sync new results for soup_XL_100M_work_dir

.gitattributes

@@ -53,3 +53,4 @@ soup_B_100M_work_dir/wandb/run-20260708_090206-4s2ksbsn/run-4s2ksbsn.wandb filte
 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_XL_100M_work_dir/wandb/run-20260708_090752-bs3i5bjq/run-bs3i5bjq.wandb filter=lfs diff=lfs merge=lfs -text

soup_XL_100M_work_dir/ckpt/0.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:57fbb768e799e4e0904604b4d2401ccd7c64c3b3d3192ed3c36d05df6d20417b
+size 1254669611

soup_XL_100M_work_dir/ckpt/10_000_000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bb0187b6c14f3ca979dc66db18624714dcebb5894a9af02cd893a6227ef4664c
+size 1308819645

soup_XL_100M_work_dir/ckpt/15_000_000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b006226dfd27137c8466601ff224698c11f400114e932d0f41aabc3cd6b9186
+size 1308819645

soup_XL_100M_work_dir/ckpt/20_000_000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f344d9c3d21faae93471da63d782025e589262454ed47cd2705beb6275823d8c
+size 1308819645

soup_XL_100M_work_dir/ckpt/5_000_000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:38d644bac8d8dae5accee808cae303c51d282d22f541d9d9c41ae00533d202f9
+size 1308819019

soup_XL_100M_work_dir/wandb/debug-internal.log

@@ -0,0 +1,13 @@
+{"time":"2026-07-08T09:07:52.33321276+08:00","level":"INFO","msg":"stream: starting","core version":"0.22.1"}
+{"time":"2026-07-08T09:07:52.879779385+08:00","level":"INFO","msg":"stream: created new stream","id":"bs3i5bjq"}
+{"time":"2026-07-08T09:07:52.879841771+08:00","level":"INFO","msg":"handler: started","stream_id":"bs3i5bjq"}
+{"time":"2026-07-08T09:07:52.88015593+08:00","level":"INFO","msg":"stream: started","id":"bs3i5bjq"}
+{"time":"2026-07-08T09:07:52.880170829+08:00","level":"INFO","msg":"sender: started","stream_id":"bs3i5bjq"}
+{"time":"2026-07-08T09:07:52.880173317+08:00","level":"INFO","msg":"writer: started","stream_id":"bs3i5bjq"}
+{"time":"2026-07-08T10:34:11.115390756+08:00","level":"INFO","msg":"flowcontrol: backed up, offloading to disk","recordNumber":28926}
+{"time":"2026-07-08T10:34:11.158966111+08:00","level":"INFO","msg":"flowcontrol: unblocked","totalOffloaded":13}
+{"time":"2026-07-08T14:26:17.46615665+08:00","level":"INFO","msg":"api: retrying HTTP error","status":429,"url":"https://api.wandb.ai/files/wenxuan-yuan-imperial-college-london/newt/bs3i5bjq/file_stream","body":"{\"error\":\"rate limit exceeded: per_run limit on filestream requests\"}"}
+{"time":"2026-07-08T14:26:26.12017532+08:00","level":"INFO","msg":"api: retrying HTTP error","status":429,"url":"https://api.wandb.ai/files/wenxuan-yuan-imperial-college-london/newt/bs3i5bjq/file_stream","body":"{\"error\":\"rate limit exceeded: per_run limit on filestream requests\"}"}
+{"time":"2026-07-08T14:26:33.988391415+08:00","level":"INFO","msg":"api: retrying HTTP error","status":429,"url":"https://api.wandb.ai/files/wenxuan-yuan-imperial-college-london/newt/bs3i5bjq/file_stream","body":"{\"error\":\"rate limit exceeded: per_run limit on filestream requests\"}"}
+{"time":"2026-07-08T14:26:43.19340454+08:00","level":"INFO","msg":"api: retrying HTTP error","status":429,"url":"https://api.wandb.ai/files/wenxuan-yuan-imperial-college-london/newt/bs3i5bjq/file_stream","body":"{\"error\":\"rate limit exceeded: per_run limit on filestream requests\"}"}
+{"time":"2026-07-08T14:27:02.94526155+08:00","level":"INFO","msg":"api: retrying HTTP error","status":429,"url":"https://api.wandb.ai/files/wenxuan-yuan-imperial-college-london/newt/bs3i5bjq/file_stream","body":"{\"error\":\"rate limit exceeded: per_run limit on filestream requests\"}"}

