      (0): NormedLinear(in_features=912, out_features=256, bias=True, act=Mish)
+      (1): NormedLinear(in_features=256, out_features=256, bias=True, act=Mish)
+      (2): Linear(in_features=256, out_features=1, bias=True)
+    )
+    Policy prior (304,672): Sequential(
+      (0): NormedLinear(in_features=896, out_features=256, bias=True, act=Mish)
+      (1): NormedLinear(in_features=256, out_features=256, bias=True, act=Mish)
+      (2): Linear(in_features=256, out_features=32, bias=True)
+    )
+    Q-functions (979,503): QEnsemble(
+      (_Qs): ModuleList(
+        (0-2): 3 x Sequential(
+          (0): NormedLinear(in_features=912, out_features=256, bias=True, act=Mish)
+          (1): NormedLinear(in_features=256, out_features=256, bias=True, act=Mish)
+          (2): Linear(in_features=256, out_features=101, bias=True)
+        )
+      )
+    )
+    Learnable parameters: 2,444,085
+  )
+)
+Update frequency: 200,000
+Episodes per update frequency: 1,933
+No checkpoint found, training from scratch.
+[Rank 0] Pretrain start
+Pretraining agent on demonstrations...
+[Rank 0] prior_coef is 10.0, setting to 1.0 for pretraining.
+Pretraining:   0%|                                                              | 1/100000 [00:27<768:20:37, 27.66s/it][rank0]:V0710 08:17:23.132000 42540 site-packages/torch/_dynamo/guards.py:3508] [0/1] [__recompiles] Recompiling function _loss_fn in /media/damoxing/che-liu-fileset/cxy_worldmodel/newt/tdmpc2/tdmpc2.py:488
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.02664
+ reward_loss           4.30745
+ value_loss            4.30745
+ total_loss            2.26892
+ bc_loss               0.57926
+ entropy_loss          0.00212
+ pi_prior_loss         0.18152
+ pi_entropy            -1.33223
+ pi_scaled_entropy     -21.23870
+ pi_std                0.75908
+ pi_max_std            1.00000
+ contrastive_loss      0.69315
+ contrastive_pos_logit 0.00000
+ contrastive_neg_logit 0.00000
+ contrastive_mean      0.00000
+ contrastive_std       0.99000
+ grad_norm             3.19082
+ lr_enc                0.00000
+ lr                    0.00000
+ lr_pi                 0.00000
+------------------------------
+[rank0]:V0710 08:17:23.132000 42540 site-packages/torch/_dynamo/guards.py:3508] [0/1] [__recompiles]     triggered by the following guard failure(s):
+[rank0]:V0710 08:17:23.132000 42540 site-packages/torch/_dynamo/guards.py:3508] [0/1] [__recompiles]     - 0/0: len(G['__import_tensordict_dot_utils']._TENSORCLASS_MEMO) != 15
+Pretraining:   4%|██▌                                                            | 3998/100000 [02:48<53:19, 30.01it/s]
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00259
+ reward_loss           0.78169
+ value_loss            0.71815
+ total_loss            0.93576
+ bc_loss               0.29717
+ entropy_loss          -0.00054
+ pi_prior_loss         0.09564
+ pi_entropy            2.23338
+ pi_scaled_entropy     5.40488
+ pi_std                0.76386
+ pi_max_std            1.00000
+ contrastive_loss      0.63842
+ contrastive_pos_logit 0.22763
+ contrastive_neg_logit -0.26581
+ contrastive_mean      0.00385
+ contrastive_std       0.70800
+ grad_norm             2.74639
+ lr_enc                0.00004
+ lr                    0.00012
+ lr_pi                 0.00012
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00227
+ reward_loss           0.59577
+ value_loss            0.63975
+ total_loss            0.89530
+ bc_loss               0.25575
+ entropy_loss          -0.00088
+ pi_prior_loss         0.08167
+ pi_entropy            2.34466
+ pi_scaled_entropy     8.81103
+ pi_std                0.76259
+ pi_max_std            1.00000
+ contrastive_loss      0.64476
+ contrastive_pos_logit 0.23244
+ contrastive_neg_logit -0.20899
+ contrastive_mean      0.02887
+ contrastive_std       0.74820
+ grad_norm             2.27961
+ lr_enc                0.00007
+ lr                    0.00024
+ lr_pi                 0.00024
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00184
+ reward_loss           0.53415
+ value_loss            0.50791
+ total_loss            0.86666
+ bc_loss               0.24264
+ entropy_loss          -0.00108
+ pi_prior_loss         0.07707
+ pi_entropy            2.50180
+ pi_scaled_entropy     10.80242
+ pi_std                0.76514
+ pi_max_std            1.00000
+ contrastive_loss      0.64853
+ contrastive_pos_logit 0.25223
+ contrastive_neg_logit -0.15956
+ contrastive_mean      0.02954
+ contrastive_std       0.74681
+ grad_norm             1.89157
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00164
+ reward_loss           0.50293
+ value_loss            0.51031
+ total_loss            0.85603
+ bc_loss               0.22221
+ entropy_loss          -0.00095
+ pi_prior_loss         0.07063
+ pi_entropy            2.39860
+ pi_scaled_entropy     9.47708
+ pi_std                0.76815
+ pi_max_std            1.00000
+ contrastive_loss      0.65121
+ contrastive_pos_logit 0.23474
+ contrastive_neg_logit -0.16281
+ contrastive_mean      0.02996
+ contrastive_std       0.76961
+ grad_norm             1.50229
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00191
+ reward_loss           0.53532
+ value_loss            0.51259
+ total_loss            0.84877
+ bc_loss               0.24007
+ entropy_loss          -0.00076
+ pi_prior_loss         0.07555
+ pi_entropy            1.78326
+ pi_scaled_entropy     7.62204
+ pi_std                0.75820
+ pi_max_std            1.00000
+ contrastive_loss      0.63028
+ contrastive_pos_logit 0.21581
+ contrastive_neg_logit -0.29171
+ contrastive_mean      0.03120
+ contrastive_std       0.78897
+ grad_norm             1.83430
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00188
+ reward_loss           0.46828
+ value_loss            0.53909
+ total_loss            0.84177
+ bc_loss               0.24596
+ entropy_loss          -0.00112
+ pi_prior_loss         0.07821
+ pi_entropy            2.53198
+ pi_scaled_entropy     11.15504
+ pi_std                0.76593
+ pi_max_std            1.00000
+ contrastive_loss      0.62521
+ contrastive_pos_logit 0.31021
+ contrastive_neg_logit -0.25196
+ contrastive_mean      0.03210
+ contrastive_std       0.80436
+ grad_norm             1.39921
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00198
+ reward_loss           0.48752
+ value_loss            0.56500
+ total_loss            0.85788
+ bc_loss               0.25185
+ entropy_loss          -0.00119
+ pi_prior_loss         0.07946
+ pi_entropy            2.75077
+ pi_scaled_entropy     11.91470
+ pi_std                0.76531
+ pi_max_std            1.03704
+ contrastive_loss      0.63348
+ contrastive_pos_logit 0.44267
+ contrastive_neg_logit -0.11679
+ contrastive_mean      0.03368
+ contrastive_std       0.82618
+ grad_norm             2.42877
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00186
+ reward_loss           0.45139
+ value_loss            0.50598
+ total_loss            0.83678
+ bc_loss               0.22312
+ entropy_loss          -0.00112
+ pi_prior_loss         0.06902
+ pi_entropy            2.12574
+ pi_scaled_entropy     11.17131
+ pi_std                0.76657
+ pi_max_std            1.00000
+ contrastive_loss      0.63483
+ contrastive_pos_logit 0.31112
+ contrastive_neg_logit -0.23578
+ contrastive_mean      0.03405
+ contrastive_std       0.83697
+ grad_norm             1.67646
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00198
+ reward_loss           0.50595
+ value_loss            0.55908
+ total_loss            0.85246
+ bc_loss               0.21695
+ entropy_loss          -0.00065
+ pi_prior_loss         0.06949
+ pi_entropy            1.68768
+ pi_scaled_entropy     6.45575
+ pi_std                0.75257
+ pi_max_std            1.00000
+ contrastive_loss      0.63694
+ contrastive_pos_logit 0.38407
