BruceYuan commited on
Commit
e28f7d2
·
verified ·
1 Parent(s): 1f7e341

sync new results for soup_XL_100M_work_dir

Browse files
.gitattributes CHANGED
@@ -53,3 +53,4 @@ soup_B_100M_work_dir/wandb/run-20260708_090206-4s2ksbsn/run-4s2ksbsn.wandb filte
53
  soup_S_100M_work_dir/wandb/run-20260708_114417-8pbrlu7w/run-8pbrlu7w.wandb filter=lfs diff=lfs merge=lfs -text
54
  soup_S_100M_work_dir/wandb/run-20260710_081650-8qh9ccql/run-8qh9ccql.wandb filter=lfs diff=lfs merge=lfs -text
55
  soup_S_100M_work_dir/wandb/run-20260710_200207-wzv4r5ph/run-wzv4r5ph.wandb filter=lfs diff=lfs merge=lfs -text
 
 
53
  soup_S_100M_work_dir/wandb/run-20260708_114417-8pbrlu7w/run-8pbrlu7w.wandb filter=lfs diff=lfs merge=lfs -text
54
  soup_S_100M_work_dir/wandb/run-20260710_081650-8qh9ccql/run-8qh9ccql.wandb filter=lfs diff=lfs merge=lfs -text
55
  soup_S_100M_work_dir/wandb/run-20260710_200207-wzv4r5ph/run-wzv4r5ph.wandb filter=lfs diff=lfs merge=lfs -text
56
+ 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 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:57fbb768e799e4e0904604b4d2401ccd7c64c3b3d3192ed3c36d05df6d20417b
3
+ size 1254669611
soup_XL_100M_work_dir/ckpt/10_000_000.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bb0187b6c14f3ca979dc66db18624714dcebb5894a9af02cd893a6227ef4664c
3
+ size 1308819645
soup_XL_100M_work_dir/ckpt/15_000_000.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9b006226dfd27137c8466601ff224698c11f400114e932d0f41aabc3cd6b9186
3
+ size 1308819645
soup_XL_100M_work_dir/ckpt/20_000_000.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f344d9c3d21faae93471da63d782025e589262454ed47cd2705beb6275823d8c
3
+ size 1308819645
soup_XL_100M_work_dir/ckpt/5_000_000.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:38d644bac8d8dae5accee808cae303c51d282d22f541d9d9c41ae00533d202f9
3
+ size 1308819019
soup_XL_100M_work_dir/torch_eval_metrics.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
soup_XL_100M_work_dir/wandb/debug-internal.log ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {"time":"2026-07-08T09:07:52.33321276+08:00","level":"INFO","msg":"stream: starting","core version":"0.22.1"}
2
+ {"time":"2026-07-08T09:07:52.879779385+08:00","level":"INFO","msg":"stream: created new stream","id":"bs3i5bjq"}
3
+ {"time":"2026-07-08T09:07:52.879841771+08:00","level":"INFO","msg":"handler: started","stream_id":"bs3i5bjq"}
4
+ {"time":"2026-07-08T09:07:52.88015593+08:00","level":"INFO","msg":"stream: started","id":"bs3i5bjq"}
5
+ {"time":"2026-07-08T09:07:52.880170829+08:00","level":"INFO","msg":"sender: started","stream_id":"bs3i5bjq"}
6
+ {"time":"2026-07-08T09:07:52.880173317+08:00","level":"INFO","msg":"writer: started","stream_id":"bs3i5bjq"}
7
+ {"time":"2026-07-08T10:34:11.115390756+08:00","level":"INFO","msg":"flowcontrol: backed up, offloading to disk","recordNumber":28926}
8
+ {"time":"2026-07-08T10:34:11.158966111+08:00","level":"INFO","msg":"flowcontrol: unblocked","totalOffloaded":13}
9
+ {"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\"}"}
10
+ {"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\"}"}
11
+ {"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\"}"}
12
+ {"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\"}"}
13
+ {"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/debug.log ADDED
The diff for this file is too large to render. See raw diff
 
soup_XL_100M_work_dir/wandb/run-20260623_100535-g9l145ew/files/output.log ADDED
@@ -0,0 +1,1180 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ Logs will be synced with wandb.
2
+ Architecture: DDPWrapper(
3
+ (_module): DistributedDataParallel(
4
+ (module): Newt World Model
5
+ Encoder (11,161,664): ModuleDict(
6
+ (state): Sequential(
7
+ (0): NormedLinear(in_features=640, out_features=2048, bias=True, act=Mish)
8
+ (1): NormedLinear(in_features=2048, out_features=2048, bias=True, act=Mish)
9
+ (2): NormedLinear(in_features=2048, out_features=2048, bias=True, act=Mish)
10
+ (3): NormedLinear(in_features=2048, out_features=704, bias=True, act=SimNorm)
11
+ )
12
+ )
13
+ Dynamics (8,173,632): Sequential(
14
+ (0): NormedLinear(in_features=1232, out_features=2048, bias=True, act=Mish)
15
+ (1): NormedLinear(in_features=2048, out_features=2048, bias=True, act=Mish)
16
+ (2): NormedLinear(in_features=2048, out_features=704, bias=True, act=SimNorm)
17
+ )
18
+ Reward (6,936,677): Sequential(
19
+ (0): NormedLinear(in_features=1232, out_features=2048, bias=True, act=Mish)
20
+ (1): NormedLinear(in_features=2048, out_features=2048, bias=True, act=Mish)
21
+ (2): Linear(in_features=2048, out_features=101, bias=True)
22
+ )
23
+ Contrastive F (6,731,777): Sequential(
24
+ (0): NormedLinear(in_features=1232, out_features=2048, bias=True, act=Mish)
25
+ (1): NormedLinear(in_features=2048, out_features=2048, bias=True, act=Mish)
26
+ (2): Linear(in_features=2048, out_features=1, bias=True)
27
+ )
28
+ Policy prior (6,762,528): Sequential(
29
+ (0): NormedLinear(in_features=1216, out_features=2048, bias=True, act=Mish)
30
+ (1): NormedLinear(in_features=2048, out_features=2048, bias=True, act=Mish)
31
+ (2): Linear(in_features=2048, out_features=32, bias=True)
32
+ )
33
+ Q-functions (48,556,739): QEnsemble(
34
+ (_Qs): ModuleList(
35
+ (0-6): 7 x Sequential(
36
+ (0): NormedLinear(in_features=1232, out_features=2048, bias=True, act=Mish)
37
+ (1): NormedLinear(in_features=2048, out_features=2048, bias=True, act=Mish)
38
+ (2): Linear(in_features=2048, out_features=101, bias=True)
39
+ )
40
+ )
41
+ )
42
+ Learnable parameters: 88,323,017
43
+ )
44
+ )
45
+ Update frequency: 200,000
46
+ Episodes per update frequency: 1,933
47
+ No checkpoint found, training from scratch.
48
+ Pretraining agent on demonstrations...
49
+ prior_coef is 10.0, setting to 1.0 for pretraining.