soup_XL_100M_work_dir/wandb/run-20260623_100535-g9l145ew/files/output.log

@@ -0,0 +1,1180 @@
+Logs will be synced with wandb.
+Architecture: DDPWrapper(
+  (_module): DistributedDataParallel(
+    (module): Newt World Model
+    Encoder (11,161,664): ModuleDict(
+      (state): Sequential(
+        (0): NormedLinear(in_features=640, out_features=2048, bias=True, act=Mish)
+        (1): NormedLinear(in_features=2048, out_features=2048, bias=True, act=Mish)
+        (2): NormedLinear(in_features=2048, out_features=2048, bias=True, act=Mish)
+        (3): NormedLinear(in_features=2048, out_features=704, bias=True, act=SimNorm)
+      )
+    )
+    Dynamics (8,173,632): Sequential(
+      (0): NormedLinear(in_features=1232, out_features=2048, bias=True, act=Mish)
+      (1): NormedLinear(in_features=2048, out_features=2048, bias=True, act=Mish)
+      (2): NormedLinear(in_features=2048, out_features=704, bias=True, act=SimNorm)
+    )
+    Reward (6,936,677): Sequential(
+      (0): NormedLinear(in_features=1232, out_features=2048, bias=True, act=Mish)
+      (1): NormedLinear(in_features=2048, out_features=2048, bias=True, act=Mish)
+      (2): Linear(in_features=2048, out_features=101, bias=True)
+    )
+    Contrastive F (6,731,777): Sequential(
+      (0): NormedLinear(in_features=1232, out_features=2048, bias=True, act=Mish)
+      (1): NormedLinear(in_features=2048, out_features=2048, bias=True, act=Mish)
+      (2): Linear(in_features=2048, out_features=1, bias=True)
+    )
+    Policy prior (6,762,528): Sequential(
+      (0): NormedLinear(in_features=1216, out_features=2048, bias=True, act=Mish)
+      (1): NormedLinear(in_features=2048, out_features=2048, bias=True, act=Mish)
+      (2): Linear(in_features=2048, out_features=32, bias=True)
+    )
+    Q-functions (48,556,739): QEnsemble(
+      (_Qs): ModuleList(
+        (0-6): 7 x Sequential(
+          (0): NormedLinear(in_features=1232, out_features=2048, bias=True, act=Mish)
+          (1): NormedLinear(in_features=2048, out_features=2048, bias=True, act=Mish)
+          (2): Linear(in_features=2048, out_features=101, bias=True)
+        )
+      )
+    )
+    Learnable parameters: 88,323,017
+  )
+)
+Update frequency: 200,000
+Episodes per update frequency: 1,933
+No checkpoint found, training from scratch.
+Pretraining agent on demonstrations...
+prior_coef is 10.0, setting to 1.0 for pretraining.
+Pretraining:   0%|          | 1/200000 [00:41<2314:40:06, 41.66s/it][rank0]:V0623 10:06:20.452000 298 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      0.02467
+ reward_loss           4.30745
+ value_loss            4.30745
+ total_loss            2.33911
+ bc_loss               0.90417
+ entropy_loss          -0.00223
+ pi_prior_loss         0.29106
+ pi_entropy            2.72638
+ pi_scaled_entropy     22.25638
+ pi_std                0.99740
+ pi_max_std            5.66591
+ contrastive_loss      0.69315
+ contrastive_pos_logit 0.00000
+ contrastive_neg_logit 0.00000
+ contrastive_mean      0.00000
+ contrastive_std       0.99000
+ grad_norm             3.97725
+ lr_enc                0.00000
+ lr                    0.00000
+ lr_pi                 0.00000
+------------------------------
+[rank0]:V0623 10:06:20.452000 298 site-packages/torch/_dynamo/guards.py:3508] [0/1] [__recompiles]     triggered by the following guard failure(s):
+[rank0]:V0623 10:06:20.452000 298 site-packages/torch/_dynamo/guards.py:3508] [0/1] [__recompiles]     - 0/0: len(G['__import_tensordict_dot_utils']._TENSORCLASS_MEMO) != 15
+Pretraining:   4%|▍         | 7999/200000 [13:26<5:04:03, 10.52it/s]
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00265
+ reward_loss           0.49153
+ value_loss            0.49874
+ total_loss            0.87740
+ bc_loss               0.25458
+ entropy_loss          -0.00088
+ pi_prior_loss         0.08286
+ pi_entropy            2.00650
+ pi_scaled_entropy     8.76924
+ pi_std                0.76371
+ pi_max_std            1.03175
+ contrastive_loss      0.64243
+ contrastive_pos_logit 0.18010
+ contrastive_neg_logit -0.21694
+ contrastive_mean      0.01501
+ contrastive_std       0.63365
+ grad_norm             0.78231
+ lr_enc                0.00007
+ lr                    0.00024
+ lr_pi                 0.00024
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00228
+ reward_loss           0.54483
+ value_loss            0.49017
+ total_loss            0.87847
+ bc_loss               0.25321
+ entropy_loss          -0.00117
+ pi_prior_loss         0.07979
+ pi_entropy            2.47168
+ pi_scaled_entropy     11.73463
+ pi_std                0.76287
+ pi_max_std            1.00000
+ contrastive_loss      0.64949
+ contrastive_pos_logit 0.37591
+ contrastive_neg_logit -0.07815
+ contrastive_mean      0.02642
+ contrastive_std       0.71706
+ grad_norm             0.74900
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00208
+ reward_loss           0.52917
+ value_loss            0.90180
+ total_loss            0.90592
+ bc_loss               0.25956
+ entropy_loss          -0.00073
+ pi_prior_loss         0.08144
+ pi_entropy            2.27827
+ pi_scaled_entropy     7.33548
+ pi_std                0.76353
+ pi_max_std            1.00000
+ contrastive_loss      0.63975
+ contrastive_pos_logit 0.38069
+ contrastive_neg_logit -0.11297
+ contrastive_mean      0.03023
+ contrastive_std       0.78676
+ grad_norm             0.54999
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00208
+ reward_loss           0.48864
+ value_loss            0.53216
+ total_loss            0.83344
+ bc_loss               0.20776
+ entropy_loss          -0.00098
+ pi_prior_loss         0.06697
+ pi_entropy            2.43512
+ pi_scaled_entropy     9.79741
+ pi_std                0.76530
+ pi_max_std            1.20144
+ contrastive_loss      0.62286
+ contrastive_pos_logit 0.36886
+ contrastive_neg_logit -0.15992
+ contrastive_mean      0.03071
+ contrastive_std       0.81851
+ grad_norm             0.40577
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00196
+ reward_loss           0.49436
+ value_loss            0.54299
+ total_loss            0.82382
+ bc_loss               0.20190
+ entropy_loss          -0.00094
+ pi_prior_loss         0.06465
+ pi_entropy            1.89240