+ contrastive_neg_logit -0.21594
+ contrastive_mean      0.03370
+ contrastive_std       0.83950
+ grad_norm             1.48361
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00191
+ reward_loss           0.53109
+ value_loss            0.60807
+ total_loss            0.84234
+ bc_loss               0.22194
+ entropy_loss          -0.00113
+ pi_prior_loss         0.06921
+ pi_entropy            2.30159
+ pi_scaled_entropy     11.30224
+ pi_std                0.76479
+ pi_max_std            1.26189
+ contrastive_loss      0.62102
+ contrastive_pos_logit 0.31549
+ contrastive_neg_logit -0.27424
+ contrastive_mean      0.03165
+ contrastive_std       0.84478
+ grad_norm             1.41632
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00175
+ reward_loss           0.50058
+ value_loss            0.61537
+ total_loss            0.84267
+ bc_loss               0.21912
+ entropy_loss          -0.00103
+ pi_prior_loss         0.06852
+ pi_entropy            1.90958
+ pi_scaled_entropy     10.29218
+ pi_std                0.76145
+ pi_max_std            1.56850
+ contrastive_loss      0.62759
+ contrastive_pos_logit 0.31835
+ contrastive_neg_logit -0.26594
+ contrastive_mean      0.03415
+ contrastive_std       0.86751
+ grad_norm             1.28091
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00182
+ reward_loss           0.48323
+ value_loss            0.53940
+ total_loss            0.83241
+ bc_loss               0.22759
+ entropy_loss          -0.00067
+ pi_prior_loss         0.07148
+ pi_entropy            1.28435
+ pi_scaled_entropy     6.66865
+ pi_std                0.76254
+ pi_max_std            1.18614
+ contrastive_loss      0.62219
+ contrastive_pos_logit 0.39614
+ contrastive_neg_logit -0.26147
+ contrastive_mean      0.03797
+ contrastive_std       0.92562
+ grad_norm             1.52791
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00171
+ reward_loss           0.51020
+ value_loss            0.54567
+ total_loss            0.81399
+ bc_loss               0.22030
+ entropy_loss          -0.00073
+ pi_prior_loss         0.06988
+ pi_entropy            0.98537
+ pi_scaled_entropy     7.28556
+ pi_std                0.76327
+ pi_max_std            1.82948
+ contrastive_loss      0.60429
+ contrastive_pos_logit 0.39282
+ contrastive_neg_logit -0.35536
+ contrastive_mean      0.03892
+ contrastive_std       0.95558
+ grad_norm             1.52608
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00178
+ reward_loss           0.45863
+ value_loss            0.52432
+ total_loss            0.83585
+ bc_loss               0.21254
+ entropy_loss          -0.00106
+ pi_prior_loss         0.06694
+ pi_entropy            2.27777
+ pi_scaled_entropy     10.55574
+ pi_std                0.76773
+ pi_max_std            1.14628
+ contrastive_loss      0.63500
+ contrastive_pos_logit 0.26335
+ contrastive_neg_logit -0.28001
+ contrastive_mean      0.03629
+ contrastive_std       0.91625
+ grad_norm             1.23293
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00193
+ reward_loss           0.46337
+ value_loss            0.56523
+ total_loss            0.83622
+ bc_loss               0.21775
+ entropy_loss          -0.00116
+ pi_prior_loss         0.06936
+ pi_entropy            2.26948
+ pi_scaled_entropy     11.64528
+ pi_std                0.76667
+ pi_max_std            1.31803
+ contrastive_loss      0.62540
+ contrastive_pos_logit 0.27070
+ contrastive_neg_logit -0.33924
+ contrastive_mean      0.03497
+ contrastive_std       0.91106
+ grad_norm             1.36921
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00178
+ reward_loss           0.42939
+ value_loss            0.53988
+ total_loss            0.83141
+ bc_loss               0.21653
+ entropy_loss          -0.00114
+ pi_prior_loss         0.06808
+ pi_entropy            1.83388
+ pi_scaled_entropy     11.35065
+ pi_std                0.77175
+ pi_max_std            1.44726
+ contrastive_loss      0.63078
+ contrastive_pos_logit 0.27010
+ contrastive_neg_logit -0.30720
+ contrastive_mean      0.03489
+ contrastive_std       0.92341
+ grad_norm             1.27465
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00191
+ reward_loss           0.44941
+ value_loss            0.56154
+ total_loss            0.83179
+ bc_loss               0.22251
+ entropy_loss          -0.00110
+ pi_prior_loss         0.06905
+ pi_entropy            1.74174
+ pi_scaled_entropy     10.97754
+ pi_std                0.76304
+ pi_max_std            1.22696
+ contrastive_loss      0.62354
+ contrastive_pos_logit 0.41751
+ contrastive_neg_logit -0.30831
+ contrastive_mean      0.03613
+ contrastive_std       0.93376
+ grad_norm             1.75312
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00172
+ reward_loss           0.48089
+ value_loss            0.53814
+ total_loss            0.81969
+ bc_loss               0.20860
+ entropy_loss          -0.00116
+ pi_prior_loss         0.06647
+ pi_entropy            1.94584
+ pi_scaled_entropy     11.60215
+ pi_std                0.76640
+ pi_max_std            1.41345
+ contrastive_loss      0.61695
+ contrastive_pos_logit 0.40548
+ contrastive_neg_logit -0.30869
+ contrastive_mean      0.03511
+ contrastive_std       0.93682
+ grad_norm             1.17280
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00190
+ reward_loss           0.45477
+ value_loss            0.53899
+ total_loss            0.82628
+ bc_loss               0.22070
+ entropy_loss          -0.00128
+ pi_prior_loss         0.06829
+ pi_entropy            1.46511
+ pi_scaled_entropy     12.75576
+ pi_std                0.75822
+ pi_max_std            1.29616
+ contrastive_loss      0.62062
+ contrastive_pos_logit 0.38516
+ contrastive_neg_logit -0.31150
+ contrastive_mean      0.03475
+ contrastive_std       0.94747
+ grad_norm             1.36236
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00179
+ reward_loss           0.48311
+ value_loss            0.53150
+ total_loss            0.83079
+ bc_loss               0.21670
+ entropy_loss          -0.00124
+ pi_prior_loss         0.06675
+ pi_entropy            1.80925
+ pi_scaled_entropy     12.44067
+ pi_std                0.76405
+ pi_max_std            1.35030
+ contrastive_loss      0.62682
+ contrastive_pos_logit 0.42226
+ contrastive_neg_logit -0.24624
+ contrastive_mean      0.03550
+ contrastive_std       0.95631
+ grad_norm             1.52002
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00165
+ reward_loss           0.45003
+ value_loss            0.52857
+ total_loss            0.80464
+ bc_loss               0.19276
+ entropy_loss          -0.00125
+ pi_prior_loss         0.06095
+ pi_entropy            1.78618
+ pi_scaled_entropy     12.54707
+ pi_std                0.76573
+ pi_max_std            1.59105
+ contrastive_loss      0.61285
+ contrastive_pos_logit 0.34110
+ contrastive_neg_logit -0.36968
+ contrastive_mean      0.03483
+ contrastive_std       0.96636
+ grad_norm             1.33766
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00181
+ reward_loss           0.47048
+ value_loss            0.55196
+ total_loss            0.82175
+ bc_loss               0.19155
+ entropy_loss          -0.00128
+ pi_prior_loss         0.05915
+ pi_entropy            1.70762
+ pi_scaled_entropy     12.84855
+ pi_std                0.75986
+ pi_max_std            1.57230
+ contrastive_loss      0.62415
+ contrastive_pos_logit 0.41114
+ contrastive_neg_logit -0.25414
+ contrastive_mean      0.03376
+ contrastive_std       0.97012
+ grad_norm             1.55513
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00157
+ reward_loss           0.47305
+ value_loss            0.54084
+ total_loss            0.81619
+ bc_loss               0.20741
+ entropy_loss          -0.00127
+ pi_prior_loss         0.06490
+ pi_entropy            1.96287