50
+ 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
51
+ ------------------------------
52
+ Pretraining metrics:
53
+ consistency_loss 0.02467
54
+ reward_loss 4.30745
55
+ value_loss 4.30745
56
+ total_loss 2.33911
57
+ bc_loss 0.90417
58
+ entropy_loss -0.00223
59
+ pi_prior_loss 0.29106
60
+ pi_entropy 2.72638
61
+ pi_scaled_entropy 22.25638
62
+ pi_std 0.99740
63
+ pi_max_std 5.66591
64
+ contrastive_loss 0.69315
65
+ contrastive_pos_logit 0.00000
66
+ contrastive_neg_logit 0.00000
67
+ contrastive_mean 0.00000
68
+ contrastive_std 0.99000
69
+ grad_norm 3.97725
70
+ lr_enc 0.00000
71
+ lr 0.00000
72
+ lr_pi 0.00000
73
+ ------------------------------
74
+ [rank0]:V0623 10:06:20.452000 298 site-packages/torch/_dynamo/guards.py:3508] [0/1] [__recompiles] triggered by the following guard failure(s):
75
+ [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
76
+ Pretraining: 4%|▍ | 7999/200000 [13:26<5:04:03, 10.52it/s]
77
+ ------------------------------
78
+ Pretraining metrics:
79
+ consistency_loss 0.00265
80
+ reward_loss 0.49153
81
+ value_loss 0.49874
82
+ total_loss 0.87740
83
+ bc_loss 0.25458
84
+ entropy_loss -0.00088
85
+ pi_prior_loss 0.08286
86
+ pi_entropy 2.00650
87
+ pi_scaled_entropy 8.76924
88
+ pi_std 0.76371
89
+ pi_max_std 1.03175
90
+ contrastive_loss 0.64243
91
+ contrastive_pos_logit 0.18010
92
+ contrastive_neg_logit -0.21694
93
+ contrastive_mean 0.01501
94
+ contrastive_std 0.63365
95
+ grad_norm 0.78231
96
+ lr_enc 0.00007
97
+ lr 0.00024
98
+ lr_pi 0.00024
99
+ ------------------------------
100
+ ------------------------------
101
+ Pretraining metrics:
102
+ consistency_loss 0.00228
103
+ reward_loss 0.54483
104
+ value_loss 0.49017
105
+ total_loss 0.87847
106
+ bc_loss 0.25321
107
+ entropy_loss -0.00117
108
+ pi_prior_loss 0.07979
109
+ pi_entropy 2.47168
110
+ pi_scaled_entropy 11.73463
111
+ pi_std 0.76287
112
+ pi_max_std 1.00000
113
+ contrastive_loss 0.64949
114
+ contrastive_pos_logit 0.37591
115
+ contrastive_neg_logit -0.07815
116
+ contrastive_mean 0.02642
117
+ contrastive_std 0.71706
118
+ grad_norm 0.74900
119
+ lr_enc 0.00009
120
+ lr 0.00030
121
+ lr_pi 0.00030
122
+ ------------------------------
123
+ ------------------------------
124
+ Pretraining metrics:
125
+ consistency_loss 0.00208
126
+ reward_loss 0.52917
127
+ value_loss 0.90180
128
+ total_loss 0.90592
129
+ bc_loss 0.25956
130
+ entropy_loss -0.00073
131
+ pi_prior_loss 0.08144
132
+ pi_entropy 2.27827
133
+ pi_scaled_entropy 7.33548
134
+ pi_std 0.76353
135
+ pi_max_std 1.00000
136
+ contrastive_loss 0.63975
137
+ contrastive_pos_logit 0.38069
138
+ contrastive_neg_logit -0.11297
139
+ contrastive_mean 0.03023
140
+ contrastive_std 0.78676
141
+ grad_norm 0.54999
142
+ lr_enc 0.00009
143
+ lr 0.00030
144
+ lr_pi 0.00030
145
+ ------------------------------
146
+ ------------------------------
147
+ Pretraining metrics:
148
+ consistency_loss 0.00208
149
+ reward_loss 0.48864
150
+ value_loss 0.53216
151
+ total_loss 0.83344
152
+ bc_loss 0.20776
153
+ entropy_loss -0.00098
154
+ pi_prior_loss 0.06697
155
+ pi_entropy 2.43512
156
+ pi_scaled_entropy 9.79741
157
+ pi_std 0.76530
158
+ pi_max_std 1.20144
159
+ contrastive_loss 0.62286
160
+ contrastive_pos_logit 0.36886
161
+ contrastive_neg_logit -0.15992
162
+ contrastive_mean 0.03071
163
+ contrastive_std 0.81851
164
+ grad_norm 0.40577
165
+ lr_enc 0.00009
166
+ lr 0.00030
167
+ lr_pi 0.00030
168
+ ------------------------------
169
+ ------------------------------
170
+ Pretraining metrics:
171
+ consistency_loss 0.00196
172
+ reward_loss 0.49436
173
+ value_loss 0.54299
174
+ total_loss 0.82382
175
+ bc_loss 0.20190
176
+ entropy_loss -0.00094
177
+ pi_prior_loss 0.06465
178
+ pi_entropy 1.89240
179
+ pi_scaled_entropy 9.39937
180
+ pi_std 0.75816
181
+ pi_max_std 1.20815
182
+ contrastive_loss 0.61620
183
+ contrastive_pos_logit 0.38863
184
+ contrastive_neg_logit -0.24144
185
+ contrastive_mean 0.03054
186
+ contrastive_std 0.88668
187
+ grad_norm 0.44075
188
+ lr_enc 0.00009
189
+ lr 0.00030
190
+ lr_pi 0.00030
191
+ ------------------------------
192
+ ------------------------------
193
+ Pretraining metrics:
194
+ consistency_loss 0.00231
195
+ reward_loss 0.45311
196
+ value_loss 0.55078
197
+ total_loss 0.83577
198
+ bc_loss 0.22282
199
+ entropy_loss -0.00128
200
+ pi_prior_loss 0.07127
201
+ pi_entropy 2.54053
202
+ pi_scaled_entropy 12.77746
203
+ pi_std 0.77247
204
+ pi_max_std 1.38735
205
+ contrastive_loss 0.61788
206
+ contrastive_pos_logit 0.27619
207
+ contrastive_neg_logit -0.33753
208
+ contrastive_mean 0.02916
209
+ contrastive_std 0.91868
210
+ grad_norm 0.51016
211
+ lr_enc 0.00009
212
+ lr 0.00030
213
+ lr_pi 0.00030
214
+ ------------------------------
215
+ ------------------------------
216
+ Pretraining metrics:
217
+ consistency_loss 0.00226
218
+ reward_loss 0.50228
219
+ value_loss 0.45888
220
+ total_loss 0.82558
221
+ bc_loss 0.20616
222
+ entropy_loss -0.00129
223
+ pi_prior_loss 0.06469
224
+ pi_entropy 2.34219
225
+ pi_scaled_entropy 12.87433
226
+ pi_std 0.76522
227
+ pi_max_std 1.05497
228
+ contrastive_loss 0.61958
229
+ contrastive_pos_logit 0.40802
230
+ contrastive_neg_logit -0.30381
231
+ contrastive_mean 0.03019
232
+ contrastive_std 0.96939
233
+ grad_norm 0.41051
234
+ lr_enc 0.00009
235
+ lr 0.00030
236
+ lr_pi 0.00030
237
+ ------------------------------
238
+ ------------------------------
239
+ Pretraining metrics:
240
+ consistency_loss 0.00227
241
+ reward_loss 0.49466
242
+ value_loss 0.55087
243
+ total_loss 0.81287
244
+ bc_loss 0.19208
245
+ entropy_loss -0.00124
246
+ pi_prior_loss 0.06074
247
+ pi_entropy 2.29622
248
+ pi_scaled_entropy 12.39907
249
+ pi_std 0.76821
250
+ pi_max_std 2.06816
251
+ contrastive_loss 0.60227
252
+ contrastive_pos_logit 0.44011
253
+ contrastive_neg_logit -0.41351
254
+ contrastive_mean 0.02361
255
+ contrastive_std 1.03155
256
+ grad_norm 0.46515
257
+ lr_enc 0.00009
258
+ lr 0.00030
259
+ lr_pi 0.00030
260
+ ------------------------------
261
+ ------------------------------
262
+ Pretraining metrics:
263
+ consistency_loss 0.00236
264
+ reward_loss 0.45463
265
+ value_loss 0.57505
266
+ total_loss 0.83072
267
+ bc_loss 0.20341
268
+ entropy_loss -0.00143
269
+ pi_prior_loss 0.06346
270
+ pi_entropy 2.57812
271
+ pi_scaled_entropy 14.29637
272
+ pi_std 0.77729
273
+ pi_max_std 1.64060
274
+ contrastive_loss 0.61704
275
+ contrastive_pos_logit 0.45206
276
+ contrastive_neg_logit -0.28638
277
+ contrastive_mean 0.02154
278
+ contrastive_std 1.09277
279
+ grad_norm 0.50279
280
+ lr_enc 0.00009
281
+ lr 0.00030
282
+ lr_pi 0.00030
283
+ ------------------------------
284
+ ------------------------------
285
+ Pretraining metrics:
286
+ consistency_loss 0.00238
287
+ reward_loss 0.44701
288
+ value_loss 0.52478
289
+ total_loss 0.81326
290
+ bc_loss 0.23596
291
+ entropy_loss -0.00130
292
+ pi_prior_loss 0.07654
293
+ pi_entropy 2.69820
294
+ pi_scaled_entropy 13.01992