+ pi_scaled_entropy     9.39937
+ pi_std                0.75816
+ pi_max_std            1.20815
+ contrastive_loss      0.61620
+ contrastive_pos_logit 0.38863
+ contrastive_neg_logit -0.24144
+ contrastive_mean      0.03054
+ contrastive_std       0.88668
+ grad_norm             0.44075
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00231
+ reward_loss           0.45311
+ value_loss            0.55078
+ total_loss            0.83577
+ bc_loss               0.22282
+ entropy_loss          -0.00128
+ pi_prior_loss         0.07127
+ pi_entropy            2.54053
+ pi_scaled_entropy     12.77746
+ pi_std                0.77247
+ pi_max_std            1.38735
+ contrastive_loss      0.61788
+ contrastive_pos_logit 0.27619
+ contrastive_neg_logit -0.33753
+ contrastive_mean      0.02916
+ contrastive_std       0.91868
+ grad_norm             0.51016
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00226
+ reward_loss           0.50228
+ value_loss            0.45888
+ total_loss            0.82558
+ bc_loss               0.20616
+ entropy_loss          -0.00129
+ pi_prior_loss         0.06469
+ pi_entropy            2.34219
+ pi_scaled_entropy     12.87433
+ pi_std                0.76522
+ pi_max_std            1.05497
+ contrastive_loss      0.61958
+ contrastive_pos_logit 0.40802
+ contrastive_neg_logit -0.30381
+ contrastive_mean      0.03019
+ contrastive_std       0.96939
+ grad_norm             0.41051
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00227
+ reward_loss           0.49466
+ value_loss            0.55087
+ total_loss            0.81287
+ bc_loss               0.19208
+ entropy_loss          -0.00124
+ pi_prior_loss         0.06074
+ pi_entropy            2.29622
+ pi_scaled_entropy     12.39907
+ pi_std                0.76821
+ pi_max_std            2.06816
+ contrastive_loss      0.60227
+ contrastive_pos_logit 0.44011
+ contrastive_neg_logit -0.41351
+ contrastive_mean      0.02361
+ contrastive_std       1.03155
+ grad_norm             0.46515
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00236
+ reward_loss           0.45463
+ value_loss            0.57505
+ total_loss            0.83072
+ bc_loss               0.20341
+ entropy_loss          -0.00143
+ pi_prior_loss         0.06346
+ pi_entropy            2.57812
+ pi_scaled_entropy     14.29637
+ pi_std                0.77729
+ pi_max_std            1.64060
+ contrastive_loss      0.61704
+ contrastive_pos_logit 0.45206
+ contrastive_neg_logit -0.28638
+ contrastive_mean      0.02154
+ contrastive_std       1.09277
+ grad_norm             0.50279
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00238
+ reward_loss           0.44701
+ value_loss            0.52478
+ total_loss            0.81326
+ bc_loss               0.23596
+ entropy_loss          -0.00130
+ pi_prior_loss         0.07654
+ pi_entropy            2.69820
+ pi_scaled_entropy     13.01992
+ pi_std                0.77659
+ pi_max_std            1.25145
+ contrastive_loss      0.59197
+ contrastive_pos_logit 0.39765
+ contrastive_neg_logit -0.49633
+ contrastive_mean      0.02081
+ contrastive_std       1.13147
+ grad_norm             0.40804
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00256
+ reward_loss           0.48533
+ value_loss            0.53448
+ total_loss            0.81000
+ bc_loss               0.19828
+ entropy_loss          -0.00112
+ pi_prior_loss         0.06219
+ pi_entropy            1.72597
+ pi_scaled_entropy     11.24058
+ pi_std                0.75782
+ pi_max_std            1.56250
+ contrastive_loss      0.59457
+ contrastive_pos_logit 0.45619
+ contrastive_neg_logit -0.44733
+ contrastive_mean      0.01845
+ contrastive_std       1.16010
+ grad_norm             0.52094
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00221
+ reward_loss           0.46861
+ value_loss            0.46767
+ total_loss            0.81188
+ bc_loss               0.20666
+ entropy_loss          -0.00150
+ pi_prior_loss         0.06662
+ pi_entropy            3.15087
+ pi_scaled_entropy     15.03450
+ pi_std                0.77937
+ pi_max_std            1.25600
+ contrastive_loss      0.60739
+ contrastive_pos_logit 0.44643
+ contrastive_neg_logit -0.39911
+ contrastive_mean      0.01690
+ contrastive_std       1.22122
+ grad_norm             0.46720
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00250
+ reward_loss           0.44491
+ value_loss            0.55384
+ total_loss            0.80343
+ bc_loss               0.17535
+ entropy_loss          -0.00128
+ pi_prior_loss         0.05722
+ pi_entropy            2.14758
+ pi_scaled_entropy     12.75763
+ pi_std                0.77828
+ pi_max_std            2.39449
+ contrastive_loss      0.59635
+ contrastive_pos_logit 0.50806
+ contrastive_neg_logit -0.48557
+ contrastive_mean      0.01545
+ contrastive_std       1.23719
+ grad_norm             0.63250
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00292
+ reward_loss           0.46144
+ value_loss            0.47883
+ total_loss            0.80751
+ bc_loss               0.19689
+ entropy_loss          -0.00145
+ pi_prior_loss         0.06260
+ pi_entropy            2.44735
+ pi_scaled_entropy     14.48672
+ pi_std                0.77586
+ pi_max_std            2.20730
+ contrastive_loss      0.59246
+ contrastive_pos_logit 0.62278
+ contrastive_neg_logit -0.36566
+ contrastive_mean      0.01638
+ contrastive_std       1.26936
+ grad_norm             0.58808
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00293
+ reward_loss           0.44249
+ value_loss            0.48634
+ total_loss            0.77170
+ bc_loss               0.19799
+ entropy_loss          -0.00143
+ pi_prior_loss         0.06052
+ pi_entropy            2.18767
+ pi_scaled_entropy     14.31973
+ pi_std                0.77055
+ pi_max_std            2.20363
+ contrastive_loss      0.55961
+ contrastive_pos_logit 0.54377
+ contrastive_neg_logit -0.64203
+ contrastive_mean      0.01358
+ contrastive_std       1.29616
+ grad_norm             0.42968
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00287
+ reward_loss           0.43705
+ value_loss            0.48313
+ total_loss            0.78217
+ bc_loss               0.17958
+ entropy_loss          -0.00174
+ pi_prior_loss         0.05721
+ pi_entropy            1.76759