+ pi_scaled_entropy     12.72554
+ pi_std                0.77111
+ pi_max_std            1.48796
+ contrastive_loss      0.61855
+ contrastive_pos_logit 0.38222
+ contrastive_neg_logit -0.34818
+ contrastive_mean      0.03351
+ contrastive_std       0.99397
+ grad_norm             1.37827
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00163
+ reward_loss           0.47797
+ value_loss            0.54548
+ total_loss            0.80520
+ bc_loss               0.20688
+ entropy_loss          -0.00127
+ pi_prior_loss         0.06405
+ pi_entropy            1.79782
+ pi_scaled_entropy     12.72388
+ pi_std                0.76706
+ pi_max_std            2.36596
+ contrastive_loss      0.60618
+ contrastive_pos_logit 0.45484
+ contrastive_neg_logit -0.34109
+ contrastive_mean      0.03317
+ contrastive_std       0.98662
+ grad_norm             1.25539
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00175
+ reward_loss           0.45341
+ value_loss            0.49732
+ total_loss            0.82067
+ bc_loss               0.21590
+ entropy_loss          -0.00124
+ pi_prior_loss         0.06678
+ pi_entropy            1.53585
+ pi_scaled_entropy     12.39355
+ pi_std                0.75887
+ pi_max_std            1.47605
+ contrastive_loss      0.62378
+ contrastive_pos_logit 0.29674
+ contrastive_neg_logit -0.36380
+ contrastive_mean      0.03234
+ contrastive_std       0.99255
+ grad_norm             1.28433
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00169
+ reward_loss           0.46283
+ value_loss            0.53173
+ total_loss            0.81638
+ bc_loss               0.19953
+ entropy_loss          -0.00121
+ pi_prior_loss         0.06172
+ pi_entropy            1.42026
+ pi_scaled_entropy     12.06843
+ pi_std                0.76062
+ pi_max_std            1.23713
+ contrastive_loss      0.62133
+ contrastive_pos_logit 0.40979
+ contrastive_neg_logit -0.28568
+ contrastive_mean      0.03179
+ contrastive_std       1.00199
+ grad_norm             1.04073
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00183
+ reward_loss           0.42951
+ value_loss            0.55113
+ total_loss            0.81788
+ bc_loss               0.21517
+ entropy_loss          -0.00132
+ pi_prior_loss         0.06789
+ pi_entropy            1.99165
+ pi_scaled_entropy     13.21604
+ pi_std                0.76384
+ pi_max_std            1.12050
+ contrastive_loss      0.61534
+ contrastive_pos_logit 0.41846
+ contrastive_neg_logit -0.33600
+ contrastive_mean      0.03058
+ contrastive_std       1.00491
+ grad_norm             1.22041
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00214
+ reward_loss           0.48963
+ value_loss            0.77980
+ total_loss            0.86636
+ bc_loss               0.20966
+ entropy_loss          -0.00115
+ pi_prior_loss         0.06603
+ pi_entropy            1.61523
+ pi_scaled_entropy     11.47172
+ pi_std                0.76500
+ pi_max_std            1.09717
+ contrastive_loss      0.63065
+ contrastive_pos_logit 0.23008
+ contrastive_neg_logit -0.41728
+ contrastive_mean      0.03305
+ contrastive_std       1.01866
+ grad_norm             1.47189
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00174
+ reward_loss           0.46284
+ value_loss            0.52650
+ total_loss            0.81615
+ bc_loss               0.18371
+ entropy_loss          -0.00141
+ pi_prior_loss         0.05677
+ pi_entropy            1.89063
+ pi_scaled_entropy     14.09429
+ pi_std                0.76852
+ pi_max_std            1.41150
+ contrastive_loss      0.62560
+ contrastive_pos_logit 0.41475
+ contrastive_neg_logit -0.28545
+ contrastive_mean      0.03155
+ contrastive_std       1.02095
+ grad_norm             1.20462
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00180
+ reward_loss           0.46288
+ value_loss            0.57003
+ total_loss            0.81123
+ bc_loss               0.19053
+ entropy_loss          -0.00126
+ pi_prior_loss         0.05954
+ pi_entropy            1.86114
+ pi_scaled_entropy     12.58255
+ pi_std                0.76878
+ pi_max_std            1.67445
+ contrastive_loss      0.61239
+ contrastive_pos_logit 0.38485
+ contrastive_neg_logit -0.39374
+ contrastive_mean      0.03215
+ contrastive_std       1.03639
+ grad_norm             1.40942
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00192
+ reward_loss           0.46871
+ value_loss            0.57886
+ total_loss            0.81632
+ bc_loss               0.20533
+ entropy_loss          -0.00112
+ pi_prior_loss         0.06249
+ pi_entropy            1.55885
+ pi_scaled_entropy     11.24827
+ pi_std                0.76100
+ pi_max_std            1.33898
+ contrastive_loss      0.61072
+ contrastive_pos_logit 0.40711
+ contrastive_neg_logit -0.34028
+ contrastive_mean      0.03125
+ contrastive_std       1.03561
+ grad_norm             1.18667
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00173
+ reward_loss           0.45842
+ value_loss            0.51514
+ total_loss            0.81196
+ bc_loss               0.19475
+ entropy_loss          -0.00122
+ pi_prior_loss         0.06107
+ pi_entropy            1.99828
+ pi_scaled_entropy     12.21405
+ pi_std                0.76428
+ pi_max_std            1.31261
+ contrastive_loss      0.61899
+ contrastive_pos_logit 0.35448
+ contrastive_neg_logit -0.36824
+ contrastive_mean      0.03069
+ contrastive_std       1.04026
+ grad_norm             1.54843
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00175
+ reward_loss           0.43922
+ value_loss            0.49828
+ total_loss            0.80480
+ bc_loss               0.18810
+ entropy_loss          -0.00116
+ pi_prior_loss         0.05922
+ pi_entropy            1.72962
+ pi_scaled_entropy     11.59478
+ pi_std                0.76050
+ pi_max_std            1.55868
+ contrastive_loss      0.61691
+ contrastive_pos_logit 0.39979
+ contrastive_neg_logit -0.39879
+ contrastive_mean      0.03132
+ contrastive_std       1.04734
+ grad_norm             1.59321
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00184
+ reward_loss           0.47127
+ value_loss            0.58090
+ total_loss            0.82060
+ bc_loss               0.19944
+ entropy_loss          -0.00115
+ pi_prior_loss         0.06261
+ pi_entropy            1.84984
+ pi_scaled_entropy     11.50996
+ pi_std                0.76758
+ pi_max_std            2.14387
+ contrastive_loss      0.61593
+ contrastive_pos_logit 0.34325
+ contrastive_neg_logit -0.39324
+ contrastive_mean      0.02955
+ contrastive_std       1.05432
+ grad_norm             1.39971
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00168
+ reward_loss           0.44298
+ value_loss            0.54228
+ total_loss            0.80110
+ bc_loss               0.18208
+ entropy_loss          -0.00121
+ pi_prior_loss         0.05753
+ pi_entropy            1.96926
+ pi_scaled_entropy     12.13498
+ pi_std                0.76870
+ pi_max_std            1.33152
+ contrastive_loss      0.61151
+ contrastive_pos_logit 0.40478
+ contrastive_neg_logit -0.39514
+ contrastive_mean      0.03099
+ contrastive_std       1.06299
+ grad_norm             1.25464
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00179
+ reward_loss           0.42663
+ value_loss            0.58410
+ total_loss            0.81757
+ bc_loss               0.19562
+ entropy_loss          -0.00103
+ pi_prior_loss         0.06108
+ pi_entropy            1.62736
+ pi_scaled_entropy     10.26910
+ pi_std                0.76058
+ pi_max_std            1.75362
+ contrastive_loss      0.61959
+ contrastive_pos_logit 0.42459
+ contrastive_neg_logit -0.28966
+ contrastive_mean      0.02965
+ contrastive_std       1.08824
+ grad_norm             1.18906
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00183
+ reward_loss           0.43197
+ value_loss            0.54107