295
+ pi_std 0.77659
296
+ pi_max_std 1.25145
297
+ contrastive_loss 0.59197
298
+ contrastive_pos_logit 0.39765
299
+ contrastive_neg_logit -0.49633
300
+ contrastive_mean 0.02081
301
+ contrastive_std 1.13147
302
+ grad_norm 0.40804
303
+ lr_enc 0.00009
304
+ lr 0.00030
305
+ lr_pi 0.00030
306
+ ------------------------------
307
+ ------------------------------
308
+ Pretraining metrics:
309
+ consistency_loss 0.00256
310
+ reward_loss 0.48533
311
+ value_loss 0.53448
312
+ total_loss 0.81000
313
+ bc_loss 0.19828
314
+ entropy_loss -0.00112
315
+ pi_prior_loss 0.06219
316
+ pi_entropy 1.72597
317
+ pi_scaled_entropy 11.24058
318
+ pi_std 0.75782
319
+ pi_max_std 1.56250
320
+ contrastive_loss 0.59457
321
+ contrastive_pos_logit 0.45619
322
+ contrastive_neg_logit -0.44733
323
+ contrastive_mean 0.01845
324
+ contrastive_std 1.16010
325
+ grad_norm 0.52094
326
+ lr_enc 0.00009
327
+ lr 0.00030
328
+ lr_pi 0.00030
329
+ ------------------------------
330
+ ------------------------------
331
+ Pretraining metrics:
332
+ consistency_loss 0.00221
333
+ reward_loss 0.46861
334
+ value_loss 0.46767
335
+ total_loss 0.81188
336
+ bc_loss 0.20666
337
+ entropy_loss -0.00150
338
+ pi_prior_loss 0.06662
339
+ pi_entropy 3.15087
340
+ pi_scaled_entropy 15.03450
341
+ pi_std 0.77937
342
+ pi_max_std 1.25600
343
+ contrastive_loss 0.60739
344
+ contrastive_pos_logit 0.44643
345
+ contrastive_neg_logit -0.39911
346
+ contrastive_mean 0.01690
347
+ contrastive_std 1.22122
348
+ grad_norm 0.46720
349
+ lr_enc 0.00009
350
+ lr 0.00030
351
+ lr_pi 0.00030
352
+ ------------------------------
353
+ ------------------------------
354
+ Pretraining metrics:
355
+ consistency_loss 0.00250
356
+ reward_loss 0.44491
357
+ value_loss 0.55384
358
+ total_loss 0.80343
359
+ bc_loss 0.17535
360
+ entropy_loss -0.00128
361
+ pi_prior_loss 0.05722
362
+ pi_entropy 2.14758
363
+ pi_scaled_entropy 12.75763
364
+ pi_std 0.77828
365
+ pi_max_std 2.39449
366
+ contrastive_loss 0.59635
367
+ contrastive_pos_logit 0.50806
368
+ contrastive_neg_logit -0.48557
369
+ contrastive_mean 0.01545
370
+ contrastive_std 1.23719
371
+ grad_norm 0.63250
372
+ lr_enc 0.00009
373
+ lr 0.00030
374
+ lr_pi 0.00030
375
+ ------------------------------
376
+ ------------------------------
377
+ Pretraining metrics:
378
+ consistency_loss 0.00292
379
+ reward_loss 0.46144
380
+ value_loss 0.47883
381
+ total_loss 0.80751
382
+ bc_loss 0.19689
383
+ entropy_loss -0.00145
384
+ pi_prior_loss 0.06260
385
+ pi_entropy 2.44735
386
+ pi_scaled_entropy 14.48672
387
+ pi_std 0.77586
388
+ pi_max_std 2.20730
389
+ contrastive_loss 0.59246
390
+ contrastive_pos_logit 0.62278
391
+ contrastive_neg_logit -0.36566
392
+ contrastive_mean 0.01638
393
+ contrastive_std 1.26936
394
+ grad_norm 0.58808
395
+ lr_enc 0.00009
396
+ lr 0.00030
397
+ lr_pi 0.00030
398
+ ------------------------------
399
+ ------------------------------
400
+ Pretraining metrics:
401
+ consistency_loss 0.00293
402
+ reward_loss 0.44249
403
+ value_loss 0.48634
404
+ total_loss 0.77170
405
+ bc_loss 0.19799
406
+ entropy_loss -0.00143
407
+ pi_prior_loss 0.06052
408
+ pi_entropy 2.18767
409
+ pi_scaled_entropy 14.31973
410
+ pi_std 0.77055
411
+ pi_max_std 2.20363
412
+ contrastive_loss 0.55961
413
+ contrastive_pos_logit 0.54377
414
+ contrastive_neg_logit -0.64203
415
+ contrastive_mean 0.01358
416
+ contrastive_std 1.29616
417
+ grad_norm 0.42968
418
+ lr_enc 0.00009
419
+ lr 0.00030
420
+ lr_pi 0.00030
421
+ ------------------------------
422
+ ------------------------------
423
+ Pretraining metrics:
424
+ consistency_loss 0.00287
425
+ reward_loss 0.43705
426
+ value_loss 0.48313
427
+ total_loss 0.78217
428
+ bc_loss 0.17958
429
+ entropy_loss -0.00174
430
+ pi_prior_loss 0.05721
431
+ pi_entropy 1.76759
432
+ pi_scaled_entropy 17.39634
433
+ pi_std 0.76911
434
+ pi_max_std 1.98494
435
+ contrastive_loss 0.57556
436
+ contrastive_pos_logit 0.61791
437
+ contrastive_neg_logit -0.53700
438
+ contrastive_mean 0.01630
439
+ contrastive_std 1.31230
440
+ grad_norm 0.56348
441
+ lr_enc 0.00009
442
+ lr 0.00030
443
+ lr_pi 0.00030
444
+ ------------------------------
445
+ ------------------------------
446
+ Pretraining metrics:
447
+ consistency_loss 0.00236
448
+ reward_loss 0.44345
449
+ value_loss 0.48383
450
+ total_loss 0.75249
451
+ bc_loss 0.16993
452
+ entropy_loss -0.00149
453
+ pi_prior_loss 0.05343
454
+ pi_entropy 1.53880
455
+ pi_scaled_entropy 14.94003
456
+ pi_std 0.76094
457
+ pi_max_std 1.95329
458
+ contrastive_loss 0.55912
459
+ contrastive_pos_logit 0.84188
460
+ contrastive_neg_logit -0.57424
461
+ contrastive_mean 0.01488
462
+ contrastive_std 1.31971
463
+ grad_norm 0.87350
464
+ lr_enc 0.00009
465
+ lr 0.00030
466
+ lr_pi 0.00030
467
+ ------------------------------
468
+ ------------------------------
469
+ Pretraining metrics:
470
+ consistency_loss 0.00241
471
+ reward_loss 0.43501
472
+ value_loss 0.50453
473
+ total_loss 0.77288
474
+ bc_loss 0.19233
475
+ entropy_loss -0.00151
476
+ pi_prior_loss 0.06041
477
+ pi_entropy 2.09735
478
+ pi_scaled_entropy 15.05507
479
+ pi_std 0.77025
480
+ pi_max_std 1.85363
481
+ contrastive_loss 0.57035
482
+ contrastive_pos_logit 0.67769
483
+ contrastive_neg_logit -0.62795
484
+ contrastive_mean 0.01325
485
+ contrastive_std 1.36544
486
+ grad_norm 0.63929
487
+ lr_enc 0.00009
488
+ lr 0.00030
489
+ lr_pi 0.00030
490
+ ------------------------------
491
+ ------------------------------
492
+ Pretraining metrics:
493
+ consistency_loss 0.00195
494
+ reward_loss 0.49201
495
+ value_loss 0.56737
496
+ total_loss 0.76104
497
+ bc_loss 0.18665
498
+ entropy_loss -0.00160
499
+ pi_prior_loss 0.05807
500
+ pi_entropy 1.93475
501
+ pi_scaled_entropy 16.01232
502
+ pi_std 0.77151
503
+ pi_max_std 2.56334
504
+ contrastive_loss 0.55800
505
+ contrastive_pos_logit 0.77683
506
+ contrastive_neg_logit -0.51331
507
+ contrastive_mean 0.00966
508
+ contrastive_std 1.41337
509
+ grad_norm 0.59012
510
+ lr_enc 0.00009
511
+ lr 0.00030
512
+ lr_pi 0.00030
513
+ ------------------------------
514
+ ------------------------------
515
+ Pretraining metrics:
516
+ consistency_loss 0.00190
517
+ reward_loss 0.44406
518
+ value_loss 0.48654
519
+ total_loss 0.74514
520
+ bc_loss 0.18871
521
+ entropy_loss -0.00145
522
+ pi_prior_loss 0.05730
523
+ pi_entropy 2.42120
524
+ pi_scaled_entropy 14.52395
525
+ pi_std 0.77601
526
+ pi_max_std 2.95923
527
+ contrastive_loss 0.55682
528
+ contrastive_pos_logit 0.59401
529
+ contrastive_neg_logit -0.78931
530
+ contrastive_mean 0.00669
531
+ contrastive_std 1.44798
532
+ grad_norm 0.57467
533
+ lr_enc 0.00009
534
+ lr 0.00030
535
+ lr_pi 0.00030
536
+ ------------------------------
537
+ ------------------------------
538
+ Pretraining metrics:
539
+ consistency_loss 0.00180
540
+ reward_loss 0.50125
541
+ value_loss 0.52190
542
+ total_loss 0.75844
543
+ bc_loss 0.18673
544
+ entropy_loss -0.00156
545
+ pi_prior_loss 0.05980