+ pi_scaled_entropy     17.39634
+ pi_std                0.76911
+ pi_max_std            1.98494
+ contrastive_loss      0.57556
+ contrastive_pos_logit 0.61791
+ contrastive_neg_logit -0.53700
+ contrastive_mean      0.01630
+ contrastive_std       1.31230
+ grad_norm             0.56348
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00236
+ reward_loss           0.44345
+ value_loss            0.48383
+ total_loss            0.75249
+ bc_loss               0.16993
+ entropy_loss          -0.00149
+ pi_prior_loss         0.05343
+ pi_entropy            1.53880
+ pi_scaled_entropy     14.94003
+ pi_std                0.76094
+ pi_max_std            1.95329
+ contrastive_loss      0.55912
+ contrastive_pos_logit 0.84188
+ contrastive_neg_logit -0.57424
+ contrastive_mean      0.01488
+ contrastive_std       1.31971
+ grad_norm             0.87350
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00241
+ reward_loss           0.43501
+ value_loss            0.50453
+ total_loss            0.77288
+ bc_loss               0.19233
+ entropy_loss          -0.00151
+ pi_prior_loss         0.06041
+ pi_entropy            2.09735
+ pi_scaled_entropy     15.05507
+ pi_std                0.77025
+ pi_max_std            1.85363
+ contrastive_loss      0.57035
+ contrastive_pos_logit 0.67769
+ contrastive_neg_logit -0.62795
+ contrastive_mean      0.01325
+ contrastive_std       1.36544
+ grad_norm             0.63929
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00195
+ reward_loss           0.49201
+ value_loss            0.56737
+ total_loss            0.76104
+ bc_loss               0.18665
+ entropy_loss          -0.00160
+ pi_prior_loss         0.05807
+ pi_entropy            1.93475
+ pi_scaled_entropy     16.01232
+ pi_std                0.77151
+ pi_max_std            2.56334
+ contrastive_loss      0.55800
+ contrastive_pos_logit 0.77683
+ contrastive_neg_logit -0.51331
+ contrastive_mean      0.00966
+ contrastive_std       1.41337
+ grad_norm             0.59012
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00190
+ reward_loss           0.44406
+ value_loss            0.48654
+ total_loss            0.74514
+ bc_loss               0.18871
+ entropy_loss          -0.00145
+ pi_prior_loss         0.05730
+ pi_entropy            2.42120
+ pi_scaled_entropy     14.52395
+ pi_std                0.77601
+ pi_max_std            2.95923
+ contrastive_loss      0.55682
+ contrastive_pos_logit 0.59401
+ contrastive_neg_logit -0.78931
+ contrastive_mean      0.00669
+ contrastive_std       1.44798
+ grad_norm             0.57467
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00180
+ reward_loss           0.50125
+ value_loss            0.52190
+ total_loss            0.75844
+ bc_loss               0.18673
+ entropy_loss          -0.00156
+ pi_prior_loss         0.05980
+ pi_entropy            2.05099
+ pi_scaled_entropy     15.55122
+ pi_std                0.77235
+ pi_max_std            2.58286
+ contrastive_loss      0.56024
+ contrastive_pos_logit 0.67183
+ contrastive_neg_logit -0.61166
+ contrastive_mean      0.00522
+ contrastive_std       1.47438
+ grad_norm             0.63365
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00201
+ reward_loss           0.42350
+ value_loss            0.62962
+ total_loss            0.75449
+ bc_loss               0.18789
+ entropy_loss          -0.00168
+ pi_prior_loss         0.05855
+ pi_entropy            1.92834
+ pi_scaled_entropy     16.80657
+ pi_std                0.77128
+ pi_max_std            1.68869
+ contrastive_loss      0.55043
+ contrastive_pos_logit 0.80425
+ contrastive_neg_logit -0.57811
+ contrastive_mean      0.00663
+ contrastive_std       1.50137
+ grad_norm             0.55352
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00162
+ reward_loss           0.45005
+ value_loss            0.47480
+ total_loss            0.74196
+ bc_loss               0.16953
+ entropy_loss          -0.00175
+ pi_prior_loss         0.05332
+ pi_entropy            2.15380
+ pi_scaled_entropy     17.48677
+ pi_std                0.78500
+ pi_max_std            2.47720
+ contrastive_loss      0.56380
+ contrastive_pos_logit 0.79463
+ contrastive_neg_logit -0.62770
+ contrastive_mean      0.00302
+ contrastive_std       1.52429
+ grad_norm             0.61682
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00161
+ reward_loss           0.44772
+ value_loss            0.55312
+ total_loss            0.73870
+ bc_loss               0.19995
+ entropy_loss          -0.00148
+ pi_prior_loss         0.06279
+ pi_entropy            2.09332
+ pi_scaled_entropy     14.79656
+ pi_std                0.77205
+ pi_max_std            2.17110
+ contrastive_loss      0.54353
+ contrastive_pos_logit 0.86714
+ contrastive_neg_logit -0.63477
+ contrastive_mean      0.00245
+ contrastive_std       1.54013
+ grad_norm             0.58738
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00162
+ reward_loss           0.40333
+ value_loss            0.53891
+ total_loss            0.74960
+ bc_loss               0.18556
+ entropy_loss          -0.00170
+ pi_prior_loss         0.05662
+ pi_entropy            1.99021
+ pi_scaled_entropy     17.02247
+ pi_std                0.76803
+ pi_max_std            2.14561
+ contrastive_loss      0.56625
+ contrastive_pos_logit 0.66684
+ contrastive_neg_logit -0.80069
+ contrastive_mean      -0.00048
+ contrastive_std       1.57910
+ grad_norm             0.65196
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00146
+ reward_loss           0.42006
+ value_loss            0.48144
+ total_loss            0.71740
+ bc_loss               0.18199
+ entropy_loss          -0.00167
+ pi_prior_loss         0.05630
+ pi_entropy            1.56013
+ pi_scaled_entropy     16.66305
+ pi_std                0.76846
+ pi_max_std            3.45237
+ contrastive_loss      0.54170
+ contrastive_pos_logit 0.80791
+ contrastive_neg_logit -0.69993
+ contrastive_mean      -0.00374
+ contrastive_std       1.59853
+ grad_norm             0.76680
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00168
+ reward_loss           0.43532
+ value_loss            0.50606
+ total_loss            0.74382
+ bc_loss               0.17647
+ entropy_loss          -0.00174
+ pi_prior_loss         0.05552