+ total_loss            0.80569
+ bc_loss               0.20177
+ entropy_loss          -0.00115
+ pi_prior_loss         0.06319
+ pi_entropy            1.82488
+ pi_scaled_entropy     11.48520
+ pi_std                0.76846
+ pi_max_std            1.73967
+ contrastive_loss      0.60860
+ contrastive_pos_logit 0.41046
+ contrastive_neg_logit -0.34530
+ contrastive_mean      0.02946
+ contrastive_std       1.07720
+ grad_norm             1.52161
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00173
+ reward_loss           0.46208
+ value_loss            0.58044
+ total_loss            0.82361
+ bc_loss               0.20675
+ entropy_loss          -0.00129
+ pi_prior_loss         0.06440
+ pi_entropy            1.74289
+ pi_scaled_entropy     12.92520
+ pi_std                0.76034
+ pi_max_std            1.71931
+ contrastive_loss      0.62041
+ contrastive_pos_logit 0.26840
+ contrastive_neg_logit -0.51616
+ contrastive_mean      0.02758
+ contrastive_std       1.09043
+ grad_norm             1.99145
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00160
+ reward_loss           0.43752
+ value_loss            0.51966
+ total_loss            0.80497
+ bc_loss               0.18736
+ entropy_loss          -0.00129
+ pi_prior_loss         0.05920
+ pi_entropy            1.97286
+ pi_scaled_entropy     12.92685
+ pi_std                0.77502
+ pi_max_std            2.00692
+ contrastive_loss      0.61813
+ contrastive_pos_logit 0.36490
+ contrastive_neg_logit -0.37364
+ contrastive_mean      0.02771
+ contrastive_std       1.10016
+ grad_norm             1.15560
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00169
+ reward_loss           0.44111
+ value_loss            0.52682
+ total_loss            0.78321
+ bc_loss               0.19211
+ entropy_loss          -0.00127
+ pi_prior_loss         0.06013
+ pi_entropy            1.44680
+ pi_scaled_entropy     12.74360
+ pi_std                0.76644
+ pi_max_std            2.14511
+ contrastive_loss      0.59256
+ contrastive_pos_logit 0.56919
+ contrastive_neg_logit -0.40108
+ contrastive_mean      0.03045
+ contrastive_std       1.09880
+ grad_norm             1.10512
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00181
+ reward_loss           0.44795
+ value_loss            0.57934
+ total_loss            0.81308
+ bc_loss               0.20173
+ entropy_loss          -0.00122
+ pi_prior_loss         0.06363
+ pi_entropy            1.75204
+ pi_scaled_entropy     12.24858
+ pi_std                0.76558
+ pi_max_std            1.72986
+ contrastive_loss      0.61053
+ contrastive_pos_logit 0.40706
+ contrastive_neg_logit -0.43737
+ contrastive_mean      0.02794
+ contrastive_std       1.11129
+ grad_norm             1.33966
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00170
+ reward_loss           0.45337
+ value_loss            0.52684
+ total_loss            0.80367
+ bc_loss               0.18781
+ entropy_loss          -0.00118
+ pi_prior_loss         0.05870
+ pi_entropy            2.11432
+ pi_scaled_entropy     11.78429
+ pi_std                0.76879
+ pi_max_std            1.33044
+ contrastive_loss      0.61299
+ contrastive_pos_logit 0.44720
+ contrastive_neg_logit -0.33875
+ contrastive_mean      0.02632
+ contrastive_std       1.10925
+ grad_norm             1.06123
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00189
+ reward_loss           0.45533
+ value_loss            0.55874
+ total_loss            0.80819
+ bc_loss               0.19541
+ entropy_loss          -0.00118
+ pi_prior_loss         0.06109
+ pi_entropy            1.76177
+ pi_scaled_entropy     11.76148
+ pi_std                0.76173
+ pi_max_std            1.33254
+ contrastive_loss      0.60782
+ contrastive_pos_logit 0.45600
+ contrastive_neg_logit -0.41991
+ contrastive_mean      0.02528
+ contrastive_std       1.12720
+ grad_norm             1.32462
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00166
+ reward_loss           0.44232
+ value_loss            0.58384
+ total_loss            0.81021
+ bc_loss               0.21284
+ entropy_loss          -0.00114
+ pi_prior_loss         0.06484
+ pi_entropy            2.07171
+ pi_scaled_entropy     11.38629
+ pi_std                0.76677
+ pi_max_std            1.32738
+ contrastive_loss      0.60959
+ contrastive_pos_logit 0.44568
+ contrastive_neg_logit -0.31854
+ contrastive_mean      0.02838
+ contrastive_std       1.12935
+ grad_norm             1.16828
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00171
+ reward_loss           0.42687
+ value_loss            0.51339
+ total_loss            0.80586
+ bc_loss               0.20802
+ entropy_loss          -0.00116
+ pi_prior_loss         0.06443
+ pi_entropy            2.03830
+ pi_scaled_entropy     11.57798
+ pi_std                0.76816
+ pi_max_std            2.10361
+ contrastive_loss      0.61316
+ contrastive_pos_logit 0.47734
+ contrastive_neg_logit -0.34071
+ contrastive_mean      0.02656
+ contrastive_std       1.12950
+ grad_norm             1.00684
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00189
+ reward_loss           0.43948
+ value_loss            0.55059
+ total_loss            0.78399
+ bc_loss               0.17744
+ entropy_loss          -0.00118
+ pi_prior_loss         0.05524
+ pi_entropy            1.07512
+ pi_scaled_entropy     11.76301
+ pi_std                0.75255
+ pi_max_std            1.68608
+ contrastive_loss      0.59202
+ contrastive_pos_logit 0.62881
+ contrastive_neg_logit -0.41368
+ contrastive_mean      0.02666
+ contrastive_std       1.13551
+ grad_norm             1.28072
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00158
+ reward_loss           0.46290
+ value_loss            0.53558
+ total_loss            0.80924
+ bc_loss               0.19987
+ entropy_loss          -0.00139
+ pi_prior_loss         0.06152
+ pi_entropy            2.19146
+ pi_scaled_entropy     13.89413
+ pi_std                0.77493
+ pi_max_std            2.14582
+ contrastive_loss      0.61620
+ contrastive_pos_logit 0.38271
+ contrastive_neg_logit -0.38618
+ contrastive_mean      0.02540
+ contrastive_std       1.14522
+ grad_norm             1.43177
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00159
+ reward_loss           0.45398
+ value_loss            0.55985
+ total_loss            0.79627
+ bc_loss               0.19555
+ entropy_loss          -0.00125
+ pi_prior_loss         0.06041
+ pi_entropy            1.66878
+ pi_scaled_entropy     12.46852
+ pi_std                0.76819
+ pi_max_std            2.32762
+ contrastive_loss      0.60274
+ contrastive_pos_logit 0.44672
+ contrastive_neg_logit -0.43363
+ contrastive_mean      0.02720
+ contrastive_std       1.16323
+ grad_norm             1.17770
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00174
+ reward_loss           0.43541
+ value_loss            0.55141
+ total_loss            0.79200
+ bc_loss               0.18089
+ entropy_loss          -0.00121
+ pi_prior_loss         0.05567
+ pi_entropy            1.58996
+ pi_scaled_entropy     12.14836
+ pi_std                0.76500
+ pi_max_std            1.63239
+ contrastive_loss      0.60286
+ contrastive_pos_logit 0.44940
+ contrastive_neg_logit -0.45131
+ contrastive_mean      0.02436
+ contrastive_std       1.16168
+ grad_norm             1.06489
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+[Rank 0] Set prior_coef to 10.0 after pretraining.
+Pretraining complete.
+[Rank 0] Pretrain end
+[Rank 0] Entering barrier: post_pretrain_updates
+[Rank 0] Training crashed with exception: RuntimeError('ProcessGroupWrapper: Monitored Barrier encountered error running collective: CollectiveFingerPrint(SequenceNumber=4OpType=BARRIER). Error: \n[/pytorch/third_party/gloo/gloo/transport/tcp/pair.cc:544] Connection closed by peer [172.17.254.122]:40678')
+Training interrupted