546
+ pi_entropy 2.05099
547
+ pi_scaled_entropy 15.55122
548
+ pi_std 0.77235
549
+ pi_max_std 2.58286
550
+ contrastive_loss 0.56024
551
+ contrastive_pos_logit 0.67183
552
+ contrastive_neg_logit -0.61166
553
+ contrastive_mean 0.00522
554
+ contrastive_std 1.47438
555
+ grad_norm 0.63365
556
+ lr_enc 0.00009
557
+ lr 0.00030
558
+ lr_pi 0.00030
559
+ ------------------------------
560
+ ------------------------------
561
+ Pretraining metrics:
562
+ consistency_loss 0.00201
563
+ reward_loss 0.42350
564
+ value_loss 0.62962
565
+ total_loss 0.75449
566
+ bc_loss 0.18789
567
+ entropy_loss -0.00168
568
+ pi_prior_loss 0.05855
569
+ pi_entropy 1.92834
570
+ pi_scaled_entropy 16.80657
571
+ pi_std 0.77128
572
+ pi_max_std 1.68869
573
+ contrastive_loss 0.55043
574
+ contrastive_pos_logit 0.80425
575
+ contrastive_neg_logit -0.57811
576
+ contrastive_mean 0.00663
577
+ contrastive_std 1.50137
578
+ grad_norm 0.55352
579
+ lr_enc 0.00009
580
+ lr 0.00030
581
+ lr_pi 0.00030
582
+ ------------------------------
583
+ ------------------------------
584
+ Pretraining metrics:
585
+ consistency_loss 0.00162
586
+ reward_loss 0.45005
587
+ value_loss 0.47480
588
+ total_loss 0.74196
589
+ bc_loss 0.16953
590
+ entropy_loss -0.00175
591
+ pi_prior_loss 0.05332
592
+ pi_entropy 2.15380
593
+ pi_scaled_entropy 17.48677
594
+ pi_std 0.78500
595
+ pi_max_std 2.47720
596
+ contrastive_loss 0.56380
597
+ contrastive_pos_logit 0.79463
598
+ contrastive_neg_logit -0.62770
599
+ contrastive_mean 0.00302
600
+ contrastive_std 1.52429
601
+ grad_norm 0.61682
602
+ lr_enc 0.00009
603
+ lr 0.00030
604
+ lr_pi 0.00030
605
+ ------------------------------
606
+ ------------------------------
607
+ Pretraining metrics:
608
+ consistency_loss 0.00161
609
+ reward_loss 0.44772
610
+ value_loss 0.55312
611
+ total_loss 0.73870
612
+ bc_loss 0.19995
613
+ entropy_loss -0.00148
614
+ pi_prior_loss 0.06279
615
+ pi_entropy 2.09332
616
+ pi_scaled_entropy 14.79656
617
+ pi_std 0.77205
618
+ pi_max_std 2.17110
619
+ contrastive_loss 0.54353
620
+ contrastive_pos_logit 0.86714
621
+ contrastive_neg_logit -0.63477
622
+ contrastive_mean 0.00245
623
+ contrastive_std 1.54013
624
+ grad_norm 0.58738
625
+ lr_enc 0.00009
626
+ lr 0.00030
627
+ lr_pi 0.00030
628
+ ------------------------------
629
+ ------------------------------
630
+ Pretraining metrics:
631
+ consistency_loss 0.00162
632
+ reward_loss 0.40333
633
+ value_loss 0.53891
634
+ total_loss 0.74960
635
+ bc_loss 0.18556
636
+ entropy_loss -0.00170
637
+ pi_prior_loss 0.05662
638
+ pi_entropy 1.99021
639
+ pi_scaled_entropy 17.02247
640
+ pi_std 0.76803
641
+ pi_max_std 2.14561
642
+ contrastive_loss 0.56625
643
+ contrastive_pos_logit 0.66684
644
+ contrastive_neg_logit -0.80069
645
+ contrastive_mean -0.00048
646
+ contrastive_std 1.57910
647
+ grad_norm 0.65196
648
+ lr_enc 0.00009
649
+ lr 0.00030
650
+ lr_pi 0.00030
651
+ ------------------------------
652
+ ------------------------------
653
+ Pretraining metrics:
654
+ consistency_loss 0.00146
655
+ reward_loss 0.42006
656
+ value_loss 0.48144
657
+ total_loss 0.71740
658
+ bc_loss 0.18199
659
+ entropy_loss -0.00167
660
+ pi_prior_loss 0.05630
661
+ pi_entropy 1.56013
662
+ pi_scaled_entropy 16.66305
663
+ pi_std 0.76846
664
+ pi_max_std 3.45237
665
+ contrastive_loss 0.54170
666
+ contrastive_pos_logit 0.80791
667
+ contrastive_neg_logit -0.69993
668
+ contrastive_mean -0.00374
669
+ contrastive_std 1.59853
670
+ grad_norm 0.76680
671
+ lr_enc 0.00009
672
+ lr 0.00030
673
+ lr_pi 0.00030
674
+ ------------------------------
675
+ ------------------------------
676
+ Pretraining metrics:
677
+ consistency_loss 0.00168
678
+ reward_loss 0.43532
679
+ value_loss 0.50606
680
+ total_loss 0.74382
681
+ bc_loss 0.17647
682
+ entropy_loss -0.00174
683
+ pi_prior_loss 0.05552
684
+ pi_entropy 2.37219
685
+ pi_scaled_entropy 17.35526
686
+ pi_std 0.78565
687
+ pi_max_std 1.85732
688
+ contrastive_loss 0.56055
689
+ contrastive_pos_logit 0.67123
690
+ contrastive_neg_logit -0.77808
691
+ contrastive_mean -0.00751
692
+ contrastive_std 1.62419
693
+ grad_norm 0.64379
694
+ lr_enc 0.00009
695
+ lr 0.00030
696
+ lr_pi 0.00030
697
+ ------------------------------
698
+ ------------------------------
699
+ Pretraining metrics:
700
+ consistency_loss 0.00149
701
+ reward_loss 0.43328
702
+ value_loss 0.52474
703
+ total_loss 0.72200
704
+ bc_loss 0.17227
705
+ entropy_loss -0.00154
706
+ pi_prior_loss 0.05387
707
+ pi_entropy 1.72626
708
+ pi_scaled_entropy 15.40534
709
+ pi_std 0.76947
710
+ pi_max_std 3.13370
711
+ contrastive_loss 0.54255
712
+ contrastive_pos_logit 0.72010
713
+ contrastive_neg_logit -0.86726
714
+ contrastive_mean -0.00482
715
+ contrastive_std 1.64503
716
+ grad_norm 0.64796
717
+ lr_enc 0.00009
718
+ lr 0.00030
719
+ lr_pi 0.00030
720
+ ------------------------------
721
+ ------------------------------
722
+ Pretraining metrics:
723
+ consistency_loss 0.00152
724
+ reward_loss 0.43261
725
+ value_loss 0.49984
726
+ total_loss 0.72449
727
+ bc_loss 0.16684
728
+ entropy_loss -0.00160
729
+ pi_prior_loss 0.05249
730
+ pi_entropy 1.84485
731
+ pi_scaled_entropy 16.04462
732
+ pi_std 0.77362
733
+ pi_max_std 2.38854
734
+ contrastive_loss 0.54828
735
+ contrastive_pos_logit 0.88922
736
+ contrastive_neg_logit -0.77415
737
+ contrastive_mean -0.00869
738
+ contrastive_std 1.68556
739
+ grad_norm 1.02152
740
+ lr_enc 0.00009
741
+ lr 0.00030
742
+ lr_pi 0.00030
743
+ ------------------------------
744
+ ------------------------------
745
+ Pretraining metrics:
746
+ consistency_loss 0.00163
747
+ reward_loss 0.42202
748
+ value_loss 0.56579
749
+ total_loss 0.73394
750
+ bc_loss 0.18851
751
+ entropy_loss -0.00177
752
+ pi_prior_loss 0.05784
753
+ pi_entropy 1.95919
754
+ pi_scaled_entropy 17.69400
755
+ pi_std 0.77332
756
+ pi_max_std 3.94518
757
+ contrastive_loss 0.54472
758
+ contrastive_pos_logit 0.67652
759
+ contrastive_neg_logit -0.90501
760
+ contrastive_mean -0.01030
761
+ contrastive_std 1.70954
762
+ grad_norm 0.70580
763
+ lr_enc 0.00009
764
+ lr 0.00030
765
+ lr_pi 0.00030
766
+ ------------------------------
767
+ ------------------------------
768
+ Pretraining metrics:
769
+ consistency_loss 0.00162
770
+ reward_loss 0.41959
771
+ value_loss 0.50879
772
+ total_loss 0.71718
773
+ bc_loss 0.17984
774
+ entropy_loss -0.00098
775
+ pi_prior_loss 0.05675
776
+ pi_entropy 0.33626
777
+ pi_scaled_entropy 9.84783
778
+ pi_std 0.77612
779
+ pi_max_std 4.09453
780
+ contrastive_loss 0.53517
781
+ contrastive_pos_logit 0.99810
782
+ contrastive_neg_logit -0.70211
783
+ contrastive_mean -0.01244
784
+ contrastive_std 1.71976
785
+ grad_norm 1.08714
786
+ lr_enc 0.00009
787
+ lr 0.00030
788
+ lr_pi 0.00030
789
+ ------------------------------
790
+ ------------------------------
791
+ Pretraining metrics:
792
+ consistency_loss 0.00142
793
+ reward_loss 0.38849
794
+ value_loss 0.53266
795
+ total_loss 0.70777
796
+ bc_loss 0.18814