+ pi_entropy            2.37219
+ pi_scaled_entropy     17.35526
+ pi_std                0.78565
+ pi_max_std            1.85732
+ contrastive_loss      0.56055
+ contrastive_pos_logit 0.67123
+ contrastive_neg_logit -0.77808
+ contrastive_mean      -0.00751
+ contrastive_std       1.62419
+ grad_norm             0.64379
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00149
+ reward_loss           0.43328
+ value_loss            0.52474
+ total_loss            0.72200
+ bc_loss               0.17227
+ entropy_loss          -0.00154
+ pi_prior_loss         0.05387
+ pi_entropy            1.72626
+ pi_scaled_entropy     15.40534
+ pi_std                0.76947
+ pi_max_std            3.13370
+ contrastive_loss      0.54255
+ contrastive_pos_logit 0.72010
+ contrastive_neg_logit -0.86726
+ contrastive_mean      -0.00482
+ contrastive_std       1.64503
+ grad_norm             0.64796
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00152
+ reward_loss           0.43261
+ value_loss            0.49984
+ total_loss            0.72449
+ bc_loss               0.16684
+ entropy_loss          -0.00160
+ pi_prior_loss         0.05249
+ pi_entropy            1.84485
+ pi_scaled_entropy     16.04462
+ pi_std                0.77362
+ pi_max_std            2.38854
+ contrastive_loss      0.54828
+ contrastive_pos_logit 0.88922
+ contrastive_neg_logit -0.77415
+ contrastive_mean      -0.00869
+ contrastive_std       1.68556
+ grad_norm             1.02152
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00163
+ reward_loss           0.42202
+ value_loss            0.56579
+ total_loss            0.73394
+ bc_loss               0.18851
+ entropy_loss          -0.00177
+ pi_prior_loss         0.05784
+ pi_entropy            1.95919
+ pi_scaled_entropy     17.69400
+ pi_std                0.77332
+ pi_max_std            3.94518
+ contrastive_loss      0.54472
+ contrastive_pos_logit 0.67652
+ contrastive_neg_logit -0.90501
+ contrastive_mean      -0.01030
+ contrastive_std       1.70954
+ grad_norm             0.70580
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00162
+ reward_loss           0.41959
+ value_loss            0.50879
+ total_loss            0.71718
+ bc_loss               0.17984
+ entropy_loss          -0.00098
+ pi_prior_loss         0.05675
+ pi_entropy            0.33626
+ pi_scaled_entropy     9.84783
+ pi_std                0.77612
+ pi_max_std            4.09453
+ contrastive_loss      0.53517
+ contrastive_pos_logit 0.99810
+ contrastive_neg_logit -0.70211
+ contrastive_mean      -0.01244
+ contrastive_std       1.71976
+ grad_norm             1.08714
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00142
+ reward_loss           0.38849
+ value_loss            0.53266
+ total_loss            0.70777
+ bc_loss               0.18814
+ entropy_loss          -0.00169
+ pi_prior_loss         0.05722
+ pi_entropy            2.29765
+ pi_scaled_entropy     16.86834
+ pi_std                0.78256
+ pi_max_std            3.00517
+ contrastive_loss      0.53004
+ contrastive_pos_logit 0.70572
+ contrastive_neg_logit -0.95326
+ contrastive_mean      -0.01656
+ contrastive_std       1.74067
+ grad_norm             0.60207
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00138
+ reward_loss           0.42173
+ value_loss            0.53435
+ total_loss            0.70711
+ bc_loss               0.17326
+ entropy_loss          -0.00180
+ pi_prior_loss         0.05472
+ pi_entropy            1.61974
+ pi_scaled_entropy     18.00428
+ pi_std                0.77371
+ pi_max_std            2.93966
+ contrastive_loss      0.52919
+ contrastive_pos_logit 0.82301
+ contrastive_neg_logit -1.10370
+ contrastive_mean      -0.01617
+ contrastive_std       1.76528
+ grad_norm             0.64897
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00151
+ reward_loss           0.43184
+ value_loss            0.51705
+ total_loss            0.71353
+ bc_loss               0.15308
+ entropy_loss          -0.00167
+ pi_prior_loss         0.05100
+ pi_entropy            1.72766
+ pi_scaled_entropy     16.73644
+ pi_std                0.77384
+ pi_max_std            3.93693
+ contrastive_loss      0.53750
+ contrastive_pos_logit 0.81595
+ contrastive_neg_logit -0.85151
+ contrastive_mean      -0.01714
+ contrastive_std       1.78801
+ grad_norm             0.55102
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00156
+ reward_loss           0.39618
+ value_loss            0.52558
+ total_loss            0.69449
+ bc_loss               0.18399
+ entropy_loss          -0.00174
+ pi_prior_loss         0.05984
+ pi_entropy            1.70508
+ pi_scaled_entropy     17.43760
+ pi_std                0.77928
+ pi_max_std            4.88674
+ contrastive_loss      0.51134
+ contrastive_pos_logit 0.89907
+ contrastive_neg_logit -0.99066
+ contrastive_mean      -0.02252
+ contrastive_std       1.81520
+ grad_norm             0.57423
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00137
+ reward_loss           0.42567
+ value_loss            0.48445
+ total_loss            0.69784
+ bc_loss               0.15756
+ entropy_loss          -0.00174
+ pi_prior_loss         0.04930
+ pi_entropy            1.94695
+ pi_scaled_entropy     17.43790
+ pi_std                0.77889
+ pi_max_std            2.18126
+ contrastive_loss      0.53015
+ contrastive_pos_logit 0.90961
+ contrastive_neg_logit -0.87386
+ contrastive_mean      -0.02177
+ contrastive_std       1.84699
+ grad_norm             0.78900
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00151
+ reward_loss           0.41609
+ value_loss            0.47456
+ total_loss            0.69935
+ bc_loss               0.16518
+ entropy_loss          -0.00151
+ pi_prior_loss         0.05001
+ pi_entropy            1.91174
+ pi_scaled_entropy     15.10952
+ pi_std                0.77819
+ pi_max_std            6.14075
+ contrastive_loss      0.53014
+ contrastive_pos_logit 0.84130
+ contrastive_neg_logit -0.97208
+ contrastive_mean      -0.02812
+ contrastive_std       1.84886
+ grad_norm             0.93767
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00151
+ reward_loss           0.41233
+ value_loss            0.52579
+ total_loss            0.69342
+ bc_loss               0.15781
+ entropy_loss          -0.00188
+ pi_prior_loss         0.04662