soup_S_100M_work_dir/wandb/run-20260710_081650-8qh9ccql/files/requirements.txt

@@ -0,0 +1,180 @@
+nvidia-cudnn-cu12==9.10.2.21
+pure_eval==0.2.3
+smmap==5.0.3
+GitPython==3.1.50
+nvidia-nvtx-cu12==12.8.90
+blessed==1.44.0
+gymnasium==0.29.1
+optax==0.2.8
+AutoROM==0.6.1
+pynvml==13.0.1
+importlib_metadata==9.0.0
+numpy==1.26.4
+rich==15.0.0
+nvidia-cusparselt-cu12==0.7.1
+pydantic==2.13.4
+mdurl==0.1.2
+box2d-py==2.3.5
+six==1.17.0
+setuptools==69.5.1
+stack-data==0.6.3
+networkx==3.6.1
+nvidia-cuda-cccl-cu12==12.9.27
+hydra-submitit-launcher==1.2.0
+prometheus_client==0.25.0
+omegaconf==2.3.0
+aiofiles==25.1.0
+tokenizers==0.22.2
+sympy==1.14.0
+termcolor==3.1.0
+markdown-it-py==4.2.0
+jinxed==2.0.4
+typing-inspection==0.4.2
+hydra-core==1.3.2
+nvidia-cusolver-cu12==11.7.3.90
+dacite==1.9.2
+importlib_resources==7.1.0
+orbax-checkpoint==0.12.0
+nvidia-ml-py==13.610.43
+pyparsing==3.3.2
+kornia==0.8.1
+transforms3d==0.4.2
+sapien==3.0.3
+orjson==3.11.9
+wandb==0.22.1
+lxml==6.1.1
+Pygments==2.20.0
+labmaze==1.0.6
+mpmath==1.3.0
+dm-env==1.6
+ipython==8.37.0
+torchvision==0.23.0+cu128
+ogbench==1.1.5
+submitit==1.5.3
+dm_control==1.0.34
+nvidia-nvshmem-cu12==3.6.5
+fsspec==2026.4.0
+kornia_rs==0.1.14
+requests==2.34.2
+swig==4.4.1
+pydantic_core==2.46.4
+pyperclip==1.11.0
+nvidia-cufile-cu12==1.13.1.3
+matplotlib==3.10.9
+docstring_parser==0.18.0
+pyvers==0.1.0
+gym==0.26.2
+attrs==26.1.0
+nvidia-cuda-runtime-cu12==12.8.90
+idna==3.18
+ale-py==0.10.0
+jax==0.7.1
+safetensors==0.7.0
+nvidia-cusparse-cu12==12.5.8.93
+scipy==1.17.1
+etils==1.14.0
+humanize==4.15.0
+gpustat==1.1.1
+tabulate==0.10.0
+asttokens==3.0.1
+antlr4-python3-runtime==4.9.3
+torchaudio==2.8.0+cu128
+decorator==4.4.2
+proglog==0.1.12
+contourpy==1.3.3
+pytorch-seed==0.2.0
+simplejson==4.1.1
+tensordict==0.10.0
+torch==2.8.0+cu128
+packaging==25.0
+PyYAML==6.0.3
+Farama-Notifications==0.0.6
+ipython_pygments_lexers==1.1.1
+jax-cuda12-plugin==0.7.1
+pandas==3.0.3
+absl-py==2.4.0
+tyro==1.0.13
+torchrl==0.10.0
+pexpect==4.9.0
+tensorstore==0.1.84
+pillow==12.2.0
+fast_kinematics==0.2.2
+nvidia-nvjitlink-cu12==12.8.93
+urllib3==2.7.0
+zipp==4.1.0
+parso==0.8.7
+gitdb==4.0.12
+transformers==4.56.2
+click==8.4.1
+opt_einsum==3.4.0
+jaxlib==0.7.1
+nvidia-cufft-cu12==11.3.3.83
+hf-xet==1.5.0
+jedi==0.20.0
+AutoROM.accept-rom-license==0.6.1
+h5py==3.14.0
+Jinja2==3.1.6
+pip==25.2
+tqdm==4.67.1
+regex==2026.5.9
+moviepy==1.0.3
+robodesk==1.0.0
+flax==0.12.0
+executing==2.2.1
+triton==3.4.0
+platformdirs==4.10.0
+cycler==0.12.1
+ptyprocess==0.7.0
+typeguard==4.5.2
+cloudpickle==3.1.2
+protobuf==5.29.6
+msgpack==1.1.2
+prompt_toolkit==3.0.52
+mani-skill-nightly==2025.9.19.39
+nvidia-cuda-cupti-cu12==12.8.90
+imageio-ffmpeg==0.6.0
+wcwidth==0.7.0
+opencv-python==4.11.0.86
+certifi==2026.5.20
+typing_extensions==4.15.0
+mplib==0.1.1
+python-dateutil==2.9.0.post0
+charset-normalizer==3.4.7
+annotated-types==0.7.0
+psutil==7.2.2
+pytorch-kinematics==0.7.5
+matplotlib-inline==0.2.2
+metaworld==2.0.0
+mujoco==3.3.6
+traitlets==5.15.1
+nvidia-curand-cu12==10.3.9.90
+treescope==0.1.10
+toppra==0.6.3
+kiwisolver==1.5.0
+nvidia-cuda-nvrtc-cu12==12.8.93
+PyOpenGL==3.1.10
+imageio==2.37.0
+dm-tree==0.1.10
+nvidia-nccl-cu12==2.27.3
+fonttools==4.63.0
+jax-cuda12-pjrt==0.7.1
+glfw==2.10.0
+pygame==2.6.1
+arm_pytorch_utilities==0.5.0
+nvidia-cuda-nvcc-cu12==12.9.86
+huggingface_hub==0.36.2
+wheel==0.45.1
+MarkupSafe==3.0.3
+ml_dtypes==0.5.4
+uvloop==0.22.1
+filelock==3.29.0
+trimesh==4.12.2
+wrapt==2.2.1
+nvidia-cublas-cu12==12.8.4.1
+sentry-sdk==2.61.1
+toolz==1.1.0
+pluggy==1.6.0
+iniconfig==2.3.0
+pytest==9.1.0
+flashbax==0.1.3
+chex==0.1.92