797
+ entropy_loss -0.00169
798
+ pi_prior_loss 0.05722
799
+ pi_entropy 2.29765
800
+ pi_scaled_entropy 16.86834
801
+ pi_std 0.78256
802
+ pi_max_std 3.00517
803
+ contrastive_loss 0.53004
804
+ contrastive_pos_logit 0.70572
805
+ contrastive_neg_logit -0.95326
806
+ contrastive_mean -0.01656
807
+ contrastive_std 1.74067
808
+ grad_norm 0.60207
809
+ lr_enc 0.00009
810
+ lr 0.00030
811
+ lr_pi 0.00030
812
+ ------------------------------
813
+ ------------------------------
814
+ Pretraining metrics:
815
+ consistency_loss 0.00138
816
+ reward_loss 0.42173
817
+ value_loss 0.53435
818
+ total_loss 0.70711
819
+ bc_loss 0.17326
820
+ entropy_loss -0.00180
821
+ pi_prior_loss 0.05472
822
+ pi_entropy 1.61974
823
+ pi_scaled_entropy 18.00428
824
+ pi_std 0.77371
825
+ pi_max_std 2.93966
826
+ contrastive_loss 0.52919
827
+ contrastive_pos_logit 0.82301
828
+ contrastive_neg_logit -1.10370
829
+ contrastive_mean -0.01617
830
+ contrastive_std 1.76528
831
+ grad_norm 0.64897
832
+ lr_enc 0.00009
833
+ lr 0.00030
834
+ lr_pi 0.00030
835
+ ------------------------------
836
+ ------------------------------
837
+ Pretraining metrics:
838
+ consistency_loss 0.00151
839
+ reward_loss 0.43184
840
+ value_loss 0.51705
841
+ total_loss 0.71353
842
+ bc_loss 0.15308
843
+ entropy_loss -0.00167
844
+ pi_prior_loss 0.05100
845
+ pi_entropy 1.72766
846
+ pi_scaled_entropy 16.73644
847
+ pi_std 0.77384
848
+ pi_max_std 3.93693
849
+ contrastive_loss 0.53750
850
+ contrastive_pos_logit 0.81595
851
+ contrastive_neg_logit -0.85151
852
+ contrastive_mean -0.01714
853
+ contrastive_std 1.78801
854
+ grad_norm 0.55102
855
+ lr_enc 0.00009
856
+ lr 0.00030
857
+ lr_pi 0.00030
858
+ ------------------------------
859
+ ------------------------------
860
+ Pretraining metrics:
861
+ consistency_loss 0.00156
862
+ reward_loss 0.39618
863
+ value_loss 0.52558
864
+ total_loss 0.69449
865
+ bc_loss 0.18399
866
+ entropy_loss -0.00174
867
+ pi_prior_loss 0.05984
868
+ pi_entropy 1.70508
869
+ pi_scaled_entropy 17.43760
870
+ pi_std 0.77928
871
+ pi_max_std 4.88674
872
+ contrastive_loss 0.51134
873
+ contrastive_pos_logit 0.89907
874
+ contrastive_neg_logit -0.99066
875
+ contrastive_mean -0.02252
876
+ contrastive_std 1.81520
877
+ grad_norm 0.57423
878
+ lr_enc 0.00009
879
+ lr 0.00030
880
+ lr_pi 0.00030
881
+ ------------------------------
882
+ ------------------------------
883
+ Pretraining metrics:
884
+ consistency_loss 0.00137
885
+ reward_loss 0.42567
886
+ value_loss 0.48445
887
+ total_loss 0.69784
888
+ bc_loss 0.15756
889
+ entropy_loss -0.00174
890
+ pi_prior_loss 0.04930
891
+ pi_entropy 1.94695
892
+ pi_scaled_entropy 17.43790
893
+ pi_std 0.77889
894
+ pi_max_std 2.18126
895
+ contrastive_loss 0.53015
896
+ contrastive_pos_logit 0.90961
897
+ contrastive_neg_logit -0.87386
898
+ contrastive_mean -0.02177
899
+ contrastive_std 1.84699
900
+ grad_norm 0.78900
901
+ lr_enc 0.00009
902
+ lr 0.00030
903
+ lr_pi 0.00030
904
+ ------------------------------
905
+ ------------------------------
906
+ Pretraining metrics:
907
+ consistency_loss 0.00151
908
+ reward_loss 0.41609
909
+ value_loss 0.47456
910
+ total_loss 0.69935
911
+ bc_loss 0.16518
912
+ entropy_loss -0.00151
913
+ pi_prior_loss 0.05001
914
+ pi_entropy 1.91174
915
+ pi_scaled_entropy 15.10952
916
+ pi_std 0.77819
917
+ pi_max_std 6.14075
918
+ contrastive_loss 0.53014
919
+ contrastive_pos_logit 0.84130
920
+ contrastive_neg_logit -0.97208
921
+ contrastive_mean -0.02812
922
+ contrastive_std 1.84886
923
+ grad_norm 0.93767
924
+ lr_enc 0.00009
925
+ lr 0.00030
926
+ lr_pi 0.00030
927
+ ------------------------------
928
+ ------------------------------
929
+ Pretraining metrics:
930
+ consistency_loss 0.00151
931
+ reward_loss 0.41233
932
+ value_loss 0.52579
933
+ total_loss 0.69342
934
+ bc_loss 0.15781
935
+ entropy_loss -0.00188
936
+ pi_prior_loss 0.04662
937
+ pi_entropy 2.27530
938
+ pi_scaled_entropy 18.81979
939
+ pi_std 0.78573
940
+ pi_max_std 2.26755
941
+ contrastive_loss 0.52285
942
+ contrastive_pos_logit 0.84257
943
+ contrastive_neg_logit -0.98326
944
+ contrastive_mean -0.02606
945
+ contrastive_std 1.84816
946
+ grad_norm 0.50376
947
+ lr_enc 0.00009
948
+ lr 0.00030
949
+ lr_pi 0.00030
950
+ ------------------------------
951
+ ------------------------------
952
+ Pretraining metrics:
953
+ consistency_loss 0.00155
954
+ reward_loss 0.39902
955
+ value_loss 0.52228
956
+ total_loss 0.70428
957
+ bc_loss 0.16217
958
+ entropy_loss -0.00156
959
+ pi_prior_loss 0.05199
960
+ pi_entropy 1.67820
961
+ pi_scaled_entropy 15.58255
962
+ pi_std 0.77093
963
+ pi_max_std 2.93290
964
+ contrastive_loss 0.52925
965
+ contrastive_pos_logit 0.97850
966
+ contrastive_neg_logit -0.95934
967
+ contrastive_mean -0.02979
968
+ contrastive_std 1.88737
969
+ grad_norm 0.60612
970
+ lr_enc 0.00009
971
+ lr 0.00030
972
+ lr_pi 0.00030
973
+ ------------------------------
974
+ ------------------------------
975
+ Pretraining metrics:
976
+ consistency_loss 0.00207
977
+ reward_loss 0.51395
978
+ value_loss 0.65641
979
+ total_loss 0.75321
980
+ bc_loss 0.15766
981
+ entropy_loss -0.00170
982
+ pi_prior_loss 0.04875
983
+ pi_entropy 1.00188
984
+ pi_scaled_entropy 16.98432
985
+ pi_std 0.75847
986
+ pi_max_std 4.29939
987
+ contrastive_loss 0.54596
988
+ contrastive_pos_logit 0.85145
989
+ contrastive_neg_logit -0.76286
990
+ contrastive_mean 0.00489
991
+ contrastive_std 1.54713
992
+ grad_norm 0.67703
993
+ lr_enc 0.00009
994
+ lr 0.00030
995
+ lr_pi 0.00030
996
+ ------------------------------
997
+ ------------------------------
998
+ Pretraining metrics:
999
+ consistency_loss 0.00206
1000
+ reward_loss 0.48397
1001
+ value_loss 0.61864
1002
+ total_loss 0.76904
1003
+ bc_loss 0.12850
1004
+ entropy_loss -0.00148
1005
+ pi_prior_loss 0.04050
1006
+ pi_entropy 0.75657
1007
+ pi_scaled_entropy 14.76664
1008
+ pi_std 0.74742
1009
+ pi_max_std 5.44932
1010
+ contrastive_loss 0.57699
1011
+ contrastive_pos_logit 0.54443
1012
+ contrastive_neg_logit -0.70636
1013
+ contrastive_mean 0.00643
1014
+ contrastive_std 1.41769
1015
+ grad_norm 0.82272
1016
+ lr_enc 0.00009
1017
+ lr 0.00030
1018
+ lr_pi 0.00030
1019
+ ------------------------------
1020
+ ------------------------------
1021
+ Pretraining metrics:
1022
+ consistency_loss 0.00182
1023
+ reward_loss 0.45756
1024
+ value_loss 0.56549
1025
+ total_loss 0.73838
1026
+ bc_loss 0.13289
1027
+ entropy_loss -0.00150
1028
+ pi_prior_loss 0.04029
1029
+ pi_entropy 0.72241
1030
+ pi_scaled_entropy 14.96662
1031
+ pi_std 0.74472
1032
+ pi_max_std 2.31815
1033
+ contrastive_loss 0.55940
1034
+ contrastive_pos_logit 0.65299
1035
+ contrastive_neg_logit -0.62610
1036
+ contrastive_mean 0.00556
1037
+ contrastive_std 1.40730
1038
+ grad_norm 0.71446
1039
+ lr_enc 0.00009
1040
+ lr 0.00030
1041
+ lr_pi 0.00030
1042
+ ------------------------------
1043
+ ------------------------------
1044
+ Pretraining metrics:
1045
+ consistency_loss 0.00200