+ pi_entropy            2.27530
+ pi_scaled_entropy     18.81979
+ pi_std                0.78573
+ pi_max_std            2.26755
+ contrastive_loss      0.52285
+ contrastive_pos_logit 0.84257
+ contrastive_neg_logit -0.98326
+ contrastive_mean      -0.02606
+ contrastive_std       1.84816
+ grad_norm             0.50376
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00155
+ reward_loss           0.39902
+ value_loss            0.52228
+ total_loss            0.70428
+ bc_loss               0.16217
+ entropy_loss          -0.00156
+ pi_prior_loss         0.05199
+ pi_entropy            1.67820
+ pi_scaled_entropy     15.58255
+ pi_std                0.77093
+ pi_max_std            2.93290
+ contrastive_loss      0.52925
+ contrastive_pos_logit 0.97850
+ contrastive_neg_logit -0.95934
+ contrastive_mean      -0.02979
+ contrastive_std       1.88737
+ grad_norm             0.60612
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00207
+ reward_loss           0.51395
+ value_loss            0.65641
+ total_loss            0.75321
+ bc_loss               0.15766
+ entropy_loss          -0.00170
+ pi_prior_loss         0.04875
+ pi_entropy            1.00188
+ pi_scaled_entropy     16.98432
+ pi_std                0.75847
+ pi_max_std            4.29939
+ contrastive_loss      0.54596
+ contrastive_pos_logit 0.85145
+ contrastive_neg_logit -0.76286
+ contrastive_mean      0.00489
+ contrastive_std       1.54713
+ grad_norm             0.67703
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00206
+ reward_loss           0.48397
+ value_loss            0.61864
+ total_loss            0.76904
+ bc_loss               0.12850
+ entropy_loss          -0.00148
+ pi_prior_loss         0.04050
+ pi_entropy            0.75657
+ pi_scaled_entropy     14.76664
+ pi_std                0.74742
+ pi_max_std            5.44932
+ contrastive_loss      0.57699
+ contrastive_pos_logit 0.54443
+ contrastive_neg_logit -0.70636
+ contrastive_mean      0.00643
+ contrastive_std       1.41769
+ grad_norm             0.82272
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00182
+ reward_loss           0.45756
+ value_loss            0.56549
+ total_loss            0.73838
+ bc_loss               0.13289
+ entropy_loss          -0.00150
+ pi_prior_loss         0.04029
+ pi_entropy            0.72241
+ pi_scaled_entropy     14.96662
+ pi_std                0.74472
+ pi_max_std            2.31815
+ contrastive_loss      0.55940
+ contrastive_pos_logit 0.65299
+ contrastive_neg_logit -0.62610
+ contrastive_mean      0.00556
+ contrastive_std       1.40730
+ grad_norm             0.71446
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00200
+ reward_loss           0.50762
+ value_loss            0.64698
+ total_loss            0.76427
+ bc_loss               0.11413
+ entropy_loss          -0.00140
+ pi_prior_loss         0.03480
+ pi_entropy            0.73301
+ pi_scaled_entropy     14.01148
+ pi_std                0.75582
+ pi_max_std            6.17229
+ contrastive_loss      0.57410
+ contrastive_pos_logit 0.63857
+ contrastive_neg_logit -0.66174
+ contrastive_mean      0.00268
+ contrastive_std       1.42234
+ grad_norm             0.59466
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00186
+ reward_loss           0.45367
+ value_loss            0.53240
+ total_loss            0.75426
+ bc_loss               0.12570
+ entropy_loss          -0.00145
+ pi_prior_loss         0.03836
+ pi_entropy            0.99112
+ pi_scaled_entropy     14.49046
+ pi_std                0.76392
+ pi_max_std            4.16374
+ contrastive_loss      0.58011
+ contrastive_pos_logit 0.62795
+ contrastive_neg_logit -0.58038
+ contrastive_mean      0.00352
+ contrastive_std       1.45118
+ grad_norm             0.64345
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00227
+ reward_loss           0.50325
+ value_loss            0.66261
+ total_loss            0.78265
+ bc_loss               0.12116
+ entropy_loss          -0.00150
+ pi_prior_loss         0.04005
+ pi_entropy            -0.09769
+ pi_scaled_entropy     15.03090
+ pi_std                0.74821
+ pi_max_std            6.42646
+ contrastive_loss      0.58067
+ contrastive_pos_logit 0.64783
+ contrastive_neg_logit -0.62724
+ contrastive_mean      0.00027
+ contrastive_std       1.47970
+ grad_norm             1.06581
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00201
+ reward_loss           0.46791
+ value_loss            0.58735
+ total_loss            0.72808
+ bc_loss               0.12404
+ entropy_loss          -0.00168
+ pi_prior_loss         0.03803
+ pi_entropy            1.18617
+ pi_scaled_entropy     16.82698
+ pi_std                0.77043
+ pi_max_std            6.80987
+ contrastive_loss      0.54422
+ contrastive_pos_logit 0.77575
+ contrastive_neg_logit -0.61826
+ contrastive_mean      -0.00020
+ contrastive_std       1.49326
+ grad_norm             0.80897
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00214
+ reward_loss           0.51148
+ value_loss            0.56932
+ total_loss            0.71622
+ bc_loss               0.12203
+ entropy_loss          -0.00138
+ pi_prior_loss         0.03894
+ pi_entropy            0.65649
+ pi_scaled_entropy     13.77774
+ pi_std                0.74909
+ pi_max_std            2.54864
+ contrastive_loss      0.52638
+ contrastive_pos_logit 0.83636
+ contrastive_neg_logit -0.78305
+ contrastive_mean      -0.00340
+ contrastive_std       1.52052
+ grad_norm             1.39178
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------
+------------------------------
+Pretraining metrics:
+ consistency_loss      0.00189
+ reward_loss           0.49172
+ value_loss            0.59165
+ total_loss            0.72787
+ bc_loss               0.12664
+ entropy_loss          -0.00139
+ pi_prior_loss         0.04071
+ pi_entropy            0.30423
+ pi_scaled_entropy     13.92600
+ pi_std                0.75045
+ pi_max_std            6.72984
+ contrastive_loss      0.54110
+ contrastive_pos_logit 0.90457
+ contrastive_neg_logit -0.67507
+ contrastive_mean      0.00001
+ contrastive_std       1.56416
+ grad_norm             1.03444
+ lr_enc                0.00009
+ lr                    0.00030
+ lr_pi                 0.00030
+------------------------------