soup_S_100M_work_dir/wandb/run-20260710_081650-8qh9ccql/files/wandb-metadata.json

@@ -0,0 +1,66 @@
+{
+  "os":  "Linux-5.15.0-72-generic-x86_64-with-glibc2.35",
+  "python":  "CPython 3.11.15",
+  "startedAt":  "2026-07-10T00:16:50.572137Z",
+  "args":  [
+    "task=soup",
+    "model_size=S",
+    "steps=100000000",
+    "demo_steps=100000",
+    "train_eval_freq=5000000",
+    "checkpoint_save_freq=5000000",
+    "diffusion_final_rerank=True",
+    "compile=True",
+    "diffusion_compile=True",
+    "multiproc=True",
+    "use_score_network=False",
+    "eval_at_start=False",
+    "env_mode=async",
+    "planner_type=diffusion",
+    "exp_name=soup_S_100M",
+    "seed=2",
+    "work_dir=/media/datasets/cheliu21/cxy_worldmodel/newt/soup_S_100M_work_dir"
+  ],
+  "program":  "/media/damoxing/che-liu-fileset/cxy_worldmodel/newt/tdmpc2/train.py",
+  "codePath":  "tdmpc2/train.py",
+  "codePathLocal":  "train.py",
+  "git":  {
+    "remote":  "git@github.com:Wenxuan52/newt.git",
+    "commit":  "0c15f2e94dcb5661c338ee4d4a9852c1f74447a2"
+  },
+  "email":  "wenxuan.yuan@qq.com",
+  "root":  "/media/datasets/cheliu21/cxy_worldmodel/newt/soup_S_100M_work_dir",
+  "host":  "aibox-r7a456c2731d-9f9cbb9ff-4ngg5",
+  "executable":  "/opt/conda/envs/newt-jax/bin/python",
+  "cpu_count":  64,
+  "cpu_count_logical":  128,
+  "gpu":  "NVIDIA A800-SXM4-80GB",
+  "gpu_count":  2,
+  "disk":  {
+    "/":  {
+      "total":  "105089261568",
+      "used":  "21334499328"
+    }
+  },
+  "memory":  {
+    "total":  "2162270285824"
+  },
+  "gpu_nvidia":  [
+    {
+      "name":  "NVIDIA A800-SXM4-80GB",
+      "memoryTotal":  "85899345920",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-6b222491-6e06-169f-f553-a20ca5a8aec4"
+    },
+    {
+      "name":  "NVIDIA A800-SXM4-80GB",
+      "memoryTotal":  "85899345920",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-6ed225ef-4f20-a7fa-e9dc-bbf84414725f"
+    }
+  ],
+  "cudaVersion":  "12.4",
+  "writerId":  "myi5i98pyh991xh9g4ow8fk7u0es34r1"
+}

soup_S_100M_work_dir/wandb/run-20260710_081650-8qh9ccql/files/wandb-summary.json

@@ -0,0 +1 @@
+{"_wandb":{"runtime":33296},"_runtime":33296}

soup_S_100M_work_dir/wandb/run-20260710_081650-8qh9ccql/logs/debug-core.log

@@ -0,0 +1,14 @@
+{"time":"2026-07-10T08:16:50.655814719+08:00","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmpkrf9kmt7/port-42540.txt","pid":42540,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2026-07-10T08:16:50.656213389+08:00","level":"INFO","msg":"server: will exit if parent process dies","ppid":42540}
+{"time":"2026-07-10T08:16:50.656193651+08:00","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-42540-93941-1585397787/socket","Net":"unix"}}
+{"time":"2026-07-10T08:16:50.844634997+08:00","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2026-07-10T08:16:50.84765988+08:00","level":"INFO","msg":"handleInformInit: received","streamId":"8qh9ccql","id":"1(@)"}
+{"time":"2026-07-10T08:16:51.398386454+08:00","level":"INFO","msg":"handleInformInit: stream started","streamId":"8qh9ccql","id":"1(@)"}
+{"time":"2026-07-10T17:31:52.226962198+08:00","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"1(@)"}
+{"time":"2026-07-10T17:31:52.227029221+08:00","level":"INFO","msg":"connection: closing","id":"1(@)"}
+{"time":"2026-07-10T17:31:52.22708053+08:00","level":"INFO","msg":"connection: closed successfully","id":"1(@)"}
+{"time":"2026-07-10T17:31:52.227042677+08:00","level":"INFO","msg":"server is shutting down"}
+{"time":"2026-07-10T17:31:52.2271423+08:00","level":"INFO","msg":"server: listener closed","addr":{"Name":"/tmp/wandb-42540-93941-1585397787/socket","Net":"unix"}}
+{"time":"2026-07-10T17:31:58.085153921+08:00","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"1(@)"}
+{"time":"2026-07-10T17:31:58.085186003+08:00","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"1(@)"}
+{"time":"2026-07-10T17:31:58.085198464+08:00","level":"INFO","msg":"server is closed"}