1046
+ reward_loss 0.50762
1047
+ value_loss 0.64698
1048
+ total_loss 0.76427
1049
+ bc_loss 0.11413
1050
+ entropy_loss -0.00140
1051
+ pi_prior_loss 0.03480
1052
+ pi_entropy 0.73301
1053
+ pi_scaled_entropy 14.01148
1054
+ pi_std 0.75582
1055
+ pi_max_std 6.17229
1056
+ contrastive_loss 0.57410
1057
+ contrastive_pos_logit 0.63857
1058
+ contrastive_neg_logit -0.66174
1059
+ contrastive_mean 0.00268
1060
+ contrastive_std 1.42234
1061
+ grad_norm 0.59466
1062
+ lr_enc 0.00009
1063
+ lr 0.00030
1064
+ lr_pi 0.00030
1065
+ ------------------------------
1066
+ ------------------------------
1067
+ Pretraining metrics:
1068
+ consistency_loss 0.00186
1069
+ reward_loss 0.45367
1070
+ value_loss 0.53240
1071
+ total_loss 0.75426
1072
+ bc_loss 0.12570
1073
+ entropy_loss -0.00145
1074
+ pi_prior_loss 0.03836
1075
+ pi_entropy 0.99112
1076
+ pi_scaled_entropy 14.49046
1077
+ pi_std 0.76392
1078
+ pi_max_std 4.16374
1079
+ contrastive_loss 0.58011
1080
+ contrastive_pos_logit 0.62795
1081
+ contrastive_neg_logit -0.58038
1082
+ contrastive_mean 0.00352
1083
+ contrastive_std 1.45118
1084
+ grad_norm 0.64345
1085
+ lr_enc 0.00009
1086
+ lr 0.00030
1087
+ lr_pi 0.00030
1088
+ ------------------------------
1089
+ ------------------------------
1090
+ Pretraining metrics:
1091
+ consistency_loss 0.00227
1092
+ reward_loss 0.50325
1093
+ value_loss 0.66261
1094
+ total_loss 0.78265
1095
+ bc_loss 0.12116
1096
+ entropy_loss -0.00150
1097
+ pi_prior_loss 0.04005
1098
+ pi_entropy -0.09769
1099
+ pi_scaled_entropy 15.03090
1100
+ pi_std 0.74821
1101
+ pi_max_std 6.42646
1102
+ contrastive_loss 0.58067
1103
+ contrastive_pos_logit 0.64783
1104
+ contrastive_neg_logit -0.62724
1105
+ contrastive_mean 0.00027
1106
+ contrastive_std 1.47970
1107
+ grad_norm 1.06581
1108
+ lr_enc 0.00009
1109
+ lr 0.00030
1110
+ lr_pi 0.00030
1111
+ ------------------------------
1112
+ ------------------------------
1113
+ Pretraining metrics:
1114
+ consistency_loss 0.00201
1115
+ reward_loss 0.46791
1116
+ value_loss 0.58735
1117
+ total_loss 0.72808
1118
+ bc_loss 0.12404
1119
+ entropy_loss -0.00168
1120
+ pi_prior_loss 0.03803
1121
+ pi_entropy 1.18617
1122
+ pi_scaled_entropy 16.82698
1123
+ pi_std 0.77043
1124
+ pi_max_std 6.80987
1125
+ contrastive_loss 0.54422
1126
+ contrastive_pos_logit 0.77575
1127
+ contrastive_neg_logit -0.61826
1128
+ contrastive_mean -0.00020
1129
+ contrastive_std 1.49326
1130
+ grad_norm 0.80897
1131
+ lr_enc 0.00009
1132
+ lr 0.00030
1133
+ lr_pi 0.00030
1134
+ ------------------------------
1135
+ ------------------------------
1136
+ Pretraining metrics:
1137
+ consistency_loss 0.00214
1138
+ reward_loss 0.51148
1139
+ value_loss 0.56932
1140
+ total_loss 0.71622
1141
+ bc_loss 0.12203
1142
+ entropy_loss -0.00138
1143
+ pi_prior_loss 0.03894
1144
+ pi_entropy 0.65649
1145
+ pi_scaled_entropy 13.77774
1146
+ pi_std 0.74909
1147
+ pi_max_std 2.54864
1148
+ contrastive_loss 0.52638
1149
+ contrastive_pos_logit 0.83636
1150
+ contrastive_neg_logit -0.78305
1151
+ contrastive_mean -0.00340
1152
+ contrastive_std 1.52052
1153
+ grad_norm 1.39178
1154
+ lr_enc 0.00009
1155
+ lr 0.00030
1156
+ lr_pi 0.00030
1157
+ ------------------------------
1158
+ ------------------------------
1159
+ Pretraining metrics:
1160
+ consistency_loss 0.00189
1161
+ reward_loss 0.49172
1162
+ value_loss 0.59165
1163
+ total_loss 0.72787
1164
+ bc_loss 0.12664
1165
+ entropy_loss -0.00139
1166
+ pi_prior_loss 0.04071
1167
+ pi_entropy 0.30423
1168
+ pi_scaled_entropy 13.92600
1169
+ pi_std 0.75045
1170
+ pi_max_std 6.72984
1171
+ contrastive_loss 0.54110
1172
+ contrastive_pos_logit 0.90457
1173
+ contrastive_neg_logit -0.67507
1174
+ contrastive_mean 0.00001
1175
+ contrastive_std 1.56416
1176
+ grad_norm 1.03444
1177
+ lr_enc 0.00009
1178
+ lr 0.00030
1179
+ lr_pi 0.00030
1180
+ ------------------------------
soup_XL_100M_work_dir/wandb/run-20260623_100535-g9l145ew/files/requirements.txt ADDED
@@ -0,0 +1,174 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ fast_kinematics==0.2.2
2
+ pyparsing==3.3.2
3
+ python-dateutil==2.9.0.post0
4
+ wcwidth==0.7.0
5
+ parso==0.8.7
6
+ kornia_rs==0.1.14
7
+ ml_dtypes==0.5.4
8
+ filelock==3.29.0
9
+ fonttools==4.63.0
10
+ jax==0.7.1
11
+ stack-data==0.6.3
12
+ termcolor==3.1.0
13
+ ogbench==1.1.5
14
+ tokenizers==0.22.2
15
+ cycler==0.12.1
16
+ nvidia-cufile-cu12==1.13.1.3
17
+ asttokens==3.0.1
18
+ prompt_toolkit==3.0.52
19
+ nvidia-nvshmem-cu12==3.6.5
20
+ PyYAML==6.0.3
21
+ absl-py==2.4.0
22
+ ipython_pygments_lexers==1.1.1
23
+ MarkupSafe==3.0.3
24
+ gitdb==4.0.12
25
+ attrs==26.1.0
26
+ executing==2.2.1
27
+ dm-env==1.6
28
+ pytorch-kinematics==0.7.5
29
+ uvloop==0.22.1
30
+ dacite==1.9.2
31
+ huggingface_hub==0.36.2
32
+ PyOpenGL==3.1.10
33
+ importlib_metadata==9.0.0
34
+ matplotlib==3.10.9
35
+ pyvers==0.1.0
36
+ docstring_parser==0.18.0
37
+ mani-skill-nightly==2025.9.19.39
38
+ typing-inspection==0.4.2
39
+ sentry-sdk==2.61.1
40
+ tqdm==4.67.1
41
+ traitlets==5.15.1
42
+ nvidia-cufft-cu12==11.3.3.83
43
+ prometheus_client==0.25.0
44
+ importlib_resources==7.1.0
45
+ matplotlib-inline==0.2.2
46
+ pytorch-seed==0.2.0
47
+ tensorstore==0.1.84
48
+ wrapt==2.2.1
49
+ GitPython==3.1.50
50
+ nvidia-cuda-runtime-cu12==12.8.90
51
+ metaworld==2.0.0
52
+ nvidia-cuda-cupti-cu12==12.8.90
53
+ jaxlib==0.7.1
54
+ annotated-types==0.7.0
55
+ setuptools==69.5.1
56
+ swig==4.4.1
57
+ pydantic_core==2.46.4
58
+ pillow==12.2.0
59
+ trimesh==4.12.2
60
+ moviepy==1.0.3
61
+ opencv-python==4.11.0.86
62
+ scipy==1.17.1
63
+ nvidia-cusparselt-cu12==0.7.1
64
+ dm_control==1.0.34
65
+ typeguard==4.5.2
66
+ sympy==1.14.0
67
+ tyro==1.0.13
68
+ rich==15.0.0
69
+ charset-normalizer==3.4.7
70
+ nvidia-ml-py==13.610.43
71
+ hf-xet==1.5.0
72
+ antlr4-python3-runtime==4.9.3
73
+ pyperclip==1.11.0
74
+ psutil==7.2.2
75
+ mdurl==0.1.2
76
+ mpmath==1.3.0
77
+ click==8.4.1
78
+ nvidia-cusolver-cu12==11.7.3.90
79
+ kiwisolver==1.5.0
80
+ gym==0.26.2
81
+ AutoROM==0.6.1
82
+ contourpy==1.3.3
83
+ pynvml==13.0.1
84
+ torchvision==0.23.0+cu128
85
+ glfw==2.10.0
86
+ torchaudio==2.8.0+cu128
87
+ numpy==1.26.4
88
+ nvidia-cudnn-cu12==9.10.2.21
89
+ nvidia-nvjitlink-cu12==12.8.93
90
+ jinxed==2.0.4
91
+ Jinja2==3.1.6
92
+ jax-cuda12-pjrt==0.7.1
93
+ cloudpickle==3.1.2
94
+ orbax-checkpoint==0.12.0
95
+ humanize==4.15.0
96
+ transforms3d==0.4.2
97
+ tabulate==0.10.0
98
+ pure_eval==0.2.3
99
+ ipython==8.37.0
100
+ nvidia-cuda-nvrtc-cu12==12.8.93
101
+ labmaze==1.0.6
102
+ zipp==4.1.0
103
+ arm_pytorch_utilities==0.5.0
104
+ requests==2.34.2
105
+ networkx==3.6.1
106
+ treescope==0.1.10
107
+ robodesk==1.0.0
108
+ blessed==1.44.0
109
+ triton==3.4.0
110
+ proglog==0.1.12
111
+ Pygments==2.20.0
112
+ tensordict==0.10.0
113
+ opt_einsum==3.4.0
114
+ pexpect==4.9.0
115
+ simplejson==4.1.1
116
+ Farama-Notifications==0.0.6
117
+ msgpack==1.1.2
118
+ decorator==4.4.2
119
+ packaging==25.0
120
+ pip==25.2
121
+ kornia==0.8.1
122