soup_XL_100M_work_dir/wandb/run-20260623_100535-g9l145ew/files/requirements.txt

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

soup_XL_100M_work_dir/wandb/run-20260623_100535-g9l145ew/files/wandb-metadata.json

@@ -0,0 +1,79 @@
+{
+  "os":  "Linux-5.15.0-72-generic-x86_64-with-glibc2.35",
+  "python":  "CPython 3.11.15",
+  "startedAt":  "2026-06-23T02:05:35.256271Z",
+  "args":  [
+    "task=soup",
+    "model_size=XL",
+    "steps=100000000",
+    "demo_steps=200000",
+    "train_eval_freq=5000000",
+    "checkpoint_save_freq=5000000",
+    "compile=True",
+    "diffusion_compile=True",
+    "multiproc=True",
+    "use_score_network=False",
+    "eval_at_start=False",
+    "env_mode=async",
+    "planner_type=diffusion",
+    "exp_name=soup_XL_100M",
+    "seed=80",
+    "work_dir=/media/datasets/cheliu21/cxy_worldmodel/newt/soup_XL_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":  "c48718ddb1e373c92801c0f36a1aa23d5300c980"
+  },
+  "email":  "wenxuan.yuan@qq.com",
+  "root":  "/media/datasets/cheliu21/cxy_worldmodel/newt/soup_XL_100M_work_dir",
+  "host":  "job-mx9zwvdf8b4m-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":  "528147169280",
+      "used":  "342780809216"
+    }
+  },
+  "memory":  {
+    "total":  "2162270285824"
+  },
+  "gpu_nvidia":  [
+    {
+      "name":  "NVIDIA A800-SXM4-80GB",
+      "memoryTotal":  "85899345920",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-e33a5d5c-ece1-16b1-ae70-38577ceaea24"
+    },
+    {
+      "name":  "NVIDIA A800-SXM4-80GB",
+      "memoryTotal":  "85899345920",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-cd62cf34-2832-7925-6fb2-50d0de797c7e"
+    },
+    {
+      "name":  "NVIDIA A800-SXM4-80GB",
+      "memoryTotal":  "85899345920",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-f15c4618-c2e7-f14e-f27e-623dda73692a"
+    },
+    {
+      "name":  "NVIDIA A800-SXM4-80GB",
+      "memoryTotal":  "85899345920",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-4f0f1b5f-f5bc-5f62-8cc8-a837e6d1f600"
+    }
+  ],
+  "cudaVersion":  "12.4",
+  "writerId":  "a1u3ipa3gcrnsm9xeh1vqzqpwmoccbed"
+}

soup_XL_100M_work_dir/wandb/run-20260623_100535-g9l145ew/logs/debug-internal.log

@@ -0,0 +1,10 @@
+{"time":"2026-06-23T10:05:35.531478601+08:00","level":"INFO","msg":"stream: starting","core version":"0.22.1"}
+{"time":"2026-06-23T10:05:36.061131519+08:00","level":"INFO","msg":"stream: created new stream","id":"g9l145ew"}
+{"time":"2026-06-23T10:05:36.061182108+08:00","level":"INFO","msg":"handler: started","stream_id":"g9l145ew"}
+{"time":"2026-06-23T10:05:36.06148986+08:00","level":"INFO","msg":"stream: started","id":"g9l145ew"}
+{"time":"2026-06-23T10:05:36.061502919+08:00","level":"INFO","msg":"writer: started","stream_id":"g9l145ew"}
+{"time":"2026-06-23T10:05:36.061503589+08:00","level":"INFO","msg":"sender: started","stream_id":"g9l145ew"}
+{"time":"2026-06-23T10:25:24.014381245+08:00","level":"INFO","msg":"flowcontrol: backed up, offloading to disk","recordNumber":6339}
+{"time":"2026-06-23T10:25:24.054476658+08:00","level":"INFO","msg":"flowcontrol: unblocked","totalOffloaded":13}
+{"time":"2026-06-23T14:12:24.105579943+08:00","level":"INFO","msg":"flowcontrol: backed up, offloading to disk","recordNumber":81739}
+{"time":"2026-06-23T14:12:24.265935776+08:00","level":"INFO","msg":"flowcontrol: unblocked","totalOffloaded":13}