soup_S_100M_work_dir/wandb/run-20260710_081650-8qh9ccql/logs/debug-internal.log

@@ -0,0 +1,19 @@
+{"time":"2026-07-10T08:16:50.848130327+08:00","level":"INFO","msg":"stream: starting","core version":"0.22.1"}
+{"time":"2026-07-10T08:16:51.397813017+08:00","level":"INFO","msg":"stream: created new stream","id":"8qh9ccql"}
+{"time":"2026-07-10T08:16:51.397870202+08:00","level":"INFO","msg":"handler: started","stream_id":"8qh9ccql"}
+{"time":"2026-07-10T08:16:51.398375797+08:00","level":"INFO","msg":"stream: started","id":"8qh9ccql"}
+{"time":"2026-07-10T08:16:51.398392925+08:00","level":"INFO","msg":"sender: started","stream_id":"8qh9ccql"}
+{"time":"2026-07-10T08:16:51.398391473+08:00","level":"INFO","msg":"writer: started","stream_id":"8qh9ccql"}
+{"time":"2026-07-10T17:18:55.739020378+08:00","level":"INFO","msg":"flowcontrol: backed up, offloading to disk","recordNumber":9157}
+{"time":"2026-07-10T17:18:56.258273218+08:00","level":"INFO","msg":"flowcontrol: unblocked","totalOffloaded":17}
+{"time":"2026-07-10T17:19:40.962303684+08:00","level":"INFO","msg":"flowcontrol: backed up, offloading to disk","recordNumber":9693}
+{"time":"2026-07-10T17:19:41.255350443+08:00","level":"INFO","msg":"flowcontrol: unblocked","totalOffloaded":17}
+{"time":"2026-07-10T17:20:25.988665329+08:00","level":"INFO","msg":"flowcontrol: backed up, offloading to disk","recordNumber":10231}
+{"time":"2026-07-10T17:20:26.060867135+08:00","level":"INFO","msg":"flowcontrol: unblocked","totalOffloaded":15}
+{"time":"2026-07-10T17:26:25.835734511+08:00","level":"INFO","msg":"flowcontrol: backed up, offloading to disk","recordNumber":14410}
+{"time":"2026-07-10T17:26:26.247732836+08:00","level":"INFO","msg":"flowcontrol: unblocked","totalOffloaded":15}
+{"time":"2026-07-10T17:31:52.227029879+08:00","level":"INFO","msg":"stream: closing","id":"8qh9ccql"}
+{"time":"2026-07-10T17:31:54.557609085+08:00","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2026-07-10T17:31:58.08116313+08:00","level":"INFO","msg":"handler: closed","stream_id":"8qh9ccql"}
+{"time":"2026-07-10T17:31:58.084574607+08:00","level":"INFO","msg":"sender: closed","stream_id":"8qh9ccql"}
+{"time":"2026-07-10T17:31:58.084595227+08:00","level":"INFO","msg":"stream: closed","id":"8qh9ccql"}

soup_S_100M_work_dir/wandb/run-20260710_081650-8qh9ccql/run-8qh9ccql.wandb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:363b929cd66551d4ba4cc6b697c630706b95dc3fedb4af55050c555e7267e61f
+size 7915555

soup_S_100M_work_dir/wandb/run-20260710_200207-wzv4r5ph/files/requirements.txt

@@ -0,0 +1,180 @@
+nvidia-cudnn-cu12==9.10.2.21
+pure_eval==0.2.3
+smmap==5.0.3
+GitPython==3.1.50
+nvidia-nvtx-cu12==12.8.90
+blessed==1.44.0
+gymnasium==0.29.1
+optax==0.2.8
+AutoROM==0.6.1
+pynvml==13.0.1
+importlib_metadata==9.0.0
+numpy==1.26.4
+rich==15.0.0
+nvidia-cusparselt-cu12==0.7.1
+pydantic==2.13.4
+mdurl==0.1.2
+box2d-py==2.3.5
+six==1.17.0
+setuptools==69.5.1
+stack-data==0.6.3
+networkx==3.6.1
+nvidia-cuda-cccl-cu12==12.9.27
+hydra-submitit-launcher==1.2.0
+prometheus_client==0.25.0
+omegaconf==2.3.0
+aiofiles==25.1.0
+tokenizers==0.22.2
+sympy==1.14.0
+termcolor==3.1.0
+markdown-it-py==4.2.0
+jinxed==2.0.4
+typing-inspection==0.4.2
+hydra-core==1.3.2
+nvidia-cusolver-cu12==11.7.3.90
+dacite==1.9.2
+importlib_resources==7.1.0
+orbax-checkpoint==0.12.0
+nvidia-ml-py==13.610.43
+pyparsing==3.3.2
+kornia==0.8.1
+transforms3d==0.4.2
+sapien==3.0.3
+orjson==3.11.9
+wandb==0.22.1
+lxml==6.1.1
+Pygments==2.20.0
+labmaze==1.0.6
+mpmath==1.3.0
+dm-env==1.6
+ipython==8.37.0
+torchvision==0.23.0+cu128
+ogbench==1.1.5
+submitit==1.5.3
+dm_control==1.0.34
+nvidia-nvshmem-cu12==3.6.5
+fsspec==2026.4.0
+kornia_rs==0.1.14
+requests==2.34.2
+swig==4.4.1
+pydantic_core==2.46.4
+pyperclip==1.11.0
+nvidia-cufile-cu12==1.13.1.3
+matplotlib==3.10.9
+docstring_parser==0.18.0
+pyvers==0.1.0
+gym==0.26.2
+attrs==26.1.0
+nvidia-cuda-runtime-cu12==12.8.90
+idna==3.18
+ale-py==0.10.0
+jax==0.7.1
+safetensors==0.7.0
+nvidia-cusparse-cu12==12.5.8.93
+scipy==1.17.1
+etils==1.14.0
+humanize==4.15.0
+gpustat==1.1.1
+tabulate==0.10.0
+asttokens==3.0.1
+antlr4-python3-runtime==4.9.3
+torchaudio==2.8.0+cu128
+decorator==4.4.2
+proglog==0.1.12
+contourpy==1.3.3
+pytorch-seed==0.2.0
+simplejson==4.1.1
+tensordict==0.10.0
+torch==2.8.0+cu128
+packaging==25.0
+PyYAML==6.0.3
+Farama-Notifications==0.0.6
+ipython_pygments_lexers==1.1.1
+jax-cuda12-plugin==0.7.1
+pandas==3.0.3
+absl-py==2.4.0
+tyro==1.0.13
+torchrl==0.10.0
+pexpect==4.9.0
+tensorstore==0.1.84
+pillow==12.2.0
+fast_kinematics==0.2.2
+nvidia-nvjitlink-cu12==12.8.93
+urllib3==2.7.0
+zipp==4.1.0
+parso==0.8.7
+gitdb==4.0.12
+transformers==4.56.2
+click==8.4.1
+opt_einsum==3.4.0
+jaxlib==0.7.1
+nvidia-cufft-cu12==11.3.3.83
+hf-xet==1.5.0
+jedi==0.20.0
+AutoROM.accept-rom-license==0.6.1
+h5py==3.14.0
+Jinja2==3.1.6
+pip==25.2
+tqdm==4.67.1
+regex==2026.5.9
+moviepy==1.0.3
+robodesk==1.0.0
+flax==0.12.0
+executing==2.2.1
+triton==3.4.0
+platformdirs==4.10.0
+cycler==0.12.1
+ptyprocess==0.7.0
+typeguard==4.5.2
+cloudpickle==3.1.2
+protobuf==5.29.6
+msgpack==1.1.2
+prompt_toolkit==3.0.52
+mani-skill-nightly==2025.9.19.39
+nvidia-cuda-cupti-cu12==12.8.90
+imageio-ffmpeg==0.6.0
+wcwidth==0.7.0
+opencv-python==4.11.0.86
+certifi==2026.5.20
+typing_extensions==4.15.0
+mplib==0.1.1
+python-dateutil==2.9.0.post0
+charset-normalizer==3.4.7
+annotated-types==0.7.0
+psutil==7.2.2
+pytorch-kinematics==0.7.5
+matplotlib-inline==0.2.2
+metaworld==2.0.0
+mujoco==3.3.6
+traitlets==5.15.1
+nvidia-curand-cu12==10.3.9.90
+treescope==0.1.10
+toppra==0.6.3
+kiwisolver==1.5.0
+nvidia-cuda-nvrtc-cu12==12.8.93
+PyOpenGL==3.1.10
+imageio==2.37.0
+dm-tree==0.1.10
+nvidia-nccl-cu12==2.27.3
+fonttools==4.63.0
+jax-cuda12-pjrt==0.7.1
+glfw==2.10.0
+pygame==2.6.1
+arm_pytorch_utilities==0.5.0
+nvidia-cuda-nvcc-cu12==12.9.86
+huggingface_hub==0.36.2
+wheel==0.45.1
+MarkupSafe==3.0.3
+ml_dtypes==0.5.4
+uvloop==0.22.1
+filelock==3.29.0
+trimesh==4.12.2
+wrapt==2.2.1
+nvidia-cublas-cu12==12.8.4.1
+sentry-sdk==2.61.1
+toolz==1.1.0
+pluggy==1.6.0
+iniconfig==2.3.0
+pytest==9.1.0
+flashbax==0.1.3
+chex==0.1.92