+ optax==0.2.8
123
+ orjson==3.11.9
124
+ urllib3==2.7.0
125
+ pandas==3.0.3
126
+ idna==3.18
127
+ hydra-core==1.3.2
128
+ pygame==2.6.1
129
+ nvidia-curand-cu12==10.3.9.90
130
+ etils==1.14.0
131
+ certifi==2026.5.20
132
+ mujoco==3.3.6
133
+ imageio==2.37.0
134
+ aiofiles==25.1.0
135
+ imageio-ffmpeg==0.6.0
136
+ typing_extensions==4.15.0
137
+ protobuf==5.29.6
138
+ sapien==3.0.3
139
+ AutoROM.accept-rom-license==0.6.1
140
+ fsspec==2026.4.0
141
+ nvidia-cuda-cccl-cu12==12.9.27
142
+ dm-tree==0.1.10
143
+ mplib==0.1.1
144
+ h5py==3.14.0
145
+ nvidia-cublas-cu12==12.8.4.1
146
+ wheel==0.45.1
147
+ ptyprocess==0.7.0
148
+ submitit==1.5.3
149
+ hydra-submitit-launcher==1.2.0
150
+ markdown-it-py==4.2.0
151
+ omegaconf==2.3.0
152
+ nvidia-cuda-nvcc-cu12==12.9.86
153
+ safetensors==0.7.0
154
+ jax-cuda12-plugin==0.7.1
155
+ smmap==5.0.3
156
+ six==1.17.0
157
+ torch==2.8.0+cu128
158
+ toppra==0.6.3
159
+ platformdirs==4.10.0
160
+ lxml==6.1.1
161
+ box2d-py==2.3.5
162
+ gymnasium==0.29.1
163
+ regex==2026.5.9
164
+ torchrl==0.10.0
165
+ gpustat==1.1.1
166
+ nvidia-nvtx-cu12==12.8.90
167
+ flax==0.12.0
168
+ nvidia-cusparse-cu12==12.5.8.93
169
+ ale-py==0.10.0
170
+ nvidia-nccl-cu12==2.27.3
171
+ wandb==0.22.1
172
+ pydantic==2.13.4
173
+ jedi==0.20.0
174
+ transformers==4.56.2
soup_XL_100M_work_dir/wandb/run-20260623_100535-g9l145ew/files/wandb-metadata.json ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "os": "Linux-5.15.0-72-generic-x86_64-with-glibc2.35",
3
+ "python": "CPython 3.11.15",
4
+ "startedAt": "2026-06-23T02:05:35.256271Z",
5
+ "args": [
6
+ "task=soup",
7
+ "model_size=XL",
8
+ "steps=100000000",
9
+ "demo_steps=200000",
10
+ "train_eval_freq=5000000",
11
+ "checkpoint_save_freq=5000000",
12
+ "compile=True",
13
+ "diffusion_compile=True",
14
+ "multiproc=True",
15
+ "use_score_network=False",
16
+ "eval_at_start=False",
17
+ "env_mode=async",
18
+ "planner_type=diffusion",
19
+ "exp_name=soup_XL_100M",
20
+ "seed=80",
21
+ "work_dir=/media/datasets/cheliu21/cxy_worldmodel/newt/soup_XL_100M_work_dir"
22
+ ],
23
+ "program": "/media/damoxing/che-liu-fileset/cxy_worldmodel/newt/tdmpc2/train.py",
24
+ "codePath": "tdmpc2/train.py",
25
+ "codePathLocal": "train.py",
26
+ "git": {
27
+ "remote": "git@github.com:Wenxuan52/newt.git",
28
+ "commit": "c48718ddb1e373c92801c0f36a1aa23d5300c980"
29
+ },
30
+ "email": "wenxuan.yuan@qq.com",
31
+ "root": "/media/datasets/cheliu21/cxy_worldmodel/newt/soup_XL_100M_work_dir",
32
+ "host": "job-mx9zwvdf8b4m-master-0",
33
+ "executable": "/opt/conda/envs/newt-jax/bin/python",
34
+ "cpu_count": 64,
35
+ "cpu_count_logical": 128,
36
+ "gpu": "NVIDIA A800-SXM4-80GB",
37
+ "gpu_count": 4,
38
+ "disk": {
39
+ "/": {
40
+ "total": "528147169280",
41
+ "used": "342780809216"
42
+ }
43
+ },
44
+ "memory": {
45
+ "total": "2162270285824"
46
+ },
47
+ "gpu_nvidia": [
48
+ {
49
+ "name": "NVIDIA A800-SXM4-80GB",
50
+ "memoryTotal": "85899345920",
51
+ "cudaCores": 6912,
52
+ "architecture": "Ampere",
53
+ "uuid": "GPU-e33a5d5c-ece1-16b1-ae70-38577ceaea24"
54
+ },
55
+ {
56
+ "name": "NVIDIA A800-SXM4-80GB",
57
+ "memoryTotal": "85899345920",
58
+ "cudaCores": 6912,
59
+ "architecture": "Ampere",
60
+ "uuid": "GPU-cd62cf34-2832-7925-6fb2-50d0de797c7e"
61
+ },
62
+ {
63
+ "name": "NVIDIA A800-SXM4-80GB",
64
+ "memoryTotal": "85899345920",
65
+ "cudaCores": 6912,
66
+ "architecture": "Ampere",
67
+ "uuid": "GPU-f15c4618-c2e7-f14e-f27e-623dda73692a"
68
+ },
69
+ {
70
+ "name": "NVIDIA A800-SXM4-80GB",
71
+ "memoryTotal": "85899345920",
72
+ "cudaCores": 6912,
73
+ "architecture": "Ampere",
74
+ "uuid": "GPU-4f0f1b5f-f5bc-5f62-8cc8-a837e6d1f600"
75
+ }
76
+ ],
77
+ "cudaVersion": "12.4",
78
+ "writerId": "a1u3ipa3gcrnsm9xeh1vqzqpwmoccbed"
79
+ }
soup_XL_100M_work_dir/wandb/run-20260623_100535-g9l145ew/logs/debug-internal.log ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {"time":"2026-06-23T10:05:35.531478601+08:00","level":"INFO","msg":"stream: starting","core version":"0.22.1"}
2
+ {"time":"2026-06-23T10:05:36.061131519+08:00","level":"INFO","msg":"stream: created new stream","id":"g9l145ew"}
3
+ {"time":"2026-06-23T10:05:36.061182108+08:00","level":"INFO","msg":"handler: started","stream_id":"g9l145ew"}
4
+ {"time":"2026-06-23T10:05:36.06148986+08:00","level":"INFO","msg":"stream: started","id":"g9l145ew"}
5
+ {"time":"2026-06-23T10:05:36.061502919+08:00","level":"INFO","msg":"writer: started","stream_id":"g9l145ew"}
6
+ {"time":"2026-06-23T10:05:36.061503589+08:00","level":"INFO","msg":"sender: started","stream_id":"g9l145ew"}
7
+ {"time":"2026-06-23T10:25:24.014381245+08:00","level":"INFO","msg":"flowcontrol: backed up, offloading to disk","recordNumber":6339}
8
+ {"time":"2026-06-23T10:25:24.054476658+08:00","level":"INFO","msg":"flowcontrol: unblocked","totalOffloaded":13}
9
+ {"time":"2026-06-23T14:12:24.105579943+08:00","level":"INFO","msg":"flowcontrol: backed up, offloading to disk","recordNumber":81739}
10
+ {"time":"2026-06-23T14:12:24.265935776+08:00","level":"INFO","msg":"flowcontrol: unblocked","totalOffloaded":13}
soup_XL_100M_work_dir/wandb/run-20260623_100535-g9l145ew/logs/debug.log ADDED
The diff for this file is too large to render. See raw diff
 
soup_XL_100M_work_dir/wandb/run-20260708_090752-bs3i5bjq/files/output.log ADDED
The diff for this file is too large to render. See raw diff
 
soup_XL_100M_work_dir/wandb/run-20260708_090752-bs3i5bjq/files/requirements.txt ADDED
@@ -0,0 +1,174 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ fast_kinematics==0.2.2
2
+ pyparsing==3.3.2
3
+ python-dateutil==2.9.0.post0
4
+ wcwidth==0.7.0
5
+ parso==0.8.7
6
+ kornia_rs==0.1.14
7
+ ml_dtypes==0.5.4
8
+ filelock==3.29.0
9
+ fonttools==4.63.0
10
+ jax==0.7.1
11
+ stack-data==0.6.3
12
+ termcolor==3.1.0
13
+ ogbench==1.1.5
14
+ tokenizers==0.22.2
15
+ cycler==0.12.1
16
+ nvidia-cufile-cu12==1.13.1.3
17
+ asttokens==3.0.1
18
+ prompt_toolkit==3.0.52
19
+ nvidia-nvshmem-cu12==3.6.5
20
+ PyYAML==6.0.3
21
+ absl-py==2.4.0
22
+ ipython_pygments_lexers==1.1.1
23
+ MarkupSafe==3.0.3
24
+ gitdb==4.0.12
25
+ attrs==26.1.0
26
+ executing==2.2.1
27
+ dm-env==1.6
28
+ pytorch-kinematics==0.7.5
29
+ uvloop==0.22.1
30
+ dacite==1.9.2
31
+ huggingface_hub==0.36.2
32
+ PyOpenGL==3.1.10
33
+ importlib_metadata==9.0.0
34
+ matplotlib==3.10.9
35
+ pyvers==0.1.0
36
+ docstring_parser==0.18.0
37
+ mani-skill-nightly==2025.9.19.39
38
+ typing-inspection==0.4.2
39
+ sentry-sdk==2.61.1
40
+ tqdm==4.67.1
41
+ traitlets==5.15.1
42
+ nvidia-cufft-cu12==11.3.3.83
43
+ prometheus_client==0.25.0
44
+ importlib_resources==7.1.0
45
+ matplotlib-inline==0.2.2
46
+ pytorch-seed==0.2.0
47
+ tensorstore==0.1.84
48
+ wrapt==2.2.1
49
+ GitPython==3.1.50
50
+ nvidia-cuda-runtime-cu12==12.8.90
51
+ metaworld==2.0.0
52
+ nvidia-cuda-cupti-cu12==12.8.90
53
+ jaxlib==0.7.1
54
+ annotated-types==0.7.0
55
+ setuptools==69.5.1
56
+ swig==4.4.1
57
+ pydantic_core==2.46.4
58
+ pillow==12.2.0
59
+ trimesh==4.12.2
60