soup_XL_100M_work_dir/wandb/run-20260708_090752-bs3i5bjq/files/requirements.txt

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

soup_XL_100M_work_dir/wandb/run-20260708_090752-bs3i5bjq/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":  "2026-07-08T01:07:52.060893Z",
+  "args":  [
+    "task=soup",
+    "model_size=XL",
+    "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_XL_100M",
+    "seed=5",
+    "work_dir=/media/datasets/cheliu21/cxy_worldmodel/newt/soup_XL_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_XL_100M_work_dir",
+  "host":  "job-xp37erxymhbm-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":  "528147169280",
+      "used":  "316705570816"
+    }
+  },
+  "memory":  {
+    "total":  "2162270285824"
+  },
+  "gpu_nvidia":  [
+    {
+      "name":  "NVIDIA A800-SXM4-80GB",
+      "memoryTotal":  "85899345920",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-e33a5d5c-ece1-16b1-ae70-38577ceaea24"
+    },
+    {
+      "name":  "NVIDIA A800-SXM4-80GB",
+      "memoryTotal":  "85899345920",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-cd62cf34-2832-7925-6fb2-50d0de797c7e"
+    },
+    {
+      "name":  "NVIDIA A800-SXM4-80GB",
+      "memoryTotal":  "85899345920",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-f15c4618-c2e7-f14e-f27e-623dda73692a"
+    },
+    {
+      "name":  "NVIDIA A800-SXM4-80GB",
+      "memoryTotal":  "85899345920",
+      "cudaCores":  6912,
+      "architecture":  "Ampere",
+      "uuid":  "GPU-4f0f1b5f-f5bc-5f62-8cc8-a837e6d1f600"
+    }
+  ],
+  "cudaVersion":  "12.4",
+  "writerId":  "txxjku83d69hgbel7az1qjfuqjpo139f"
+}

soup_XL_100M_work_dir/wandb/run-20260708_090752-bs3i5bjq/logs/debug-internal.log

@@ -0,0 +1,13 @@
+{"time":"2026-07-08T09:07:52.33321276+08:00","level":"INFO","msg":"stream: starting","core version":"0.22.1"}
+{"time":"2026-07-08T09:07:52.879779385+08:00","level":"INFO","msg":"stream: created new stream","id":"bs3i5bjq"}
+{"time":"2026-07-08T09:07:52.879841771+08:00","level":"INFO","msg":"handler: started","stream_id":"bs3i5bjq"}
+{"time":"2026-07-08T09:07:52.88015593+08:00","level":"INFO","msg":"stream: started","id":"bs3i5bjq"}
+{"time":"2026-07-08T09:07:52.880170829+08:00","level":"INFO","msg":"sender: started","stream_id":"bs3i5bjq"}
+{"time":"2026-07-08T09:07:52.880173317+08:00","level":"INFO","msg":"writer: started","stream_id":"bs3i5bjq"}
+{"time":"2026-07-08T10:34:11.115390756+08:00","level":"INFO","msg":"flowcontrol: backed up, offloading to disk","recordNumber":28926}
+{"time":"2026-07-08T10:34:11.158966111+08:00","level":"INFO","msg":"flowcontrol: unblocked","totalOffloaded":13}
+{"time":"2026-07-08T14:26:17.46615665+08:00","level":"INFO","msg":"api: retrying HTTP error","status":429,"url":"https://api.wandb.ai/files/wenxuan-yuan-imperial-college-london/newt/bs3i5bjq/file_stream","body":"{\"error\":\"rate limit exceeded: per_run limit on filestream requests\"}"}
+{"time":"2026-07-08T14:26:26.12017532+08:00","level":"INFO","msg":"api: retrying HTTP error","status":429,"url":"https://api.wandb.ai/files/wenxuan-yuan-imperial-college-london/newt/bs3i5bjq/file_stream","body":"{\"error\":\"rate limit exceeded: per_run limit on filestream requests\"}"}
+{"time":"2026-07-08T14:26:33.988391415+08:00","level":"INFO","msg":"api: retrying HTTP error","status":429,"url":"https://api.wandb.ai/files/wenxuan-yuan-imperial-college-london/newt/bs3i5bjq/file_stream","body":"{\"error\":\"rate limit exceeded: per_run limit on filestream requests\"}"}
+{"time":"2026-07-08T14:26:43.19340454+08:00","level":"INFO","msg":"api: retrying HTTP error","status":429,"url":"https://api.wandb.ai/files/wenxuan-yuan-imperial-college-london/newt/bs3i5bjq/file_stream","body":"{\"error\":\"rate limit exceeded: per_run limit on filestream requests\"}"}
+{"time":"2026-07-08T14:27:02.94526155+08:00","level":"INFO","msg":"api: retrying HTTP error","status":429,"url":"https://api.wandb.ai/files/wenxuan-yuan-imperial-college-london/newt/bs3i5bjq/file_stream","body":"{\"error\":\"rate limit exceeded: per_run limit on filestream requests\"}"}

soup_XL_100M_work_dir/wandb/run-20260708_090752-bs3i5bjq/run-bs3i5bjq.wandb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d1eb9e010c63c1825f3288ff899e7717ff9be5a74cfd9915a09115b5b018688e
+size 50298880