soup_S_100M_work_dir/wandb/run-20260710_200207-wzv4r5ph/files/wandb-metadata.json

@@ -0,0 +1,66 @@
+{
+  "os":  "Linux-5.15.0-72-generic-x86_64-with-glibc2.35",
+  "python":  "CPython 3.11.15",
+  "startedAt":  "2026-07-10T12:02:07.771625Z",
+  "args":  [
+    "task=soup",
+    "model_size=S",
+    "steps=100000000",
+    "demo_steps=100000",
+    "train_eval_freq=5000000",
+    "checkpoint_save_freq=5000000",
+    "diffusion_final_rerank=True",
+    "compile=True",
+    "diffusion_compile=True",
+    "multiproc=True",
+    "use_score_network=False",
+    "eval_at_start=False",
+    "env_mode=async",
+    "planner_type=diffusion",
+    "exp_name=soup_S_100M",
+    "seed=2",
+    "work_dir=/media/datasets/cheliu21/cxy_worldmodel/newt/soup_S_100M_work_dir"
+  ],
+  "program":  "/media/damoxing/che-liu-fileset/cxy_worldmodel/newt/tdmpc2/train.py",
+  "codePath":  "tdmpc2/train.py",
+  "codePathLocal":  "train.py",
+  "git":  {
+    "remote":  "git@github.com:Wenxuan52/newt.git",
+    "commit":  "0c15f2e94dcb5661c338ee4d4a9852c1f74447a2"
+  },
+  "email":  "wenxuan.yuan@qq.com",
+  "root":  "/media/datasets/cheliu21/cxy_worldmodel/newt/soup_S_100M_work_dir",
+  "host":  "aibox-r7a456c2731d-9f9cbb9ff-4ngg5",
+  "executable":  "/opt/conda/envs/newt-jax/bin/python",
+  "cpu_count":  64,
+  "cpu_count_logical":  128,
+  "gpu":  "NVIDIA A800-SXM4-80GB",
+  "gpu_count":  2,
+  "disk":  {
+    "/":  {
+      "total":  "105089261568",
+      "used":  "37952913408"
+    }
+  },
+  "memory":  {
+    "total":  "2162270285824"
+  },
+  "gpu_nvidia":  [
+    {
+      "name":  "NVIDIA A800-SXM4-80GB",
+      "memoryTotal":  "85899345920",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-6b222491-6e06-169f-f553-a20ca5a8aec4"
+    },
+    {
+      "name":  "NVIDIA A800-SXM4-80GB",
+      "memoryTotal":  "85899345920",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-6ed225ef-4f20-a7fa-e9dc-bbf84414725f"
+    }
+  ],
+  "cudaVersion":  "12.4",
+  "writerId":  "vk8lkx53xv34va7k1mrw2i9dra9zw04u"
+}

soup_S_100M_work_dir/wandb/run-20260710_200207-wzv4r5ph/logs/debug-core.log

@@ -0,0 +1,6 @@
+{"time":"2026-07-10T20:02:07.848646032+08:00","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmpf2_2ql5y/port-291934.txt","pid":291934,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false,"enable-dcgm-profiling":false}
+{"time":"2026-07-10T20:02:07.849023994+08:00","level":"INFO","msg":"server: will exit if parent process dies","ppid":291934}
+{"time":"2026-07-10T20:02:07.849030289+08:00","level":"INFO","msg":"server: accepting connections","addr":{"Name":"/tmp/wandb-291934-345136-2539605336/socket","Net":"unix"}}
+{"time":"2026-07-10T20:02:08.038660869+08:00","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"1(@)"}
+{"time":"2026-07-10T20:02:08.041780311+08:00","level":"INFO","msg":"handleInformInit: received","streamId":"wzv4r5ph","id":"1(@)"}
+{"time":"2026-07-10T20:02:08.572363158+08:00","level":"INFO","msg":"handleInformInit: stream started","streamId":"wzv4r5ph","id":"1(@)"}

soup_S_100M_work_dir/wandb/run-20260710_200207-wzv4r5ph/logs/debug-internal.log

@@ -0,0 +1,10 @@
+{"time":"2026-07-10T20:02:08.042270494+08:00","level":"INFO","msg":"stream: starting","core version":"0.22.1"}
+{"time":"2026-07-10T20:02:08.571791737+08:00","level":"INFO","msg":"stream: created new stream","id":"wzv4r5ph"}
+{"time":"2026-07-10T20:02:08.571857921+08:00","level":"INFO","msg":"handler: started","stream_id":"wzv4r5ph"}
+{"time":"2026-07-10T20:02:08.572352151+08:00","level":"INFO","msg":"stream: started","id":"wzv4r5ph"}
+{"time":"2026-07-10T20:02:08.572360008+08:00","level":"INFO","msg":"writer: started","stream_id":"wzv4r5ph"}
+{"time":"2026-07-10T20:02:08.572362278+08:00","level":"INFO","msg":"sender: started","stream_id":"wzv4r5ph"}
+{"time":"2026-07-10T20:12:29.33661486+08:00","level":"INFO","msg":"flowcontrol: backed up, offloading to disk","recordNumber":5608}
+{"time":"2026-07-10T20:12:29.799465299+08:00","level":"INFO","msg":"flowcontrol: unblocked","totalOffloaded":17}
+{"time":"2026-07-10T20:34:17.674522565+08:00","level":"INFO","msg":"flowcontrol: backed up, offloading to disk","recordNumber":17923}
+{"time":"2026-07-10T20:34:20.955546777+08:00","level":"INFO","msg":"flowcontrol: unblocked","totalOffloaded":31}

soup_S_100M_work_dir/wandb/run-20260710_200207-wzv4r5ph/run-wzv4r5ph.wandb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50ea0289218e1d2a63685b6360b75b46167034551b599e9ef543b0815addf859
+size 18841600