+ moviepy==1.0.3
61
+ opencv-python==4.11.0.86
62
+ scipy==1.17.1
63
+ nvidia-cusparselt-cu12==0.7.1
64
+ dm_control==1.0.34
65
+ typeguard==4.5.2
66
+ sympy==1.14.0
67
+ tyro==1.0.13
68
+ rich==15.0.0
69
+ charset-normalizer==3.4.7
70
+ nvidia-ml-py==13.610.43
71
+ hf-xet==1.5.0
72
+ antlr4-python3-runtime==4.9.3
73
+ pyperclip==1.11.0
74
+ psutil==7.2.2
75
+ mdurl==0.1.2
76
+ mpmath==1.3.0
77
+ click==8.4.1
78
+ nvidia-cusolver-cu12==11.7.3.90
79
+ kiwisolver==1.5.0
80
+ gym==0.26.2
81
+ AutoROM==0.6.1
82
+ contourpy==1.3.3
83
+ pynvml==13.0.1
84
+ torchvision==0.23.0+cu128
85
+ glfw==2.10.0
86
+ torchaudio==2.8.0+cu128
87
+ numpy==1.26.4
88
+ nvidia-cudnn-cu12==9.10.2.21
89
+ nvidia-nvjitlink-cu12==12.8.93
90
+ jinxed==2.0.4
91
+ Jinja2==3.1.6
92
+ jax-cuda12-pjrt==0.7.1
93
+ cloudpickle==3.1.2
94
+ orbax-checkpoint==0.12.0
95
+ humanize==4.15.0
96
+ transforms3d==0.4.2
97
+ tabulate==0.10.0
98
+ pure_eval==0.2.3
99
+ ipython==8.37.0
100
+ nvidia-cuda-nvrtc-cu12==12.8.93
101
+ labmaze==1.0.6
102
+ zipp==4.1.0
103
+ arm_pytorch_utilities==0.5.0
104
+ requests==2.34.2
105
+ networkx==3.6.1
106
+ treescope==0.1.10
107
+ robodesk==1.0.0
108
+ blessed==1.44.0
109
+ triton==3.4.0
110
+ proglog==0.1.12
111
+ Pygments==2.20.0
112
+ tensordict==0.10.0
113
+ opt_einsum==3.4.0
114
+ pexpect==4.9.0
115
+ simplejson==4.1.1
116
+ Farama-Notifications==0.0.6
117
+ msgpack==1.1.2
118
+ decorator==4.4.2
119
+ packaging==25.0
120
+ pip==25.2
121
+ kornia==0.8.1
122
+ optax==0.2.8
123
+ orjson==3.11.9
124
+ urllib3==2.7.0
125
+ pandas==3.0.3
126
+ idna==3.18
127
+ hydra-core==1.3.2
128
+ pygame==2.6.1
129
+ nvidia-curand-cu12==10.3.9.90
130
+ etils==1.14.0
131
+ certifi==2026.5.20
132
+ mujoco==3.3.6
133
+ imageio==2.37.0
134
+ aiofiles==25.1.0
135
+ imageio-ffmpeg==0.6.0
136
+ typing_extensions==4.15.0
137
+ protobuf==5.29.6
138
+ sapien==3.0.3
139
+ AutoROM.accept-rom-license==0.6.1
140
+ fsspec==2026.4.0
141
+ nvidia-cuda-cccl-cu12==12.9.27
142
+ dm-tree==0.1.10
143
+ mplib==0.1.1
144
+ h5py==3.14.0
145
+ nvidia-cublas-cu12==12.8.4.1
146
+ wheel==0.45.1
147
+ ptyprocess==0.7.0
148
+ submitit==1.5.3
149
+ hydra-submitit-launcher==1.2.0
150
+ markdown-it-py==4.2.0
151
+ omegaconf==2.3.0
152
+ nvidia-cuda-nvcc-cu12==12.9.86
153
+ safetensors==0.7.0
154
+ jax-cuda12-plugin==0.7.1
155
+ smmap==5.0.3
156
+ six==1.17.0
157
+ torch==2.8.0+cu128
158
+ toppra==0.6.3
159
+ platformdirs==4.10.0
160
+ lxml==6.1.1
161
+ box2d-py==2.3.5
162
+ gymnasium==0.29.1
163
+ regex==2026.5.9
164
+ torchrl==0.10.0
165
+ gpustat==1.1.1
166
+ nvidia-nvtx-cu12==12.8.90
167
+ flax==0.12.0
168
+ nvidia-cusparse-cu12==12.5.8.93
169
+ ale-py==0.10.0
170
+ nvidia-nccl-cu12==2.27.3
171
+ wandb==0.22.1
172
+ pydantic==2.13.4
173
+ jedi==0.20.0
174
+ transformers==4.56.2
soup_XL_100M_work_dir/wandb/run-20260708_090752-bs3i5bjq/files/wandb-metadata.json ADDED
@@ -0,0 +1,80 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "os": "Linux-5.15.0-72-generic-x86_64-with-glibc2.35",
3
+ "python": "CPython 3.11.15",
4
+ "startedAt": "2026-07-08T01:07:52.060893Z",
5
+ "args": [
6
+ "task=soup",
7
+ "model_size=XL",
8
+ "steps=100000000",
9
+ "demo_steps=100000",
10
+ "train_eval_freq=5000000",
11
+ "checkpoint_save_freq=5000000",
12
+ "diffusion_final_rerank=True",
13
+ "compile=True",
14
+ "diffusion_compile=True",
15
+ "multiproc=True",
16
+ "use_score_network=False",
17
+ "eval_at_start=False",
18
+ "env_mode=async",
19
+ "planner_type=diffusion",
20
+ "exp_name=soup_XL_100M",
21
+ "seed=5",
22
+ "work_dir=/media/datasets/cheliu21/cxy_worldmodel/newt/soup_XL_100M_work_dir"
23
+ ],
24
+ "program": "/media/damoxing/che-liu-fileset/cxy_worldmodel/newt/tdmpc2/train.py",
25
+ "codePath": "tdmpc2/train.py",
26
+ "codePathLocal": "train.py",
27
+ "git": {
28
+ "remote": "git@github.com:Wenxuan52/newt.git",
29
+ "commit": "0c15f2e94dcb5661c338ee4d4a9852c1f74447a2"
30
+ },
31
+ "email": "wenxuan.yuan@qq.com",
32
+ "root": "/media/datasets/cheliu21/cxy_worldmodel/newt/soup_XL_100M_work_dir",
33
+ "host": "job-xp37erxymhbm-master-0",
34
+ "executable": "/opt/conda/envs/newt-jax/bin/python",
35
+ "cpu_count": 64,
36
+ "cpu_count_logical": 128,
37
+ "gpu": "NVIDIA A800-SXM4-80GB",
38
+ "gpu_count": 4,
39
+ "disk": {
40
+ "/": {
41
+ "total": "528147169280",
42
+ "used": "316705570816"
43
+ }
44
+ },
45
+ "memory": {
46
+ "total": "2162270285824"
47
+ },
48
+ "gpu_nvidia": [
49
+ {
50
+ "name": "NVIDIA A800-SXM4-80GB",
51
+ "memoryTotal": "85899345920",
52
+ "cudaCores": 6912,
53
+ "architecture": "Ampere",
54
+ "uuid": "GPU-e33a5d5c-ece1-16b1-ae70-38577ceaea24"
55
+ },
56
+ {
57
+ "name": "NVIDIA A800-SXM4-80GB",
58
+ "memoryTotal": "85899345920",
59
+ "cudaCores": 6912,
60
+ "architecture": "Ampere",
61
+ "uuid": "GPU-cd62cf34-2832-7925-6fb2-50d0de797c7e"
62
+ },
63
+ {
64
+ "name": "NVIDIA A800-SXM4-80GB",
65
+ "memoryTotal": "85899345920",
66
+ "cudaCores": 6912,
67
+ "architecture": "Ampere",
68
+ "uuid": "GPU-f15c4618-c2e7-f14e-f27e-623dda73692a"
69
+ },
70
+ {
71
+ "name": "NVIDIA A800-SXM4-80GB",
72
+ "memoryTotal": "85899345920",
73
+ "cudaCores": 6912,
74
+ "architecture": "Ampere",
75
+ "uuid": "GPU-4f0f1b5f-f5bc-5f62-8cc8-a837e6d1f600"
76
+ }
77
+ ],
78
+ "cudaVersion": "12.4",
79
+ "writerId": "txxjku83d69hgbel7az1qjfuqjpo139f"
80
+ }
soup_XL_100M_work_dir/wandb/run-20260708_090752-bs3i5bjq/logs/debug-internal.log ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {"time":"2026-07-08T09:07:52.33321276+08:00","level":"INFO","msg":"stream: starting","core version":"0.22.1"}
2
+ {"time":"2026-07-08T09:07:52.879779385+08:00","level":"INFO","msg":"stream: created new stream","id":"bs3i5bjq"}
3
+ {"time":"2026-07-08T09:07:52.879841771+08:00","level":"INFO","msg":"handler: started","stream_id":"bs3i5bjq"}
4
+ {"time":"2026-07-08T09:07:52.88015593+08:00","level":"INFO","msg":"stream: started","id":"bs3i5bjq"}
5
+ {"time":"2026-07-08T09:07:52.880170829+08:00","level":"INFO","msg":"sender: started","stream_id":"bs3i5bjq"}
6
+ {"time":"2026-07-08T09:07:52.880173317+08:00","level":"INFO","msg":"writer: started","stream_id":"bs3i5bjq"}
7
+ {"time":"2026-07-08T10:34:11.115390756+08:00","level":"INFO","msg":"flowcontrol: backed up, offloading to disk","recordNumber":28926}
8
+ {"time":"2026-07-08T10:34:11.158966111+08:00","level":"INFO","msg":"flowcontrol: unblocked","totalOffloaded":13}
9
+ {"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\"}"}
10
+ {"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\"}"}
11
+ {"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\"}"}
12
+ {"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\"}"}
13
+ {"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/logs/debug.log ADDED
The diff for this file is too large to render. See raw diff
 
soup_XL_100M_work_dir/wandb/run-20260708_090752-bs3i5bjq/run-bs3i5bjq.wandb ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d1eb9e010c63c1825f3288ff899e7717ff9be5a74cfd9915a09115b5b018688e
3
+ size 50298880